Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Manuscript number: RC-2025-02860

Corresponding author(s): Duncan, Sproul

[The "revision plan" should delineate the revisions that authors intend to carry out in response to the points raised by the referees. It also provides the authors with the opportunity to explain their view of the paper and of the referee reports.

• *

The document is important for the editors of affiliate journals when they make a first decision on the transferred manuscript. It will also be useful to readers of the reprint and help them to obtain a balanced view of the paper.

• *

If you wish to submit a full revision, please use our "Full Revision" template. It is important to use the appropriate template to clearly inform the editors of your intentions.]

1. General Statements [optional]

This section is optional. Insert here any general statements you wish to make about the goal of the study or about the reviews.

We thank the reviewers for recognizing that our work contributes 'both conceptually and mechanistically to our understanding of how DNA methylation patterns are regulated during cancer development' and their insightful suggestions to improve the manuscript. We note that the reviewers suggest that the data are 'comprehensive', 'well-controlled', 'rigorously done' and 'diligently analysed'.

Our planned revisions focus on further elucidating the broader implications of our findings for partially methylated domain formation in cancer, the effects of the methylation changes we observe on gene expression and the potential mechanisms underpinning the formation of the hypermethylated domains we observe.

2. Description of the planned revisions

Insert here a point-by-point reply that explains what revisions, additional experimentations and analyses are planned to address the points raised by the referees.

We have reproduced the reviewer's comments in their entirety and highlighted them in blue italics.

February 21, 2025
*RE: Review Commons Refereed Preprint #RC-2025-02860 *

*Kafetzopoulos *

DNMT1 loss leads to hypermethylation of a subset of late replicating domains by DNMT3A

------------------------------------------------------------------------------

*Reviewer #1 (Evidence, reproducibility and clarity (Required)): *

The DNA methylation landscape is frequently altered in cancers, which may contribute to genome misregulation and cancer cell behavior. One phenomenon is the emergence of "partially methylated domains (PMDs)": intermediately methylated regions of the genome that are generally heterochromatic and late replicating. The prevailing explanation is that the DNA methyltransferase, DNMT1, is not able to maintain DNAme levels at late replicating sites in proliferating cancer cells. This could result in genome instability. In this study, Kafetzopoulos and colleagues interrogated this possibility using a common laboratory colorectal cancer cell line (HCT116). Additionally, they utilized a DNMT1 mutant line that they refer to as a knockout, even though, more accurately, it is a hypomorphic truncation. They performed several genomic assays, such as whole genome bisulfite sequencing, ChIP and repli-seq, in order to assess the effect of reduced DNMT1 activity. While expectedly, global DNAme levels are decreased, they discovered a subset of PMDs gain DNA methylation, which they term hyperPMDs. There seems to be no impact on DNA replication timing, but the authors did go on to show that the de novo DNA methyltransferase, DNMT3Α, is likely responsible for this counterintuitive increase in DNAme levels.

*Reviewer #1 (Significance (Required)): *

Overall, I found the data well-presented and diligently analyzed, as we have come to expect from the Sproul group. However, I am somewhat at a loss to understand both the rationale for the experimental set-up and the meaning of the results. The HCT116 cell line is already transformed but was treated as though it was a wild-type control. I was more curious to see how the PMD chromatin state and replication compare to a healthy cell.

We focused on the comparison between WT and DNMT1 KO cells as we wanted to understand the role of DNMT1 in maintaining the organisation of the cancer methylome. We agree that, strictly, this could differ from its role in normal cells. However, we are unaware of a suitable cell line to test the consequences of DNMT1 KO in normal colon cells and testing this in vivo would be beyond the time-scale of a manuscript revision.

To further understand the relevance of our findings in the context of carcinogenesis, we propose to analyse data derived from normal and cancerous colon tissue in the revised manuscript. Preliminary analysis shows that HCT116 PMDs are hypomethylated in a colorectal tumour but not in the normal colon (revision plan figure 1). This suggests that HCT116 cells are a model that can be used to understand PMD formation in tumours and we will extend this analysis in the revised manuscript. We will also add discussion of the caveat that DNMT1 may function differently in normal tissues and cancer cells.

Note, revision plan figure 1 was included with the full submission but cannot be uploaded in this format.

Revision plan figure 1. HCT116 PMDs are hypomethylated in colorectal tumours. Heatmaps and pileup plots of HCT116, normal colon and colorectal tumour DNA methylation levels for HCT116 PMDs (n=546 domains) and HMDs (n=558 domains). DNA methylation levels are mean % mCpG. PMDs and HMDs are aligned and scaled to the start and end points of each domain and ranked based on their mean methylation levels in HCT116 cells. Colon and tumour data re-analysed from a previous publication (Berman et al 2011, PMID: 22120008).

Moreover, the link between late replication and PMDs would indicate that a DNMT1 gain-of- function line would potentially be more interesting: could more increased DNMT activity rescue the PMDs, and how would this impact the chromatin and replication states? Perhaps this is not trivial to create; I do not know if simply overexpressing DNMT1 and/or UHRF1 could act as a gain-of-function.

We agree with the reviewer that a DNMT1 overexpression or a gain-of-function mutation cell line would be interesting to analyse and potentially informative as to the mechanism of PMD formation. However, as the reviewer notes, this is a complex experiment that could require the overexpression of partners such as UHRF1 or generation of an unknown gain-of-function mutation. In addition, the full dissection of the implications of this separate experimental strategy would entail the repetition of the majority of our experiments in DNMT1 KO cells. Instead, in the revised manuscript, we will focus on a related experiment suggested by reviewer 2 and ask whether re-expression of DNMT1 rescues DNA methylation patterns DNMT1 KO cells.

Nevertheless, the appearance of hyperPMDs was a curious finding worth publishing. However, it is unclear what the biological relevance is. There is no effect on replication timing, and no assessment on cell behavior (eg, proliferation assays).* In other words, is DNMT3A performing some kind of compensatory action, or is it just a curiosity? Below in the significance section, I have highlighted some additional specific points *

PMDs are important to study because cancer-associated hypomethylation is believed to drive carcinogenesis through genomic instability (Eden et al 2003, PMID: 12702868). However, the mechanisms underpinning their formation remain unclear. At present the predominant hypothesis is that PMDs emerge in heterochromatin because their late replication timing leaves insufficient time for re-methylation following DNA replication (Zhou et al 2018, PMID: 29610480 and Petryk et al 2021, PMID: 33300031). We believe that our observations of hypermethylated PMDs in DNMT1 KO cells provides important evidence contrary to this hypothesis because they disconnect domain-level methylation patterns from the replication timing program. Our work instead suggests that the localization of de novo DNMTs plays a key role in the formation of PMDs by protecting euchromatin from hypomethylation.

To further explore this hypothesis, we propose to analyze data derived from tumours in our revised manuscript to understand the degree to which our findings are reflected in vivo. As shown above, our preliminary analysis suggests that HCT116 cell PMDs are also hypomethylated in a colorectal tumour but not the normal colon (revision plan figure 1). We will also analyze how the changes in methylome affect gene expression using our RNA-seq data.

- Why were DNMT3A and 3B transgenes used for ChIP instead of endogenous proteins? I know the authors cited work justifying this strategy, but this still merits explanation. Also, the expression level of transgenes compared to the endogenes was not shown (neither protein nor RNA level).

DNMT3A and B transgenes were used because antibodies against the endogenous proteins are not suitable for ChIP. Furthermore, performing these experiments using endogenously tagged proteins, required generating 3 knock-in tagged lines (we have already generated HCT116 cells with tagged DNMT3B, Masalmeh et al 2021, PMID: 33514701).

We have previously shown that our constructs do indeed result in overexpression of DNMT3B compared to endogenous protein in this system (Masalmeh et al 2021, PMID: 33514701). However, our previous results also demonstrate that overexpressed DNMT3B recapitulates the localization of the endogenously tagged protein to the genome (Taglini et al 2024, PMID: 38291337). Others have similarly demonstrated that ectopically expressed DNMT3A and DNMT3B can be used to understand their localization on the genome (Baubec et al 2015, PMID: 25607372 and Weinberg et al 2019, PMID: 31485078).

To address this point, we propose to add further justification of our approach and discussion of this potential limitation to a revised version of the manuscript.

- The DNMT3A binding profile appears as though it is on the edges of the PMDs and fairly depleted within (Fig 4A,D). Could the authors comment on this?

This is an interesting point. We note that although mean DNMT3A signal is indeed higher at the edges of hypermethylated PMDs than inside these domains, its levels are both above background and the levels observed in HCT116 cells. As suggested by reviewer 3, this could be consistent with H3K36me2 and DNMT3A spreading in from the boundaries of hypermethylated PMDs in DNMT1 KO cells. We propose to add discussion of this possibility to the revised version of the manuscript.

- A more compelling experiment would be to assess the loss of DNMT3A genetically. How would this affect PMD DNA methylation? Maybe in this case there would be an effect on replication timing. Could a KO or KD (eg, siRNA) strategy be employed to assess this on top of either the HCT116 or DNMT1 KO.

As the reviewer suggests, functional experiments aimed at understanding the role of DNMT3A in our system are likely to be informative. We therefore propose to include such experiments in a revised version of the manuscript.

- What is the major H3K36me2 methylatransferase in these cells? Could an Nsd1 KO or KD strategy be used, for example, to show that indeed H3K36 methylation is required for HyperPMDs? This would complement the DNMT3A experiment above.

H3K36 methylation is thought to be deposited in the mammalian genome by at least 8 different methyltransferase enzymes, NSD1, NSD2, NSD3, ASH1L, SETD2, SETMAR, SMYD2 and SETD3 (Wagner and Carpenter 2023, PMID: 22266761). To understand which of these might be responsible for the deposition of H3K36me2 in hypermethylated PMDs, we have examined their expression in HCT116 and DNMT1 KO cells using our RNA-seq data. This suggests that 5 of these enzymes are highly expressed in HCT116 cells and their expression levels are similar in DNMT1 KO cellsrevision plan figure 2). The other 3 putative methyltransferases have lower expression levels and, although SMYD2 is significantly upregulated in DNMT1 KO cells, its expression remains low (revision plan figure 2). It is currently unclear whether SMYD2 is a bona fide H3K36 methyltransferase (Wagner and Carpenter 2023, PMID: 22266761). We also note that in a recent study, cells lacking NSD1, NSD2, NSD3, ASH1L and SETD2 had no detectable H3K36 methylation, although expression levels of SMYD2 were not reported (Shipman et al, 2024. PMID: 39390582). Based on this analysis, it is therefore unclear which enzyme(s) might be responsible for H3K36me2 deposition in hypermethylated PMDs and delineation of this enzyme would require multiple perturbation and sequencing experiments. We therefore suggest that assessing the consequences of knocking out H3K36me2 methyltransferase activity on hypermethylated PMDs is beyond the scope of a manuscript revision. We propose to include discussion of the expression of the different H3K36me2 depositing enzymes in the revised manuscript.

Note, revision plan figure 2 was included with the full submission but cannot be uploaded in this format.

Revision plan figure 2. HCT116 cells express multiple H3K36 methyltransferases. Barplot of mean expression levels for putative mammalian H3K36 methyltransferases in HCT116 and DNMT1 KO cells. Expression levels are counts per million (CPM) derived from RNA-seq. Mean expression levels are derived from 9 and 4 independent cultures of HCT116 and DNMT1 KO cells respectively.

- Based on Figure 2C, it seems that a general predictive pattern of hyperPMDs is H3K9me3-enriched and H3K27me3-depleted. Is this an accurate interpretation? Given the authors' expertise in the relationship between DNMT3A and polycomb, could they perhaps give an explanation for this phenomenon?

The reviewer is correct. In HCT116 cells, those PMDs that become hypermethylated in DNMT1 KO cells are marked by H3K9me3 and are H3K27me3-depleted (except at their boundaries). DNMT3A is recruited to polycomb-marked regions associated with H3K27me3 through interaction of its N-terminal region with H2AK119ub. However, this mark is depleted from hypermethylated-PMDs in DNMT1 KO cells (current manuscript Figure S5D) meaning that this pathway of recruitment is unlikely to explain DNMT3A's localisation to these regions in DNMT1 KO cells. This is discussed in the current manuscript:

We and others have reported that DNMT3A is also recruited to the polycomb-associated H2AK119ub mark through its N-terminal region (Chen et al, 2024; Gretarsson et al, 2024; Gu et al, 2022; Wapenaar et al, 2024; Weinberg et al, 2021). However, we do not observe the polycomb-associated H3K27me3 mark, which is generally tightly correlated with H2AK119ub (Ku et al, 2008), at hypermethylated PMDs suggesting that H2AK119ub does not play a role in the recruitment of DNMT3A to these regions.

Furthermore, DNMT3A's localisation is predominantly driven by its PWWP-dependent H3K36me2 recruitment pathway unless its PWWP domain is mutated (Heyn et al 2019, PMID: 30478443, Sendžikaitė et al 2019, PMID: 31015495, Kibe et al 2021, PMID: 34048432 and Weinberg et al, 2021, PMID: 33986537). Our observations of DNMT3A at hypermethylated PMDs marked by H3K36me2 is therefore consistent with previous findings. We propose to discuss this point in the revised manuscript.

- This is a minor point, but calling the DNMT1 mutant a "KO" seemed a bit misleading, as it is a truncation mutant. Perhaps there is a more accurate way to describe this line.

We propose to amend the manuscript to reflect this point as suggested by the reviewer. To ensure our responses are consistent with the reviewer comments we continue to refer to this line as DNMT1 KO cells in our revision plan.

*Reviewer #2 (Evidence, reproducibility and clarity (Required)): *

*In this study, Kafetzopoulos et al. investigated the role of DNMT1-mediated methylation maintenance in cancer partially methylated domains (PMDs) using DNMT1 knockout HCT116 colorectal cancer cells. They used a range of sequencing-based approaches, including whole-genome bisulfite sequencing (WGBS), chromatin immunoprecipitation sequencing (ChIP-seq), and replication timing sequencing (Repli-seq), to define the dynamics of DNA methylation loss and gain in PMDs during DNA synthesis. Interestingly, they demonstrate that specific PMDs marked by H3K9me3 undergo a gain of DNA methylation in DNMT1-deficient HCT116 cells. This increase in methylation is associated with the loss of H3K9me3, an enrichment of H3K36me2, and the recruitment of DNMT3A. These findings suggest that de novo methyltransferase activity plays a critical role in determining which genomic regions become PMDs in cancer. *

*The authors use a comprehensive and well-controlled set of sequencing-based techniques. While the sequencing depth for DNA methylation is somewhat limited, the inclusion of multiple biological replicates strengthens the reliability of the data. The study effectively integrates multiple layers of epigenomic information, providing a nuanced view of PMD regulation in the context of DNMT1 loss. *

*Overall, this paper provides valuable insights into the epigenetic regulation of PMDs in cancer, and its conclusions are well supported by the data. It significantly advances our understanding of how DNMT1 loss reshapes the epigenome and highlights the interplay between de novo and maintenance methylation mechanisms in cancers. *

------------------------------------------------------------------------------

*Reviewer #2 (Significance (Required)): *

General assessment

-The main strength of the study lies in the clear presentation of the data, which follows a cohesive and well-defined storyline.

*-The authors demonstrate that both hypomethylated and hypermethylated domains occur at the late replication stage. They further investigate the dynamics of histone modifications and DNA methylation, focusing on the acquisition and loss of these marks, particularly in relation to DNMT3A and DNMT3B. *

Limitation

-Although the study is compelling, its primary limitation is the correlative nature of most of the data. While the high-level representations (e.g., tracks, heat maps) are convincing, the study would have been more informative if it had explored the impact of these changes on a specific set of genes or regions critical to cancer initiation and progression. For example, in the DNMT1 knockout model, how does the loss of H3K9me3, the gain of H3K36me2, and the recruitment of DNMT3A in hypermethylated PMDs affect the expression of key genes involved in colorectal cancer?

To understand how the remodeling of DNA methylation and chromatin structure in DNMT1 KO cells affects gene expression, we propose to include an analysis of our RNA-seq data in the revised manuscript. We will also cross reference these results and our ChIP-seq with lists of colorectal cancer genes.

Additional experiments that could provide deeper insights

-Cross-validation in other cancer cell lines would have enable to define if these signatures are observed beyond HCT116.

As the reviewer suggests, we propose to undertake analyses of additional samples in the revised manuscript to understand how our findings relate to domain-level methylation patterns beyond HCT116 cells. As noted above in response to reviewer 1, our preliminary analysis suggests our findings are relevant for primary colorectal tumours (revision plan figure 1).

-Are the observed signatures permanent, or could they be reversed by reinstating the full activity of DNMT1? Since DNMT1 might be dysregulated but never completely deleted.

To address this suggestion, we propose to include the results of a DNMT1 rescue experiment in the revised manuscript.

-Use knockdown and overexpression experiments to track the dynamics and occurrence of these molecular events over time, providing insight into the progression and reversibility of epigenetic changes.

This is an interesting suggestion. As the reviewer suggests, we propose to analyse data derived from time-course experiments to understand the dynamics of changes in different genomic compartments following perturbation of DNMT1.

Advances

-The study provides new insights into the establishment of PMD types in colorectal cancer cell lines.

-These findings contribute both conceptually and mechanistically to our understanding of how DNA methylation patterns are regulated during cancer development.

Audience:

-This study will appeal to a broad audience, from researchers primarily focused on epigenetics and cancer biology to those interested in the mechanistic underpinnings of DNA methylation and its role in cancer progression. It will also be relevant to those exploring therapeutic strategies targeting epigenetic regulators in cancer.

We thank the reviewer for their kind comments on our manuscript.

------------------------------------------------------------------------------

*Reviewer #3 (Evidence, reproducibility and clarity (Required)): *

Summary:
*Cancer is linked to the acquisition of an atypical DNA methylation landscape, with broad domains of partial DNA methylation (termed PMDs). This study investigates PMDs in a colorectal cancer cell line and evaluates the contribution of DNMT1 in maintaining PMDs, using a DNMT1 KO line. The authors find that PMDs preferentially lose DNA methylation upon loss of DNMT1, but they find a number of domains that paradoxically gain DNA methylation (hyperPMDs). They attribute this gain of methylation to the action of DNMT3A through the accumulation of H3K36me2 and loss of heterochromatin mark H3K9me3. Together this work sheds light on the dynamic mechanisms regulating the atypical DNA methylation landscape in colorectal cancer cells. *

General comments:
The introduction is informative and well written. Additionally, the work is rigorously done and analyses are clear. However, the conclusions and summary figure largely focus on the relationship between PMDs with H3K9me3 and H3K36me2, but I think the role for H3K27me3 should be revisited based on the results presented. H3K9me3 is present at PMDs and hyperPMDs, but H3K27me3 level appears to be a much more defining feature of whether they lose or gain methylation upon loss of DNMT1 (Figure 2, Figure S2C- D). There is a reported interplay between PRC2 and DNMT3A activity at DNA methylation valleys in other cell contexts (e.g., mouse embryogenesis, hematopoietic cells), so couldn't H3K27me3 be performing a 'boundary' function at PMDs and when sufficiently low, permits spread of H3K36me2 in the absence of DNMT1? I think it is worth further exploring the H3K27me3 data.

The reviewer makes an interesting point about the potential for H3K27me3 to act as a boundary preventing H3K36me2 spread into PMDs. Multiple studies have shown that H3K36me2 restricts H3K27me3 deposition in the genome (Streubel et al 2018, PMID: 29606589, Shirane et al 2020, PMID: 32929285 and Farhangdoost et al 2021, PMID: 3362635). The structural nature of this inhibitory effect has also been resolved, demonstrating that the PRC2 catalytic subunit, EZH2 directly binds H3K36 and this is inhibited when the residue is methylated (Jani et al 2019, PMID: 30967505, Finogenova et al 2020, PMID: 33211010 and Cookis et al 2025, PMID: 39774834). The effect of H3K27me3 on H3K36me2 is less well characterised. However, previous work has suggested that inhibiting EZH2 leads to elevated H3K36me2 being established on newly replicated chromatin (Alabert et al 2020, PMID: 31995760). Expression of the EZH2-inhibiting oncohistone H3.3K27M has also been reported to lead to increased H3K36me2 dependent on NSD1/2 in diffuse intrinsic pontine gliomas (DIPG) (Stafford et al 2018, PMID: 30402543 and Yu et al 2021, PMID: 34261657). However, this increase was not reported by an independent study of H3.3K27M DIPG cells (Harutyunyan et al 2020, PMID: 33207202) and the molecular basis of the effect of H3K27me3 on H3K36me2 remains unclear.

As the reviewer suggests, we propose to explore the relationship between H3K27me3 and H3K36me2 further in a revised manuscript along with the including further discussion of previous findings in this area.

Additionally, a key point that is illustrated in the summary figure, is the localization of H3K36me2 at HMDs and its mutual exclusivity with H3K9me3 (a mark typically associated with high DNA methylation). However, because the H3K36me2 is introduced quite late in the analysis, I feel that a rigorous evaluation of its enrichment and anti-correlation with H3K9me3 at highly methylated domains (HMDs) is missing.

The relationship between H3K36me2 and H3K9me3 is far less explored than that of H3K27me3 and H3K36me2. Interestingly, we note that a recent study reported that depletion of H3K36me2 results in H3K9me3 re-distribution suggesting that H3K9me3 is restricted by H3K36me2 (Padilla et al 2024, DOI: 10.1101/2024.08.10.607446, also cited in the original manuscript).

To understand this relationship further, we therefore propose to explore the relationship between H3K9me3 and H3K36me2 in our datasets as part of revised manuscript along with including additional discussion of relevant experimental findings.

In general, I also found that I was jumping between figures a lot and needed to look at the supplements to gain the full picture. It may be beneficial to re-organize the figures.

In accordance with the reviewer's suggestion, we propose to re-organise the revised manuscript to make it easier to follow.

Specific Comments/Questions:

• An expanded explanation of the truncated DNMT1 in the DNMT1 KO cells would be helpful for context**
* As suggested by the reviewer, we will amend the manuscript to include an expanded discussion of the DNMT1 truncation present in the cell line.

• Does the DNMT expression in HCT116 cells reflect the levels seen in primary colorectal cancers? Hence, do you think these cultured cells reflect aspects of DNA methylation dynamics that would be seen in tumors?**
*

While differences between cancer cell line and tumour methylation patterns have previously been noted (for example Anne Rogers et al 2018, PMID: 30559935), we have previously demonstrated that HCT116 cells recapitulate CpG island methylation patterns observed in colorectal tumours (Masalmeh et al 2021, PMID: 33514701). As stated above in response to reviewer 1, we have now examined the methylation status of HCT116 PMDs in a colorectal tumour. This analysis shows that HCT116 PMDs have reduced methylation levels in a colorectal tumour but not in the normal colon (revision plan figure 1). We propose to extend this analysis of colorectal tumour samples and add them to the revised manuscript to address this point.

Regarding the expression of DNMTs in colorectal tumours, DNMT1 is ubiquitously expressed to our knowledge. DNMT3B is reported to be overexpressed in 15-20% of cases of colorectal cancer, often as a result of amplification (Nosho et al 2009, PMID: 19470733, Ibrahim et al 2011, PMID: PMID: 21068132, Zhang et al 2018, PMID: 30468428 and Mackenzie et al 2020, PMID: 32058953). DNMT3A expression in colorectal tumours is less studied but one report suggests upregulation in at least some tumours (Robertson et al 1999, PMID: 10325416 and Zhang et al 2018, PMID: 30468428). We propose to add additional discussion of DNMT expression in colorectal cancer to the revised manuscript to clarify the degree to which our results reflect methylation regulation in primary colorectal tumours.

• Although DNMT3A/B mRNA levels are similar between DNMT1 KO and HCT116 cells, is the protein abundance altered? I think there would be value in showing a Western blot analysis, as the loss of DNMT1 protein may alter the stability of the de novo DNMTs. Is a similar level of expression of the ectopic T7-DNMT3A and T7-DNMT3B achieved in HCT116 and DNMT1 KO cells? A western blot showing this would also be valuable.**
*

As part of our work towards revising the manuscript, we have undertaken blots of DNMT3A in our cell lines. This shows that DNMT3A levels in DNMT1 KO cells are similar to those in HCT116 cells which (revision plan figure 3). We propose to include this in the revised manuscript alongside a similar analysis of DNMT3B. We will also include an analysis of T7-DNMT3A and T7-DNMT3B levels to understand whether they are expressed to similar levels in HCT116 and DNMT1 KO cells.

Note, revision plan figure 3 was included with the full submission but cannot be uploaded in this format.

Revision plan figure 3. DNMT3A protein levels are similar in HCT116 and DNMT1 KO cells. Left, representative DNMT3A Western blot. Right, bar plot quantifying relative DNMT3A levels. The bar height indicates the mean levels observed in protein extracts from 3 independent cell cultures. Individual points indicate the level of each replicate.

• Do you think that the increase in DNMT3A over HyperPMD compared to H3K9me3-marked PMDs is related to an increase in protein bound at these domains or an altered residence time?*

The reviewer makes an interesting point with regard to a potential alteration of DNMT3A residence at hypermethylated PMDs. Given that ChIP-seq signal is affected by residence time (Schmiedeberg et al 2009, PMID: 19247482), it is possible that our findings could reflect this rather than increased DNMT3A localisation. We propose to add discussion of this point as a limitation of the current study to the manuscript.

It would also be valuable to move the plot showing levels of DNMT3A/3B at HMDs, from the S4C/D to the main Figure 4, for reference. It would also be interesting to see the enrichment of DNMT3A/B at all PMDs (not just H3K9me3-marked PMDs).*
*

As the reviewer suggests, we will include the data on HMDs to the main Figure 4 and include enrichments at all PMDs in the supplementary figures.

• It appears that the same genomic locus is used multiple times across figures Fig 1A, Fig 2B, Fig 3A, Fig 4A, Fig 5B to illustrate the trends reported from the global analyses. While this has value in showing the dynamics across datasets at this region, I think it is important to illustrate that these trends can be observed elsewhere. Please add or replace some plots with additional loci. Furthermore, please add the genomic region coordinates to the figure or figure legend.*

We had shown a single locus for consistency and to not overcomplicate figures which already contain multiple panels. As the reviewer suggests, we will add additional loci in the supplementary figures of our revised manuscript. We had also included the chromosome co-ordinates in the figures. In the revised version we will ensure that the precise co-ordinates are included in the legends.

• The ChIP-seq data is quantified as IP/input. This quantitation can be prone to introducing artefacts into analyses if the input coverage is substantially uneven over AT-rich regions or CpG islands, or if the sequencing depth is insufficient. I would encourage the authors to check that the trends observed are still present if quantified without correcting against the inputs. If using IP/input, in the supplementary figures, I think it would be valuable to show the uncorrected quantitation of inputs across PMDs, to demonstrate that there is even coverage and this isn't contributing to any of the changes observed.**
*

We thank the reviewer for this point and we propose to examine the quantification of the ChIP-seq without normalizing to input to ensure that uneven input signal does not substantially contribute to our results.

• Generally, the n numbers for different groups of probes can be confusing and increased clarity would be helpful.*

We will clarify the explanation of n numbers in the revised manuscript.

*Reviewer #3 (Significance (Required)): *

This study adds to the accumulating body of evidence that DNMT3A recruitment is mediated primarily through H3K36me2 across cell contexts, shedding light on the interplay between histone modifications and de novo DNA methylation. Understanding these mechanisms is important to appreciate the role for DNMT3A in establishing DNA methylation in development and disease contexts. It does remain unclear why, upon loss of DNMT1 in colorectal cancer cells, some PMDs accumulate H3K36me2 and consequently DNA methylation, while others do not. Further study into the chromatin dynamics will be valuable in understanding determinants of the DNA methylation landscape in cancer.

We thank the reviewer for their insightful comments and believe that our proposed revisions will further clarify the points they raise.

3. Description of the revisions that have already been incorporated in the transferred manuscript

Please insert a point-by-point reply describing the revisions that were already carried out and included in the transferred manuscript. If no revisions have been carried out yet, please leave this section empty.

We have not yet incorporated revisions into the manuscript.

4. Description of analyses that authors prefer not to carry out

Please include a point-by-point response explaining why some of the requested data or additional analyses might not be necessary or cannot be provided within the scope of a revision. This can be due to time or resource limitations or in case of disagreement about the necessity of such additional data given the scope of the study. Please leave empty if not applicable.

As stated in our responses to the reviewer comments above, we plan to address all comments. However, we suggest that two experiments proposed by the reviewers are beyond the scope of a manuscript revision and we will instead address these comments in the following manner:

Analysis of a DNMT1 gain-of-function line (Reviewer 1). As suggested by the reviewer such a line is non-trivial to generate. It would also require extensive profiling of this new line to fully understand its implications for our findings. We therefore believe it is outwith the scope of a manuscript revision. Instead, we propose to address this comment by undertaking the related experiment suggested by Reviewer 2 and perform a DNMT1 rescue experiment in the DNMT1 KO line. Analysis of H3K36me2 methyltransferase knockout cells (Reviewer 1). Our preliminary analysis suggests that HCT116 cells express multiple H3K36 methyltransferases and that their expression does not vary greatly in DNMT1 KO cels (revision plan figure 2). This means that it is unclear which enzyme(s) might be responsible for depositing H3K36me2 in hypermethylated PMDs. Delineation of this would require the generation and analysis of multiple knockouts and we suggest it is therefore outwith the scope of a manuscript revision. To address this point we will instead include discussion of the spectrum of H3K36 methyltransferases expressed in our cells in the revised manuscript as detailed in the specific response above.

This work has been published in GigaByte Journal under a CC-BY 4.0 license (https://doi.org/10.46471/gigabyte.153). These reviews (including a protocol review) are as follows.

Reviewer 1. Yanshuo Liang

The manuscript by Conn et al. detail the high-quality genome assembly of Acropora pulchra, a Acropora of ecological and evolutionary significance, and also analyzes its genome-wide DNA methylation characteristics. These data complement the genetic resources of the Acropora genome. This manuscript is well written and represents a valuable contribution to the field. I have some comments below for the authors to address but look forward to seeing this research published. Q1: In the first sentence of the second paragraph of the Context: This is the first study to utilize PacBio long-read HiFi sequencing to generate a high quality genome with high BUSCO completeness, in tandem with its DNA methylome for scleractinian corals. Language such as "new", "first", "unprecedented", etc, should be avoided because it often leads to unproductive controversy. As far as I know, the genome you assembled is not the first stony coral to be sequenced using PacBio long-read HiFi sequencing. Back in 2024, He et al. assembled Pocillopora verrucosa (Scleractinia) to the chromosome level using PacBio HiFi long-read sequencing and Hi-C technology. Here I would suggest please rephrase. Reference： He CP, Han TY, Huang WL, et al. Deciphering omics atlases to aid stony corals in response to global change, 11 March 2024, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-4037544/v1]. Q2: In this sentence: “On 23 October 2022, sperm samples were collected from the spawning of A.pulchra and preserved in Zymo DNA/RNA shield.” Please “A.pulchra” to “A. pulchra”. Q3: Please change all “k-mer” into “k-mer” in the manuscript. Q4: Please change “Long-Tandem Repeats” to “Long Terminal Repeats” Q5: In this sentence: “Funannotate train uses Trinity [18] and PASA [19] for ab initio predictions. Funannotate predict was then run to assign gene models using AUGUSTUS [20], GeneMark [21], and Evidence Modeler [19] to estimate final gene models.” Please write versions of these software. Q6: [20] Later references do not correspond well in the manuscript, please check!

Reference 2. Jason Selwyn

Is the language of sufficient quality? Yes. There are some minor grammatical issues throughout that warrent a closer reading to correct. E.g. Abstract: "...urgency to identify how genetic, epigenetic, and environmental...", "...management and and conservation...". Context: "...we aim to provide..." etc. Are all data available and do they match the descriptions in the paper? Yes. The link to the OSF repository in the PDF did not work. However, the link to the OSF repository from the github did work. Is the data acquisition clear, complete and methodologically sound? No. It isn't mentioned in the manuscript where the RNAseq data used to annotate the genome is from, nor any quality filtering steps that may have been applied to the RNA data prior to its use for annotation. Is there sufficient detail in the methods and data-processing steps to allow reproduction? Yes. Excluding the above comment about the RNA data. Additional Comments: This is a well assembled, and annotated genome that will contribute to the growing database of Acropora genomes. The manuscript could do with a simple pass to identify and correct some relatively minor grammatical issues and inconsistencies (Table 1 includes a thousands comma separator in some instances and not others) and needs to include details about the source of the RNA data used to train the ab initio gene predictors. There also appears to be a problem with the citation numbering after 20.

**Reviewer 3. Benjamin Young ** Are all data available and do they match the descriptions in the paper? Yes. Raw reads, metadata, and genome assembly are publicly available and have a NCBI project number in which they are all linked. Is the data acquisition clear, complete and methodologically sound? Yes. Collection of sperm samples, HMW DNA extraction, and SMRT Bell Library prep are written clearly. I have asked for a few clarifications on wording in this section in the attached edited pdf document. Is there sufficient detail in the methods and data-processing steps to allow reproduction? Yes. I think the pipeline used for de-novo genome generation (including raw read cleaning and assembly), repeat masking, and gene prediction and annotation is of high quality and best practices. With the inclusion of the GitHub and all analyses scripts, it is possible to reproduce the assembly generated. Is there sufficient data validation and statistical analyses of data quality? Yes. This is not super relevant for a genome assembly paper so I have no additional comments here. Is the validation suitable for this type of data? Yes. The authors use tools such as GenomeScope2 and BUSCO for validation of their data. It would be nice to see the tool they used to identify N50 and L50 (maybe Quast) included in the methods. Additionally, I would like to see a Merqury analysis of the HifiAsm primary and alternate assemblies to show that duplicate purging was successful. Additional Comments: I would first like to commend the authors for a well assembled genome resource for a coral species that will be greatly beneficial to the wider coral and scientific community. I have provided a PDF with comments throughout for the authors to address. The majority of these are easy fixes, including things such as sentence structure, inconsistent capitalisation of subheadings, additional references for methods, clarification of statements, and other suggestions. I do have a few larger requests for this to be published, and these are the reasons for selecting the major revision option as there may need to be figure updates, and quick additional analyses to be run. 1. Can you please correct the verbiage around BUSCO analysis throughout the manuscript. It is often stated "BUSCO completeness of xx%". BUSCO doesn't directly measure completeness, rather completeness of single copy orthologs against a specific database. I have left comments throughout on potential rewording for these instances. Please also specify the exact database you used (i.e. odb10_metazoa). Finally, can you please be more specific when stating BUSCO results, specifically when you use 96.9% this is single copy and duplicated complete BUSCOS. I have left comments in the pdf again for this. 2. In the results for Genome Assembly section can you please include results (i.e. length, N50, L50, number contigs/scaffolds) for the primary assembly and the scaffolded assembly. 3. I think it would be not much work and provide additional information to show successful duplicate purging to run a Merqury analysis on the primary and alternative assemblies from HiFiAsm. 4. Can you include some additional information in the "Structural and Functional Annotation section". Specifically, can you provide information on the results from the funannoatate predict step, and then how funannotate update improved this (if at all). 5. Please double check the methods section for funannotate. From reading the funannoatate documentation I think there may be some confusion on what each step (train, predict, update, annotate) is doing. I have provided comments in the pdf to help clarify, and have also linked the funnannotate documentation. 6. On NCBI I see that an additional Acropora pulchra genome has just been made available (29th Jan 2025), with this to the chromosome level (https://www.ncbi.nlm.nih.gov/datasets/genome/GCA_965118205.1/). I think it would be prudent to include this assemblies statistics in your Table 1, and also run a BUSCO analysis on this other assembly to compare with your one. While they got to chromosome level, you do have markedly less contigs. I do not think this is necessary for this manuscript, but future work you could look to use their chromosome assembly to get your scaffolded assembly to chromosome level. Again, I want to say this is a wonderful resource for the coral and wider scientific community, and the pipeline for de-novo assembly and annotation is best practices in my opinion. Annotated additional file: https://gigabyte-review.rivervalleytechnologies.comdownload-api-file?ZmlsZV9wYXRoPXVwbG9hZHMvZ3gvRFIvNTk0L2Nvbm5ldGFsMjAyNV9yZXZpZXdjb21tZW50cy5wZGY=

Re-review:

The authors have addressed all my comments and queries, and included nearly all recommendations. Thank you ! A few quick notes to fix before publication -

"The input created Funannotate train uses Trinity v.2.15.2 [22] and PASA v.2.5.3 [23] for transcript assembly prior to ab initio predictions". This sentence reads weird, reword before publishing. I think maybe just remove "created Funannotate train" and then it reads correctly. Or "Funnannotate trains uses .....". - "PFAM v.37.0 [28], CAZyme [29], UniProtKB v[30] and GO [31]." Missing a few version numbers, and UniProt just has a v. - "The mitochondrial genome was successfully assembled and circularized using MitoHifi v3.2.2 The final assembled A. pulchra mitogenome is". Just missing a period i think before "The final assembly". Great job and a very useful resource for the coral community !!

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

We thank the reviewers for their feedback on our paper. We have taken all their comments into account in revising the manuscript. We provide a point-by-point response to their comments, below.

Reviewer #1

Major comments:

The manuscript is clearly written with a level of detail that allows others to reproduce the imaging and cell-tracking pipeline. Of the 22 movies recorded one was used for cell tracking. One movie seems sufficient for the second part of the manuscript, as this manuscript presents a proof-of-principle pipeline for an imaging experiment followed by cell tracking and molecular characterisation of the cells by HCR. In addition, cell tracking in a 5-10 day time-lapse movie is an enormous time commitment.

My only major comment is regarding "Suppl_data_5_spineless_tracking". The image file does not load. It looks like the wrong file is linked to the mastodon dataset. The "Current BDV dataset path" is set to "Beryl_data_files/BLB mosaic cut movie-02.xml", but this file does not exist in the folder. Please link it to the correct file.

We have corrected the file path in the updated version of Suppl. Data 5.

Minor comments:

The authors state that their imaging settings aim to reduce photo damage. Do they see cell death in the regenerating legs? Is the cell death induced by the light exposure or can they tell if the same cells die between the movies? That is, do they observe cell death in the same phases of regeneration and/or in the same regions of the regenerating legs?

Yes, we observe cell death during Parhyale leg regeneration. We have added the following sentence to explain this in the revised manuscript: "During the course of regeneration some cells undergo apoptosis (reported in Alwes et al., 2016). Using the H2B-mRFPruby marker, apoptotic cells appear as bright pyknotic nuclei that break up and become engulfed by circulating phagocytes (see bright specks in Figure 2F)."

We now also document apoptosis in regenerated legs that have not been subjected to live imaging in a new supplementary figure (Suppl. Figure 3), and we refer to these observations as follows: "While some cell death might be caused by photodamage, apoptosis can also be observed in similar numbers in regenerating legs that have not been subjected to live imaging (Suppl. Figure 3)."

Based on 22 movies, the authors divide the regeneration process into three phases and they describe that the timing of leg regeneration varies between individuals. Are the phases proportionally the same length between regenerating legs or do the authors find differences between fast/slow regenerating legs? If there is a difference in the proportions, why might this be?

Both early and late phases contribute to variation in the speed of regeneration, but there is no clear relationship between the relative duration of each phase and the speed of regeneration. We now present graphs supporting these points in a new supplementary figure (Suppl. Figure 2).

To clarify this point, we have added the following sentence in the manuscript: "We find that the overall speed of leg regeneration is determined largely by variation in the speed of the early (wound closure) phase of regeneration, and to a lesser extent by variation in later phases when leg morphogenesis takes place (Suppl. Figure 2 A,B). There is no clear relationship between the relative duration of each phase and the speed of regeneration (Suppl. Figure 2 A',B')."

Based on their initial cell tracing experiment, could the authors elaborate more on what kind of biological information can be extracted from the cell lineages, apart from determining which is the progenitor of a cell? What does it tell us about the cell population in the tissue? Is there indication of multi- or pluripotent stem cells? What does it say about the type of regeneration that is taking place in terms of epimorphosis and morphallaxis, the old concepts of regeneration?

In the first paragraph of Future Directions we describe briefly the kind of biological information that could be gained by applying our live imaging approach with appropriate cell-type markers (see below). We do not comment further, as we do not currently have this information at hand. Regarding the concepts of epimorphosis and morphallaxis, as we explain in Alwes et al. 2016, these terms describe two extreme conditions that do not capture what we observe during Parhyale leg regeneration. Our current work does not bring new insights on this topic.

Page 5. The authors mention the possibility of identifying the cell ID based on transcriptomic profiling data. Can they suggest how many and which cell types they expect to find in the last stage based on their transcriptomic data?

We have added this sentence: "Using single-nucleus transcriptional profiling, we have identified approximately 15 transcriptionally-distinct cell types in adult Parhyale legs (Almazán et al., 2022), including epidermis, muscle, neurons, hemocytes, and a number of still unidentified cell types."

Page 6. Correction: "..molecular and other makers.." should be "..molecular and other markers.."

Corrected

Page 8. The HCR in situ protocol probably has another important advantage over the conventional in situ protocol, which is not mentioned in this study. The hybridisation step in HCR is performed at a lower temperature (37˚C) than in conventional in situ hybridisation (65˚C, Rehm et al., 2009). In other organisms, a high hybridisation temperature affects the overall tissue morphology and cell location (tissue shrinkage). A lower hybridisation temperature has less impact on the tissue and makes manual cell alignment between the live imaging movie and the fixed HCR in situ stained specimen easier and more reliable. If this is also the case in Parhyale, the authors must mention it.

This may be correct, but all our specimens were treated at 37˚C, so we cannot assess whether hybridisation temperature affects morphological preservation in our specimens.

Page 9. The authors should include more information on the spineless study. What been is spineless? What do the cell lineages tell about the spineless progenitors, apart from them being spread in the tissue at the time of amputation? Do spineless progenitors proliferate during regeneration? Do any spineless expressing cells share a common progenitor cell?

We now point out that spineless encodes a transcription factor. We provide a summary of the lineages generating spineless-expressing cells in Suppl. Figure 6, and we explain that "These epidermal progenitors undergo 0, 1 or 2 cell divisions, and generate mostly spineless-expressing cells (Suppl. Figure 5)."

Page 10. Regarding the imaging temperature, the Materials and Methods state "... a temperature control chamber set to 26 or 27˚C..."; however, in Suppl. Data 1, 26˚C and 29˚C are indicated as imaging temperatures. Which is correct?

We corrected the Methods by adding "with the exception of dataset li51, imaged at 29{degree sign}C"

Page 10. Regarding the imaging step size, the Materials and Methods state "...step size of 1-2.46 µm..."; however, Suppl. Data 1 indicate a step size between 1.24 - 2.48 µm. Which is correct?

We corrected the Methods.

Page 11. Correct "...as the highest resolution data..." to "...at the highest resolution data..."

The original text is correct ("standardised to the same dimensions as the highest resolution data").

Page 11. Indicate which supplementary data set is referred to: "Using Mastodon, we generated ground truth annotations on the original image dataset, consisting of 278 cell tracks, including 13,888 spots and 13,610 links across 55 time points (see Supplementary Data)."

Corrected

p. 15. Indicate which supplementary data set is referred to: "In this study we used HCR probes for the Parhyale orthologues of futsch (MSTRG.441), nompA (MSTRG.6903) and spineless (MSTRG.197), ordered from Molecular Instruments (20 oligonucleotides per probe set). The transcript sequences targeted by each probe set are given in the Supplementary Data."

Corrected

Figure 3. Suggestion to the overview schematics: The authors might consider adding "molting" as the end point of the red bar (representing differentiation).

The time of molting is not known in the majority of these datasets, because the specimens were fixed and stained prior to molting. We added the relevant information in the figure legend: "Datasets li-13 and li-16 were recorded until the molt; the other recordings were stopped before molting."

Figure 4B': Please indicate that the nuclei signal is DAPI.

Corrected

Supplementary figure 1A. Word is missing in the figure legend: ...the image also shows weak...

Corrected

Supplementary Figure 2: Please indicate the autofluorescence in the granular cells. Does it correspond to the yellow cells?

Corrected

Video legend for video 1 and 2. Please correct "H2B-mREFruby" to "H2B-mRFPruby".

Corrected

Reviewer #2

Major comments:

MC 1. Given that most of the technical advances necessary to achieve the work described in this manuscript have been published previously, it would be helpful for the authors to more clearly identify the primary novelty of this manuscript. The abstract and introduction to the manuscript focus heavily on the technical details of imaging and analysis optimization and some additional summary of the implications of these advances should be included here to aid the reader.

This paper describes a technical advance. While previous work (Alwes et al. 2016) established some key elements of our live imaging approach, we were not at that time able to record the entire time course of leg regeneration (the longest recordings were 3.5 days long). Here we present a method for imaging the entire course of leg regeneration (up to 10 days of imaging), optimised to reduce photodamage and to improve cell tracking. We also develop a method of in situ staining in cuticularised adult legs (an important technical breakthrough in this experimental system), which we combine with live imaging to determine the fate of tracked cells. We have revised the abstract and introduction of the paper to point out these novelties, in relation to our previous publications.

In the abstract we explain: "Building on previous work that allowed us to image different parts of the process of leg regeneration in the crustacean Parhyale hawaiensis, we present here a method for live imaging that captures the entire process of leg regeneration, spanning up to 10 days, at cellular resolution. Our method includes (1) mounting and long-term live imaging of regenerating legs under conditions that yield high spatial and temporal resolution but minimise photodamage, (2) fixing and in situ staining of the regenerated legs that were imaged, to identify cell fates, and (3) computer-assisted cell tracking to determine the cell lineages and progenitors of identified cells. The method is optimised to limit light exposure while maximising tracking efficiency."

The introduction includes the following text: "Our first systematic study using this approach presented continuous live imaging over periods of 2-3 days, capturing key events of leg regeneration such as wound closure, cell proliferation and morphogenesis of regenerating legs with single-cell resolution (Alwes et al., 2016). Here, we extend this work by developing a method for imaging the entire course of leg regeneration, optimised to reduce photodamage and to improve cell tracking. We also develop a method of in situ staining of gene expression in cuticularised adult legs, which we combine with live imaging to determine the fate of tracked cells."

MC 2. The description of the regeneration time course is nicely detailed but also very qualitative. A major advantage of continuous recording and automated cell tracking in the manner presented in this manuscript would be to enable deeper quantitative characterization of cellular and tissue dynamics during regeneration. Rather than providing movies and manually annotated timelines, some characterization of the dynamics of the regeneration process (the heterogeneity in this is very very interesting, but not analyzed at all) and correlating them against cellular behaviors would dramatically increase the impact of the work and leverage the advances presented here. For example, do migration rates differ between replicates? Division rates? Division synchrony? Migration orientation? This seems to be an incredibly rich dataset that would be fascinating to explore in greater detail, which seems to me to be the primary advance presented in this manuscript. I can appreciate that the authors may want to segregate some biological findings from the method, but I believe some nominal effort highlighting the quantitative nature of what this method enables would strengthen the impact of the paper and be useful for the reader. Selecting a small number of simple metrics (eg. Division frequency, average cell migration speed) and plotting them alongside the qualitative phases of the regeneration timeline that have already been generated would be a fairly modest investment of effort using tools that already exist in the Mastodon interface, I would roughly estimate on the order of an hour or two per dataset. I believe that this effort would be well worth it and better highlight a major strength of the approach.

The primary goal of this work was to establish a robust method for continuous long-term live imaging of regeneration, but we do appreciate that a more quantitative analysis would add value to the data we are presenting. We tried to address this request in three steps:

First, we examined whether clear temporal patterns in cell division, cell movements or other cellular features can be observed in an accurately tracked dataset (li13-t4, tracked in Sugawara et al. 2022). To test this we used the feature extraction functions now available on the Mastodon platform (see link). We could discern a meaningful temporal pattern for cell divisions (see below); the other features showed no interpretable pattern of variation.

Second, we asked whether we could use automated cell tracking to analyse the patterns of cell division in all our datasets. Using an Elephant deep learning model trained on the tracks of the li13-t4 dataset, we performed automated cell tracking in the same dataset, and compared the pattern of cell divisions from the automated cell track predictions with those coming from manually validated cell tracks. We observed that the automated tracks gave very imprecise results, with a high background of false positives obscuring the real temporal pattern (see images below, with validated data on the left, automated tracking on the right). These results show that the automated cell tracking is not accurate enough to provide a meaningful picture on the pattern of cell divisions.

Third, we tried to improve the accuracy of detection of dividing cells by additional training of Elephant models on each dataset (to lower the rate of false positives), followed by manual proofreading. Given how labour intensive this is, we could only apply this approach to 4 additional datasets. The results of this analysis are presented in Figure 4.

MC 3. The authors describe the challenges faced by their described approach: Using this mode of semi-automated and manual cell tracking, we find that most cells in the upper slices of our image stacks (top 30 microns) can be tracked with a high degree of confidence. A smaller proportion of cell lineages are trackable in the deeper layers.

Given that the authors quantify this in Table 1, it would aid the reader to provide metrics in the manuscript text at this point. Furthermore, the metrics provided in Table 1 appear to be for overall performance, but the text describes that performance appears to be heavily depth dependent. Segregating the performance metrics further, for example providing DET, TRA, precision and recall for superficial layers only and for the overall dataset, would help support these arguments and better highlight performance a potential adopter of the method might expect.

In the revised manuscript we have added data on the tracking performance of Elephant in relation to imaging depth in Suppl. Figure 3. These data confirm our original statement (which was based on manual tracking) that nuclei are more challenging to track in deeper layers.

We point to these new results in two parts of the paper, as follows: "A smaller proportion of cells are trackable in the deeper layers (see Suppl. Figure 3)", and "Our results, summarised in Table 1A, show that the detection of nuclei can be enhanced by doubling the z resolution at the expense of xy resolution and image quality. This improvement is particularly evident in the deeper layers of the imaging stacks, which are usually the most challenging to track (Suppl. Figure 3)."

MC 4. Performance characterization in Table 1 appears to derive from a single dataset that is then subsampled and processed in different ways to assess the impact of these changes on cell tracking and detection performance. While this is a suitable strategy for this type of optimization it leaves open the question of performance consistency across datasets. I fully recognize that this type of quantification can be onerous and time consuming, but some attempt to assess performance variability across datasets would be valuable. Manual curation over a short time window over a random sampling of the acquired data would be sufficient to assess this.

We think that similar trade-offs will apply to all our datasets because tracking performance is constrained by the same features, which are intrinsic to our system; e.g. by the crowding of nuclei in relation to axial resolution, or the speed of mitosis in relation to the temporal resolution of imaging. We therefore do not see a clear rationale for repeating this analysis. On a practical level, our existing image datasets could not be subsampled to generate the various conditions tested in Table 1, so proving this point experimentally would require generating new recordings, and tracking these to generate ground truth data. This would require months of additional work.

A second, related question is whether Elephant would perform equally well in detecting and tracking nuclei across different datasets. This point has been addressed in the Sugawara et al. 2022 paper, where the performance of Elephant was tested on diverse fluorescence datasets.

Reviewer #3

Major comments:

The authors should clearly specify what are the key technical improvements compared to their previous studies (Alwes et al. 2016, Elife; Konstantinides & Averof 2014, Science). There, the approaches for mounting, imaging, and cell tracking are already introduced, and the imaging is reported to run for up to 7 days in some cases.

In Konstantinides and Averof (2014) we did not present any live imaging at cellular resolution. In Alwes et al. (2016) we described key elements of our live imaging approach, but we were never able to record the entire time course of leg regeneration. The longest recordings in that work were 3.5 days long.

We have revised the abstract and introduction to clarify the novelty of this work, in relation to our previous publications. Please see our response to comment MC1 of reviewer 2.

While the authors mention testing the effect of imaging parameters (such as scanning speed and line averaging) on the imaging/tracking outcome, very little or no information is provided on how this was done beyond the parameters that they finally arrived to.

Scan speed and averaging parameters were determined by measuring contrast and signal-to-noise ratios in images captured over a range of settings. We have now added these data in Supplementary Figure 1.

The authors claim that, using the acquired live imaging data across entire regeneration time course, they are now able to confirm and extend their description of leg regeneration. However, many claims about the order and timing of various cellular events during regeneration are supported only by references to individual snapshots in figures or supplementary movies. Presenting a more quantitative description of cellular processes during regeneration from the acquired data would significantly enhance the manuscript and showcase the usefulness of the improved workflow.

The events we describe can be easily observed in the maximum projections, available in Suppl. Data 2. Regarding the quantitative analysis, please see our response to comment MC2 of reviewer 2.

Table 1 summarizes the performance of cell tracking using simulated datasets of different quality. However only averages and/or maxima are given for the different metrics, which makes it difficult to evaluate the associated conclusions. In some cases, only 1 or 2 test runs were performed.

The metrics extracted from each of the three replicates, per dataset, are now included in Suppl. Data 4.

We consistently used 3 replicates to measure tracking performance with each of the datasets. The "replicates" column label in Table 1 referred to the number of scans that were averaged to generate the image, not to the replicates used for estimating the tracking performance. To avoid confusion, we changed that label to "averaging".

OPTIONAL: An imaging approach that allows using the current mounting strategy but could help with some of the tradeoffs is using a spinning-disk confocal microscope instead of a laser scanning one. If the authors have such a system available, it could be interesting to compare it with their current scanning confocal setup.

Preliminary experiments that we carried out several years ago on a spinning disk confocal (with a 20x objective and the CSU-W1 spinning disk) were not very encouraging, and we therefore did not pursue this approach further. The main problem was bad image quality in deeper tissue layers.

Minor comments:

The presented imaging protocol was optimized for one laser wavelength only (561 nm) - this should be mentioned when discussing the technical limitations since animals tend to react differently to different wavelengths. Same settings might thus not be applicable for imaging a different fluorescent protein.

In the second paragraph of the Results section, we explain that we perform the imaging at long wavelengths in order to minimise photodamage. It should be clear to the readers that changing the excitation wavelength will have an impact for long-term live imaging.

For transferability, it would be useful if the intensity of laser illumination was measured and given in the Methods, instead of just a relative intensity setting from the imaging software. Similarly,more details of the imaging system should be provided where appropriate (e.g., detector specifications).

We have now measured the intensity of the laser illumination and added this information in the Methods: "Laser power was typically set to 0.3% to 0.8%, which yields 0.51 to 1.37 µW at 561 nm (measured with a ThorLabs Microscope Slide Power Sensor, #S170C)."

Regarding the imaging system and the detector, we provide all the information that is available to us on the microscope's technical sheets.

The versions of analysis scripts associated with the manuscript should be uploaded to an online repository that permanently preserves the respective version.

The scripts are now available on gitbub and online repositories. The relevant links are included in the revised manuscript.

Welcome back, and in this lesson, I want to cover EC2 purchase options. EC2 purchase options are often referred to as launch types, but the official way to refer to them from AWS is purchase options, and so to be consistent, I think it's worth focusing on that name. So, EC2 purchase options. Let's step through all of the main types with a focus on the situations where you would and wouldn't use each of them. So, let's jump in and get started.

The first purchase option that I want to talk about is the default, which is on demand, and on demand is simple to explain because it's entirely unremarkable in every way. It's the default because it's the average of anything with no specific pros or cons. Now, the way that it works, let's start with two EC2 hosts. Obviously, AWS has more, but it's easy to diagram with just the two. Now, instances of different sizes when launched using on demand will run on these EC2 hosts, and different AWS customers, they're all mixed up on the shared pool of EC2 hosts. So, even though instances are isolated and protected, different AWS customers launch instances which share the same pool of underlying hardware. This means that AWS can efficiently allocate resources, which is why the starting price for on demand in EC2 is so reasonable.

In terms of the price, on demand uses per second billing, and this happens while instances are running, so you're paying for the resources that you consume. If you shut an instance down logically, you don't pay for those resources. Other associated services such as storage, which do consume resources regardless of if the instance is running or in a shutdown state, do charge constantly while those resources are being consumed. So, remember this: while instances only charge while in the running state, other associated resources may charge regardless. This is how on demand works, but what types of situations should it be used for? Well, it's the default purchase option, and so you should always start your evaluation process by considering on demand as your default. For all projects, assume on demand and move to something else if you can justify that alternative purchase option.

With on demand, there are no interruptions. You launch an instance, you pay a per second charge, and barring any failures, the instance runs until you decide otherwise. You don't receive any capacity reservations with on demand. If AWS has a major failure and capacity is limited, the reserved purchase option receives highest provisioning priority on whatever capacity remains, and so if something is critical to your business, then you should consider an alternative rather than using on demand. So, on demand does not give you any priority access to remaining capacity if there are any major failures.

On demand offers predictable pricing, it's defined upfront, you pay a constant price, but there are no specific discounts. This consistent pricing applies to the duration that you use instances. So, on demand is suitable for short term workloads. Anything which you just need to provision, perform a workload and then terminate is ideal for on demand. If you're unsure about the duration or the type of workload, then again, on demand is ideal. And then lastly, if you have short term or unknown workloads, which definitely can't tolerate any interruption, then on demand is the perfect purchase option.

Next, let's talk about spot pricing, and spot is the cheapest way to get access to EC2 capacity. Let's look at how this works visually. Let's start with the same two EC2 hosts. On the left, we have A and on the right B. Then, on these EC2 hosts, we're currently running four EC2 instances, two per host. And let's assume for this example that all of these four instances are using the on demand purchase option. So, right now, with what you see on screen, the hosts are wasting capacity. Enough capacity for four additional instances on each host is being wasted. Spot pricing is AWS selling that spare capacity at a discounted rate.

The way that it works is that within each region for each type of instance, there is a given amount of free capacity on EC2 hosts at any time. AWS tracks this and it publishes a price for how much it costs to use that capacity, and this price is the spot price. Now, you can offer to pay more than the spot price, but this is a maximum. You'll only ever pay the current spot price for the type of instance in the specific region where you provision services. So, let's say that there are two different customers who want to provision four instances each. The first customer sets a maximum price of four gold coins, and the other customer sets a maximum price of two gold coins. Now, obviously, AWS doesn't charge in gold coins, and there are more than two EC2 hosts, but it's just easier to represent it in this way.

Now, because the current spot price set by AWS is only two gold coins, then both customers are only paying two gold coins a second for their instances. Even though customer one has offered to pay more, this is their maximum and they only ever pay the current spot price. So, let's say now that the free capacity is getting a little bit on the low side. AWS are getting nervous, they know that they need to free up capacity for the normal on demand instances, which they know are about to launch, and so they up the spot price to four gold coins. Now, customer one is fine because they've set a maximum price of four coins, and so now they start paying four coins because that's what the current spot price is. Customer two, they've set their maximum price at two coins, and so their instances are terminated.

If the spot price goes above your maximum price, then any spot instances which you have are terminated. That's the critical part to understand because spot instances should not be viewed as reliable. At this point in our example, maybe another customer decides to launch four on demand instances. AWS sell that capacity at the normal on demand rates, which are higher, and no capacity is wasted. Spot pricing offers up to a 90% reduction versus the price of on demand, and there are some significant trade offs that you need to be aware of.

You should never use the spot purchase option for workloads which can't tolerate interruptions. No matter how well you manage your maximum spot price, there are going to be periods when instances are terminated. If you run workloads where that's a problem, don't use spot. This means that workloads such as domain controllers, mail servers, traditional websites, or even flight control systems are all bad fits for spot instances. The types of scenarios which are good fits for using spot instances are things which are not time critical. Since the spot price changes throughout each day and throughout days of the week, if you're able to process workloads around this, then you can take advantage of the maximum cost benefits for using spot. Anything which can tolerate interruption and just rerun is ideal for spot instances.

So, if you have highly parallel workloads which can be broken into hundreds or thousands of pieces, maybe scientific analysis, and if any parts which fail can be rerun, then spot is ideal. Anything which has a bursty capacity need, maybe media processing, image processing, any cost sensitive workloads which wouldn't be economical to do using normal on-demand instances, assuming they can tolerate interruption, these are ideal for spot. Anything which is stateless where the state of the user session is not stored on the instances themselves, meaning they can handle disruption, again, ideal for using spot. Don't use spot for anything that's long-term, anything that requires consistent, reliable compute, any business critical things, or things which cannot tolerate disruption. For those type of workloads, you should not use spot. It's an anti-pattern.

OK, so this is the end of part one of this lesson. It was getting a little bit on the long side, and I wanted to give you the opportunity to take a small break, maybe stretch your legs or make a coffee. Now, part two will continue immediately from this point, so go ahead, complete this video, and when you're ready, I look forward to you joining me in part two.

Welcome back. In this lesson, now that we've covered virtualization at a high level, I want to focus on the architecture of the EC2 product in more detail. EC2 is one of the services you'll use most often in AWS since one which features on a lot of exam questions, so let's get started.

First thing, let's cover some key, high level architectural points about EC2. EC2 instances are virtual machines, so this means an operating system plus an allocation of resources such as virtual CPU, memory, potential some local storage, maybe some network storage, and access to other hardware such as networking and graphics processing units. EC2 instances run on EC2 hosts, and these are physical servers hardware which AWS manages. These hosts are either shared hosts or dedicated hosts.

Shared hosts are hosts which are shared across different AWS customers, so you don't get any ownership of the hardware and you pay for the individual instances based on how long you run them for and what resources they have allocated. It's important to understand, though, that every customer when using shared hosts are isolated from each other, so there's no visibility of it being shared, there's no interaction between different customers, even if you're using the same shared host, and shared hosts are the default.

With dedicated hosts, you're paying for the entire host, not the instances which run on it. It's yours, it's dedicated to your account, and you don't have to share it with any other customers. So if you pay for a dedicated host, you pay for that entire host, you don't pay for any instances running on it, and you don't share it with other AWS customers.

EC2 is an availability zone resilient service. The reason for this is that hosts themselves run inside a single availability zone, so if that availability zone fails, the hosts inside that availability zone could fail, and any instances running on any hosts that fail will themselves fail. So as a solutions architect, you have to assume if an AZ fails, then at least some and probably all of the instances that are running inside that availability zone will also fail or be heavily impacted.

Now let's look at how this looks visually. So this is a simplification of the US East One region, I've only got two AZs represented, AZA and AZB, and in AZA, I've represented that I've got two subnet, subnet A and subnet B. Now inside each of these availability zones is an EC2 host. Now these EC2 hosts, they run within a single AZ, I'm going to keep repeating that because it's critical for the exam and you're thinking about EC2 in the exam.

Keep thinking about it being an AZ resilient service, if you see EC2 mentioned in an exam, see if you can locate the availability zone details because that might factor into the correct answer. Now EC2 hosts have some local hardware, logically CPU and memory, which you should be aware of, but also they have some local storage called the instance store. The instance store is temporary, if an instance is running on a particular host, depending on the type of the instance, it might be able to utilize this instance store, but if the instance moves off this host to another one, then that storage is lost.

And they also have two types of networking, storage networking and data networking. When instances are provisioned into a specific subnet within a VPC, what's actually happening is that a primary elastic network interface is provisioned in a subnet, which maps to the physical hardware on the EC2 host. Remember, subnets are also in one specific availability zone. Instances can have multiple network interfaces, even in different subnets, as long as they're in the same availability zone. Everything about EC2 is focused around this architecture, the fact that it runs in one specific availability zone.

Now EC2 can make use of remote storage so an EC2 host can connect to the elastic block store, which is known as EBS. The elastic block store service also runs inside a specific availability zone, so the service running inside availability zone A is different than the one running inside availability zone B, and you can't access them cross zone. EBS lets you allocate volumes and volumes of portions of persistent storage, and these can be allocated to instances in the same availability zone, so again, it's another area where the availability zone matters.

What I'm trying to do by keeping repeating availability zone over and over again is to paint a picture of a service which is very reliant on the availability zone that it's running in. The host is in an availability zone, the network is per availability zone, the persistent storage is per availability zone, even availability zone in AWS experiences major issues, it impacts all of those things.

Now an instance runs on a specific host, and if you restart the instance, it will stay on a host. Instances stay on a host until one of two things happen: firstly, the host fails or is taken down for maintenance for some reason by AWS; or secondly, if an instance is stopped and then started, and that's different than just restarting, so I'm focusing on an instance being stopped and then being started, so not just a restart. If either of those things happen, then an instance will be relocated to another host, but that host will also be in the same availability zone.

Instances cannot natively move between availability zones. Everything about them, their hardware, networking and storage is locked inside one specific availability zone. Now there are ways you can do a migration, but it essentially means taking a copy of an instance and creating a brand new one in a different availability zone, and I'll be covering that later in this section where I talk about snapshots and AMIs.

What you can never do is connect network interfaces or EBS storage located in one availability zone to an EC2 instance located in another. EC2 and EBS are both availability zone services, they're isolated, you cannot cross AZs with instances or with EBS volumes. Now instances running on an EC2 host share the resources of that host. And instances of different sizes can share a host, but generally instances of the same type and generation will occupy the same host.

And I'll be talking in much more detail about instance types and sizes and generations in a lesson that's coming up very soon. But when you think about an EC2 host, think that it's from a certain year and includes a certain class of processor and a certain type of memory and a certain type and configuration of storage. And instances are also created with different generations, different versions that you apply specific types of CPU memory and storage, so it's logical that if you provision two different types of instances, they may well end up on two different types of hosts.

So a host generally has lots of different instances from different customers of the same type, but different sizes. So before we finish up this lesson, I want to answer a question. That question is what's EC2 good for? So what types of situations might you use EC2 for? And this is equally valuable when you're evaluating a technical architecture while you're answering questions in the exam.

So first, EC2 is great when you've got a traditional OS and application compute need, so if you've got an application that requires to be running on a certain operating system at a certain runtime with certain configuration, maybe your internal technical staff are used to that configuration, or maybe your vendor has a certain set of support requirements, EC2 is a perfect use case for this type of scenario.

And it's also great for any long running compute needs. There are lots of other services inside AWS that provide compute services, but many of these have got runtime limits, so you can't leave these things running consistently for one year or two years. With EC2, it's designed for persistent, long running compute requirements. So if you have an application that runs constantly 24/7, 365, and needs to be running on a normal operating system, Linux or Windows, then EC2 is the default and obvious choice for this.

If you have any applications, which is server style applications, so traditional applications they expect to be running in an operating system, waiting for incoming connections, then again, EC2 is a perfect service for this. And it's perfect for any applications or services that need burst requirements or steady state requirements. There are different types of EC2 instances, which are suitable for low levels of normal loads with occasional bursts, as well as steady state load.

So again, if your application needs an operating system, and it's not bursty needs or consistent steady state load, then EC2 should be the first thing that you review. EC2 is also great for monolithic application stack, so if your monolithic application requires certain components, a stack, maybe a database, maybe some middleware, maybe other runtime based components, and especially if it needs to be running on a traditional operating system, EC2 should be the first thing that you look at.

And EC2 is also ideally suited for migrating application workloads, so application workloads, which expect a traditional virtual machine or server style environment, or if you're performing disaster recovery. So if you have existing traditional systems which run on virtual servers, and you want to provision a disaster recovery environment, then EC2 is perfect for that.

In general, EC2 tends to be the default compute service within AWS. There are lots of niche requirements that you might have, and if you do have those, there are other compute services such as the elastic container service or Lambda. But generally, if you've got traditional style workloads, or you're looking for something that's consistent, or if it requires an operating system, or if it's monolithic, or if you migrated into AWS, then EC2 is a great default first option.

Now in this section of the course, I'm covering the basic architectural components of EC2, so I'm gonna be introducing the basics and let you get some exposure to it, and I'm gonna be teaching you all the things that you'll need for the exam.

Author response:

The following is the authors’ response to the original reviews

eLife Assessment

Examination of (a)periodic brain activity has gained particular interest in the last few years in the neuroscience fields relating to cognition, disorders, and brain states. Using large EEG/MEG datasets from younger and older adults, the current study provides compelling evidence that age-related differences in aperiodic EEG/MEG signals can be driven by cardiac rather than brain activity. Their findings have important implications for all future research that aims to assess aperiodic neural activity, suggesting control for the influence of cardiac signals is essential.

We want to thank the editors for their assessment of our work and highlighting its importance for the understanding of aperiodic neural activity. Additionally, we want to thank the three present and four former reviewers (at a different journal) whose comments and ideas were critical in shaping this manuscript to its current form. We hope that this paper opens up many more questions that will guide us - as a field - to an improved understanding of how “cortical” and “cardiac” changes in aperiodic activity are linked and want to invite readers to engage with our work through eLife’s comment function.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The present study addresses whether physiological signals influence aperiodic brain activity with a focus on age-related changes. The authors report age effects on aperiodic cardiac activity derived from ECG in low and high-frequency ranges in roughly 2300 participants from four different sites. Slopes of the ECGs were associated with common heart variability measures, which, according to the authors, shows that ECG, even at higher frequencies, conveys meaningful information. Using temporal response functions on concurrent ECG and M/EEG time series, the authors demonstrate that cardiac activity is instantaneously reflected in neural recordings, even after applying ICA analysis to remove cardiac activity. This was more strongly the case for EEG than MEG data. Finally, spectral parameterization was done in large-scale resting-state MEG and ECG data in individuals between 18 and 88 years, and age effects were tested. A steepening of spectral slopes with age was observed particularly for ECG and, to a lesser extent, in cleaned MEG data in most frequency ranges and sensors investigated. The authors conclude that commonly observed age effects on neural aperiodic activity can mainly be explained by cardiac activity.

Strengths:

Compared to previous investigations, the authors demonstrate the effects of aging on the spectral slope in the currently largest MEG dataset with equal age distribution available. Their efforts of replicating observed effects in another large MEG dataset and considering potential confounding by ocular activity, head movements, or preprocessing methods are commendable and valuable to the community. This study also employs a wide range of fitting ranges and two commonly used algorithms for spectral parameterization of neural and cardiac activity, hence providing a comprehensive overview of the impact of methodological choices. Based on their findings, the authors give recommendations for the separation of physiological and neural sources of aperiodic activity.

Weaknesses:

While the aim of the study is well-motivated and analyses rigorously conducted, the overall structure of the manuscript, as it stands now, is partially misleading. Some of the described results are not well-embedded and lack discussion.

We want to thank the reviewer for their comments focussed on improving the overall structure of the manuscript. We agree with their suggestions that some results could be more clearly contextualized and restructured the manuscript accordingly.

Reviewer #2 (Public review):

I previously reviewed this important and timely manuscript at a previous journal where, after two rounds of review, I recommended publication. Because eLife practices an open reviewing format, I will recapitulate some of my previous comments here, for the scientific record.

In that previous review, I revealed my identity to help reassure the authors that I was doing my best to remain unbiased because I work in this area and some of the authors' results directly impact my prior research. I was genuinely excited to see the earlier preprint version of this paper when it first appeared. I get a lot of joy out of trying to - collectively, as a field - really understand the nature of our data, and I continue to commend the authors here for pushing at the sources of aperiodic activity!

In their manuscript, Schmidt and colleagues provide a very compelling, convincing, thorough, and measured set of analyses. Previously I recommended that the push even further, and they added the current Figure 5 analysis of event-related changes in the ECG during working memory. In my opinion this result practically warrants a separate paper its own!

The literature analysis is very clever, and expanded upon from any other prior version I've seen.

In my previous review, the broadest, most high-level comment I wanted to make was that authors are correct. We (in my lab) have tried to be measured in our approach to talking about aperiodic analyses - including adopting measuring ECG when possible now - because there are so many sources of aperiodic activity: neural, ECG, respiration, skin conductance, muscle activity, electrode impedances, room noise, electronics noise, etc. The authors discuss this all very clearly, and I commend them on that. We, as a field, should move more toward a model where we can account for all of those sources of noise together. (This was less of an action item, and more of an inclusion of a comment for the record.)

I also very much appreciate the authors' excellent commentary regarding the physiological effects that pharmacological challenges such as propofol and ketamine also have on non-neural (autonomic) functions such as ECG. Previously I also asked them to discuss the possibility that, while their manuscript focuses on aperiodic activity, it is possible that the wealth of literature regarding age-related changes in "oscillatory" activity might be driven partly by age-related changes in neural (or non-neural, ECG-related) changes in aperiodic activity. They have included a nice discussion on this, and I'm excited about the possibilities for cognitive neuroscience as we move more in this direction.

Finally, I previously asked for recommendations on how to proceed. The authors convinced me that we should care about how the ECG might impact our field potential measures, but how do I, as a relative novice, proceed. They now include three strong recommendations at the end of their manuscript that I find to be very helpful.

As was obvious from previous review, I consider this to be an important and impactful cautionary report, that is incredibly well supported by multiple thorough analyses. The authors have done an excellent job responding to all my previous comments and concerns and, in my estimation, those of the previous reviewers as well.

We want to thank the reviewer for agreeing to review our manuscript again and for recapitulating on their previous comments and the progress the manuscript has made over the course of the last ~2 years. The reviewer's comments have been essential in shaping the manuscript into its current form. Their feedback has made the review process truly feel like a collaborative effort, focused on strengthening the manuscript and refining its conclusions and resulting recommendations.

Reviewer #3 (Public review):

Summary:

Schmidt et al., aimed to provide an extremely comprehensive demonstration of the influence cardiac electromagnetic fields have on the relationship between age and the aperiodic slope measured from electroencephalographic (EEG) and magnetoencephalographic (MEG) data.

Strengths:

Schmidt et al., used a multiverse approach to show that the cardiac influence on this relationship is considerable, by testing a wide range of different analysis parameters (including extensive testing of different frequency ranges assessed to determine the aperiodic fit), algorithms (including different artifact reduction approaches and different aperiodic fitting algorithms), and multiple large datasets to provide conclusions that are robust to the vast majority of potential experimental variations.

The study showed that across these different analytical variations, the cardiac contribution to aperiodic activity measured using EEG and MEG is considerable, and likely influences the relationship between aperiodic activity and age to a greater extent than the influence of neural activity.

Their findings have significant implications for all future research that aims to assess aperiodic neural activity, suggesting control for the influence of cardiac fields is essential.

We want to thank the reviewer for their thorough engagement with our work and the resultant substantive amount of great ideas both mentioned in the section of Weaknesses and Authors Recommendations below. Their suggestions have sparked many ideas in us on how to move forward in better separating peripheral- from neuro-physiological signals that are likely to greatly influence our future attempts to better extract both cardiac and muscle activity from M/EEG recordings. So we want to thank them for their input, time and effort!

Weaknesses:

Figure 4I: The regressions explained here seem to contain a very large number of potential predictors. Based on the way it is currently written, I'm assuming it includes all sensors for both the ECG component and ECG rejected conditions?

I'm not sure about the logic of taking a complete signal, decomposing it with ICA to separate out the ECG and non-ECG signals, then including these latent contributions to the full signal back into the same regression model. It seems that there could be some circularity or redundancy in doing so. Can the authors provide a justification for why this is a valid approach?

After observing significant effects both in the MEG_{ECG component} and MEG_{ECG rejected} conditions in similar frequency bands we wanted to understand whether or not these age-related changes are statistically independent. To test this we added both variables as predictors in a regression model (thereby accounting for the influence of the other in relation to age). The regression models we performed were therefore actually not very complex. They were built using only two predictors, namely the data (in a specific frequency range) averaged over channels on which we noticed significant effects in the ECG rejected and ECG components data respectively (Wilkinson notation: age ~ 1 + ECG rejected + ECG components). This was also described in the results section stating that: “To see if MEG_{ECG rejected} and MEG_{ECG component} explain unique variance in aging at frequency ranges where we noticed shared effects, we averaged the spectral slope across significant channels and calculated a multiple regression model with MEG_{ECG component} and MEG_{ECG rejected} as predictors for age (to statistically control for the effect of MEG_{ECG component}s and MEG_{ECG rejected} on age). This analysis was performed to understand whether the observed shared age-related effects (MEG_{ECG rejected} and MEG_{ECG component}) are in(dependent).”  

We hope this explanation solves the previous misunderstanding.

I'm not sure whether there is good evidence or rationale to support the statement in the discussion that the presence of the ECG signal in reference electrodes makes it more difficult to isolate independent ECG components. The ICA algorithm will still function to detect common voltage shifts from the ECG as statistically independent from other voltage shifts, even if they're spread across all electrodes due to the referencing montage. I would suggest there are other reasons why the ICA might lead to imperfect separation of the ECG component (assumption of the same number of source components as sensors, non-Gaussian assumption, assumption of independence of source activities).

The inclusion of only 32 channels in the EEG data might also have reduced the performance of ICA, increasing the chances of imperfect component separation and the mixing of cardiac artifacts into the neural components, whereas the higher number of sensors in the MEG data would enable better component separation. This could explain the difference between EEG and MEG in the ability to clean the ECG artifact (and perhaps higher-density EEG recordings would not show the same issue).

The reviewer is making a good argument suggesting that our initial assumption that the presence of cardiac activity on the reference electrode influences the performance of the ICA may be wrong. After rereading and rethinking upon the matter we think that the reviewer is correct and that their assumptions for why the ECG signal was not so easily separable from our EEG recordings are more plausible and better grounded in the literature than our initial suggestion. We therefore now highlight their view as a main reason for why the ECG rejection was more challenging in EEG data. However, we also note that understanding the exact reason probably ends up being an empirical question that demands further research stating that:

“Difficulties in removing ECG related components from EEG signals via ICA might be attributable to various reasons such as the number of available sensors or assumptions related to the non-gaussianity of the underlying sources. Further understanding of this matter is highly important given that ICA is the most widely used procedure to separate neural from peripheral physiological sources. ”

In addition to the inability to effectively clean the ECG artifact from EEG data, ICA and other component subtraction methods have also all been shown to distort neural activity in periods that aren't affected by the artifact due to the ubiquitous issue of imperfect component separation (https://doi.org/10.1101/2024.06.06.597688). As such, component subtraction-based (as well as regression-based) removal of the cardiac artifact might also distort the neural contributions to the aperiodic signal, so even methods to adequately address the cardiac artifact might not solve the problem explained in the study. This poses an additional potential confound to the "M/EEG without ECG" conditions.

The reviewer is correct in stating that, if an “artifactual” signal is not always present but appears and disappears (like e.g. eye-blinks) neural activity may be distorted in periods where the “artifactual” signal is absent. However, while this plausibly presents a problem for ocular activity, there is no obvious reason to believe that this applies to cardiac activity. While the ECG signal is non-stationary in nature, it is remarkably more stable than eye-movements in the healthy populations we analyzed (especially at rest). Therefore, the presence of the cardiac “artifact” was consistently present across the entirety of the MEG recordings we visually inspected.

Literature Analysis, Page 23: was there a method applied to address studies that report reducing artifacts in general, but are not specific to a single type of artifact? For example, there are automated methods for cleaning EEG data that use ICLabel (a machine learning algorithm) to delete "artifact" components. Within these studies, the cardiac artifact will not be mentioned specifically, but is included under "artifacts".

The literature analysis was largely performed automatically and solely focussed on ECG related activity as described in the methods section under Literature Analysis, if no ECG related terms were used in the context of artifact rejection a study was flagged as not having removed cardiac activity. This could have been indeed better highlighted by us and we apologize for the oversight on our behalf. We now additionally link to these details stating that:

“However, an analysis of openly accessible M/EEG articles (N_Articles=279; see Methods - Literature Analysis for further details) that investigate aperiodic activity revealed that only 17.1% of EEG studies explicitly mention that cardiac activity was removed and only 16.5% measure ECG (45.9% of MEG studies removed cardiac activity and 31.1% of MEG studies mention that ECG was measured; see Figure 1EF).”

The reviewer makes a fair point that there is some uncertainty here and our results probably present a lower bound of ECG handling in M/EEG research as, when I manually rechecked the studies that were not initially flagged in studies it was often solely mentioned that “artifacts” were rejected. However, this information seemed too ambiguous to assume that cardiac activity was in fact accounted for. However, again this could have been mentioned more clearly in writing and we apologize for this oversight. Now this is included as part of the methods section Literature Analysis stating that:

“All valid word contexts were then manually inspected by scanning the respective word context to ensure that the removal of “artifacts” was related specifically to cardiac and not e.g. ocular activity or the rejection of artifacts in general (without specifying which “artifactual” source was rejected in which case the manuscript was marked as invalid). This means that the results of our literature analysis likely present a lower bound for the rejection of cardiac activity in the M/EEG literature investigating aperiodic activity.”

Statistical inferences, page 23: as far as I can tell, no methods to control for multiple comparisons were implemented. Many of the statistical comparisons were not independent (or even overlapped with similar analyses in the full analysis space to a large extent), so I wouldn't expect strong multiple comparison controls. But addressing this point to some extent would be useful (or clarifying how it has already been addressed if I've missed something).

In the present study we tried to minimize the risk of type 1 errors by several means, such as A) weakly informative priors, B) robust regression models and C) by specifying a region of practical equivalence (ROPE, see Methods Statistical Inference for further Information) to define meaningful effects.

Weakly informative priors can lower the risk of type 1 errors arising from multiple testing by shrinking parameter estimates towards zero (see e.g. Lemoine, 2019). Robust regression models use a Student T distribution to describe the distribution of the data. This distribution features heavier tails, meaning it allocates more probability to extreme values, which in turn minimizes the influence of outliers. The ROPE criterion ensures that only effects exceeding a negligible size are considered meaningful, representing a strict and conservative approach to interpreting our findings (see Kruschke 2018, Cohen, 1988).

Furthermore, and more generally we do not selectively report “significant” effects in the situations in which multiple analyses were conducted on the same family of data (e.g. Figure 2 & 4). Instead we provide joint inference across several plausible analysis options (akin to a specification curve analysis, Simonsohn, Simmons & Nelson 2020) to provide other researchers with an overview of how different analysis choices impact the association between cardiac and neural aperiodic activity.

Lemoine, N. P. (2019). Moving beyond noninformative priors: why and how to choose weakly informative priors in Bayesian analyses. Oikos, 128(7), 912-928.

Simonsohn, U., Simmons, J. P., & Nelson, L. D. (2020). Specification curve analysis. Nature Human Behaviour, 4(11), 1208-1214.

Methods:

Applying ICA components from 1Hz high pass filtered data back to the 0.1Hz filtered data leads to worse artifact cleaning performance, as the contribution of the artifact in the 0.1Hz to 1Hz frequency band is not addressed (see Bailey, N. W., Hill, A. T., Biabani, M., Murphy, O. W., Rogasch, N. C., McQueen, B., ... & Fitzgerald, P. B. (2023). RELAX part 2: A fully automated EEG data cleaning algorithm that is applicable to Event-Related-Potentials. Clinical Neurophysiology, result reported in the supplementary materials). This might explain some of the lower frequency slope results (which include a lower frequency limit <1Hz) in the EEG data - the EEG cleaning method is just not addressing the cardiac artifact in that frequency range (although it certainly wouldn't explain all of the results).

We want to thank the reviewer for suggesting this interesting paper, showing that lower high-pass filters may be preferable to the more commonly used >1Hz high-pass filters for detection of ICA components that largely contain peripheral physiological activity. However, the results presented by Bailey et al. contradict the more commonly reported findings by other researchers that >1Hz high-pass filter is actually preferable (e.g. Winkler et al. 2015; Dimingen, 2020 or Klug & Gramann, 2021) and recommendations in widely used packages for M/EEG analysis (e.g. https://mne.tools/1.8/generated/mne.preprocessing.ICA.html). Yet, the fact that there seems to be a discrepancy suggests that further research is needed to better understand which type of high-pass filtering is preferable in which situation. Furthermore, it is notable that all the findings for high-pass filtering in ICA component detection and removal that we are aware of relate to ocular activity. Given that ocular and cardiac activity have very different temporal and spectral patterns it is probably worth further investigating whether the classic 1Hz high-pass filter is really also the best option for the detection and removal of cardiac activity. However, in our opinion this requires a dedicated investigation on its own..

We therefore highlight this now in our manuscript stating that:

“Additionally, it is worth noting that the effectiveness of an ICA crucially depends on the quality of the extracted components(63,64) and even widely suggested settings e.g. high-pass filtering at 1Hz before fitting an ICA may not be universally applicable (see supplementary material of (64)).

Winkler, S. Debener, K. -R. Müller and M. Tangermann, "On the influence of high-pass filtering on ICA-based artifact reduction in EEG-ERP," 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Milan, Italy, 2015, pp. 4101-4105, doi: 10.1109/EMBC.2015.7319296.

Dimigen, O. (2020). Optimizing the ICA-based removal of ocular EEG artifacts from free viewing experiments. NeuroImage, 207, 116117.

Klug, M., & Gramann, K. (2021). Identifying key factors for improving ICA‐based decomposition of EEG data in mobile and stationary experiments. European Journal of Neuroscience, 54(12), 8406-8420.

It looks like no methods were implemented to address muscle artifacts. These can affect the slope of EEG activity at higher frequencies. Perhaps the Riemannian Potato addressed these artifacts, but I suspect it wouldn't eliminate all muscle activity. As such, I would be concerned that remaining muscle artifacts affected some of the results, particularly those that included high frequency ranges in the aperiodic estimate. Perhaps if muscle activity were left in the EEG data, it could have disrupted the ability to detect a relationship between age and 1/f slope in a way that didn't disrupt the same relationship in the cardiac data (although I suspect it wouldn't reverse the overall conclusions given the number of converging results including in lower frequency bands). Is there a quick validity analysis the authors can implement to confirm muscle artifacts haven't negatively affected their results?

I note that an analysis of head movement in the MEG is provided on page 32, but it would be more robust to show that removing ICA components reflecting muscle doesn't change the results. The results/conclusions of the following study might be useful for objectively detecting probable muscle artifact components: Fitzgibbon, S. P., DeLosAngeles, D., Lewis, T. W., Powers, D. M. W., Grummett, T. S., Whitham, E. M., ... & Pope, K. J. (2016). Automatic determination of EMG-contaminated components and validation of independent component analysis using EEG during pharmacologic paralysis. Clinical neurophysiology, 127(3), 1781-1793.

We thank the reviewer for their suggestion. Muscle activity can indeed be a potential concern, for the estimation of the spectral slope. This is precisely why we used head movements (as also noted by the reviewer) as a proxy for muscle activity. We also agree with the reviewer that this is not a perfect estimate. Additionally, also the riemannian potato would probably only capture epochs that contain transient, but not persistent patterns of muscle activity.

The paper recommended by the reviewer contains a clever approach of using the steepness of the spectral slope (or lack thereof) as an indicator whether or not an independent component (IC) is driven by muscle activity. In order to determine an optimal threshold Fitzgibbon et al. compared paralyzed to temporarily non paralyzed subjects. They determined an expected “EMG-free” threshold for their spectral slope on paralyzed subjects and used this as a benchmark to detect IC’s that were contaminated by muscle activity in non paralyzed subjects.

This is a great idea, but unfortunately would go way beyond what we are able to sensibly estimate with our data for the following reasons. The authors estimated their optimal threshold on paralyzed subjects for EEG data and show that this is a feasible threshold to be applied across different recordings. So for EEG data it might be feasible, at least as a first shot, to use their threshold on our data. However, we are measuring MEG and as alluded to in our discussion section under “Differences in aperiodic activity between magnetic and electric field recordings” the spectral slope differs greatly between MEG and EEG recordings for non-trivial reasons. Furthermore, the spectral slope even seems to also differ across different MEG devices. We noticed this when we initially tried to pool the data recorded in Salzburg with the Cambridge dataset. This means we would need to do a complete validation of this procedure for the MEG data recorded in Cambridge and in Salzburg, which is not feasible considering that we A) don’t have direct access to one of the recording sites and B) would even if we had access face substantial hurdles to get ethical approval for the experiment performed by Fitzgibbon et al..

However, we think the approach brought forward by Fitzgibbon and colleagues is a clever way to remove muscle activity from EEG recordings, whenever EMG was not directly recorded. We therefore suggested in the Discussion section that ideally also EMG should be recorded stating that:

“It is worth noting that, apart from cardiac activity, muscle activity can also be captured in (non-)invasive recordings and may drastically influence measures of the spectral slope(72). To ensure that persistent muscle activity does not bias our results we used changes in head movement velocity as a control analysis (see Supplementary Figure S9). However, it should be noted that this is only a proxy for the presence of persistent muscle activity. Ideally, studies investigating aperiodic activity should also be complemented by measurements of EMG. Whenever such measurements are not available creative approaches that use the steepness of the spectral slope (or the lack thereof) as an indicator to detect whether or not e.g. an independent component is driven by muscle activity are promising(72,73). However, these approaches may require further validation to determine how well myographic aperiodic thresholds are transferable across the wide variety of different M/EEG devices.”

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) As outlined above, I recommend rephrasing the last section of the introduction to briefly summarize/introduce all main analysis steps undertaken in the study and why these were done (for example, it is only mentioned that the Cam-CAN dataset was used to study the impact of cardiac on MEG activity although the author used a variety of different datasets). Similarly, I am missing an overview of all main findings in the context of the study goals in the discussion. I believe clarifying the structure of the paper would not only provide a red thread to the reader but also highlight the efforts/strength of the study as described above.

This is a good call! As suggested by the reviewer we now try to give a clearer overview of what was investigated why. We do that both at the end of the introduction stating that: “Using the publicly available Cam-CAN dataset(28,29), we find that the aperiodic signal measured using M/EEG originates from multiple physiological sources. In particular, significant portions of age-related changes in aperiodic activity –normally attributed to neural processes– can be better explained by cardiac activity. This observation holds across a wide range of processing options and control analyses (see Supplementary S1), and was replicable on a separate MEG dataset. However, the extent to which cardiac activity accounts for age-related changes in aperiodic activity varies with the investigated frequency range and recording site. Importantly, in some frequency ranges and sensor locations, age-related changes in neural aperiodic activity still prevail. But does the influence of cardiac activity on the aperiodic spectrum extend beyond age? In a preliminary analysis, we demonstrate that working memory load modulates the aperiodic spectrum of “pure” ECG recordings. The direction of this working memory effect mirrors previous findings on EEG data(5) suggesting that the impact of cardiac activity goes well beyond aging. In sum, our results highlight the complexity of aperiodic activity while cautioning against interpreting it as solely “neural“ without considering physiological influences.”

and at the beginning of the discussion section:

“Difficulties in removing ECG related components from EEG signals via ICA might be attributable to various reasons such as the number of available sensors or assumptions related to the non-gaussianity of the underlying sources. Further understanding of this matter is highly important given that ICA is the most widely used procedure to separate neural from peripheral physiological sources (see Figure 1EF). Additionally, it is worth noting that the effectiveness of an ICA crucially depends on the quality of the extracted components(63,64) and even widely suggested settings e.g. high-pass filtering at 1Hz before fitting an ICA may not be universally applicable (see supplementary material of (64)). “

(2) I found it interesting that the spectral slopes of ECG activity at higher frequency ranges (> 10 Hz) seem mostly related to HRV measures such as fractal and time domain indices and less so with frequency-domain indices. Do the authors have an explanation for why this is the case? Also, the analysis of the HRV measures and their association with aperiodic ECG activity is not explained in any of the method sections.

We apologize for the oversight in not mentioning the HRV analysis in more detail in our methods section. We added a subsection to the Methods section entitled ECG Processing - Heart rate variability analysis to further describe the HRV analyses.

“ECG Processing - Heart rate variability analysis

Heart rate variability (HRV) was computed using the NeuroKit2 toolbox, a high level tool for the analysis of physiological signals. First, the raw electrocardiogram (ECG) data were preprocessed, by highpass filtering the signal at 0.5Hz using an infinite impulse response (IIR) butterworth filter(order=5) and by smoothing the signal with a moving average kernel with the width of one period of 50Hz to remove the powerline noise (default settings of neurokit.ecg.ecg_clean). Afterwards, QRS complexes were detected based on the steepness of the absolute gradient of the ECG signal. Subsequently, R-Peaks were detected as local maxima in the QRS complexes (default settings of neurokit.ecg.ecg_peaks; see (98) for a validation of the algorithm). From the cleaned R-R intervals, 90 HRV indices were derived, encompassing time-domain, frequency-domain, and non-linear measures. Time-domain indices included standard metrics such as the mean and standard deviation of the normalized R-R intervals , the root mean square of successive differences, and other statistical descriptors of interbeat interval variability. Frequency-domain analyses were performed using power spectral density estimation, yielding for instance low frequency (0.04-0.15Hz) and high frequency (0.15-0.4Hz) power components. Additionally, non-linear dynamics were characterized through measures such as sample entropy, detrended fluctuation analysis and various Poincaré plot descriptors. All these measures were then related to the slopes of the low frequency (0.25 – 20 Hz) and high frequency (10 – 145 Hz) aperiodic spectrum of the raw ECG.”

With regards to association of the ECG’s spectral slopes at high frequencies and frequency domain indices of heart rate variability. Common frequency domain indices of heart rate variability fall in the range of 0.01-.4Hz. Which probably explains why we didn’t notice any association at higher frequency ranges (>10Hz).

This is also stated in the related part of the results section:

“In the higher frequency ranges (10 - 145 Hz) spectral slopes were most consistently related to fractal and time domain indices of heart rate variability, but not so much to frequency-domain indices assessing spectral power in frequency ranges < 0.4 Hz.”

(3) Related to the previous point - what is being reflected in the ECG at higher frequency ranges, with regard to biological mechanisms? Results are being mentioned, but not further discussed. However, this point seems crucial because the age effects across the four datasets differ between low and high-frequency slope limits (Figure 2C).

This is a great question that definitely also requires further attention and investigation in general (see also Tereshchenko & Josephson, 2015). We investigated the change of the slope across frequency ranges that are typically captured in common ECG setups for adults (0.05 - 150Hz, Tereshchenko & Josephson, 2015; Kusayama, Wong, Liu et al. 2020). While most of the physiological significant spectral information of an ECG recording rests between 1-50Hz (Clifford & Azuaje, 2006), meaningful information can be extracted at much higher frequencies. For instance, ventricular late potentials have a broader frequency band (~40-250Hz) that falls straight in our spectral analysis window. However, that’s not all, as further meaningful information can be extracted at even higher frequencies (>100Hz). Yet, the exact physiological mechanisms underlying so-called high-frequency QRS remain unclear (HF-QRS; see Tereshchenko & Josephson, 2015; Qiu et al. 2024 for a review discussing possible mechanisms). Yet, at the same time the HF-QRS seems to be highly informative for the early detection of myocardial ischemia and other cardiac abnormalities that may not yet be evident in the standard frequency range (Schlegel et al. 2004; Qiu et al. 2024). All optimism aside, it is also worth noting that ECG recordings at higher frequencies can capture skeletal muscle activity with an overlapping frequency range up to 400Hz (Kusayama, Wong, Liu et al. 2020). We highlight all of this now when introducing this analysis in the results sections as outstanding research question stating that:

“However, substantially less is known about aperiodic activity above 0.4Hz in the ECG. Yet, common ECG setups for adults capture activity at a broad bandwidth of 0.05 - 150Hz(33,34).

Importantly, a lot of the physiological meaningful spectral information rests between 1-50Hz(35), similarly to M/EEG recordings. Furthermore, meaningful information can be extracted at much higher frequencies. For instance, ventricular late potentials have a broader frequency band (~40-250Hz(35)). However, that’s not all, as further meaningful information can be extracted at even higher frequencies (>100Hz). For instance, the so-called high-frequency QRS seems to be highly informative for the early detection of myocardial ischemia and other cardiac abnormalities that may not yet be evident in the standard frequency range(36,37). Yet, the exact physiological mechanisms underlying the high-frequency QRS remain unclear (see (37) for a review discussing possible mechanisms). ”

Tereshchenko, L. G., & Josephson, M. E. (2015). Frequency content and characteristics of ventricular conduction. Journal of electrocardiology, 48(6), 933-937.

Kusayama, T., Wong, J., Liu, X. et al. Simultaneous noninvasive recording of electrocardiogram and skin sympathetic nerve activity (neuECG). Nat Protoc 15, 1853–1877 (2020). https://doi.org/10.1038/s41596-020-0316-6

Clifford, G. D., & Azuaje, F. (2006). Advanced methods and tools for ECG data analysis (Vol. 10). P. McSharry (Ed.). Boston: Artech house.

Qiu, S., Liu, T., Zhan, Z., Li, X., Liu, X., Xin, X., ... & Xiu, J. (2024). Revisiting the diagnostic and prognostic significance of high-frequency QRS analysis in cardiovascular diseases: a comprehensive review. Postgraduate Medical Journal, qgae064.

Schlegel, T. T., Kulecz, W. B., DePalma, J. L., Feiveson, A. H., Wilson, J. S., Rahman, M. A., & Bungo, M. W. (2004, March). Real-time 12-lead high-frequency QRS electrocardiography for enhanced detection of myocardial ischemia and coronary artery disease. In Mayo Clinic Proceedings (Vol. 79, No. 3, pp. 339-350). Elsevier.

(4) Page 10: At first glance, it is not quite clear what is meant by "processing option" in the text. Please clarify.

Thank you for catching this! Upon re-reading this is indeed a bit oblivious. We now swapped “processing options” with “slope fits” to make it clearer that we are talking about the percentage of effects based on the different slope fits.

(5) The authors mention previous findings on age effects on neural 1/f activity (References Nr 5,8,27,39) that seem contrary to their own findings such as e.g., the mostly steepening of the slopes with age. Also, the authors discuss thoroughly why spectral slopes derived from MEG signals may differ from EEG signals. I encourage the authors to have a closer look at these studies and elaborate a bit more on why these studies differ in their conclusions on the age effects. For example, Tröndle et al. (2022, Ref. 39) investigated neural activity in children and young adults, hence, focused on brain maturation, whereas the CamCAN set only considers the adult lifespan. In a similar vein, others report age effects on 1/f activity in much smaller samples as reported here (e.g., Voytek et al., 2015).

I believe taking these points into account by briefly discussing them, would strengthen the authors' claims and provide a more fine-grained perspective on aging effects on 1/f.

The reviewer is making a very important point. As age-related differences in (neuro-)physiological activity are not necessarily strictly comparable and entirely linear across different age-cohorts (e.g. age-related changes in alpha center frequency). We therefore, added the suggested discussion points to the discussion section.

“Differences in electric and magnetic field recordings aside, aperiodic activity may not change strictly linearly as we are ageing and studies looking at younger age groups (e.g. <22; (44) may capture different aspects of aging (e.g. brain maturation), than those looking at older subjects (>18 years; our sample). A recent report even shows some first evidence of an interesting putatively non-linear relationship with age in the sensorimotor cortex for resting recordings(59)”

(6) The analysis of the working memory paradigm as described in the outlook-section of the discussion comes as a bit of a surprise as it has not been introduced before. If the authors want to convey with this study that, in general, aperiodic neural activity could be influenced by aperiodic cardiac activity, I recommend introducing this analysis and the results earlier in the manuscript than only in the discussion to strengthen their message.

The reviewer is correct. This analysis really comes a bit out of the blue. However, this was also exactly the intention for placing this analysis in the discussion. As the reviewer correctly noted, the aim was to suggest “that, in general, aperiodic neural activity could be influenced by aperiodic cardiac activity”. We placed this outlook directly after the discussion of “(neuro-)physiological origins of aperiodic activity”, where we highlight the potential challenges of interpreting drug induced changes to M/EEG recordings. So the aim was to get the reader to think about whether age is the only feature affected by cardiac activity and then directly present some evidence that this might go beyond age.

However, we have been rethinking this approach based on the reviewers comments and moved that paragraph to the end of the results section accordingly and introduce it already at the end of the introduction stating that:

“But does the influence of cardiac activity on the aperiodic spectrum extend beyond age? In a preliminary analysis, we demonstrate that working memory load modulates the aperiodic spectrum of “pure” ECG recordings. The direction of this working memory effect mirrors previous findings on EEG data(5) suggesting that the impact of cardiac activity goes well beyond aging.”

(7) The font in Figure 2 is a bit hard to read (especially in D). I recommend increasing the font sizes where necessary for better readability.

We agree with the Reviewer and increased the font sizes accordingly.

(8) Text in the discussion: Figure 3B on page 10 => shouldn't it be Figure 4?

Thank you for catching this oversight. We have now corrected this mistake.

(9) In the third section on page 10, the Figure labels seem to be confused. For example, Figure 4 E is supposed to show "steepening effects", which should be Figure 4B I believe.

Please check the figure labels in this section to avoid confusion.

Thank you for catching this oversight. We have now corrected this mistake.

(10) Figure Legend 4 I), please check the figure labels in the text

Thank you for catching this oversight. We have now corrected this mistake.

Reviewer #3 (Recommendations for the authors):

I have a number of suggestions for improving the manuscript, which I have divided by section in the following:

ABSTRACT:

I would suggest re-writing the first sentences to make it easier to read for non-expert readers: "The power of electrophysiologically measured cortical activity decays with an approximately 1/fX function. The slope of this decay (i.e. the spectral exponent, X) is modulated..."

Thank you for the suggestion. We adjusted the sentence as suggested to make it easier for less technical readers to understand that “X” refers to the exponent.

Including the age range that was studied in the abstract could be informative.

Done as suggested.

As an optional recommendation, I think it would increase the impact of the article if the authors note in the abstract that the current most commonly applied cardiac artifact reduction approaches don't resolve the issue for EEG data, likely due to an imperfect ability to separate the cardiac artifact from the neural activity with independent component analysis. This would highlight to the reader that they can't just expect to address these concerns by cleaning their data with typical cleaning methods.

I think it would also be useful to convey in the abstract just how comprehensive the included analyses were (in terms of artifact reduction methods tested, different aperiodic algorithms and frequency ranges, and both MEG and EEG). Doing so would let the reader know just how robust the conclusions are likely to be.

This is a brilliant idea! As suggested we added a sentence highlighting that simply performing an ICA may not be sufficient to separate cardiac contributions to M/EEG recordings and refer to the comprehensiveness of the performed analyses.

INTRODUCTION:

I would suggest re-writing the following sentence for readability: "In the past, aperiodic neural activity, other than periodic neural activity (local peaks that rise above the "power-law" distribution), was often treated as noise and simply removed from the signal"

To something like: "In the past, aperiodic neural activity was often treated as noise and simply removed from the signal e.g. via pre-whitening, so that analyses could focus on periodic neural activity (local peaks that rise above the "power-law" distribution, which are typically thought to reflect neural oscillations).

We are happy to follow that suggestion.

Page 3: please provide the number of articles that were included in the examination of the percentage that remove cardiac activity, and note whether the included articles could be considered a comprehensive or nearly comprehensive list, or just a representative sample.

We stated the exact number of articles in the methods section under Literature Analysis. However, we added it to the Introduction on page 3 as suggested by the reviewer. The selection of articles was done automatically, dependent on a list of pre-specified terms and exclusively focussed on articles that had terms related to aperiodic activity in their title (see Literature Analysis). Therefore, I would personally be hesitant in calling it a comprehensive or nearly comprehensive list of the general M/EEG literature as the analysis of aperiodic activity is still relatively niche compared to the more commonly investigated evoked potentials or oscillations. I think whether or not a reader perceives our analysis as comprehensive should be up to them to decide and does not reflect something I want to impose on them. This is exacerbated by the fact that the analysis of neural aperiodic activity has rapidly gained traction over the last years (see Figure 1D orange) and the literature analysis was performed almost 2 years ago and therefore, in my eyes, only represents a glimpse in the rapidly evolving field related to the analysis of aperiodic activity.

Figure 1E-F: It's not completely clear that the "Cleaning Methods" part of the figure indicates just methods to clean the cardiac artifact (rather than any artifact). It also seems that ~40% of EEG studies do not apply any cleaning methods even from within the studies that do clean the cardiac artifact (if I've read the details correctly). This seems unlikely. Perhaps there should be a bar for "other methods", or "unspecified"? Having said that, I'm quite familiar with the EEG artifact reduction literature, and I would be very surprised if ~40% of studies cleaned the cardiac artifact using a different method to the methods listed in the bar graph, so I'm wondering if I've misunderstood the figure, or whether the data capture is incomplete / inaccurate (even though the conclusion that ICA is the most common method is almost certainly accurate).

The cleaning is indeed only focussed on cardiac activity specifically. This was however also mentioned in the caption of Figure 1: “We were further interested in determining which artifact rejection approaches were most commonly used to remove cardiac activity, such as independent component analysis (ICA(22)), singular value decomposition (SVD(23)), signal space separation (SSS(24)), signal space projections (SSP(25)) and denoising source separation (DSS(26)).” and in the methods section under Literature Analysis. However, we adjusted figure 1EF to make it more obvious that the described cleaning methods were only related to the ECG. Aside from using blind source separation techniques such as ICA a good amount of studies mentioned that they cleaned their data based on visual inspection (which was not further considered). Furthermore, it has to be noted that only studies were marked as having separated cardiac from neural activity, when this was mentioned explicitly.

RESULTS:

Page 6: I would delete the "from a neurophysiological perspective" clause, which makes the sentence more difficult to read and isn't so accurate (frequencies 13-25Hz would probably more commonly be considered mid-range rather than low or high). Additionally, both frequency ranges include 15Hz, but the next sentence states that the ranges were selected to avoid the knee at 15Hz, which seems to be a contradiction. Could the authors explain in more detail how the split addresses the 15Hz knee?

We removed the “from a neurophysiological perspective” clause as suggested. With regards to the “knee” at ~15Hz I would like to defer the reviewer to Supplementary Figure S1. The Knee Frequency varies substantially across subjects so splitting the data at only 1 exact Frequency did not seem appropriate. Additionally, we found only spurious significant age-related variations in Knee Frequency (i.e. only one out of the 4 datasets; not shown).

Furthermore, we wanted to better connect our findings to our MEG results in Figure 4 and also give the readers a holistic overview of how different frequency ranges in the aperiodic ECG would be affected by age. So to fulfill all of these objectives we decided to fit slopes with respective upper/lower bounds around a range of 5Hz above and below the average 15Hz Knee Frequency across datasets.

The later parts of this same paragraph refer to a vast amount of different frequency ranges, but only the "low" and "high" frequency ranges were previously mentioned. Perhaps the explanation could be expanded to note that multiple lower and upper bounds were tested within each of these low and high frequency windows?

This is a good catch we adjusted the sentence as suggested. We now write: “.. slopes were fitted individually to each subject's power spectrum in several lower (0.25 – 20 Hz) and higher (10-145 Hz) frequency ranges.”

The following two sentences seem to contradict each other: "Overall, spectral slopes in lower frequency ranges were more consistently related to heart rate variability indices(> 39.4% percent of all investigated indices)" and: "In the lower frequency range (0.25 - 20Hz), spectral slopes were consistently related to most measures of heart rate variability; i.e. significant effects were detected in all 4 datasets (see Figure 2D)." (39.4% is not "most").

The reviewer is correct in stating that 39.4% is not most. However, the 39.4% is the lowest bound and only refers to 1 dataset. In the other 3 datasets the percentage of effects was above 64% which can be categorized as “most” i.e. above 50%. We agree that this was a bit ambiguous in the sentence so we added the other percentages as well as a reference to Figure 2D to make this point clearer.

Figure 2D: it isn't clear what the percentages in the semi-circles reflect, nor why some semi-circles are more full circles while others are only quarter circles.

The percentages in the semi-circles reflect the amount of effects (marked in red) and null effects (marked in green) per dataset, when viewed as average across the different measures of HRV. Sometimes less effects were found for some frequency ranges resulting in quarters instead of semi circles.

Page 8: I think the authors could make it more clear that one of the conditions they were testing was the ECG component of the EEG data (extracted by ICA then projected back into the scalp space for the temporal response function analysis).

As suggested by the reviewer we adjusted our wording and replaced the arguably a bit ambiguous “... projected back separately” with “... projected back into the sensor space”. We thank the reviewer for this recommendation, as it does indeed make it easier to understand the procedure.

“After pre-processing (see Methods) the data was split in three conditions using an ICA(22). Independent components that were correlated (at r > 0.4; see Methods: MEG/EEG Processing - pre-processing) with the ECG electrode were either not removed from the data (Figure 3ABCD - blue), removed from the data (Figure 2ABCD - orange) or projected back into the sensor space (Figure 3ABCD - green).”

Figure 4A: standardized beta coefficients for the relationship between age and spectral slope could be noted to provide improved clarity (if I'm correct in assuming that is what they reflect).

This was indeed shown in Figure 4A and noted in the color bar as “average beta (standardized)”. We do not specifically highlight this in the text, because the exact coefficients would depend on both on the analyzed frequency range and the selected electrodes.

Figure 4I: The regressions explained at this point seems to contain a very large number of potential predictors, as I'm assuming it includes all sensors for both the ECG component and ECG rejected conditions? (if that is not the case, it could be explained in greater detail). I'm also not sure about the logic of taking a complete signal, decomposing it with ICA to separate out the ECG and non-ECG signals, then including them back into the same regression model. It seems that there could be some circularity or redundancy in doing so. However, I'm not confident that this is an issue, so would appreciate the authors explaining why it this is a valid approach (if that is the case).

After observing significant effects both in the MEG_{ECG component} and MEG_{ECG rejected} conditions in similar frequency bands we wanted to understand whether or not these age-related changes are statistically independent. To test this we added both variables as predictors in a regression model (thereby accounting for the influence of the other in relation to age). The regression models we performed were therefore actually not very complex. They were built using only two predictors, namely the data (in a specific frequency range) averaged over channels on which we noticed significant effects in the ECG rejected and ECG components data respectively (Wilkinson notation: age ~ 1 + ECG rejected + ECG components). This was also described in the results section stating that: “To see if MEG_{ECG rejected} and MEG_{ECG component} explain unique variance in aging at frequency ranges where we noticed shared effects, we averaged the spectral slope across significant channels and calculated a multiple regression model with MEG_{ECG component} and MEG_{ECG rejected} as predictors for age (to statistically control for the effect of MEG_{ECG component}s and MEG_{ECG rejected} on age). This analysis was performed to understand whether the observed shared age-related effects (MEG_{ECG rejected} and MEG_{ECG component}) are in(dependent).”  

We hope this explanation solves the previous misunderstanding.

The explanation of results for relationships between spectral slopes and aging reported in Figure 4 refers to clusters of effects, but the statistical inference methods section doesn't explain how these clusters were determined.

The wording of “cluster” was used to describe a “category” of effects e.g. null effects. We changed the wording from “cluster” to “category” to make this clearer stating now that: “This analysis, which is depicted in Figure 4, shows that over a broad amount of individual fitting ranges and sensors, aging resulted in a steepening of spectral slopes across conditions (see Figure 4E) with “steepening effects” observed in 25% of the processing options in MEG_{ECG not rejected} , 0.5% in MEG_{ECG rejected}, and 60% for MEG_{ECG components}. The second largest category of effects were “null effects” in 13% of the options for MEG_{ECG not rejected} , 30% in MEG_{ECG rejected}, and 7% for MEG_{ECG components}. ”

Page 12: can the authors clarify whether these age related steepenings of the spectral slope in the MEG are when the data include the ECG contribution, or when the data exclude the ECG? (clarifying this seems critical to the message the authors are presenting).

We apologize for not making this clearer. We now write: “This analysis also indicates that a vast majority of observed effects irrespective of condition (ECG components, ECG not rejected, ECG rejected) show a steepening of the spectral slope with age across sensors and frequency ranges.”

Page 13: I think it would be useful to describe how much variance was explained by the MEG-ECG rejected vs MEG-ECG component conditions for a range of these analyses, so the reader also has an understanding of how much aperiodic neural activity might be influenced by age (vs if the effects are really driven mostly by changes in the ECG).

With regards to the explained variance I think that the very important question of how strong age influences changes in aperiodic activity is a topic better suited for a meta analysis. As the effect sizes seems to vary largely depending on the sample e.g. for EEG in the literature results were reported at r=-0.08 (Cesnaite et al. 2023), r=-0.26 (Cellier et al. 2021), r=-0.24/r=-0.28/r=-0.35 (Hill et al. 2022) and r=0.5/r=0.7 (Voytek et al. 2015). I would defer the reader/reviewer to the standardized beta coefficients as a measure of effect size in the current study that is depicted in Figure 4A.

Cellier, D., Riddle, J., Petersen, I., & Hwang, K. (2021). The development of theta and alpha neural oscillations from ages 3 to 24 years. Developmental cognitive neuroscience, 50, 100969.

Cesnaite, E., Steinfath, P., Idaji, M. J., Stephani, T., Kumral, D., Haufe, S., ... & Nikulin, V. V. (2023). Alterations in rhythmic and non‐rhythmic resting‐state EEG activity and their link to cognition in older age. NeuroImage, 268, 119810.

Hill, A. T., Clark, G. M., Bigelow, F. J., Lum, J. A., & Enticott, P. G. (2022). Periodic and aperiodic neural activity displays age-dependent changes across early-to-middle childhood. Developmental Cognitive Neuroscience, 54, 101076.

Voytek, B., Kramer, M. A., Case, J., Lepage, K. Q., Tempesta, Z. R., Knight, R. T., & Gazzaley, A. (2015). Age-related changes in 1/f neural electrophysiological noise. Journal of Neuroscience, 35(38), 13257-13265.

Also, if there are specific M/EEG sensors where the 1/f activity does relate strongly to age, it would be worth noting these, so future research could explore those sensors in more detail.

I think it is difficult to make a clear claim about this for MEG data, as the exact location or type of the sensor may differ across manufacturers. Such a statement could be easier made for source projected data or in case EEG electrodes were available, where the location would be normed eg. according to the 10-20 system.

DISCUSSION:

Page 15: Please change the wording of the following sentence, as the way it is currently worded seems to suggest that the authors of the current manuscript have demonstrated this point (which I think is not the case): "The authors demonstrate that EEG typically integrates activity over larger volumes than MEG, resulting in differently shaped spectra across both recording methods."

Apologies for the oversight! The reviewer is correct we in fact did not show this, but the authors of the cited manuscript. We correct the sentence as suggested stating now that:

“Bénar et al. demonstrate that EEG typically integrates activity over larger volumes than MEG, resulting in differently shaped spectra across both recording methods.”

Page 16: The authors mention the results can be sensitive to the application of SSS to clean the MEG data, but not ICA. I think it would be sensitive to the application of either SSS or ICA?

This is correct and actually also supported by Figure S7, as differences in ICA thresholds affect also the detection of age-related effects. We therefore adjusted the related sentences stating now that:

“ In case of the MEG signal this may include the application of Signal-Space-Separation algorithms (SSS(24,55)), different thresholds for ICA component detection (see Figure S7), high and low pass filtering, choices during spectral density estimation (window length/type etc.), different parametrization algorithms (e.g. IRASA vs FOOOF) and selection of frequency ranges for the aperiodic slope estimation.”

It would be worth clarifying that the linked mastoid re-reference alone has been proposed to cancel out the ECG signal, rather than that a linked-mastoid re-reference improves the performance of the ICA separation (which could be inferred by the explanation as it's currently written).

This is correct and we adjusted the sentence accordingly! Stating now that:

“ Previous work(12,56) has shown that a linked mastoid reference alone was particularly effective in reducing the impact of ECG related activity on aperiodic activity measured using EEG. “

The issue of the number of EEG channels could probably just be noted as a potential limitation, as could the issue of neural activity being mixed into the ECG component (although this does pose a potential confound to the M/EEG without ECG condition, I suspect it wouldn't be critical).

This is indeed a very fair point as a higher amount of electrodes would probably make it easier to better isolate ECG components in the EEG, which may be the reason why the separation did not work so well in our case. However, this is ultimately an empirical question so we highlighted it in the discussion section stating that: “Difficulties in removing ECG related components from EEG signals via ICA might be attributable to various reasons such as the number of available sensors or assumptions related to the non-gaussianity of the underlying sources. Further understanding of this matter is highly important given that ICA is the most widely used procedure to separate neural from peripheral physiological sources. ”

OUTLOOK:

Page 19: Although there has been a recent trend to control for 1/f activity when examining oscillatory power, recent research suggests that this should only be implemented in specific circumstances, otherwise the correction causes more of a confound than the issue does. It might be worth considering this point with regards to the final recommendation in the Outlook section: Brake, N., Duc, F., Rokos, A., Arseneau, F., Shahiri, S., Khadra, A., & Plourde, G. (2024). A neurophysiological basis for aperiodic EEG and the background spectral trend. Nature Communications, 15(1), 1514.

We want to thank the reviewer for recommending this very interesting paper! The authors of said paper present compelling evidence showing that, while peak detection above an aperiodic trend using methods like FOOOF or IRASA is a prerequisite to determine the presence of oscillatory activity, it’s not necessarily straightforward to determine which detrending approach should be applied to determine the actual power of an oscillation. Furthermore, the authors suggest that wrongfully detrending may cause larger errors than not detrending at all. We therefore added a sentence stating that: “However, whether or not periodic activity (after detection) should be detrended using approaches like FOOOF or IRASA still remains disputed, as incorrectly detrending the data may cause larger errors than not detrending at all(75).”

RECOMMENDATIONS:

Page 20: "measure and account for" seems like it's missing a word, can this be re-written so the meaning is more clear?

Done as suggested. The sentence now states: “To better disentangle physiological and neural sources of aperiodic activity, we propose the following steps to (1) measure and (2) account for physiological influences.”

I would re-phrase "doing an ICA" to "reducing cardiac artifacts using ICA" (this wording could be changed in other places also).

I do not like to describe cardiac or ocular activity as artifactual per se. This is also why I used hyphens whenever I mention the word “artifact” in association with the ECG or EOG. However, I do understand that the wording of “doing an ICA” is a bit sloppy. We therefore reworded it accordingly throughout the manuscript to e.g. “separating cardiac from neural sources using an ICA” and “separating physiological from neural sources using an ICA”.

I would additionally note that even if components are identified as unambiguously cardiac, it is still likely that neural activity is mixed in, and so either subtracting or leaving the component will both be an issue (https://doi.org/10.1101/2024.06.06.597688). As such, even perfect identification of whether components are cardiac or not would still mean the issue remains (and this issue is also consistent across a considerable range of component based methods). Furthermore, current methods including wavelet transforms on the ICA component still do not provide good separation of the artifact and neural activity.

This is definitely a fair point and we also highlight this in our recommendations under 3 stating that:

“However, separating physiological from neural sources using an ICA is no guarantee that peripheral physiological activity is fully removed from the cortical signal. Even more sophisticated ICA based methods that e.g. apply wavelet transforms on the ICA components may still not provide a good separation of peripheral physiological and neural activity76,77. This turns the process of deciding whether or not an ICA component is e.g. either reflective of cardiac or neural activity into a challenging problem. For instance, when we only extract cardiac components using relatively high detection thresholds (e.g. r > 0.8), we might end up misclassifying residual cardiac activity as neural. In turn, we can’t always be sure that using lower thresholds won’t result in misinterpreting parts of the neural effects as cardiac. Both ways of analyzing the data can potentially result in misconceptions.”

Castellanos, N. P., & Makarov, V. A. (2006). Recovering EEG brain signals: Artifact suppression with wavelet enhanced independent component analysis. Journal of neuroscience methods, 158(2), 300-312.

Bailey, N. W., Hill, A. T., Godfrey, K., Perera, M. P. N., Rogasch, N. C., Fitzgibbon, B. M., & Fitzgerald, P. B. (2024). EEG is better when cleaning effectively targets artifacts. bioRxiv, 2024-06.

METHODS:

Pre-processing, page 24: I assume the symmetric setting of fastica was used (rather than the deflation setting), but this should be specified.

Indeed the reviewer is correct, we used the standard setting of fastICA implemented in MNE python, which is calling the FastICA implementation in sklearn that is per default using the “parallel” or symmetric algorithm to compute an ICA. We added this information to the text accordingly, stating that:

“For extracting physiological “artifacts” from the data, 50 independent components were calculated using the fastica algorithm(22) (implemented in MNE-Python version 1.2; with the parallel/symmetric setting; note: 50 components were selected for MEG for computational reasons for the analysis of EEG data no threshold was applied).”

Temporal response functions, page 26: can the authors please clarify whether the TRF is computed against the ECG signal for each electrode or sensory independently, or if all electrodes/sensors are included in the analysis concurrently? I'm assuming it was computed for each electrode and sensory separately, since the TRF was computed in both the forward and backwards direction (perhaps the meaning of forwards and backwards could be explained in more detail also - i.e. using the ECG to predict the EEG signal, or using the EEG signal to predict the ECG signal?).

A TRF can also be conceptualized as a multiple regression model over time lags. This means that we used all channels to compute the forward and backward models. In the case of the forward model we predicted the signal of the M/EEG channels in a multivariate regression model using the ECG electrode as predictor. In case of the backward model we predicted the ECG electrode based on the signal of all M/EEG channels. The forward model was used to depict the time window at which the ECG signal was encoded in the M/EEG recording, which appears at 0 time lags indicating volume conduction. The backward model was used to see how much information of the ECG was decodable by taking the information of all channels.

We tried to further clarify this approach in the methods section stating that:

“We calculated the same model in the forward direction (encoding model; i.e. predicting M/EEG data in a multivariate model from the ECG signal) and backward direction (decoding model; i.e. predicting the ECG signal using all M/EEG channels as predictors).”

Page 27: the ECG data was fit using a knee, but it seems the EEG and MEG data was not.

Does this different pose any potential confound to the conclusions drawn? (having said this, Figure S4 suggests perhaps a knee was tested in the M/EEG data, which should perhaps be explained in the text also).

This was indeed tested in a previous review round to ensure that our results are not dependent on the presence/absence of a knee in the data. We therefore added figure S4, but forgot to actually add a description in the text. We are sorry for this oversight and added a paragraph to S1 accordingly:

“Using FOOOF(5), we also investigated the impact of different slope fitting options (fixed vs. knee model fits) on the aperiodic age relationship (see Supplementary Figure S4). The results that we obtained from these analyses using FOOOF offer converging evidence with our main analysis using IRASA.”

Page 32: my understanding of the result reported here is that cleaning with ICA provided better sensitivity to the effects of age on 1/f activity than cleaning with SSS. Is this accurate? I think this could also be reported in the main manuscript, as it will be useful to researchers considering how to clean their M/EEG data prior to analyzing 1/f activity.

The reviewer is correct in stating that we overall detected slightly more “significant” effects, when not additionally cleaning the data using SSS. However, I am a bit wary of recommending omitting the use of SSS maxfilter solely based on this information. It can very well be that the higher quantity of effects (when not employing SSS maxfilter) stems from other physiological sources (e.g. muscle activity) that are correlated with age and removed when applying SSS maxfiltering. I think that just conditioning the decision of whether or not maxfilter is applied based on the amount or size of effects may not be the best idea. Instead I think that the applicability of maxfilter for research questions related to aperiodic activity should be the topic of additional methodological research. We therefore now write in Text S1:

“Considering that we detected less and weaker aperiodic effects when using SSS maxfilter is it advisable to omit maxfilter, when analyzing aperiodic signals? We don’t think that we can make such a judgment based on our current results. This is because it's unclear whether or not the reduction of effects stems from an additional removal of peripheral information (e.g. muscle activity; that may be correlated with aging) or is induced by the SSS maxfiltering procedure itself. As the use of maxfilter in detecting changes of aperiodic activity was not subject of analysis that we are aware of, we suggest that this should be the topic of additional methodological research.”

Page 39, Figure S6 and Figure S8: Perhaps the caption could also briefly explain the difference between maxfilter set to false vs true? I might have missed it, but I didn't gain an understanding of what varying maxfilter would mean.

Figure S6 shows the effect of ageing on the spectral slope averaged across all channels. The maxfilter set to false in AB) means that no maxfiltering using SSS was performed vs. in CD) where the data was additionally processed using the SSS maxfilter algorithm. We now describe this more clearly by writing in the caption:

“Supplementary Figure S6: Age-related changes in aperiodic brain activity are most prominent on explained by cardiac components irrespective of maxfiltering the data using signal space separation (SSS) or not AC) Age was used to predict the spectral slope (fitted at 0.1-145Hz) averaged across sensors at rest in three different conditions (ECG components not rejected [blue], ECG components rejected [orange], ECG components only [green].”

Author response:

The following is the authors’ response to the original reviews

Public reviews:

Reviewer #1 (Public Review):

Summary:

In this paper, Weber et al. investigate the role of 4 dopaminergic neurons of the Drosophila larva in mediating the association between an aversive high-salt stimulus and a neutral odor. The 4 DANs belong to the DL1 cluster and innervate non-overlapping compartments of the mushroom body, distinct from those involved in appetitive associative learning. Using specific driver lines, they show that activation of the DAN-g1 is sufficient to mimic an aversive memory and it is also necessary to form a high-salt memory of full strength, although optogenetic silencing of this neuron only partially affects the performance index. The authors use calcium imaging to show that the DAN-g1 is not the only one that responds to salt. DAN-c1 and d1 also respond to salt, but they seem to play no role in the assays tested. DAN-f1, which does not respond to salt, is able to lead to the formation of memory (if optogenetically activated), but it is not necessary for the salt-odor memory formation in normal conditions. However, silencing of DAN-f1 together with DAN-g1, enhances the memory deficit of DAN-g1.

Strengths:

The paper therefore reveals that also in the Drosophila larva as in the adult, rewards and punishments are processed by exclusive sets of DANs and that a complex interaction between a subset of DANs mediates salt-odor association.

Overall, the manuscript contributes valuable results that are useful for understanding the organization and function of the dopaminergic system. The behavioral role of the specific DANs is accessed using specific driver lines which allow for testing of their function individually and in pairs. Moreover, the authors perform calcium imaging to test whether DANs are activated by salt, a prerequisite for inducing a negative association with it. Proper genetic controls are carried across the manuscript.

Weaknesses:

The authors use two different approaches to silence dopaminergic neurons: optogenetics and induction of apoptosis. The results are not always consistent, and the authors could improve the presentation and interpretation of the data. Specifically, optogenetics seems a better approach than apoptosis, which can affect the overall development of the system, but apoptosis experiments are used to set the grounds of the paper.

The physiological data would suggest the role of a certain subset of DANs in salt-odor association, but a different partially overlapping set seems to be necessary. This should be better discussed and integrated into the author's conclusion. The EM data analysis reveals a non-trivial organization of sensory inputs into DANs and it is hard to extrapolate a link to the functional data presented in the paper.

We would like to thank reviewer 1 for the positive evaluation of our work and for the critical suggestions for improvement. In the new version of the manuscript, we have centralized the optogenetic results and moved some of the ablation experiments to the Supplement. We also discuss in detail the experimental differences in the results. In addition, we have softened our interpretation of the specificity of memory for salt. As a result, we now emphasize more the general role of DANs for aversive learning in the larva. These changes are now also summarized and explained more simply and clearly in the Discussion, along with a revised discussion of the EM data.

Reviewer #2 (Public Review):

Summary:

In this work, the authors show that dopaminergic neurons (DANs) from the DL1 cluster in Drosophila larvae are required for the formation of aversive memories. DL1 DANs complement pPAM cluster neurons which are required for the formation of attractive memories. This shows the compartmentalized network organization of how an insect learning center (the mushroom body) encodes memory by integrating olfactory stimuli with aversive or attractive teaching signals. Interestingly, the authors found that the 4 main dopaminergic DL1 neurons act redundantly, and that single-cell ablation did not result in aversive memory defects. However, ablation or silencing of a specific DL1 subset (DAN-f1,g1) resulted in reduced salt aversion learning, which was specific to salt but no other aversive teaching stimuli were tested. Importantly, activation of these DANs using an optogenetic approach was also sufficient to induce aversive learning in the presence of high salt. Together with the functional imaging of salt and fructose responses of the individual DANs and the implemented connectome analysis of sensory (and other) inputs to DL1/pPAM DANs, this represents a very comprehensive study linking the structural, functional, and behavioral role of DL1 DANs. This provides fundamental insight into the function of a simple yet efficiently organized learning center which displays highly conserved features of integrating teaching signals with other sensory cues via dopaminergic signaling.

Strengths:

This is a very careful, precise, and meticulous study identifying the main larval DANs involved in aversive learning using high salt as a teaching signal. This is highly interesting because it allows us to define the cellular substrates and pathways of aversive learning down to the single-cell level in a system without much redundancy. It therefore sets the basis to conduct even more sophisticated experiments and together with the neat connectome analysis opens the possibility of unraveling different sensory processing pathways within the DL1 cluster and integration with the higher-order circuit elements (Kenyon cells and MBONs). The authors' claims are well substantiated by the data and clearly discussed in the appropriate context. The authors also implement neat pathway analyses using the larval connectome data to its full advantage, thus providing network pathways that contribute towards explaining the obtained results.

Weaknesses:

While there is certainly room for further analysis in the future, the study is very complete as it stands. Suggestions for clarification are minor in nature.

We would like to thank reviewer 2 for the positive evaluation of our work. In fact, follow-up work is already underway to further analyze the role of the individual DL1 DANs. We have addressed the constructive and detailed suggestions for improvement in our point-by-point responses in the “Recommendations for the authors” section.

Reviewer #3 (Public Review):

The study of Weber et al. provides a thorough investigation of the roles of four individual dopamine neurons for aversive associative learning in the Drosophila larva. They focus on the neurons of the DL-1 cluster which already have been shown to signal aversive teaching signals. However, the authors go far beyond the previous publications and test whether each of these dopamine neurons responds to salt or sugar, is necessary for learning about salt, bitter, or sugar, and is sufficient to induce a memory when optogenetically activated. In addition, previously published connectomic data is used to analyze the synaptic input to each of these dopamine neurons. The authors conclude that the aversive teaching signal induced by salt is distributed across the four DL-1 dopamine neurons, with two of them, DAN-f1 and DAN-g1, being particularly important. Overall, the experiments are well designed and performed, support the authors' conclusions, and deepen our understanding of the dopaminergic punishment system.

Strengths:

(1) This study provides, at least to my knowledge, the first in vivo imaging of larval dopamine neurons in response to tastants. Although the selection of tastants is limited, the results close an important gap in our understanding of the function of these neurons.

(2) The authors performed a large number of experiments to probe for the necessity of each individual dopamine neuron, as well as combinations of neurons, for associative learning. This includes two different training regimens (1 or 3 trials), three different tastants (salt, quinine, and fructose) and two different effectors, one ablating the neuron, the other one acutely silencing it. This thorough work is highly commendable, and the results prove that it was worth it. The authors find that only one neuron, DAN-g1, is partially necessary for salt learning when acutely silenced, whereas a combination of two neurons, DAN-f1 and DAN-g1, are necessary for salt learning when either being ablated or silenced.

(3) In addition, the authors probe whether any of the DL-1 neurons is sufficient for inducing an aversive memory. They found this to be the case for three of the neurons, largely confirming previous results obtained by a different learning paradigm, parameters, and effector.

(4) This study also takes into account connectomic data to analyze the sensory input that each of the dopamine neurons receives. This analysis provides a welcome addition to previous studies and helps to gain a more complete understanding. The authors find large differences in inputs that each neuron receives, and little overlap in input that the dopamine neurons of the "aversive" DL-1 cluster and the "appetitive" pPAM cluster seem to receive.

(5) Finally, the authors try to link all the gathered information in order to describe an updated working model of how aversive teaching signals are carried by dopamine neurons to the larva's memory center. This includes important comparisons both between two different aversive stimuli (salt and nociception) and between the larval and adult stages.

Weaknesses:

(1) The authors repeatedly claim that they found/proved salt-specific memories. I think this is problematic to some extent.

(1a) With respect to the necessity of the DL-1 neurons for aversive memories, the authors' notion of salt-specificity relies on a significant reduction in salt memory after ablating DAN-f1 and g1, and the lack of such a reduction in quinine memory. However, Fig. 5K shows a quite suspicious trend of an impaired quinine memory which might have been significant with a higher sample size. I therefore think it is not fully clear yet whether DAN-f1 and DAN-g1 are really specifically necessary for salt learning, and the conclusions should be phrased carefully.

(1b) With respect to the results of the optogenetic activation of DL-1 neurons, the authors conclude that specific salt memories were established because the aversive memories were observed in the presence of salt. However, this does not prove that the established memory is specific to salt - it could be an unspecific aversive memory that potentially could be observed in the presence of any other aversive stimuli. In the case of DAN-f1, the authors show that the neuron does not even get activated by salt, but is inhibited by sugar. Why should activation of such a neuron establish a specific salt memory? At the current state, the authors clearly showed that optogenetic activation of the neurons does induce aversive memories - the "content" of those memories, however, remains unknown.

(2) In many figures (e.g. figures 4, 5, 6, supplementary figures S2, S3, S5), the same behavioural data of the effector control is plotted in several sub-figures. Were these experiments done in parallel? If not, the data should not be presented together with results not gathered in parallel. If yes, this should be clearly stated in the figure legends.

We would also like to thank reviewer 3 for his positive assessment of our work. As already mentioned by reviewer 1, we understand the criticism that the salt specificity for which the individual DANs are coded is not fully always supported by the results of the work. We have therefore rewritten the relevant passages, which are also cited by the reviewer. We have also included the second point of criticism and incorporated it into our manuscript. As the control groups were always measured in parallel with the experimental animals, we can also present the data together in a sub-figure. We clearly state this now in the revised figure legends.

Summary of recommendations to authors:

Overall, the study is commendable for its systematic approach and solid methodology. Several weaknesses were identified, prompting the need for careful revisions of the manuscript:

We thank the reviewers for the careful revision of our manuscript. In the subsequent sections, we aim to address their concerns as thoroughly as possible. A comprehensive one-to-one listing can be found below.

(1) The authors should reconsider their assertion of uncovering a salt-specific memory, as the evidence does not conclusively demonstrate the exclusive necessity of DAN-f1 and DAN-g1 for salt learning. In particular, the optogenetic activation of DAN-f1 leads to plasticity but this might not be salt-specific. The precise nature of the memory content remains elusive, warranting a nuanced rephrasing of the conclusions.

We only partially agree – optogenetic activation of DANs does not really allow to comment on its salt-specificity, true. However, we used high-salt concentrations during test. Over the years, the Gerber lab nicely demonstrated in several papers that larvae recall an aversive odor-salt memory only if salt is present during test (Gerber and Hendel, 2006; Niewalda et al 2008; Schleyer et al. 2011; Schleyer et al. 2015). The used US has to be present during test. Even at the same concentration other aversive stimuli (e.g. bitter quinine) are not able to allow the larvae to recall this particular type of memory. So, if the optogenetic activation of DAN-f1 establishes a memory that can be recalled on salt, we argue that it has to encode aspects of the salt information. On the other hand, only for DAN-g1 we see the necessity for salt learning. And – although (based on the current literature) very unlikely, we cannot fully exclude that the activation of DAN-f1 establishes a yet unknown type of memory that can be also recalled on a salt plate. Therefore, we partially agree and accordingly have rephrased the entire manuscript to avoid an over-interpretation of our data. Throughout the manuscript we avoid now to use the term salt-specific memory but rather describe the type of memory as aversive memory.

(2) A thorough examination or discussion about the potential influence of blue light aversion on behavioral observations is necessary to ensure a balanced interpretation of the findings.

To address this point every single behavioral experiment that uses optogenetic blue light activation runs with appropriate and mandatory controls. For blue light activation experiments, two genetic controls are used that either get the same blue light treatment (effector control, w1118>UAS-ChR2XXL) or no blue light treatment (dark control, XY-split-Gal4>UAS-ChR2XXL). For blue light inactivation experiments one group is added that has exactly the same genotype but did not receive food containing retinal. These experiments show that blue light exposure itself does not induce an aversive nor positive memory and blue light exposure does not impair the establishment of odor-high salt memory. In addition, we used the latest established transgenes available. ChR2^XXL is very sensitive to blue light. Only 220 lux (60 µW/cm^²) were necessary to obtain stable results. In our hands – short term exposure for up to 5 minutes with such low intensities does not induce a blue light aversion. Following the advice of the reviewer, we also address this concern by adding several sentences into the related results and methods sections.

(3) The authors should address the limitations associated with the use of rpr/hid for neuronal ablations, such as the effects of potential developmental compensation.

We agree with this concern. It is well possible that the ablation experiments induce compensatory effects during larval development. Such an effect may be the reason for differences in phenotypes when comparing hid,rpr ablation with optogenetic inhibition. This is now part of the discussion. In addition, we evaluated if the ablation worked in our experiments. So far controls were missing that show that the expression of hid,rpr really leads to the ablation of DANs. We now added these experiments and clearly show anatomically that the DANs are ablated (related to figure 4-figure supplement 6).

(4) While the connectome analysis offers valuable insights into the observed functions of specific DANs in relation to their extrinsic (sensory) and intrinsic (state) inputs, integrating this data more cohesively within the manuscript through careful rewriting would enhance the coherence of the study.

We understand this concern. Therefore, the new version of our manuscript is now intensifying the inclusion of the EM data in our interpretation of the results. Throughout the entire manuscript we have now rewritten the related parts. We have also completely revised the corresponding section in the results chapter.

(5) More generally, the authors are encouraged to discuss internal discrepancies in the results of their functional manipulation experiments.

Thank you for this suggestion. We do of course understand that we have not given the different results enough space in the discussion. We have now changed this and have been happy to comprehensively address the concern. 

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Here are some suggestions for clarification and improvement of the manuscript:

(1) The authors should discuss why the silencing experiment with TH-GAL4 (Fig. 1) does not abolish memory formation (I assume that the PI should go to zero). Does it mean that other non-TH neurons are involved in salt-odor memory formation? Are there other lines that completely abolish this type of learning?

Thank you very much for highlighting this crucial point. Indeed, the functional intervention does not completely eliminate the memory. There could be several reasons, or a combination thereof, for this outcome. For instance, it's plausible that the UAS-GtACR2 effector doesn't entirely suppress the activity of dopaminergic neurons. Additionally, the memory may comprise different types, not all of which are linked to dopamine function. It's also noteworthy that TH-Gal4 doesn't encompass all dopaminergic neurons – even a neuron from the DL1 cluster is absent (as previously reported in Selcho et al., 2009). Considering we're utilizing high salt concentrations in this experiment, it's conceivable that non gustatory-driven memories are formed based solely on the systemic effects of salt (e.g., increased osmotic pressure). These possibilities are now acknowledged in the text.

(2) The Rpr experiments in Fig. 4 do not lead to any phenotype and there is a general assumption that the system compensates during development. However, there is no demonstration that Rpr worked or that development compensated for that. What do we learn from these data? Would it make sense to move it to supplement to make the story more compact? In addition: the conclusion at L 236 "DL1.... Are not individually necessary" is later disproved by optogenetic silencing. Similarly, optogenetic silencing of f1+g1 is affecting 1X and 3X learning, but not when using Rpr. Moreover, Rpr wdid not give any phenotype in other data in the supplementary material. I'm not sure how valid these results are.

We acknowledge this concern and have actively deliberated various options for restructuring the presented ablation data. Ultimately, we reached a consensus that relocating Figure 4 to the supplement is warranted. Furthermore, corresponding adjustments have been made in the text. This decision amplifies the significance of the optogenetic results. In addition, we also addressed the other part of the concern. We examined the efficacy of hid and rpr in our experiments. Indeed, we successfully ablated specific DANs, as illustrated in the new anatomical data presented in Figure 4- figure supplement 6, which strengthens the interpretation of the hid,rpr experiments.

(3) In most figures that show data for 1X and 3X training, there is no difference between these two conditions (I would suggest moving one set as a supplement). When a difference appears (Fig.5A-D) the implications are not discussed properly. Is it known that some circuits are necessary for the 1X but not for the 3X protocol? Is that a reasonable finding? I would expect the opposite, but I might lack of knowledge here. However, the optogenetic silencing of the same neurons in Figure 7 shows the same phenotype for 1X and 3X. Again, the validity of the Rpr experiments seems debatable.

Different training protocols lead to different memory phases (STM and STM+ARM). We have shown that in the past in Widmann et al. 2016. Therefore, we are convinced that it makes sense to keep both data sets in the main manuscript. However, we agree that this was not properly introduced and discussed and therefore made the respective changes in the manuscript.

(4) In Figure 3, it is unclear what the responses were tested against. Since they are so small and noisy there would be a need for a control. Moreover, in some cases, it looks like the DF/F is normalized to the wrong value: e.g. in DAN-c1 100mM, the activity in 0-10s is always above zero, and in pPAM with fructose is always below zero. This might not have any consequence on the results but should be adjusted.

Thank you very much for your criticism, which we greatly appreciate. We have carefully re-examined the data and found that there was a mistake for the normalization of the values. We made the necessary adjustments to the evaluation, as per your suggestions. The updated figures, figure legends, and results have been incorporated into the new version of the manuscript. As noted by the reviewer, these corrections have not altered the interpretation of the data or the primary responses of the various DANs.

(5) In the abstract: "Optogenetic activation of DAN-f1 and DAN-g1 alone suffices to substitute for salt punishment... Each DAN encodes a different aspect of salt punishment". These sentences might be misleading and an overstatement: only DAN-g1 shows a clear role, while the function of the other DANs in the context of salt-odor learning remains obscure.

We have refined the respective part of the abstract accordingly. Consequently, we have reworded the related section, aiming to avoid any exaggeration.

(6) The physiology is done in L1 larvae but behavior is tested in L3 larvae. There could be a change in this time that could explain the salt responses in c1 and d1 but no role in salt-odor learning?

While we cannot dismiss the possibility of a developmental change from L1 to L3, a comparison of the anatomical data of the DL1 DANs from electron microscopy (EM) and light microscopy (LM) data indicates that their overall morphology remains consistent. However, it's important to note that this observation does not analyse the physiological aspects of these cells. Consequently, we have incorporated this concern into the discussion of the revised version of the manuscript.

(7) The introduction needs some editing starting at L 129, as it ends with a discussion of a previously published EM data analysis. I would rather suggest stating which questions are addressed in this paper and which methods will be used and perhaps a hint on the results obtained.

We understand the concern. We have added a concise paragraph to the conclusion of the introduction, highlighting the biological question, technical details, and a short hint on the acquired findings.

(8) It is clear to me that the presentation of salt during the test is necessary for recall, however in L 166 I don't understand the explanation: how is the memory used in a beneficial way in the test? The salt is present everywhere and the odor cue is actually useless to escape it.

Extensive research, exemplified by studies such as Schleyer et al. (2015) published in Elife, clearly demonstrates that the recall of odor-high salt memory occurs exclusively when tested on a high salt plate. Even when tested on a bitter quinine plate, the aversive memory is not recalled. This phenomenon is attributed to the triggering of motivation to recall the memory by the omnipresent abundance of the unconditioned stimulus (US) during the test, which in our case is high salt. Furthermore, the concentration of the stimulus plays a crucial role (Schleyer et al. 2011). The odor cue indicates where the situation could potentially be improved; however, if high salt is absent, this motivational drive diminishes as there is no memory present to enhance the already favorable situation. Additionally, the motivation to evade the omnipresent and unpleasant high salt stimulus persists throughout the entire 5-minute test period.

(9) L288: the fact that f1 shows a phenotype in this experiment does not mean that it encodes a salt signal, indeed it does not respond to salt. It perhaps induces a plasticity that can be recalled by salt, but not necessarily linked to salt. The synergy between f1 and g1 in the salt assay was postulated based on exp with Rpr, but the validity of these experiments is dubious. I'm not sure there is sufficient evidence from Figures 6 and 7 to support a synergistic action between f1 and g1.

It is true that DAN-f1 alone is not necessary for mediating a high salt teaching signal based on ablation, optogenetic inhibition and even physiology. However, optogenetic activation alone shows a memory tested on a salt plate. Given the logic explained above that is accepted by several publications, we would like to keep the statement. Especially as the joined activation with DAN-g1 gives rise to significant higher or lower values after joined optogenetic activation or inactivation (Figure 5E and F, Figure 6E and F in the new version). Nevertheless, we have modified the sentence. In the text we describe these effects now as “these results may suggest that DAN-f1 and DAN-g1 encode aspects of the natural aversive high salt teaching signal under the conditions that we tested”. We think that this is an appropriate and three-fold restricted statement. Therefore, we would like to keep it in this restricted version. However, we are happy to reconsider this if the reviewer thinks it is critical. 

(10) I find the EM analysis hard to read. First of all, because of the two different graphical representations used in Fig. 8, wouldn't one be sufficient to make the point? Secondly, I could not grasp a take-home-message: what do we learn from the EM data? Do they explain any of the results? It seems to me that they don't provide an explanation of why some DL1 neurons respond to salt and others don't.

We understand that the EM analysis is hard to read and have now carefully rewritten this part of the manuscript. See also general concern 4 above. The main take home message is not to explain why some DL1 neurons respond to salt and other do not. This cannot be resolved due to the missing information on the salt perceiving receptor cells. Unfortunately, we miss the peripheral nervous system in the EM - the first layer of salt information processing. However, our analysis shows clearly that the 4 DANs have their own identity based on their connectivity. None of them is the same – but to a certain extent similarities exist. This nicely reflects the physiological and behavioral results. We have now clarified that in the result to ease the understanding for the readership. In addition, we also clearly state that we don’t address the point why some DL1 neurons respond to salt and why others don’t respond.

(11) Do the manipulations (activation and silencing) affect odor preference in the presence of salt? Did the authors test that the two odors do not drive different behaviors on the salty plate? Or did they only test the odor preference on plain agarose? Can we exclude a role for the DAN in driving multisensory-driven innate behavior?

Innate odor preferences are not changed by the presence of salt or even other tastants (this work but see also Schleyer et al 2015, Figure 3, Elife). Even the naïve choice between two odors is the same if tested in the presence of different tastants (Schleyer et al 2015, Figure 3, Elife). This shows – at least for the tested stimuli and conditions – that are similar to the ones that we use – that there is no multisensory-driven innate odor-taste behavior. Therefore – at least to our knowledge - experiments as the ones suggested by the reviewer were never done in larval odor-taste learning studies. Therefore, we suggest that DAN activation has no effect on innate larval behavior. However, we are happy to reconsider this if the reviewer thinks it is critical. 

(12) L 280: the authors generalize the conclusion to all DL1-DANs, but it does not apply to c1 and d1.

Thanks for this comment. We deleted that sentence as suggested and thus do not anymore generalize the conclusion to all DL-DANs.

(13) L345: I do not see the described differences in Fig. 8F, presynaptic sites of both types seem to appear in rather broad regions: could the author try to clarify this?

We understand that the anatomical description of the data is often hard to read. Especially to readers that are not used to these kind of figures. We have therefore modified the text to ease the understanding and clarify the difference in the labeled brain regions for the broad readership.

(14) L373: the conclusion on c1 is unsupported by data: this neuron responds to both salt and fructose (Figure 3 ) while the conclusion is purely based on EM data analysis.

The sentence is not a conclusion but a speculation and we also list the cell's response to positive and negative gustatory stimuli. Therefore, we do not understand exactly what the reviewer means here. However, we have tried to address the criticism and have revised the sentences.

(15) L385: the data on d1 seem to be inconsistent with Eschbach 2020, but the authors do not discuss if this is due to the differential vs absolute training, or perhaps the presence of the US during the test (which does not seem to be there in Eschbach, 2020) - is the training protocol really responsible for this inconsistency? For f1 the data seem to be consistent across these studies. The authors should clarify how the exp in Fig 6 differs from Eschbach, 2020 and how one could interpret the differences.

True. This concern is correct. We now discuss the difference in more detail. Eschbach et al. used Cs-Crimson as a genetic tool, a one odor paradigm with 3 training cycles, and no gustatory cues in their approach. These differences are now discussed in the new version of the manuscript.

(16) L460-475 A long part of this paragraph discusses the similarities between c1 and d1 and corresponding PPL1 neurons in the adult fly. However, c1 and d1 do not really show any phenotype in this paper, I'm not sure what we learn from this discussion and how much this paper can contribute to it. I would have wished for a discussion of how one could possibly reconcile the observed inconsistencies.

Based on the comments of the different reviewers several paragraphs in the discussion were modified. We agree that the part on the larval-adult comparison is quite long. Thus we have shortened it as suggested by the reviewer.

Minor corrections:

L28 "resultant association" maybe resulting instead.

L55 "animals derive benefit": remove derive.

L78 "composing 12,000 neurons": composed of.

L79 what is stable in a "stable behavioral assay"?

L104: 2 times cluste.

L122: "DL1 DANs are involved" in what?

Fig. 1 please check subpanels labels, D repeats.

L 362: "But how do individual neurons contribute to the teaching signal of the complete cluster?" I don't understand the question.

L364 I did not hear before about the "labeled line hypothesis" in this context - could the author clarify?

L368: edit "combinatorically".

L390: "current suppression" maybe acute suppression.

L 400 I'm not sure what is meant by "judicious functional configuration" and "redundancy". The functions of these cells are not redundant, and no straightforward prediction of their function can be done from their physiological response to salt.

Thanks a lot for your in detail review of our manuscript. We welcome your well-taken concerns and have made the requested changes for all points that you have raised.

Reviewer #2 (Recommendations For The Authors):

(1) In Figure 1 the reconstruction of pPAM and DL1 DANs shows the compartmentalized innervation of the larval MB. However, the images are a bit low in color contrast to appreciate the innervation well. In particular in panel B, it is hard to identify the innervated MB body structure. A schematic model of the larval MB and DAN innervation domains like in Fig. 2A would help to clarify the innervation pattern to the non-specialist.

We understand this concern and have changed figure 1 as suggested by the reviewer. A schematic model of the MB and DANs is now presented already in figure 1 as well as the according supplemental figure.

(2) Blue light itself can be aversive for larvae and thus interfere with the aversive learning paradigm. Does the given Illuminance (220 lux) used in these experiments affect the behavior and learning outcome?

Yes, in former times high intensities of blue light were necessary to trigger the first generation optogenetic tools. The high intensity blue light itself was able to establish an aversive memory (e.g. Rohwedder et al. 2016). Usage of the second generation optogenetic tools allowed us to strongly reduce the applied light intensity. Now we use 220 lux (equal to 60 µW/cm²). Please note that all Gal4 and UAS controls in the manuscript are nonsignificant different from zero. The mild blue light stimulation therefore does not serve as a teaching signal and has neither an aversive nor an appetitive effect. Furthermore, we use this mild light intensity for several other behavioral paradigms (locomotion, feeding, naïve preferences) and have never seen an effect on the behavior.

(3) Fig.2: Except for MB054B-Gal4 only the MB expression pattern is shown for other lines. Is there any additional expression in other cells of the brain? In the legend in line 761, the reporter does not show endogenous expression, rather it is a fluorescent reporter signal labeling the mushroom body.

The lines were initially identified by a screen on larval MB neurons done together with Jim Truman, Marta Zlatic and Bertram Gerber. Here full brain scans were always analyzed. These images can be seen in Eschbach et al. 2020, extended figure 1. Neither in their evaluation nor in our anatomical evaluation (using a different protocol) additional expression in brain cells was detectable. We also modified the figure legend as suggested.

(4) Fig.3: Precise n numbers per experiment should be stated in the figure legend.

True, we now present n numbers per experiment whenever necessary.

(5) Fig.4: Have the authors confirmed complete ablation of the targeted neuron using rpr/hid? Ablations can be highly incomplete depending on the onset and strength of Gal4 expression, leaving some functionality intact. While the ablation experiments are largely in line with the acute silencing of single DANs during high salt learning performed later on (Fig.7), there is potentially an interesting aspect of developmental compensation hidden in this data. Not a major point, but potentially interesting to check.

We agree with this criticism. We have not tested if the expression of hid,rpr in DL1 DANs does really ablate them. Therefore we did an additional experiment to show that. The new data is now present as a supplemental figure (Figure 4- figure supplement 6). The result shows that expression of hid,rpr ablates also DL1 DANs similar to earlier experiments where we used the same effectors to ablate serotoniergic neurons (Huser et al., 2012, figure 5).

(6) The performance index in Fig. 4 and 5 sometimes seems lower and the variability is higher than in some of the other experiments shown. Is this due to the high intrinsic variability of these particular experiments, or the background effects of the rpr/hid or splitGal4 lines?

The general variability of these experiments is within the expected and known borders. In these kind of experiments there is always some variation due to several external factors (e.g. experimental time over the year). Therefore it is always important to measure controls and experimental animals at the same time. Of course that’s what we did and we only compare directly results of individual datasets. But not between different datasets. This is further hampered given that the experiments of Figure 4 (now Figure 4- figure supplement 1) and Figure 5 (now Figure 4) differ in several parameters from other learning experiments presented later in the text. Optogenetic activation uses blue light stimulation instead of “real world” high salt. Most often direct activation of specific DANs in the brain is more stable than the external high salt stimulation. Also optogenetic inactivation uses blue light stimulation and also retinal supplemented food. Both factors can affect the measurement. We thus want to argue that it is for each experiment most often the particular parameters that affect the variability of the results rather than background effects of the rpr/hid and split-Gal4 lines.

(7) Fig.7: This is a neat experiment showing the effects of acute silencing of individual DL1 DANs. As silencing DAN-f1/g1 does not result in complete suppression of aversive learning, it would be highly interesting to test (or speculate about) additive or modulatory effects by the other DANs. Dan-c-1/d-1 also responds to high salt but does not show function on its own in these assays. I am aware that this is currently genetically not feasible. It would however be a nice future experiment.

True, we were intensively screening for DL1 cluster specific driver lines that cover all 4 DL1 neurons or other combinations than the ones we tested. Unfortunately, we did not succeed in identifying them. Nevertheless, we will further screen new genetic resources (e.g. Meissner et al., 2024, bioRxiv) to expand our approach in future experiments. Please also see our comment on concern 1 of reviewer 1 for further technical limitations and biological questions that can also potentially explain the absence of complete suppression of high salt learning and memory. Some of these limitations are now also mentioned and discussed in the new version of the manuscript.

(8) The discussion is excellent. I would just amend that it is likely that larval DAN-c1, which has high interconnectivity within the larval CNS, is likely integrating state-dependent network changes, similar to the role of some DANs in innate and state-dependent preference behavior. This might contribute to modulating learned behavior depending on the present (acute) and previous environmental conditions.

Thanks a lot for bringing this up. We rewrote this part and added a discussion on recent work on DAN-c1 function in larvae as well as results on DAN function in innate and state-dependent preference behavior.

(9) Citation in line 1115 missing access information: "Schnitzer M, Huang C, Luo J, Je Woo S, Roitman L, et al. 2023. Dopamine signals integrate innate and learned valences to regulate memory dynamics. Research Square".

Unfortunately this escaped our notice. The paper is now published in Nature: Huang, C., Luo, J., Woo, S.J. et al. Dopamine-mediated interactions between short- and long-term memory dynamics. Nature 634, 1141–1149 (2024). https://doi.org/10.1038/s41586-024-07819-w. We have now changed the citation. The new citation includes the missing access information.

Reviewer #3 (Recommendations For The Authors):

Regarding my issue about salt specificity in the public review, I want to make clear that I do not suggest additional experiments, but to be very careful in phrasing the conclusions, in particular whenever referring to the experiments with optogenetic activation. This includes presenting these experiments as "(salt) substitution" experiments - inferring that the optogenetic activation would substitute for a natural salt punishment. As important and interesting as the experiments are, they simply do not allow such an interpretation at this point.

Results, line 140ff: When presenting the results regarding TH-Gal4 crossed to ChR2-XXL, please cite Schroll et al. 2006 who demonstrated the same results for the first time.

Thanks for mentioning this. We now cite Schroll et al. 2006 here in the text of the manuscript.

Figure 3: The subfigure labels (ABC) are missing.

Unfortunately this escaped our notice. Thanks a lot – we have now corrected this mistake.

Figure 5: For I and L, it reads "salt replaced with fru", but the sketch on the left shows salt in the test. I assume that fructose was not actually present in the test, and therefore the figure can be misleading. I suggest separate sketches. Also, I and L are not mentioned in the figure legend.

True, this is rather confusing. Based on the well taken concern we have changed the figure by adding a new and correct scheme for sugar reward learning that does not symbolize fructose during test.

Figure S1: The experimental sketches for E,F and G,H seem to be mixed up.

We thank the reviewer for bringing this up. In the new version we corrected this mistake.

Figure S5: There are three sub-figures labelled with B. Please correct.

Again, thanks a lot. We made the suggested correction in Figure S5.

Discussion, line 353ff: this and the following sentences can be read as if the authors have discovered the DL-1 neurons as aversive teaching mediators in this study. However, Eschbach et al. 2020 already demonstrated very similar results regarding the optogenetic activation of single DL-1 DANs. I suggest to rephrase and cite Eschbach et al. 2020 at this point.

That is correct. Our focus was on the gustatory pathway. The original discovery was made by Eschbach et al. We have now corrected this in the discussion and clarified our contribution. It was never our intention to hide this work, as the laboratory was also involved. Nevertheless, this is an annoying omission on our side.

Line 385-387: this sentence is only correct with respect to Eschbach et al. 2020. Weiglein et al. 2021 used ChR2-XXL as an effector, but another training regimen.

We understand this criticism. Therefore, we changed the sentence as suggested by the reviewer. See also our response on concern 15 of reviewer 1.

Line 389ff: I do not understand this sentence. What is meant by persistent and current suppression of activity? If this refers to the behavioural experiments, it is misleading as in the hid, reaper experiments neurons are ablated and not suppressed in activity.

We made the requested changes in the text. It is true that the ablation of a neuron throughout larval life is different from constantly blocking the output of a persisting neuron.

Methods, line 615 ff: the performance index is said to be calculated as the difference between the two preferences, but the equation shows the average of the preferences.

Thanks a lot. We are sorry for the confusion. We have carefully rewritten this part of the methods section to avoid any misunderstanding.

When discussing the organization of the DL1 cluster, on several occasions I have the impression the authors use the terms "redundant" and "combinatorial" synonymously. I suggest to be more careful here. Redundancy implies that each DAN in principle can "do the job", whereas combinatorial coding implies that only a combination of DANs together can "do the job". If "the job" is establishing an aversive salt memory, the authors' results point to redundancy: no experimental manipulation totally abolished salt learning, implying that the non-manipulated neurons in each experiment sufficed to establish a memory; and several DANs, when individually activated, can establish an aversive memory, implying that each of them indeed can "do the job".

Based on this concern we have rewritten the discussion as suggested to be more precise when talking about redundancy or combinatorial coding of the aversive teaching signal. Basically, we have removed all the combinatorial terms and replaced them by the term “redundancy”.

The authors mix parametric and non-parametric statistical tests across the experiments dependent on whether the distribution of the data is normal or not. It would help readers if the authors would clearly state for which data which tests were used.

We understand the criticism and now have added an additional supplemental file that includes all the information on the statistical tests applied and the distribution of the data.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1 (Public Review):

Summary

In this study, the authors build upon previous research that utilized non-invasive EEG and MEG by analyzing intracranial human ECoG data with high spatial resolution. They employed a receptive field mapping task to infer the retinotopic organization of the human visual system. The results present compelling evidence that the spatial distribution of human alpha oscillations is highly specific and functionally relevant, as it provides information about the position of a stimulus within the visual field.

Using state-of-the-art modeling approaches, the authors not only strengthen the existing evidence for the spatial specificity of the human dominant rhythm but also provide new quantification of its functional utility, specifically in terms of the size of the receptive field relative to the one estimated based on broad band activity.

We thank the reviewer for their positive summary.

Weakness 1.1

The present manuscript currently omits the complementary view that the retinotopic map of the visual system might be related to eye movement control. Previous research in non-human primates using microelectrode stimulation has clearly shown that neuronal circuits in the visual system possess motor properties (e.g. Schiller and Styker 1972, Schiller and Tehovnik 2001). More recent work utilizing Utah arrays, receptive field mapping, and electrical stimulation further supports this perspective, demonstrating that the retinotopic map functions as a motor map. In other words, neurons within a specific area responding to a particular stimulus location also trigger eye movements towards that location when electrically stimulated (e.g. Chen et al. 2020).

Similarly, recent studies in humans have established a link between the retinotopic variation of human alpha oscillations and eye movements (e.g., Quax et al. 2019, Popov et al. 2021, Celli et al. 2022, Liu et al. 2023, Popov et al. 2023). Therefore, it would be valuable to discuss and acknowledge this complementary perspective on the functional relevance of the presented evidence in the discussion section.

The reviewer notes that we do not discuss the oculomotor system and alpha oscillations. We agree that the literature relating eye movements and alpha oscillations are relevant.

At the Reviewer’s suggestion, we added a paragraph on this topic to the first section of the Discussion (section 3.1, “Other studies have proposed … “).

Reviewer #2 (Public Review):

Summary:

In this work, Yuasa et al. aimed to study the spatial resolution of modulations in alpha frequency oscillations (~10Hz) within the human occipital lobe. Specifically, the authors examined the receptive field (RF) tuning properties of alpha oscillations, using retinotopic mapping and invasive electroencephalogram (iEEG) recordings. The authors employ established approaches for population RF mapping, together with a careful approach to isolating and dissociating overlapping, but distinct, activities in the frequency domain. Whereby, the authors dissociate genuine changes in alpha oscillation amplitude from other superimposed changes occurring over a broadband range of the power spectrum. Together, the authors used this approach to test how spatially tuned estimated RFs were when based on alpha range activity, vs. broadband activities (focused on 70-180Hz). Consistent with a large body of work, the authors report clear evidence of spatially precise RFs based on changes in alpha range activity. However, the size of these RFs were far larger than those reliably estimated using broadband range activity at the same recording site. Overall, the work reflects a rigorous approach to a previously examined question, for which improved characterization leads to improved consistency in findings and some advance of prior work.

We thank the reviewer for the summary.

Strengths:

Overall, the authors take a careful and well-motivated approach to data analyses. The authors successfully test a clear question with a rigorous approach and provide strong supportive findings. Firstly, well-established methods are used for modeling population RFs. Secondly, the authors employ contemporary methods for dissociating unique changes in alpha power from superimposed and concomitant broadband frequency range changes. This is an important confound in estimating changes in alpha power not employed in prior studies. The authors show this approach produces more consistent and robust findings than standard band-filtering approaches. As noted below, this approach may also account for more subtle differences when compared to prior work studying similar effects.

We thank the reviewer for the positive comments.

Weaknesses:

Weakness 2.1 Theoretical framing:

The authors frame their study as testing between two alternative views on the organization, and putative functions, of occipital alpha oscillations: i) alpha oscillation amplitude reflects broad shifts in arousal state, with large spatial coherence and uniformity across cortex; ii) alpha oscillation amplitude reflects more specific perceptual processes and can be modulated at local spatial scales. However, in the introduction this framing seems mostly focused on comparing some of the first observations of alpha with more contemporary observations. Therefore, I read their introduction to more reflect the progress in studying alpha oscillations from Berger's initial observations to the present. I am not aware of a modern alternative in the literature that posits alpha to lack spatially specific modulations. I also note this framing isn't particularly returned to in the discussion.

This was helpful feedback. We have rewritten nearly the entire Introduction to frame the study differently. The emphasis is now on the fact that several intracranial studies of spatial tuning of alpha (in both human and macaque) tend to show increases in alpha due to visual stimulation, in contrast to a century of MEG/EEG studies, from Berger to the present, showing decreases. We believe that the discrepancy is due to an interaction between measurement type and brain signals. Specifically, intracranial measurements sum decreases in alpha oscillations and increases in broadband power on the same trials, and both signals can be large. In contrast, extracranial measures are less sensitive to the broadband signals and mostly just measure the alpha oscillation. Our study reconciles this discrepancy by removing the baseline broadband power increases, thereby isolating the alpha oscillation, and showing that with iEEG spatial analyses, the alpha oscillation decreases with visual stimulation, consistent with EEG and MEG results.

Weakness 2.2 A second important variable here is the spatial scale of measurement.

It follows that EEG based studies will capture changes in alpha activity up to the limits of spatial resolution of the method (i.e. limited in ability to map RFs). This methodological distinction isn't as clearly mentioned in the introduction, but is part of the author's motivation. Finally, as noted below, there are several studies in the literature specifically addressing the authors question, but they are not discussed in the introduction.

The new Introduction now explicitly contrasts EEG/MEG with intracranial studies and refers to the studies below.

Weakness 2.3 Prior studies:

There are important findings in the literature preceding the author's work that are not sufficiently highlighted or cited. In general terms, the spatio-temporal properties of the EEG/iEEG spectrum are well known (i.e. that changes in high frequency activity are more focal than changes in lower frequencies). Therefore, the observations of spatially larger RFs for alpha activities is highly predicted. Specifically, prior work has examined the impact of using different frequency ranges to estimate RF properties, for example ECoG studies in the macaque by Takura et al. NeuroImage (2016) [PubMed: 26363347], as well as prior ECoG work by the author's team of collaborators (Harvey et al., NeuroImage (2013) [PubMed: 23085107]), as well as more recent findings from other groups (Luo et al., (2022) BioRxiv: https://doi.org/10.1101/2022.08.28.505627). Also, a related literature exists for invasively examining RF mapping in the time-voltage domain, which provides some insight into the author's findings (as this signal will be dominated by low-frequency effects). The authors should provide a more modern framing of our current understanding of the spatial organization of the EEG/iEEG spectrum, including prior studies examining these properties within the context of visual cortex and RF mapping. Finally, I do note that the author's approach to these questions do reflect an important test of prior findings, via an improved approach to RF characterization and iEEG frequency isolation, which suggests some important differences with prior work.

Thank you for these references and suggestions. Some of the references were already included, and the others have been added.

There is one issue where we disagree with the Reviewer, namely that “the observations of spatially larger RFs for alpha activities is highly predicted”. We agree that alpha oscillations and other low frequency rhythms tend to be less focal than high frequency responses, but there are also low frequency non-rhythmic signals, and these can be spatially focal. We show this by demonstrating that pRFs solved using low frequency responses outside the alpha band (both below and above the alpha frequency) are small, similar to high frequency broadband pRFs, but differing from the large pRFs associated with alpha oscillations. Hence we believe the degree to which signals are focal is more related to the degree of rhythmicity than to the temporal frequency per se. While some of these results were already in the supplement, we now address the issue more directly in the main text in a new section called, “2.5 The difference in pRF size is not due to a difference in temporal frequency.”

We incorporated additional references into the Introduction, added a new section on low frequency broadband responses to the Results (section 2.5), and expanded the Discussion (section 3.2) to address these new references.

Weakness 2.4 Statistical testing:

The authors employ many important controls in their processing of data. However, for many results there is only a qualitative description or summary metric. It appears very little statistical testing was performed to establish reported differences. Related to this point, the iEEG data is highly nested, with multiple electrodes (observations) coming from each subject, how was this nesting addressed to avoid bias?

We reviewed the primary claims made in the manuscript and for each claim, we specify the supporting analyses and, where appropriate, how we address the issue of nesting. Although some of these analyses were already in the manuscript, many of them are new, including all of the analyses concerning nesting. We believe that putting this information in one place will be useful to the reader, and we now include this text as a new section in supplement, Graphical and statistical support for primary claims.

Reviewer #2 (Recommendations For The Authors):

Recommendation 2.1:

Data presentation: In several places, the authors discuss important features of cortical responses as measured with iEEG that need to be carefully considered. This is totally appropriate and a strength of the author's work, however, I feel the reader would benefit from more depiction of the time-domain responses, to help better understand the authors frequency domain approach. For example, Figure 1 would benefit from showing some form of voltage trace (ERP) and spectrogram, not just the power spectra. In addition, part (a) of Figure 1 could convey some basic information about the timing of the experimental paradigm.

We changed panel A of Figure 1 to include the timing of the experimental paradigm, and we added panels C and D to show the electrode time series before and after regression out of the ERP.

Recommendation 2.2

Update introduction to include references to prior EEG/iEEG work on spatial distribution across frequency spectrum, and importantly, prior work mapping RFs with different frequencies.

We have addressed this issue and re-written our introduction. Please refer to our response in Public Review for further details.

Recommendation 2.3

Figure 3 has several panels and should be labeled to make it easier to follow.The dashed line in lower power spectra isn't defined in a legend and is missing from the upper panel - please clarify.

We updated Figure 3 and reordered the panels to clarify how we computed the summary metrics in broadband and alpha for each stimulus location (i.e., the “ratio” values plotted in panel B). We also simplified the plot of the alpha power spectrum. It now shows a dashed line representing a baseline-corrected response to the mapping stimulus, which is defined in the legend and explained in the caption.

Recommendation 2.4

Power spectra are always shown without error shading, but they are mean estimates.

We added error shading to Figures 1, 2 and 3.

Recommendation 2.5

The authors deal with voltage transients in response to visual stimulation, by subtracting out the trail averaged mean (commonly performed). However, the efficacy of this approach depends on signal quality and so some form of depiction for this processing step is needed.

We added a depiction of the processing steps for regressing out the averaged responses in Figure 1 in an example electrode (panels C and D). We also show in the supplement the effect of regressing out the ERP on all the electrode pRFs. We have added Supplementary Figure 1-2.

Recommendation 2.6

I have a similar request for the authors latency correction of their data, where they identified a timing error and re-aligned the data without ground truth. Again, this is appropriate, but some depiction of the success of this correction is very critical for confirming the integrity of the data.

We now report more detail on the latency correction, and also point out that any small error in the estimate would not affect our conclusions (4.6 ECoG data analysis | Data epoching). The correction was important for a prior paper on temporal dynamics (Groen et al, 2022), which used data from the same participants and estimated the latency of responses. In this paper, our analyses are in the spectral domain (and discard phase), so small temporal shifts are not critical. We now also link to the public code associated with that paper, which implemented the adjustment and quantified the uncertainty in the latency adjustment.

More details on latency adjustment provided in section 4.6.

Recommendation 2.7

In many places the authors report their data shows a 'summary' value, please clarify if this means averaging or summation over a range.

For both broadband and alpha, we derive one summary value (a scalar) for trial for each stimulus. For broadband, the summary metric is the ratio of power during a given trial and power during blanks, where power in a trial is the geometric mean of the power at each frequency within the defined band). This is equation 3 in the methods, which is now referred to the first time that summary metrics are mentioned in the results.  For alpha, the summary metric is the height of the Gaussian from our model-based approach. This is in equations 1 and 2, and is also now referred to the first time summary metrics are mentioned in the results.

We added explanation of the summary metrics in the figure captions and results where they are first used, and also referred to the equations in the methods where they are defined.

Recommendation 2.8

The authors conclude: "we have discovered that spectral power changes in the alpha range reflect both suppression of alpha oscillations and elevation of broadband power." It might not have been the intention, but 'discovered' seems overstated.

We agree and changed this sentence.

Recommendation 2.9

Supp Fig 9 is a great effort by the authors to convey their findings to the reader, it should be a main figure.

We are glad you found Supplementary Figure 9 valuable. We moved this figure to the main text.

Reviewer #3 (Public Review):

Summary:

This study tackles the important subject of sensory driven suppression of alpha oscillations using a unique intracranial dataset in human patients. Using a model-based approach to separate changes in alpha oscillations from broadband power changes, the authors try to demonstrate that alpha suppression is spatially tuned, with similar center location as high broadband power changes, but much larger receptive field. They also point to interesting differences between low-order (V1-V3) and higher-order (dorsolateral) visual cortex. While I find some of the methodology convincing, I also find significant parts of the data analysis, statistics and their presentation incomplete. Thus, I find that some of the main claims are not sufficiently supported. If these aspects could be improved upon, this study could potentially serve as an important contribution to the literature with implications for invasive and non-invasive electrophysiological studies in humans.

We thank the reviewer for the summary.

Strengths:

The study utilizes a unique dataset (ECOG & high-density ECOG) to elucidate an important phenomenon of visually driven alpha suppression. The central question is important and the general approach is sound. The manuscript is clearly written and the methods are generally described transparently (and with reference to the corresponding code used to generate them). The model-based approach for separating alpha from broadband power changes is especially convincing and well-motivated. The link to exogenous attention behavioral findings (figure 8) is also very interesting. Overall, the main claims are potentially important, but they need to be further substantiated (see weaknesses).

We thank the reviewer for the positive comments.

Weaknesses:

I have three major concerns:

Weakness 3.1. Low N / no single subject results/statistics:

The crucial results of Figure 4,5 hang on 53 electrodes from four patients (Table 2). Almost half of these electrodes (25/53) are from a single subject. Data and statistical analysis seem to just pool all electrodes, as if these were statistically independent, and without taking into account subject-specific variability. The mean effect per each patient was not described in text or presented in figures. Therefore, it is impossible to know if the results could be skewed by a single unrepresentative patient. This is crucial for readers to be able to assess the robustness of the results. N of subjects should also be explicitly specified next to each result.

We have added substantial changes to deal with subject specific effects, including new results and new figures.

• Figure 4 now shows variance explained by the alpha pRF broken down by each participant for electrodes in V1 to V3. We also now show a similar figure for dorsolateral electrodes in Supplementary Figure 4-2.

• Figure 5, which shows results from individual electrodes in V1 to V3, now includes color coding of electrodes by participant to make it clear how the electrodes group with participant. Similarly, for dorsolateral electrodes, we show electrodes grouped by participant in Supplementary Figure 5-1. Same for Supplementary Figure 6-2.

• Supplementary Figure 7-2 now shows the benefits of our model-based approach for estimating alpha broken down by individual participants.

• We also now include a new section in the supplement that summarizes for every major claim, what the supporting data are and how we addressed the issue of nesting electrodes by participant, section Graphical and statistical support for primary claims.

Weakness 3.2. Separation between V1-V3 and dorsolateral electrodes:

Out of 53 electrodes, 27 were doubly assigned as both V1-V3 and dorsolateral (Table 2, Figures 4,5). That means that out of 35 V1-V3 electrodes, 27 might actually be dorsolateral. This problem is exasperated by the low N. for example all the 20 electrodes in patient 8 assigned as V1-V3 might as well be dorsolateral. This double assignment didn't make sense to me and I wasn't convinced by the authors' reasoning. I think it needlessly inflates the N for comparing the two groups and casts doubts on the robustness of these analyses.

Electrode assignment was probabilistic to reflect uncertainty in the mapping between location and retinotopic map. The probabilistic assignment is handled in two ways.

(1) For visualizing results of single electrodes, we simply go with the maximum probability, so no electrode is visualized for both groups of data. For example, Figure 5a (V1-V3) and supplementary Figure 5-1a (dorsolateral electrodes) have no electrodes in common: no electrode is in both plots.

(2) For quantitative summaries, we sample the electrodes probabilistically (for example Figures 4, 5c). So, if for example, an electrode has a 20% chance of being in V1 to V3, and 30% chance of being in dorsolateral maps, and a 50% chance of being in neither, the data from that electrode is used in only 20% of V1-V3 calculations and 30% of dorsolateral calculations. In 50% of calculations, it is not used at all. This process ensures that an electrode with uncertain assignment makes no more contribution to the results than an electrode with certain assignment. An electrode with a low probability of being in, say, V1-V3, makes little contribution to any reported results about V1-V3. This procedure is essentially a weighted mean, which the reviewer suggests in the recommendations. Thus, we believe there is not a problem of “double counting”.

The alternative would have been to use maximum probability for all calculations. However, we think that doing so would be misleading, since it would not take into account uncertainty of assignment, and would thus overstate differences in results between the maps.

We now clarify in the Results that for probabilistic calculations, the contribution of an electrode is limited by the likelihood of assignment (Section 2.3). We also now explain in the methods why we think probabilistic sampling is important.

Weakness 3.3. Alpha pRFs are larger than broadband pRFs:

First, as broadband pRF models were on average better fit to the data than alpha pRF models (dark bars in Supp Fig 3. Top row), I wonder if this could entirely explain the larger Alpha pRF (i.e. worse fits lead to larger pRFs). There was no anlaysis to rule out this possibility.

We addressed this question in a new paragraph in Discussion section 3.1 (“What is the function of the large alpha pRFs?”, paragraph beginning… “Another possible interpretation is that the poorer model fit in the alpha pRF is due to lower signal-to-noise”). This paragraph both refers to prior work on the relationship between noise and pRF size and to our own control analyses (Supplementary Figure 5-2).

Weakness 3.4 Statistics

Second, examining closely the entire 2.4 section there wasn't any formal statistical test to back up any of the claims (not a single p-value is mentioned). It is crucial in my opinion to support each of the main claims of the paper with formal statistical testing.

We agree that it is important for the reader to be able to link specific results and analyses to specific claims. We are not convinced that null hypothesis statistical testing is always the best approach. This is a topic of active debate in the scientific community.

We added a new section that concisely states each major claim and explicitly annotates the supporting evidence. (Section 4.7). Please also refer to our responses to Reviewer #2 regarding statistical testing (Reviewer weakness 2.4 “Statistical testing”)

Weakness 3.5 Summary

While I judge these issues as crucial, I can also appreciate the considerable effort and thoughtfulness that went into this study. I think that addressing these concerns will substantially raise the confidence of the readership in the study's findings, which are potentially important and interesting.

We again thank the reviewer for the positive comments.

Reviewer #3 (Recommendations For The Authors):

Suggestions for how to address the three major concerns:

Suggestion 3.1.

I am very well aware that it's very hard to have n=30 in a visual cortex ECOG study. That's fine. Best practice would be to have a linear mixed effects model with patients as a random effect. However, for some figures with just 3-4 patients (Figure 4,5) the sample size might be too small even for that. At the very minimum, I would expect to show in figures/describe in text all results per patient (perhaps one can do statistics within each patient, and show for each patient that the effect is significant). Even in primate studies with just two subjects it is expected to show that the results replicate for subject A and B. It is necessary to show that your results don't depend on a single unrepresentative subject. And if they do, at least be transparent about it.

We have addressed this thoroughly. Please see response to Weakness 3.1 (“Low N / no single subject results/statistics”).

Suggestion 3.2.

I just don't get it. I would simply assign an electrode to V1-V3 or dorsolateral cortex based on which area has the highest probability. It doesn't make sense to me that an electrode that has 60% of being in dorsolateral cortex and only 10% to be in V1-V3 would be assigned as both V1-V3 and dorsolateral. Also, what's the rationale to include such electrode in the analysis for let's say V1-V3 (we have weak evidence to believe it's there)? I would either assign electrodes based on the highest probability, or alternatively do a weighted mean based on the probability of each electrode belonging to each region group (e.g. electrode with 40% to be in V1-V3, will get twice the weight as an electrode who has 20% to be in V1-V3) but this is more complicated.

We have addressed this issue. Please refer to our response in Public Review (“Weakness 3.2 Separation between V1-V3 and dorsolateral”) for details.

Suggestion 3.3.

First, to exclude the possibility that alpha pRF are larger simply because they have a worse fit to the neural data, I would show if there is a correlation between the goodnessof-fit and pRF size (for alpha and broadband signals, separately). No [negative] correlation between goodness-of-fit and pRF size would be a good sign. I would also compare alpha & broadband receptive field size when controlling for the goodness-of-fit (selecting electrodes with similar goodness-of-fit for both signals). If the results replicate this way it would be convincing.

Second, there are no statistical tests in section 2.4, possibly also in others. Even if you employ bootstrap / Monte-Carlo resampling methods you can extract a p-value.

We have addressed this issue. Please refer to our response in Public Review Point 3.3 (“Alpha pRFs are larger than broadband pRFs”) for further details.

Suggestion 3.4.

Also, I don't understand the resampling procedure described in lines 652-660: "17.7 electrodes were assigned to V1-V3, 23.2 to dorsolateral, and 53 to either " - but 17.7 + 23.2 doesn't add up to 53. It also seems as if you assign visual areas differently in this resampling procedure than in the real data - "and randomly assigned each electrode to a visual area according to the Wang full probability distributions". If you assign in your actual data 27 electrodes to both visual areas, the same should be done in the resampling procedure (I would expect exactly 35 V1-V3 and 45 dorsolateral electrodes in every resampling, just the pRFs will be shuffled across electrodes).

We apologize for the confusion.

We fixed the sentence above, clarified the caption to Table 2, and also explained the overall strategy of probabilistic resampling better. See response to Public Review point 3.2 for details.

Suggestion 3.5.

These are rather technical comments but I believe they are crucial points to address in order to support your claims. I genuinely think your results are potentially interesting and important but these issues need to be first addressed in a revision. I also think your study may carry implications beyond just the visual domain, as alpha suppression is observed for different sensory modalities and cortical regions. Might be useful to discuss this in the discussion section.

Agree. We added a paragraph on this point to the Discussion (very end of 3.2).

Author response:

The following is the authors’ response to the original reviews

We thank the reviewers for their thoughtful feedback. We have made substantial revisions to the manuscript to address each of their comments, as we detail below. We want to highlight one major change in particular that addresses a concern raised by both reviewers: the role of the drift rate in our models. Motivated by their astute comments, we went back through our models and realized that we had made a particular assumption that deserved more scrutiny. We previously assumed that the process of encoding the observations made correct use of the objective, generative correlation, but then the process of calculating the weight of evidence used a mis-scaled, subjective version of the correlation. These assumptions led us to scale the drift rate in the model by a term that quantified how the standard deviation of the observation distribution was affected by the objective correlation (encoding), but to scale the bound height by the subjective estimate of the correlation (evidence weighing). However, we realized that encoding may also depend on the subjective correlation experienced by the participant. We have now tested several alternative models and found that the best-fitting model assumes that a single, subjective estimate of the correlation governs both encoding and evidence weighing. An important consequence of updating our models in this way is that we can now account for the behavioral data without needing the additional correlation-dependent drift terms (which, as reviewer #2 pointed out, were difficult to explain).

We also note that we changed the title slightly, replacing “weighting” with “weighing” for consistency with our usage throughout the manuscript.

Please see below for more details about this important point and our responses to the reviewers’ specific concerns. 

Public Reviews:

Reviewer #1 (Public review):

Summary:

The behavioral strategies underlying decisions based on perceptual evidence are often studied in the lab with stimuli whose elements provide independent pieces of decision-related evidence that can thus be equally weighted to form a decision. In more natural scenarios, in contrast, the information provided by these pieces is often correlated, which impacts how they should be weighted. Tardiff, Kang & Gold set out to study decisions based on correlated evidence and compare the observed behavior of human decision-makers to normative decision strategies. To do so, they presented participants with visual sequences of pairs of localized cues whose location was either uncorrelated, or positively or negatively correlated, and whose mean location across a sequence determined the correct choice. Importantly, they adjusted this mean location such that, when correctly weighted, each pair of cues was equally informative, irrespective of how correlated it was. Thus, if participants follow the normative decision strategy, their choices and reaction times should not be impacted by these correlations. While Tardiff and colleagues found no impact of correlations on choices, they did find them to impact reaction times, suggesting that participants deviated from the normative decision strategy. To assess the degree of this deviation, Tardiff et al. adjusted drift-diffusion models (DDMs) for decision-making to process correlated decision evidence. Fitting these models to the behavior of individual participants revealed that participants considered correlations when weighing evidence, but did so with a slight underestimation of the magnitude of this correlation. This finding made Tardiff et al. conclude that participants followed a close-to-normative decision strategy that adequately took into account correlated evidence.

Strengths:

The authors adjust a previously used experimental design to include correlated evidence in a simple, yet powerful way. The way it does so is easy to understand and intuitive, such that participants don't need extensive training to perform the task. Limited training makes it more likely that the observed behavior is natural and reflective of everyday decision-making. Furthermore, the design allowed the authors to make the amount of decision-related evidence equal across different correlation magnitudes, which makes it easy to assess whether participants correctly take account of these correlations when weighing evidence: if they do, their behavior should not be impacted by the correlation magnitude.

The relative simplicity with which correlated evidence is introduced also allowed the authors to fall back to the well-established DDM for perceptual decisions, which has few parameters, is known to implement the normative decision strategy in certain circumstances, and enjoys a great deal of empirical support. The authors show how correlations ought to impact these parameters, and which changes in parameters one would expect to see if participants misestimate these correlations or ignore them altogether (i.e., estimate correlations to be zero). This allowed them to assess the degree to which participants took into account correlations on the full continuum from perfect evidence weighting to complete ignorance. With this, they could show that participants in fact performed rational evidence weighting if one assumed that they slightly underestimated the correlation magnitude.

Weaknesses:

The experiment varies the correlation magnitude across trials such that participants need to estimate this magnitude within individual trials. This has several consequences:

(1) Given that correlation magnitudes are estimated from limited data, the (subjective) estimates might be biased towards their average. This implies that, while the amount of evidence provided by each 'sample' is objectively independent of the correlation magnitude, it might subjectively depend on the correlation magnitude. As a result, the normative strategy might differ across correlation magnitudes, unlike what is suggested in the paper. In fact, it might be the case that the observed correlation magnitude underestimates corresponds to the normative strategy.

We thank the reviewer for raising this interesting point, which we now address directly with new analyses including model fits (pp. 15–24). These analyses show that the participants were computing correlation-dependent weights of evidence from observation distributions that reflected suboptimal misestimates of correlation magnitudes. This strategy is normative in the sense that it is the best that they can do, given the encoding suboptimality. However, as we note in the manuscript, we do not know the source of the encoding suboptimality (pp. 23–24). We thus do not know if there might be a strategy they could have used to make the encoding more optimal.

(2) The authors link the normative decision strategy to putting a bound on the log-likelihood ratio (logLR), as implemented by the two decision boundaries in DDMs. However, as the authors also highlight in their discussion, the 'particle location' in DDMs ceases to correspond to the logLR as soon as the strength of evidence varies across trials and isn't known by the decision maker before the start of each trial. In fact, in the used experiment, the strength of evidence is modulated in two ways:

(i) by the (uncorrected) distance of the cue location mean from the decision boundary (what the authors call the evidence strength) and

(ii) by the correlation magnitude. Both vary pseudo-randomly across trials, and are unknown to the decision-maker at the start of each trial. As previous work has shown (e.g. Kiani & Shadlen (2009), Drugowitsch et al. (2012)), the normative strategy then requires averaging over different evidence strength magnitudes while forming one's belief. This averaging causes the 'particle location' to deviate from the logLR. This deviation makes it unclear if the DDM used in the paper indeed implements the normative strategy, or is even a good approximation to it.

We appreciate this subtle, but important, point. We now clarify that the DDM we use includes degrees of freedom that are consistent with normative decision processes that rely on the imperfect knowledge that participants have about the generative process on each trial, specifically: 1) a single drift-rate parameter that is fit to data across different values of the mean of the generative distribution, which is based on the standard assumption for these kinds of task conditions in which stimulus strength is varied randomly from trial-to-trial and thus prevents the use of exact logLR (which would require stimulus strength-specific scale factors; Gold and Shadlen, 2001); 2) the use of a collapsing bound, which in certain cases (including our task) is thought to support a stimulus strength-dependent calibration of the decision variable to optimize decisions (Drugowitsch et al, 2012); and 3) free parameters (one per correlation) to account for subjective estimates of the correlation, which affected the encoding of the observations that are otherwise weighed in a normative manner in the best-fitting model.

Also, to clarify our terminology, we define the objective evidence strength as the expected logLR in a given condition, which for our task is dependent on both the distance of the mean from the decision boundary and the correlation (p. 7). 

Given that participants observe 5 evidence samples per second and on average require multiple seconds to form their decisions, it might be that they are able to form a fairly precise estimate of the correlation magnitude within individual trials. However, whether this is indeed the case is not clear from the paper.

These points are now addressed directly in Results (pp. 23–24) and Figure 7 supplemental figures 1–3. Specifically, we show that, as the reviewer correctly surmised above, empirical correlations computed on each trial tended to be biased towards zero (Fig 7–figure supplement 1). However, two other analyses were not consistent with the idea that participants’ decisions were based on trial-by-trial estimates of the empirical correlations: 1) those with the shortest RTs did not have the most-biased estimates (Fig 7–figure supplement 2), and 2) there was no systematic relationship between objective and subjective fit correlations across participants (Fig 7–figure supplement 3).

Furthermore, the authors capture any underestimation of the correlation magnitude by an adjustment to the DDM bound parameter. They justify this adjustment by asking how this bound parameter needs to be set to achieve correlation-independent psychometric curves (as observed in their experiments) even if participants use a 'wrong' correlation magnitude to process the provided evidence. Curiously, however, the drift rate, which is the second critical DDM parameter, is not adjusted in the same way. If participants use the 'wrong' correlation magnitude, then wouldn't this lead to a mis-weighting of the evidence that would also impact the drift rate? The current model does not account for this, such that the provided estimates of the mis-estimated correlation magnitudes might be biased.

We appreciate this valuable comment, and we agree that we previously neglected the potential impact of correlation misestimates on evidence strength. As we now clarify, the correlation enters these models in two ways: 1) via its effect on how the observations are encoded, which involves scaling both the drift and the bound; and 2) via its effect on evidence weighing, which involves scaling only the bound (pp. 15–18). We previously assumed that only the second form of scaling might involve a subjective (mis-)estimate of the correlation. We now examine several models that also include the possibility of either or both forms using subjective correlation estimates. We show that a model that assumes that the same subjective estimate drives both encoding and weighing (the “full-rho-hat” model) best accounts for the data. This model provides better fits (after accounting for differences in numbers of parameters) than models with: 1) no correlation-dependent adjustments (“base” model), 2) separate drift parameters for each correlation condition (“drift” model), 3) optimal (correlation-dependent) encoding but suboptimal weighing (“bound-rho-hat” model, which was our previous formulation), 4) suboptimal encoding and weighing (“scaled-rho-hat” model), and 5) optimal encoding but suboptimal weighing and separate correlation-dependent adjustments to the drift rate (“boundrho-hat plus drift” model). We have substantially revised Figures 5–7 and the associated text to address these points.

Lastly, the paper makes it hard to assess how much better the participants' choices would be if they used the correct correlation magnitudes rather than underestimates thereof. This is important to know, as it only makes sense to strictly follow the normative strategy if it comes with a significant performance gain.

We now include new analyses in Fig. 7 that demonstrate how much participants' choices and RT deviate from: 1) an ideal observer using the objective correlations, and 2) an observer who failed to adjust for the fit subjective correlation when weighing the evidence (i.e., using the subjective correlation for encoding but a correlation of zero for weighing). We now indicate that participants’ performance was quite close to that predicted by the ideal observer (using the true, objective correlation) for many conditions. Thus, we agree that they might not have had the impetus to optimize the decision process further, assuming it were possible under these task conditions.

Reviewer #2 (Public review):

Summary:

This study by Tardiff, Kang & Gold seeks to: i) develop a normative account of how observers should adapt their decision-making across environments with different levels of correlation between successive pairs of observations, and ii) assess whether human decisions in such environments are consistent with this normative model.

The authors first demonstrate that, in the range of environments under consideration here, an observer with full knowledge of the generative statistics should take both the magnitude and sign of the underlying correlation into account when assigning weight in their decisions to new observations: stronger negative correlations should translate into stronger weighting (due to the greater information furnished by an anticorrelated generative source), while stronger positive correlations should translate into weaker weighting (due to the greater redundancy of information provided by a positively correlated generative source). The authors then report an empirical study in which human participants performed a perceptual decision-making task requiring accumulation of information provided by pairs of perceptual samples, under different levels of pairwise correlation. They describe a nuanced pattern of results with effects of correlation being largely restricted to response times and not choice accuracy, which could partly be captured through fits of their normative model (in this implementation, an extension of the well-known drift-diffusion model) to the participants' behaviour while allowing for misestimation of the underlying correlations.

Strengths:

As the authors point out in their very well-written paper, appropriate weighting of information gathered in correlated environments has important consequences for real-world decisionmaking. Yet, while this function has been well studied for 'high-level' (e.g. economic) decisions, how we account for correlations when making simple perceptual decisions on well-controlled behavioural tasks has not been investigated. As such, this study addresses an important and timely question that will be of broad interest to psychologists and neuroscientists. The computational approach to arrive at normative principles for evidence weighting across environments with different levels of correlation is very elegant, makes strong connections with prior work in different decision-making contexts, and should serve as a valuable reference point for future studies in this domain. The empirical study is well designed and executed, and the modelling approach applied to these data showcases a deep understanding of relationships between different parameters of the drift-diffusion model and its application to this setting. Another strength of the study is that it is preregistered.

Weaknesses:

In my view, the major weaknesses of the study center on the narrow focus and subsequent interpretation of the modelling applied to the empirical data. I elaborate on each below:

Modelling interpretation: the authors' preference for fitting and interpreting the observed behavioural effects primarily in terms of raising or lowering the decision bound is not well motivated and will potentially be confusing for readers, for several reasons. First, the entire study is conceived, in the Introduction and first part of the Results at least, as an investigation of appropriate adjustments of evidence weighting in the face of varying correlations. The authors do describe how changes in the scaling of the evidence in the drift-diffusion model are mathematically equivalent to changes in the decision bound - but this comes amidst a lengthy treatment of the interaction between different parameters of the model and aspects of the current task which I must admit to finding challenging to follow, and the motivation behind shifting the focus to bound adjustments remained quite opaque. 

We appreciate this valuable feedback. We have revised the text in several places to make these important points more clearly. For example, in the Introduction we now clarify that “The weight of evidence is computed as a scaled version of each observation (the scaling can be applied to the observations or to the bound, which are mathematically equivalent; Green and Swets, 1966) to form the logLR” (p. 3). We also provide more details and intuition in the Results section for how and why we implemented the DDM the way we did. In particular, we now emphasize that the correlation enters these models in two ways: 1) via its effect on encoding the observations, which scales both the drift and the bound; and 2) via its effect on evidence weighing, which scales only the bound (pp. 15–18).

Second, and more seriously, bound adjustments of the form modelled here do not seem to be a viable candidate for producing behavioural effects of varying correlations on this task. As the authors state toward the end of the Introduction, the decision bound is typically conceived of as being "predefined" - that is, set before a trial begins, at a level that should strike an appropriate balance between producing fast and accurate decisions. There is an abundance of evidence now that bounds can change over the course of a trial - but typically these changes are considered to be consistently applied in response to learned, predictable constraints imposed by a particular task (e.g. response deadlines, varying evidence strengths). In the present case, however, the critical consideration is that the correlation conditions were randomly interleaved across trials and were not signaled to participants in advance of each trial - and as such, what correlation the participant would encounter on an upcoming trial could not be predicted. It is unclear, then, how participants are meant to have implemented the bound adjustments prescribed by the model fits. At best, participants needed to form estimates of the correlation strength/direction (only possible by observing several pairs of samples in sequence) as each trial unfolded, and they might have dynamically adjusted their bounds (e.g. collapsing at a different rate across correlation conditions) in the process. But this is very different from the modelling approach that was taken. In general, then, I view the emphasis on bound adjustment as the candidate mechanism for producing the observed behavioural effects to be unjustified (see also next point).

We again appreciate this valuable feedback and have made a number of revisions to try to clarify these points. In addition to addressing the equivalence of scaling the evidence and the bound in the Introduction, we have added the following section to Results (Results, p.18):

“Note that scaling the bound in these formulations follows conventions of the DDM, as detailed above, to facilitate interpretation of the parameters. These formulations also raise an apparent contradiction: the “predefined” bound is scaled by subjective estimates of the correlation, but the correlation was randomized from trial to trial and thus could not be known in advance. However, scaling the bound in these ways is mathematically equivalent to using a fixed bound on each trial and scaling the observations to approximate logLR (see Methods). This equivalence implies that in the brain, effectively scaling a “predefined” bound could occur when assigning a weight of evidence to the observations as they are presented.”

We also note in Methods (pp. 40–41):

“In the DDM, this scaling of the evidence is equivalent to assuming that the decision variable accumulates momentary evidence of the form (x1 + x2) and then dividing the bound height by the appropriate scale factor. An alternative approach would be to scale both the signal and noise components of the DDM by the scale factor. However, scaling the bound is both simpler and maintains the conventional interpretation of the DDM parameters in which the bound reflects the decision-related components of the evidence accumulation process, and the drift rate represents sensory-related components.”

We believe we provide strong evidence that participants adjust their evidence weighing to account for the correlations (see response below), but we remain agnostic as to how exactly this weighing is implemented in the brain.

Modelling focus: Related to the previous point, it is stated that participants' choice and RT patterns across correlation conditions were qualitatively consistent with bound adjustments (p.20), but evidence for this claim is limited. Bound adjustments imply effects on both accuracy and RTs, but the data here show either only effects on RTs, or RT effects mixed with accuracy trends that are in the opposite direction to what would be expected from bound adjustment (i.e. slower RT with a trend toward diminished accuracy in the strong negative correlation condition; Figure 3b). Allowing both drift rate and bound to vary with correlation conditions allowed the model to provide a better account of the data in the strong correlation conditions - but from what I can tell this is not consistent with the authors' preregistered hypotheses, and they rely on a posthoc explanation that is necessarily speculative and cannot presently be tested (that the diminished drift rates for higher negative correlations are due to imperfect mapping between subjective evidence strength and the experimenter-controlled adjustment to objective evidence strengths to account for effects of correlations). In my opinion, there are other candidate explanations for the observed effects that could be tested but lie outside of the relatively narrow focus of the current modelling efforts. Both explanations arise from aspects of the task, which are not mutually exclusive. The first is that an interesting aspect of this task, which contrasts with most common 'univariate' perceptual decision-making tasks, is that participants need to integrate two pieces of information at a time, which may or may not require an additional computational step (e.g. averaging of two spatial locations before adding a single quantum of evidence to the building decision variable). There is abundant evidence that such intermediate computations on the evidence can give rise to certain forms of bias in the way that evidence is accumulated (e.g. 'selective integration' as outlined in Usher et al., 2019, Current Directions in Psychological Science; Luyckx et al., 2020, Cerebral Cortex) which may affect RTs and/or accuracy on the current task. The second candidate explanation is that participants in the current study were only given 200 ms to process and accumulate each pair of evidence samples, which may create a processing bottleneck causing certain pairs or individual samples to be missed (and which, assuming fixed decision bounds, would presumably selectively affect RT and not accuracy). If I were to speculate, I would say that both factors could be exacerbated in the negative correlation conditions, where pairs of samples will on average be more 'conflicting' (i.e. further apart) and, speculatively, more challenging to process in the limited time available here to participants. Such possibilities could be tested through, for example, an interrogation paradigm version of the current task which would allow the impact of individual pairs of evidence samples to be more straightforwardly assessed; and by assessing the impact of varying inter-sample intervals on the behavioural effects reported presently.

We thank the reviewer for this thoughtful and valuable feedback. We have thoroughly updated the modeling section to include new analysis and clearer descriptions and interpretations of our findings (including Figs. 5–7 and additional references to the Usher, Luyckx, and other studies that identified decision suboptimalities). The comment about “an additional computational step” in converting the observations to evidence was particularly useful, in that it made us realize that we were making what we now consider to be a faulty assumption in our version of the DDM. Specifically, we assumed that subjective misestimates of the correlation affected how observations were converted to evidence (logLR) to form the decision (implemented as a scaling of the bound height), but we neglected to consider how suboptimalities in encoding the observations could also lead to misestimates of the correlation. We have retained the previous best-fitting models in the text, for comparison (the “bound-rho-hat” and “bound-rho-hat + drift” models). In addition, we now include a “full-rho-hat” model that assumes that misestimates of rho affect both the encoding of the observations, which affects the drift rate and bound height, and the weighing of the evidence, which affects only the bound height. This was the best-fitting model for most participants (after accounting for different numbers of parameters associated with the different models we tested). Note that the full-rho-hat model predicts the lack of correlation-dependent choice effects and the substantial correlation-dependent RT effects that we observed, without requiring any additional adjustments to the drift rate (as we resorted to previously).

In summary, we believe that we now have a much more parsimonious account of our data, in terms of a model in which subjective estimates of the correlation are alone able to account for our patterns of choice and RT data. We fully agree that more work is needed to better understand the source of these misestimates but also think those questions are outside the scope of the present study.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

A few minor comments:

(1) Evidence can be correlated in multiple ways. It could be correlated within individual pieces of evidence in a sequence, or across elements in that sequence (e.g., across time). This distinction is important, as it determines how evidence ought to be accumulated across time. In particular, if evidence is correlated across time, simply summing it up might be the wrong thing to do. Thus, it would be beneficial to make this distinction in the Introduction, and to mention that this paper is only concerned with the first type of correlation.

We now clarify this point in the Introduction (p. 5–6).

(2) It is unclear without reading the Methods how the blue dashed line in Figure 4c is generated. To my understanding, it is a prediction of the naive DDM model. Is this correct?

We now specify the models used to make the predictions shown in Fig. 4c (which now includes an additional model that uses unscaled observations as evidence).

(3) In Methods, given the importance of the distribution of x1 + x2, it would be useful to write it out explicitly, e.g., x1 + x2 ~ N(2 mu_g, ..), specifying its mean and its variance.

Excellent suggestion, added to p. 38.

(4) From Methods and the caption of Figure 6 - Supplement 1 it becomes clear that the fitted DDM features a bound that collapses over time. I think that this should also be mentioned in the main text, as it is a not-too-unimportant feature of the model.

Excellent suggestion, added to p. 15, with reference to Fig. 6-supplement 1 on p. 20.

(5) The functional form of the bound is 2 (B - tb t). To my understanding, the effective B changes as a function of the correlation magnitude. Does tb as well? If not, wouldn't it be better if it does, to ensure that 2 (B - tb t) = 0 independent of the correlation magnitude?

In our initial modeling, we also considered whether the correlation-dependent adjustment, which is a function of both correlation sign and magnitude, should be applied to the initial bound or to the instantaneous bound (i.e., after collapse, affecting tb as well). In a pilot analysis of data from 22 participants in the 0.6 correlation-magnitude group, we found that this choice had a negligible effect on the goodness-of-fit (deltaAIC = -0.9, protected exceedance probability = 0.63, in favor of the instantaneous bound scaling). We therefore used the instantaneous bound version in the analyses reported in the manuscript but doubt this choice was critical based on these results. We have clarified our implementation of the bound in Methods (p. 43–44).

Reviewer #2 (Recommendations for the authors):

In addition to the points raised above, I have some minor suggestions/open questions that arose from my reading of the manuscript:

(1) Are the predictions outlined in the paper specific to cases where the two sources are symmetric around zero? If distributions are allowed to be asymmetric then one can imagine cases (i.e. when distribution means are sufficiently offset from one another) where positive correlations can increase evidence strength and negative correlations decrease evidence strength. There's absolutely still value and much elegance in what the authors are showing with this work, but if my intuition is correct, it should ideally be acknowledged that the predictions are restricted to a specific set of generative circumstances.

We agree that there are a lot of ways to manipulate correlations and their effect on the weight of evidence. At the end of the Discussion, we emphasize that our results apply to this particular form of correlation (p. 32).

(2) Isn't Figure 4C misleading in the sense that it collapses across the asymmetry in the effect of negative vs positive correlations on RT, which is clearly there in the data and which simply adjusting the correlation-dependent scale factor will not reproduce?

We agree that this analysis does not address any asymmetries in suboptimal estimates of positive versus negative correlations. We believe that those effects are much better addressed using the model fitting, which we present later in the Results section. We have now simplified the analyses in Fig. 4c, reporting the difference in RT between positive and negative correlation conditions instead of a linear regression.

(3) I found the transition on p.17 of the Results section from the scaling of drift rate by correlation to scaling of bound height to be quite abrupt and unclear. I suspect that many readers coming from a typical DDM modelling background will be operating under the assumption that drift rate and bound height are independent, and I think more could be done here to explain why scaling one parameter by correlation in the present case is in fact directly equivalent to scaling the other.

Thank you for the very useful feedback, we have substantially revised this text to make these points more clearly.

(4) P.3, typo: Alan *Turing*

That’s embarrassing. Fixed.

(5) P.27, typo: "participants adopt a *fixed* bound"

Fixed.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1:

The manuscript suggests the zebrafish homolog of ctla-4 and generates a new mutant in it. However, the locus that is mutated is confusingly annotated as both CD28 (current main annotation in ZFIN) and CTLA-4/CD152 (one publication from 2020), see: https://zfin.org/ZDB-GENE-070912-128. Both human CTLA-4 and CD28 align with relatively similar scores to this gene. There seem to be other orthologs of these receptors in the zebrafish genome, including CD28-like (https://zfin.org/ZDB-GENE-070912-309) which neighbors the gene annotated as CD28 (exhibiting similar synteny as human CD28 and CTLA-4). It would be helpful to provide more information to distinguish between this family of genes and to further strengthen the evidence that this mutant is in ctla-4, not cd28. Also, is one of these genes in the zebrafish genome (e.g. cd28l) potentially a second homolog of CTLA-4? Is this why this mutant is viable in zebrafish and not mammals? Some suggestions:

(a) A more extensive sequence alignment that considers both CTLA-4 and CD28, potentially identifying the best homolog of each human gene, especially taking into account any regions that are known to produce the functional differences between these receptors in mammals and effectively assigns identities to the two genes annotated as "cd28" and "cd28l" as well as the gene "si:dkey-1H24.6" that your CD28 ORF primers seem to bind to in zebrafish.

In response to the reviewer's insightful suggestions, we have conducted more extensive sequence alignment and phylogenetic analyses that consider both CTLA-4, CD28, and CD28-like molecules, taking into account key regions crucial for the functionalities and functional differences between these molecules across various species, including mammals and zebrafish.

Identification of zebrafish Ctla-4: We identified zebrafish Ctla-4 as a homolog of mammalian CTLA-4 based on key conserved structural and functional characteristics. Structurally, the Ctla-4 gene shares similar exon organization compared to mammalian CTLA-4. Ctla-4 is a type I transmembrane protein with typical immunoglobulin superfamily features. Multiple amino acid sequence alignments revealed that Ctla-4 contains a ¹¹³LFPPPY¹¹⁸ motif and a ¹²³GNGT¹²⁶ motif in the ectodomain, and a tyrosine-based ²⁰⁶YVKF²⁰⁹ motif in the distal C-terminal region. These motifs closely resemble MYPPPY, GNGT, and YVKM motifs in mammalian CTLA-4s, which are essential for binding to CD80/CD86 ligands and molecular internalization and signaling inhibition. Despite only 23.7% sequence identity to human CTLA-4, zebrafish Ctla-4 exhibits a similar tertiary structure with a two-layer β-sandwich architecture in its extracellular IgV-like domain. Four cysteine residues responsible for the formation of two pairs of disulfide bonds (Cys²⁰-Cys⁹¹/Cys⁴⁶-Cys⁶⁵ in zebrafish and Cys²¹-Cys⁹²/Cys⁴⁸-Cys⁶⁶ in humans) that connect the two-layer β-sandwich are conserved. Additionally, a separate cysteine residue (Cys¹²⁰ in zebrafish and Cys¹²⁰ in humans) involved in dimerization is also present, and Western blot analysis under reducing and non-reducing conditions confirmed Ctla-4’s dimerization. Phylogenetically, Ctla-4 clusters with other known CTLA-4 homologs from different species with high bootstrap probability, while zebrafish Cd28 groups separately with other CD28s. Functionally, Ctla-4 is predominantly expressed on CD4⁺ T and CD8⁺ T cells in zebrafish. It plays a pivotal inhibitory role in T cell activation by competing with CD28 for binding to CD80/86, as validated through a series of both in vitro and in vivo assays, including microscale thermophoresis assays which demonstrated that Ctla-4 exhibits a significantly higher affinity for Cd80/86 than Cd28 (KD = 0.50 ± 0.25 μM vs. KD = 2.64 ± 0.45 μM). These findings confirm Ctla-4 as an immune checkpoint molecule, reinforcing its identification within the CTLA-4 family.

Comparison between zebrafish Cd28 and "Cd28l": Zebrafish Cd28 contains an extracellular SYPPPF motif and an intracellular FYIQ motif. The extracellular SYPPPF motif is essential for binding to Cd80/CD86, while the intracellular FYIQ motif likely mediates kinase recruitment and co-stimulatory signaling. In contrast, the "Cd28l" molecule lacks the SYPPPF motif, which is critical for Cd80/CD86 binding, and exhibits strong similarity in its C-terminal 79 amino acids to Ctla-4 rather than Cd28. Consequently, "Cd28l" resembles an atypical Ctla-4-like molecule but fails to exhibit Cd80/CD86 binding activity.

We have incorporated the relevant analysis results into the main text of the revised manuscript and updated Supplementary Figure 1. Additionally, we provide key supplementary analyses here for the reviewer's convenience.  

Author response image 1.

Illustrates the alignment of Ctla-4 (XP_005167576.1) and Ctla-4-like (XP_005167567.1, previously referred to as "Cd28l") in zebrafish, generated using ClustalX and Jalview. Conserved and partially conserved amino acid residues are highlighted in color gradients ranging from carnation to red, respectively. The B7-binding motif is encircled with a red square.

(b) Clearer description in the main text of such an analysis to better establish that the mutated gene is a homolog of ctla-4, NOT cd28.

We appreciate the reviewer's advice. Additional confirmation of zebrafish Ctla-4 is detailed in lines 119-126 of the revised manuscript.

(c) Are there mammalian anti-ctla-4 and/or anti-cd28 antibodies that are expected to bind to these zebrafish proteins? If so, looking to see whether staining is lost (or western blotting is lost) in your mutants could be additionally informative. (Our understanding is that your mouse anti-Ctla-4 antibody is raised against recombinant protein generated from this same locus, and so is an elegant demonstration that your mutant eliminates the production of the protein, but unfortunately does not contribute additional information to help establish its homology to mammalian proteins).

This suggestion holds significant value. However, a major challenge in fish immunology research is the limited availability of antibodies suitable for use in fish species; antibodies developed for mammals are generally not applicable. We attempted to use human and mouse anti-CTLA-4 and anti-CD28 antibodies to identify Ctla-4 and Cd28 in zebrafish, but the results were inconclusive, with no expected signals. This outcome likely arises from the low sequence identity between human/mouse CTLA-4 and CD28 and their zebrafish homologs (ranging from 21.3% to 23.7% for CTLA-4 and 21.2% to 24.0% for CD28). Therefore, developing specific antibodies against zebrafish Ctla-4 is essential for advancing this research.

The methods section is generally insufficient and doesn't describe many of the experiments performed in this manuscript. Some examples:

(a) No description of antibodies used for staining or Western blots (Figure1C, 1D, 1F).

(b) No description of immunofluorescence protocol (Figure 1D, 1F).

(c) No description of Western blot protocol (Figure 1C, 2C).

(d) No description of electron microscopy approach (Figure 2K).

(e) No description of the approach for determining microbial diversity (Entirety of Figure 6).

(f) No description of PHA/CFSE/Flow experiments (Figure 7A-E).

(g) No description of AlphaFold approach (Figures 7F-G).

(h) No description of co-IP approach (Figure 7H).

(i) No description of MST assay or experiment (Figure 7I).

(j) No description of purification of recombinant proteins, generation of anti-Ctla-4 antibody, or molecular interaction assays (Figures S2 and S6).

We apologize for this oversight. The methods section was inadvertently incomplete due to an error during the file upload process at submission. This issue has been addressed in the revised manuscript. We appreciate your understanding.

Figure 5 suggests that there are more Th2 cells 1, Th2 cells 2, and NKT cells in ctla-4 mutants through scRNA-seq. However, as the cell numbers for these are low in both genotypes, there is only a single replicate for each genotype scRNA-seq experiment, and dissociation stress can skew cell-type proportions, this finding would be much more convincing if another method that does not depend on dissociation was used to verify these results. Furthermore, while Th2 cells 2 are almost absent in WT scRNA-seq, KEGG analysis suggests that a major contributor to their clustering may be ribosomal genes (Fig. 5I). Since no batch correction was described in the methods, it would be beneficial to verify the presence of this cluster in ctla-4 mutants and WT animals through other means, such as in situ hybridization or transgenic lines.   

We are grateful for the insightful comments provided by the reviewer. Given that research on T cell subpopulations in fish is still in its nascent stages, the availability of specific marker antibodies and relevant transgenic strains remains limited. Our single-cell RNA sequencing (scRNA-seq) analysis revealed that a distinct Th2 subset 2 was predominantly observed in Ctla-4 mutants but was rare in wild-type zebrafish, it suggests that this subset may primarily arise under pathological conditions associated with Ctla-4 mutation. Due to the near absence of Th2 subset 2 in wild-type samples, KEGG enrichment analysis was performed exclusively on this subset from Ctla-4-deficient intestines. The ribosome pathway was significantly enriched, suggesting that these cells may be activated to fulfill their effector functions. However, confirming the presence of Th2 subset 2 using in situ hybridization or transgenic zebrafish lines is currently challenging due to the lack of lineage-specific markers for detailed classification of Th2 cell subsets and the preliminary nature of scRNA-seq predictions.

To address the reviewers' suggestion to confirm compositional changes in Th2 and NKT cells using dissociation-independent methods, we quantified mRNA levels of Th2 (il4, il13, and gata3) and NKT (nkl.2, nkl.4, and prf1.1) cell marker genes via RT-qPCR in intestines from wild-type and mutant zebrafish. As shown in Figure S7B and S7C, these markers were significantly upregulated in Ctla-4-deficient intestines compared to wild-type controls. This indicates an overall increase in Th2 and NKT cell activity in mutant zebrafish, aligning with our scRNA-seq analysis and supports the validity of our initial findings.

Before analyzing the scRNA-seq data, we performed batch correction using the Harmony algorithm via cloud-based Cumulus v1.0 on the aggregated gene-count matrices. This methodological detail has been included in the “Materials and Methods” section of the revised manuscript. Moreover, the RT-qPCR results are presented in Supplementary Figures S7B and S7C.

Quality control (e.g., no. of UMIs, no. of genes, etc.) metrics of the scRNAseq experiments should be presented in the supplementary information for each sample to help support that observed differential expression is not merely an outcome of different sequencing depths of the two samples.

As illustrated in Fig. S5, the quality control data have been supplemented to include the effective cell number of the sample, along with pre- and post-filtering metrics such as nFeature_RNA, nCount_RNA and mitochondrial percentage (percent.mito). Furthermore, scatter plots comparing the basic information of the sample cells before and after filtering are provided.

Some references to prior research lack citations. Examples:

(a)"Given that Ctla-4 is primarily expressed on T cells (Figure 1E-F), and its absence has been shown to result in intestinal immune dysregulation, indicating a crucial role of this molecule as a conserved immune checkpoint in T cell inhibition."

The references were incorporated into line 71 of the revised manuscript.

(b) Line 83: Cite evidence/review for the high degree of conservation in adaptive immunity.

The references were incorporated into line 93 of the revised manuscript.

(c) Lines 100-102: Cite the evidence that MYPPPY is a CD80/86 binding motif.

The references were incorporated into line 117 of the revised manuscript.

The text associated with Figure 8 (Lines 280-289) does not clearly state that rescue experiments are being done in mutant zebrafish.

We have provided a clear explanation of the rescue experiments conducted in Ctla-4-deficient zebrafish. This revision has been incorporated into line 319.

Line 102: Is there evidence from other animals that LFPPPY can function as a binding site for CD80/CD86? Does CD28 also have this same motif?

The extracellular domains of CTLA-4 and CD28, which bind to CD80/CD86, are largely conserved across various species. This conservation is exemplified by a central PPP core motif, although the flanking amino acids exhibit slight variations. In mammals, both CTLA-4 and CD28 feature the conserved MYPPPY motif. By contrast, in teleost fish, such as rainbow trout, CTLA-4 contains an LYPPPY motif, while CD28 has an MYPPPI motif (Ref. 1). Grass carp CTLA-4 displays an LFPPPY motif, whereas its CD28 variant bears an IYPPPF motif. Yeast two-hybrid assays confirm that these motifs facilitate interactions between grass carp CTLA-4 and CD28 with CD80/CD86 (Ref. 2). Similarly, zebrafish Ctla-4 contains the LFPPPY motif observed in grass carp, while Cd28 exhibits a closely related SYPPPF motif.

References:

(1) Bernard, D et al. (2006) Costimulatory Receptors in a Teleost Fish: Typical CD28, Elusive CTLA-4. J Immunol. 176: 4191-4200.

(2) Lu T Z et al. (2022) Molecular and Functional Analyses of the Primordial Costimulatory Molecule CD80/86 and Its Receptors CD28 and CD152 (CTLA-4) in a Teleost Fish. Frontiers in Immunology. 13:885005.

Line 110-111: Suggest adding citation of these previously published scRNAseq data to the main text in addition to the current description in the Figure legend.

The reference has been added in line 129 in the main text.

Figure 3B: It would be helpful to label a few of the top differentially expressed genes in Panel B?

The top differentially expressed genes have been labeled in Figure 3B.

Figure 3G: It's unclear how this analysis was conducted, what this figure is supposed to demonstrate, and in its current form it is illegible.

Figure 3G displays a protein-protein interaction network constructed from differentially expressed genes. The densely connected nodes, representing physical interactions among proteins, provide valuable insights for basic scientific inquiry and biological or biomedical applications. As proteins are crucial to diverse biological functions, their interactions illuminate the molecular and cellular mechanisms that govern both healthy and diseased states in organisms. Consequently, these networks facilitate the understanding of pathogenic and physiological processes involved in disease onset and progression.

To construct this network, we first utilized the STRING database (https://string-db.org) to generate an initial network diagram using the differentially expressed genes. This diagram was subsequently imported into Cytoscape (version 3.9.1) for visualization and further analysis. Node size and color intensity reflect the density of interactions, indicating the relative importance of each protein. Figure 3G illustrates that IL1β was a central cytokine hub in the disease process of intestinal inflammation in Ctla-4-deficient zebrafish.

Expression scale labeling:

(a) Most gene expression scales are not clearly labeled: do they represent mean expression or scaled expression? Has the expression been log-transformed, and if so, which log (natural log? Log10? Log2?). See: Figure 3E, 3I, 4D, 4E, 5B, 5G, 5H, 6I.

The gene expression scales are detailed in the figure legends. Specifically, Figures 3E, 3I, and 6I present heatmaps depicting row-scaled expression levels for the corresponding genes. In contrast, Figures 4D and 4E display heatmaps illustrating the mean expression of these genes. Additionally, the dot plots in Figures 5B, 5G, and 5H visualize the mean expression levels of the respective genes.

(b) For some plots, diverging color schemes (i.e. with white/yellow in the middle) are used for non-diverging scales and would be better represented with a sequential color scale. See: 4D, 4E, and potentially others (not fully clear because of the previous point).

The color schemes in Figures 4D and 4E have been updated to a sequential color scale. The gene expression data depicted in these figures represent mean expression values and have not undergone log transformation. This information has been incorporated into the figure legend for clarity.

Lines 186-187: Though it is merely suggested, apoptotic gene expression can be upregulated as part of the dissociation process for single-cell RNAseq. This would be much stronger if supported by a staining, such as anti-Caspase 3.

Following the reviewer's insightful recommendations, we conducted a TUNEL assay to evaluate apoptosis in the posterior intestinal epithelial cells of both wild-type and Ctla-4-deficient zebrafish. As expected, our results demonstrate a significant increase in epithelial cell apoptosis in Ctla-4-deficient zebrafish compared with wild-type fish. The corresponding data are presented in Figure S6D and have been incorporated into the manuscript. Detailed protocols for the TUNEL assay have also been included in the Materials and Methods section.

Author response image 2.

Illustrates the quantification of TUNEL-positive cells per 1 × 10⁴ μm^2/⁻ in the posterior intestines of both wild-type (WT) and ctla-4^⁻/⁻ zebrafish (n = 5). The data demonstrate a comparative analysis of apoptotic cell density between the two genotypes.

Lines 248-251: This manuscript demonstrates gut inflammation and also changes in microbial diversity, but I don't think it demonstrates an association between them, which would require an experiment that for instance rescues one of these changes and shows that it ameliorates the other change, despite still being a ctla-4 mutant.

We appreciate the valuable comments from the reviewer. Recently, the relationship between inflammatory bowel disease (IBD) and gut microbial diversity has garnered considerable attention, with several key findings emerging from human IBD studies. For instance, patients with IBD (including ulcerative colitis and Crohn's disease) exhibit reduced microbial diversity, which is correlated with disease severity. This decrease in microbial richness is thought to stem from the loss of normal anaerobic bacteria, such as Bacteroides, Eubacterium, and Lactobacillus (Refs. 1-6). Research using mouse models has shown that inflammation increases oxygen and nitrate levels within the intestinal lumen, along with elevated host-derived electron acceptors, thereby promoting anaerobic respiration and overgrowth of Enterobacteriaceae (Ref 7). Consistent with these findings, our study observed a significant enrichment of Enterobacteriaceae in the inflamed intestines of Ctla-4-deficient zebrafish, which supporting the observations in mice. Despite this progress, the zebrafish model for intestinal inflammation remains under development, with limitations in available techniques for manipulating intestinal inflammation and reconstructing gut microbiota. These challenges hinder investigations into the association between intestinal inflammation and changes in microbial diversity. We plan to address these issues through ongoing technological advancements and further research. We thank the reviewer for their understanding.

References:

(1) Ott S J, Musfeldt M, Wenderoth D F, Hampe J, Brant O, Fölsch U R et al. (2004) Reduction in diversity of the colonic mucosa associated bacterial microflora in patients with active inflammatory bowel disease. Gut 53:685-693.

(2) Manichanh C, Rigottier-Gois L, Bonnaud E, Gloux K, Pelletier E, Frangeul L et al. (2006) Reduced diversity of faecal microbiota in Crohn's disease revealed by a metagenomic approach. Gut 55:205-211.

(3) Qin J J, Li R Q, Raes J, Arumugam M, Burgdorf K S, Manichanh C et al. (2010) A human gut microbial gene catalogue established by metagenomic sequencing. Nature 464:59-U70.

(4) Sha S M, Xu B, Wang X, Zhang Y G, Wang H H, Kong X Y et al. (2013) The biodiversity and composition of the dominant fecal microbiota in patients with inflammatory bowel disease. Diagn Micr Infec Dis 75:245-251.

(5) Ray K. (2015) IBD. Gut microbiota in IBD goes viral. Nat Rev Gastroenterol Hepatol 12:122.

(6) Papa E, Docktor M, Smillie C, Weber S, Preheim S P, Gevers D et al. (2012) Non-Invasive Mapping of the Gastrointestinal Microbiota Identifies Children with Inflammatory Bowel Disease. Plos One 7: e39242-39254.

(7) Hughes E R, Winter M G, Duerkop B A, Spiga L, de Carvalho T F, Zhu W H et al. (2017) Microbial Respiration and Formate Oxidation as Metabolic Signatures of Inflammation-Associated Dysbiosis. Cell Host Microbe 21:208-219.

Lines 270-272 say that interaction between Cd28/ctla-4 and Cd80/86 was demonstrated through bioinformatics, flow-cytometry, and Co-IP. Does this need to reference Fig S6D for the flow data? Figures 7F-G are very hard to read or comprehend as they are very small. Figure 7H is the most compelling evidence of this interaction and might stand out better if emphasized with a sentence referencing it on its own in the manuscript. 

In this study, we utilized an integrated approach combining bioinformatics prediction, flow cytometry, and co-immunoprecipitation (Co-IP) to comprehensively investigate and validate the interactions between Cd28/Ctla-4 and Cd80/86. Flow cytometry analysis, as depicted in Supplementary Figure 6D (revised as Supplementary Figure 8F), demonstrated the surface expression of Cd80/86 on HEK293T cells and quantified their interactions with Cd28 and Ctla-4. These experiments not only validated the interactions between Cd80/86 and Cd28/Ctla-4 but also revealed a dose-dependent relationship, providing robust supplementary evidence for the molecular interactions under investigation. Furthermore, in Figure 7F-G, the axis font sizes were enlarged to improve readability. Additionally, in response to reviewers' feedback, we have emphasized Figure 7H, which presents the most compelling evidence for molecular interactions, by including a standalone sentence in the text to enhance its prominence.

For Figure 7A-E, for non-immunologists, it is unclear what experiment was performed here - it would be helpful to add a 1-sentence summary of the assay to the main text or figure legend.

We apologize for this oversight. Figures 7A–E illustrate the functional assessment of the inhibitory role of Ctla-4 in Cd80/86 and Cd28-mediated T cell activation. A detailed description of the methodologies associated with Figures 7A–E is provided in the ‘Materials and Methods’ section of the revised manuscript.

For Figure 7F-G, it is extremely hard to read the heat map legends and the X and Y-axis. Also, what the heatmaps show and how that fits the overall narrative can be elaborated significantly.

We regret this oversight. To enhance clarity, we have increased the font size of the heatmap legends and the X and Y-axes, as shown in the following figure. Additionally, a detailed analysis of these figures is provided in lines 299–306 of the main text.

In general, the main text that accompanies Figure 7 should be expanded to more clearly describe these experiments/analyses and their results.

We have conducted a detailed analysis of the experiments and results presented in Figure 7. This analysis is described in lines 278-314.

Reviewer #2:

The scRNASeq assay is missing some basic characterization: how many WT and mutant fish were assayed in the experiment? how many WT and mutant cells were subject to sequencing? Before going to the immune cell types, are intestinal cell types comparable between the two conditions? Are there specific regions in the tSNE plot in Figure 4A abundant of WT or ctla-4 mutant cells?

In the experiment, we analyzed 30 wild-type and 30 mutant zebrafish for scRNA-seq, with an initial dataset comprising 8,047 cells in the wild-type group and 8,321 cells in the mutant group. Sample preparation details are provided on lines 620-652. Due to the relatively high expression of mitochondrial genes in intestinal tissue, quality control filtering yielded 3,263 cells in the wild-type group and 4,276 cells in the mutant group. Given that the intestinal tissues were dissociated using identical protocols, the resulting cell types are comparable between the two conditions. Both the wild-type and Ctla-4-deficient groups contained enterocytes, enteroendocrine cells, smooth muscle cells, neutrophils, macrophages, B cells, and a cluster of T/NK/ILC-like cells. Notably, no distinct regions were enriched for either condition in the tSNE plot (Figure 4A).

The cell proliferation experiment using PHA stimulation assay demonstrated the role of Ctla-4 in cell proliferation, while the transcriptomic evidence points towards activation rather than an overall expansion of T-cell numbers. This should be discussed towards a more comprehensive model of how subtypes of cells can be differentially proliferating in the disease model.

In the PHA-stimulated T cell proliferation assay, we aimed to investigate the regulatory roles of Ctla-4, Cd28, and Cd80/86 in T cell activation, focusing on validating Ctla-4's inhibitory function as an immune checkpoint. While our study examined general regulatory mechanisms, it did not specifically address the distinct roles of Ctla-4 in different T cell subsets. We appreciate the reviewer's suggestion to develop a more comprehensive model that elucidates differential T cell activation across various subsets in disease models. However, due to the nascent stage of research on fish T cell subsets and limitations in lineage-specific antibodies and transgenic strains, such investigations are currently challenging. We plan to pursue these studies in the future. Despite these constraints, our single-cell RNA sequencing data revealed an increased proportion of Th2 subset cells in Ctla-4-deficient zebrafish, as evidenced by elevated expression levels of Th2 markers (Il4, Il13, and Gata3) via RT-qPCR (see Figures S7B). Notably, recent studies in mouse models have shown that naïve T cells from CTLA-4-deficient mice tend to differentiate into Th2 cells post-proliferation, with activated Th2 cells secreting higher levels of cytokines like IL-4, IL-5, and IL-13, thereby exerting their effector functions (Refs. 1-2). Consequently, our findings align with observations in mice, suggesting conserved CTLA-4 functions across species. We have expanded the "Discussion" section to clarify these points.

References:

(1) Bour-Jordan H, Grogan J L, Tang Q Z, Auger J A, Locksley R M, Bluestone J A et al. (2003) CTLA-4 regulates the requirement for cytokine-induced signals in T_H2 lineage commitment. Nature Immunology 4: 182-188.

(2) Khattri Roli, Auger, Julie A, Griffin Matthew D, Sharpe Arlene H, Bluestone Jeffrey A et al. (1999) Lymphoproliferative Disorder in CTLA-4 Knockout Mice Is Characterized by CD28-Regulated Activation of Th2 Responses. The Journal of Immunology 162:5784-5791.

It would be nice if the authors could also demonstrate whether other tissues in the zebrafish have an inflammation response, to show whether the model is specific to IBD.

In addition to intestinal tissues, we also performed histological analysis on the liver of Ctla-4-deficient zebrafish. The results showed that Ctla-4 deficiency led to mild edema in a few hepatocytes, and lymphocyte infiltration was not significant. Compared to the liver, we consider intestinal inflammation to be more pronounced.

Some minor comments on terminology

(a) "multiomics" usually refers to omics experiments with different modalities (e.g. transcriptomics, proteomics, metabolomics etc), while the current paper only has transcriptomics assays. I wouldn't call it "multiomics" analysis.

We appreciate the reviewer's attention to this issue. The "multi-omics" has been revised to "transcriptomics".

(b) In several parts of the figure legend the author mentioned "tSNE nonlinear clustering" (Figures 4A and 5A). tSNE is an embedding method rather than a clustering method.

The "tSNE nonlinear clustering" has been revised to "tSNE embedding”.

(c) Figure 1E is a UMAP rather than tSNE.

The "tSNE" has been revised to "UMAP" in the figure legend in line 1043.

Reviewer #3: 

Line 28: The link is not directly reflected in this sentence describing CTLA-4 knockout mice.

We appreciate the reviewer for bringing this issue to our attention. We have expanded our description of CTLA-4 knockout mice on lines 77-84.

Line 80-83: There is a lack of details about the CTLA-4-deficient mice. The factor that Th2 response could be induced has been revealed in mouse model. See the reference entitled "CTLA-4 regulates the requirement for cytokine-induced signals in TH2 lineage commitment" published in Nature Immunology.

We thank the reviewer for providing valuable references. We have added descriptions detailing the differentiation of T cells into Th2 cells in CTLA-4-deficient mice on lines 78–81, and the relevant references have been cited in the revised manuscript.

To better introduce the CTLA-4 immunobiology, the paper entitled "Current Understanding of Cytotoxic T Lymphocyte Antigen-4 (CTLA-4) Signaling in T-Cell Biology and Disease Therapy" published in Molecules and Cells should be referred.

We have provided additional details on CTLA-4 immunology (lines 75-84) and have included the relevant reference in the revised manuscript.

In current results, there are many sentences that should be moved to the discussion, such as lines 123-124, lines 152-153, lines 199-200, and lines 206-207. So, the result sections just describe the results, and the discussions should be put together in the discussion.

We have relocated these sentences to the 'Discussion' section and refined the writing.

In the discussion, the zebrafish enteritis model, such as DSS/TNBS and SBMIE models, should also be compared with the current CTLA-4 knockout model. Also, the comparison between the current fish IBD model and the previous mouse model should also be included, to enlighten the usage of CTLA-4 knockout zebrafish IBD model.

We compared the phenotypes of our current Ctla-4-knockout zebrafish IBD model with other models, including DSS-induced IBD models in zebrafish and mice, as well as TNBS- and SBM-induced IBD models in zebrafish. The details are included in the "Discussion" section (lines 353-365).

As to the writing, the structure of the discussion is poor. The paragraphs are very long and hard to follow. Many findings from current results were not yet discussed. I just can't find any discussion about the alteration of intestinal microbiota.

In response to the reviewers' constructive feedback, we have revised and enhanced the discussion section. Furthermore, we have integrated the most recent research findings relevant to this study into the discussion to improve its relevance and comprehensiveness.

In the discussion, the aerobic-related bacteria in 16s rRNA sequencing results should be focused on echoing the histopathological findings, such as the emptier gut of CTLA-4 knockout zebrafish.

As mentioned above, the discussion section has been revised and expanded to provide a better understanding of the potential interplay among intestinal inflammatory pathology, gut microbiota alterations, and immune cell dysregulation in Ctla-4-deficient zebrafish. Furthermore, promising avenues for future research that warrant further investigation were also discussed.

In the current method, there are no descriptions for many used methods, which already generated results, such as WB, MLR, MST, Co-IP, AlphaFold2 prediction, and how to make currently used anti-zfCTLA4 antibody. Also, there is a lack of description of the method of the husbandry of knockout zebrafish line.

We regret these flaws. The methods section was inadvertently incomplete due to an error during the file upload process at submission. This issue has been rectified in the revised manuscript. Additionally, Ctla-4-deficient zebrafish were reared under the same conditions as wild-type zebrafish, and the rearing methods are now described in the "Generation of Ctla-4-deficient zebrafish" section of the Materials and Methods.

Line 360: the experimental zebrafish with different ages could be a risk for unstable intestinal health. See the reference entitled "The immunoregulatory role of fish-specific type II SOCS via inhibiting metaflammation in the gut-liver axis" published in Water Biology and Security. The age-related differences in zebrafish could be observed in the gut.

We appreciate the reviewers' reminders. The Ctla-4 mutant zebrafish used in our experiments were 4 months old, while the wild-type zebrafish ranged from 4 to 6 months old. These experimental fish were relatively young and uniformly distributed in age. During our study, we examined the morphological structures of the intestines in zebrafish aged 4 to 6 months and observed no significant abnormalities. These findings align with previous research indicating no significant difference in intestinal health between 3-month-old and 6-month-old wild-type zebrafish (Ref. 1). Consequently, we conclude that there is no notable aging-related change in the intestines of zebrafish aged 4 to 6 months. This reduces the risk associated with age-related variables in our study. We have added an explanation stating that the Ctla-4 mutant zebrafish used in the experiments were 4 months old (Line 449) in the revised manuscript.

Reference

(1) Shan Junwei, Wang Guangxin, Li Heng, Zhao Xuyang et al. (2023) The immunoregulatory role of fish-specific type II SOCS via inhibiting metaflammation in the gut-liver axis. Water Biology and Security 2: 100131-100144.

Section "Generation of Ctla-4-deficient zebrafish": There is a lack of description of PCR condition for the genotyping.

The target DNA sequence was amplified at 94 °C for 4 min, followed by 35 cycles at 94°C for 30 s, 58°C for 30 s and 72°C for 30 s, culminating in a final extension at 72 °C for 10 min. The polymerase chain reaction (PCR) conditions are described in lines 458-460.

How old of the used mutant fish? There should be a section "sampling" to provide the sampling details.

The "Sampling" information has been incorporated into the "Materials and Methods" section of the revised manuscript. Wild-type and Ctla-4-deficient zebrafish of varying months were housed in separate tanks, each labeled with its corresponding birth date. Experiments utilized Ctla-4-deficient zebrafish aged 4 months and wild-type zebrafish aged between 4 to 6 months.

Line 378-380: The index for the histopathological analysis should be detailed, rather than just provide a reference. I don't think these indexes are good enough to specifically describe the pathological changes of intestinal villi and mucosa. It is suggested to improve with detailed parameters. As described in the paper entitled "Pathology of Gastric Intestinal Metaplasia: Clinical Implications" published in Am J Gastroenterol., histochemical, normal gastric mucins are pH neutral, and they stain magenta with periodic acid-Schiff (PAS). In an inflamed gut, acid mucins replace the original gastric mucins and are stained blue with Alcian blue (AB). So, to reveal the pathological changes of goblet cells and involved mucin components, AB staining should be added. Also, for the number of goblet cells in the inflammatory intestine, combining PAS and AB staining is the best way to reveal all the goblet cells. In Figure 2, there were very few goblet cells. The infiltration of lymphocytes and the empty intestinal lumen could be observed. Thus, the ratio between the length of intestinal villi and the intestinal ring radius should calculated.

In response to the reviewers’ valuable suggestions, we have augmented the manuscript by providing additional parameters related to the pathological changes observed in the Ctlta-4-deficient zebrafish intestines, including the mucin component changes identified through PAS and AB-PAS staining, the variations in the number of goblet cells evaluated by AB-PAS staining, and the ratio of intestinal villi length to the intestinal ring radius, as illustrated in the following figures. These new findings are detailed in the "Materials and Methods" (lines 563-566) and "Results" (lines 143-146) sections, along with Supplementary Figure S3 of the revised manuscript.

Section "Quantitative real-time PCR": What's the machine used for qPCR? How about the qPCR validation of RNA seq data? I did not see any related description of data and methods for qPCR validation. In addition, beta-actin is not a stable internal reference gene, to analyze inflammation and immune-related gene expression. See the reference entitled "Actin, a reliable marker of internal control?" published in Clin Chim Acta. Other stable housekeeping genes, such as EF1alpha and 18s, could be better internal references.

RT-qPCR experiments were conducted using a PCR thermocycler device (CFX Connect Real-Time PCR Detection System with Precision Melt Analysis^TM Software, Bio-Rad, Cat. No. 1855200EM1). This information has been incorporated into lines 608-610 of the "Materials and Methods" section. In these experiments, key gene sequences of interest, including il13, mpx, and il1β, were extracted from RNA-seq data for RT-qPCR validation. To ensure accurate normalization, potential internal controls were evaluated, and β-actin was identified as a suitable candidate due to its consistent expression levels in the intestines of both wild-type and Ctla-4-deficient zebrafish. The use of β-actin as an internal control is further supported by its application in recent studies on intestinal inflammation (Refs 1–2).

References:

(1) Tang Duozhuang, Zeng Ting, Wang Yiting, Cui Hui et al. (2020) Dietary restriction increases protective gut bacteria to rescue lethal methotrexate-induced intestinal toxicity. Gut Microbes 12: 1714401-1714422.

(2) Malik Ankit, Sharma Deepika et al. (2023) Epithelial IFNγ signaling and compartmentalized antigen presentation orchestrate gut immunity. Nature 623: 1044-1052.

How to generate sCtla-4-Ig, Cd28-Ig and Cd80/86? No method could be found.

We apologize for the omission of these methods. The detailed protocols have now been added to the "Materials and Methods" section of the revised manuscript (lines 464-481).

Figure 5: As reviewed in the paper entitled "Teleost T and NK cell immunity" published in Fish and Shellfsh Immunology, two types of NK cell homologues have been described in fish: non-specific cytotoxic cells and NK-like cells. There is no NKT cell identified in the teleost yet. Therefore, "NKT-like" could be better to describe this cell type.

We refer to "NKT" cells as "NKT-like" cells, as suggested.

For the supplementary data of scRNA-seq, there lacks the details of expression level.

The expression levels of the corresponding genes are provided in Supplemental Table 4.

Supplemental Table 1: There are no accession numbers of amplified genes.

The accession numbers of the amplified genes are included in Supplemental Table 1.

The English needs further editing.

We have made efforts to enhance the English to meet the reviewers' expectations.

Line 32: The tense should be the past.

This tense error has been corrected.

Line 363-365: The letter of this approval should be provided as an attachment.

The approval document is provided as an attachment.

Line 376: How to distinguish the different intestinal parts? Were they judged as the first third, second third, and last third parts of the whole intestine?

The differences among the three segments of zebrafish intestine are apparent. The intestinal tube narrows progressively from the anterior to the mid-intestine and then to the posterior intestine. Moreover, the boundaries between the intestinal segments are well-defined, facilitating the isolation of each segment.

Line 404: Which version of Cytoscape was used?

The version of Cytoscape used in this study is 3.9.1. Information about the Cytoscape version is provided on line 603.

The product information of both percoll and cell strainer should be provided.

The information regarding Percoll and cell strainers has been added on lines 626 and 628, respectively.

Line 814: Here should be a full name to tell what is MST.

The acronym MST stands for "Microscale Thermophoresis", a technique that has been referenced on lines 1157-1158.

Author response:

The following is the authors’ response to the original reviews

eLife Assessment

This study presents valuable findings related to seasonal brain size plasticity in the Eurasian common shrew (Sorex araneus), which is an excellent model system for these studies. The evidence supporting the authors' claims is convincing. However, the authors should be careful when applying the term adaptive to the gene expression changes they observe; it would be challenging to demonstrate the differential fitness effects of these gene expression changes. The work will be of interest to biologists working on neuroscience, plasticity, and evolution.

We appreciate the reviewers’ suggestions and comments. For the phylogenetic ANOVA we used (EVE), which tests for a separate RNA expression optimum specific to the shrew lineage consistent with expectations for adaptive evolution of gene expression. But, as you noted, while this analysis highlights many candidate genes evolving in a manner consistent with positive selection, further functional validation is required to confirm if and how these genes contribute to Dehnel’s phenomenon. In the discussion, we now emphasize that inferred adaptive expression of these genes is putative and outline that future studies are needed to test the function of proposed adaptations. For example, cell line validations of BCL2L1 on apoptosis is a case study that tests the function of a putatively adaptive change in gene expression, and it illuminates this limitation. We also have refined our discussion to focus more on pathway-level analyses rather than on individual genes, and have addressed other issues presented, including clarity of methods and using sex as a covariate in our analyses.

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this paper, Thomas et al. set out to study seasonal brain gene expression changes in the Eurasian common shrew. This mammalian species is unusual in that it does not hibernate or migrate but instead stays active all winter while shrinking and then regrowing its brain and other organs. The authors previously examined gene expression changes in two brain regions and the liver. Here, they added data from the hypothalamus, a brain region involved in the regulation of metabolism and homeostasis. The specific goals were to identify genes and gene groups that change expression with the seasons and to identify genes with unusual expression compared to other mammalian species. The reason for this second goal is that genes that change with the season could be due to plastic gene regulation, where the organism simply reacts to environmental change using processes available to all mammals. Such changes are not necessarily indicative of adaptation in the shrew. However, if the same genes are also expression outliers compared to other species that do not show this overwintering strategy, it is more likely that they reflect adaptive changes that contribute to the shrew's unique traits.

The authors succeeded in implementing their experimental design and identified significant genes in each of their specific goals. There was an overlap between these gene lists. The authors provide extensive discussion of the genes they found.

The scope of this paper is quite narrow, as it adds gene expression data for only one additional tissue compared to the authors' previous work in a 2023 preprint. The two papers even use the same animals, which had been collected for that earlier work. As a consequence, the current paper is limited in the results it can present. This is somewhat compensated by an expansive interpretation of the results in the discussion section, but I felt that much of this was too speculative. More importantly, there are several limitations to the design, making it hard to draw stronger conclusions from the data. The main contribution of this work lies in the generated data and the formulation of hypotheses to be tested by future work.

Thank you for your interest in our manuscript and for your insights. We addressed your comments below: we now highlight the limitations of our study design in the discussion and emphasize that, while a second optimum of gene expression in shrews is consistent with adaptive evolution, we recognize that not all sources of variation in gene expression can be fully accounted for. We highlight the putative nature of these results in our revisions, especially in our new limitations section (lines 541-555).

Strengths:

The unique biological model system under study is fascinating. The data were collected in a technically sound manner, and the analyses were done well. The paper is overall very clear, well-written, and easy to follow. It does a thorough job of exploring patterns and enrichments in the various gene sets that are identified.

I specifically applaud the authors for doing a functional follow-up experiment on one of the differentially expressed genes (BCL2L1), even if the results did not support the hypothesis. It is important to report experiments like this and it is terrific to see it done here.

We are glad to hear that you found our manuscript fascinating and clearly written. While we hoped to see an effect of BCL2L1 on apoptosis as proposed, we agree that reporting null results is valuable when validating evolutionary inferences.

Weaknesses:

While the paper successfully identifies differentially expressed seasonal genes, the real question is (as explained by the authors) whether these are evolved adaptations in the shrews or whether they reflect plastic changes that also exist in other species. This question was the motivation for the inter-species analyses in the paper, but in my view, these cannot rigorously address this question. Presumably, the data from the other species were not collected in comparable environments as those experienced by the shrews studied here. Instead, they likely (it is not specified, and might not be knowable for the public data) reflect baseline gene expression. To see why this is problematic, consider this analogy: if we were to compare gene expression in the immune system of an individual undergoing an acute infection to other, uninfected individuals, we would see many, strong expression differences. However, it would not be appropriate to claim that the infected individual has unique features - the relevant physiological changes are simply not triggered in the other individuals. The same applies here: it is hard to draw conclusions from seasonal expression data in the shrews to non-seasonal data in the other species, as shrew outlier genes might still reflect physiological changes that weren't active in the other species.

There is no solution for this design flaw given the public data available to the authors except for creating matched data in the other species, which is of course not feasible. The authors should acknowledge and discuss this shortcoming in the paper.

Thank you for taking the time to provide such insightful feedback. As you noted, whiles shrews experience seasonal size changes, their environments may differ from the other species used in this experiment, leading to increased or decreased expression of certain genes and reducing our ability accurately detect selection across the phylogeny. Although we sought to control for as many sources of variation as possible, such as using only post-pubescent, wild, or non-domesticated individuals when feasible, we recognize that not all sources of variation can be fully accounted for within a practical experiment. We agree that these sources of variation can introduce both false positives and negatives into our results, and we have now highlighted this limitation within our discussion (lines 538-552).

Related to the point above: in the section "Evolutionary Divergence in Expression" it is not clear which of the shrew samples were used. Was it all of them, or only those from winter, fall, etc? One might expect different results depending on this. E.g., there could be fewer genes with inferred adaptive change when using only summer samples. The authors should specify which samples were included in these analyses, and, if all samples were used, conduct a robustness analysis to see which of their detected genes survive the exclusion of certain time points.

Thank you for this attention to detail. We used spring adults for this analysis. This decision was made as only used post pubescent individuals for all species in the analysis, and this was the only season where adult shrews were going through Dehnel’s phenomenon. We have now clarified this in both the methods and results (line 247 and line 667)

In the same section, were there also genes with lower shrew expression? None are mentioned in the text, so did the authors not test for this direction, or did they test and there were no significant hits?

We did test for decreased shrew expression compared to the rest of the species, but there were no significant genes with significant decreases. We hypothesize that there are two potential reasons for this results; 1) If a gene were to be selected for decreased expression, selection for constitutive expression of the gene across all species may be weak, and thus found in other lineages as well, or 2) decreased or no expression may relax selection on the coding regions, and thus these genes are not pulled out as we identify 1:1 orthologs. This is consistent with results provided from the original methods manuscript. Thank you for pointing out that we did not discuss this information in the text, and we now include it in our results (lines 250-251).

The Discussion is too long and detailed, given that it can ultimately only speculate about what the various expression changes might mean. Many of the specific points made (e.g. about the blood-brain-barrier being more permissive to sensing metabolic state, about cross-organ communication, the paragraphs on single, specific genes) are a stretch based on the available data. Illustrating this point, the one follow-up experiment the authors did (on BCL2L1) did not give the expected result. I really applaud the authors for having done this experiment, which goes beyond typical studies in this space. At the same time, its result highlights the dangers of reading too much into differential expression analyses.

We agree with your point, while our extensive discussion is useful for testing future hypotheses, ultimately some of the discussion may be too speculative for our readers. To amend this, we have reduced some portions of our discussion and focused more on pathways than individual genes, including removing mechanisms related to HRH2, FAM57B, GPR3, and GABAergic neurons. We hope that this highlights to the reader the speculative nature of many of our results.

There is no test of whether the five genes observed in both analyses (seasonal change and inter-species) exceed the number expected by chance. When two gene sets are drawn at random, some overlap is expected randomly. The expected overlap can be computed by repeated draws of pairs of random sets of the same size as seen in real data and by noting the overlap between the random pairs. If this random distribution often includes sets of five genes, this weakens the conclusions that can be drawn from the genes observed in the real data.

Thank you for highlighting this approach, it is greatly needed. After running this test, we found that observed overlapping genes were more than the expected overlap, yet not significant. We now show this in our methods (lines 277-278) and results (lines 719-720).

Reviewer #2 (Public review):

Summary:

Shrews go through winter by shrinking their brain and most organs, then regrow them in the spring. The gene expression changes underlying this unusual brain size plasticity were unknown. Here, the authors looked for potential adaptations underlying this trait by looking at differential expression in the hypothalamus. They found enrichments for DE in genes related to the blood-brain barrier and calcium signaling, as well as used comparative data to look at gene expression differences that are unique in shrews. This study leverages a fascinating organismal trait to understand plasticity and what might be driving it at the level of gene expression. This manuscript also lays the groundwork for further developing this interesting system.

We are glad you found our manuscript interesting and thank and thank you for your feedback. We hope that we have addressed all of your concerns as described below.

Strengths:

One strength is that the authors used OU models to look for adaptation in gene expression. The authors also added cell culture work to bolster their findings.

Weaknesses:

I think that there should be a bit more of an introduction to Dehnel's phenomenon, given how much it is used throughout.

Thank you for this insight. With a lengthy introduction and discussion, we agree that the importance of Dehnel’s phenomenon may have been overshadowed. We have shortened both sections and emphasized the background on Dehnel’s phenomenon in the first two paragraphs of the introduction, allowing this extraordinary seasonal size plasticity to stand out.

Reviewer #3 (Public review):

Summary:

In their study, the authors combine developmental and comparative transcriptomics to identify candidate genes with plastic, canalized, or lineage-specific (i.e., divergent) expression patterns associated with an unusual overwintering phenomenon (Dehnel's phenomenon - seasonal size plasticity) in the Eurasian shrew. Their focus is on the shrinkage and regrowth of the hypothalamus, a brain region that undergoes significant seasonal size changes in shrews and plays a key role in regulating metabolic homeostasis. Through combined transcriptomic analysis, they identify genes showing derived (lineage-specific), plastic (seasonally regulated), and canalized (both lineage-specific and plastic) expression patterns. The authors hypothesize that genes involved in pathways such as the blood-brain barrier, metabolic state sensing, and ion-dependent signaling will be enriched among those with notable transcriptomic patterns. They complement their transcriptomic findings with a cell culture-based functional assessment of a candidate gene believed to reduce apoptosis.

Strengths:

The study's rationale and its integration of developmental and comparative transcriptomics are well-articulated and represent an advancement in the field. The transcriptome, known for its dynamic and plastic nature, is also influenced by evolutionary history. The authors effectively demonstrate how multiple signals-evolutionary, constitutive, and plastic-can be extracted, quantified, and interpreted. The chosen phenotype and study system are particularly compelling, as it not only exemplifies an extreme case of Dehnel's phenotype, but the metabolic requirements of the shrew suggest that genes regulating metabolic homeostasis are under strong selection.

Weaknesses:

(1) In a number of places (described in detail below), the motivation for the experimental, analytical, or visualization approach is unclear and may obscure or prevent discoveries.

Thank you for finding our research and manuscript compelling, as well as the valuable feedback that will drastically improve our manuscript. We hope that we have alleviated your concerns below by following your instructions below.

(2) Temporal Expression - Figure 1 and Supplemental Figure 2 and associated text:

- It is unclear whether quantitative criteria were used to distinguish "developmental shift" clusters from "season shift" clusters. A visual inspection of Supplemental Figure 2 suggests that some clusters (e.g., clusters 2, 8, and to a lesser extent 12) show seasonal variation, not just developmental differences between stages 1 and 2. While clustering helps to visualize expression patterns, it may not be the most appropriate filter in this case, particularly since all "season shift" clusters are later combined in KEGG pathway and GO analyses (Figure 1B).

- The authors do not indicate whether they perform cluster-specific GO or KEGG pathway enrichment analyses. The current analysis picks up relevant pathways for hypothalamic control of homeostasis, which is a useful validation, but this approach might not fully address the study's key hypotheses.

Thank you for this valuable feedback. We did not want to include clusters we deemed to be related to development, as this should not be attributed to changes associated with Dehnel’s phenomenon. We did this through qualitative, visual inspection, which we realize can differ between parties (i.e., clusters 2, 8, and 12 appeared to be seasonal). Qualitatively, we were looking for extreme divergence between Stage 1 and Stage 5 individuals, as expression was related to season and not development, then the average of these stages within cluster should be relatively similar. We have now quantified this as large differences in z-score (abs(summer juvenile-summer adult)>1.25) without meaningful interseason variations determined by a second local maximum (abs(autumn-winter)<0.5 and abs(winter-summer)<0.5)), and added it both our methods (lines 699-702) and results (line 192).

Regarding the combination of clusters for pathway enrichment compared to individual pathways, we agree that combining clusters may be more informative for overall homeostasis, compared to individual clusters which may inform us on processes directly related to Dehnel’s phenomenon. Initially, we were tentative to conduct this analysis, as clusters contain small gene sets, reducing the ability to detect pathway enrichments. We have now included this analysis, which is reported in our methods (lines 703-704), results (lines 203-204)., and new supplemental table.

(3) Differential expression between shrinkage (stage 2) and regrowth (stage 4) and cell culture targets

- The rationale for selecting BCL2L1 for cell culture experiments should be clarified. While it is part of the apoptosis pathway, several other apoptosis-related genes were identified in the differential gene expression (DGE) analysis, some showing stronger differential expression or shrew-specific branch shifts. Why was BCL2L1 prioritized over these other candidates?

We agree that our rationale for validating BCL2L1 function in neural cell lines was not clearly explained in the manuscript. We selected BCL2L1 because it is the furthest downstream gene in the apoptotic pathway, thus making it the most directly involved gene in programmed cell death, whereas upstream genes could influence additional genes or alternative processes. We have clarified this choice in the revised methods section (lines 748-750).

- The authors mention maintaining (or at least attempting to maintain) a 1:1 sex ratio for the comparative analysis, but it is unclear if this was also done for the S. araneus analysis. If not, why? If so, was sex included as a covariate (e.g., a random effect) in the differential expression analysis? Sex-specific expression elevates with group variation and could impact the discovery of differentially expressed genes.

Regarding the use of sex as a covariate, we acknowledge the concerns raised. In our evolutionary analyses, we maintained a balanced sex ratio within species when possible. EVE models handle the effect of sex on gene expression as intraspecific variation. In shrews, however, we used males exclusively, as females were only found among juvenile individuals. Including those juvenile females would have introduced age effects, with perhaps a larger effect on our results. For the seasonal data, we have now included sex as a covariate in differential expression analyses. However, our design is imbalanced in relation to sex, which we have now discussed in our methods (lines 713-714) and discussion limitations (lines 544-548).

(4) Discussion: The term "adaptive" is used frequently and liberally throughout the discussion. The interpretation of seasonal changes in gene expression as indicators of adaptive evolution should be done cautiously as such changes do not necessarily imply causal or adaptive associations.

Thank you for this insight. We have reviewed our discussion and clarified that adaptations are putative (i.e. lines 146, 285, and 332), and highlighted this in our limitations section.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) I would recommend always spelling out "Dehnel's phenomenon" or even replacing this term (after crediting the DP term) with the more informative "seasonal size plasticity". Every time I saw "DP", I had to remind myself what this referred to. If the authors choose not to do so, please use the acronym consistently (e.g. line 186 has it spelled out).

We have replaced the acronym DP with either the full term or the more informative “seasonal size plasticity” throughout the text.

(2) Line 202: "DEG" has not been defined. Simply add to the line before.

Thank you for this attention to detail. We have added this to the line above (210).

(3) Please add a reference for the "AnAge" tool that was used to determine if samples were pubescent.

Thank you for identifying this oversight. We have now cited the proper paper in line 634.

(4) In the BCL2L1 section in the results, add a callout to Figure 2D.

We have now added a callout to Figure 2D within the results (line 234).

Reviewer #2 (Recommendations for the authors):

(1) Line 122: is associated? These adaptations?

Thank you for identifying that we were missing the words “associated with” here. We have fixed this in the revision.

(2) The first paragraph of the Results should be moved to the methods, except maybe the number of orthologs.

Thank you for this insight. We have removed this portion from the results section.

(3) Why a Bonferroni correction on line 188? That seems too strict.

We agree the Bonferroni correction is strict. Results when using other less strict methods for controlling false discovery rate are also not significant after correction. These corrections can be found within the data, however, we only report on the Bonferroni correction.

(4) Line 427: "is a novel candidate gene for several neurological disorders" needs some references. I see them a couple of sentences later, but that's quite a sentence with no references at the end.

We have added the proper citations for this sentence (line 524).

Reviewer #3 (Recommendations for the authors):

(1) Temporal Expression - Figure 1 and Supplemental Figure 2 and associated text Line176-193:

- The authors report the total number of genes meeting inclusion criteria (>0.5-fold change between any two stages and 2 samples >10 normalized reads), but it would be more informative to also provide the number of genes within each temporal cluster. This would offer a clearer understanding of how gene expression patterns are distributed over time.

Unfortunately, this information is difficult to depict on our figure and would use too much space in the text. We have thus added a description of the range of genes in a new supplemental table depicting this information.

- It is unclear whether quantitative criteria were used to distinguish "developmental shift" clusters from "season shift" clusters. A visual inspection of Supplemental Figure 2 suggests that some clusters (e.g., clusters 2, 8, and to a lesser extent 12) show seasonal variation, not just developmental differences between stages 1 and 2. While clustering helps to visualize expression patterns, it may not be the most appropriate filter in this case, particularly since all "season shift" clusters are later combined in KEGG pathway and GO analyses (Fig. 1B). Using a differential gene expression criterion might be more suitable. For example, do excluded genes show significant log-fold differences between late-stage comparisons?

As previously mentioned, we have now quantified seasonal shifts as large differences in z-score (abs(summer juveniles-summer adults)>1.25) without meaningful interseason variations determined by a second local maximum (abs(autumn-winter)<0.5 and abs(winter-summer)<0.5)), and added it to our methods (lines 699-702).  We then follow this up with differential expression analyses as described in Figure 2.

- Did the authors perform cluster-specific GO or KEGG pathway enrichment analyses instead of focusing on the combined set of genes across the season shift clusters? While I understand that the small number of genes in each cluster may be limiting, if pathways emerge from cluster-specific analysis, they could provide more detailed insights into the functional significance of these temporal expression patterns. The current analysis picks up relevant pathways for hypothalamic control of homeostasis, which is a useful validation, but this approach might not fully address the study's key hypotheses. Additionally, no corrections for multiple hypothesis testing were applied, as noted in the results. A more refined gene set (e.g., using differential expression criteria, described above) could be more appropriate for these analyses.

We have now included cluster-specific KEGG enrichments as previously described.

(2) Differential expression between shrinkage (stage 2) and regrowth (stage 4) and cell culture targets - Figure 2 and lines195-227:

- The rationale for selecting BCL2L1 for cell culture experiments should be clarified. While it is part of the apoptosis pathway, several other apoptosis-related genes were identified in the differential gene expression (DGE) analysis, some showing stronger differential expression or shrew-specific branch shifts. Why was BCL2L1 prioritized over these other candidates?

We have now included the reasoning for further validation of BCL2L1 as described above.

- The relevance of the "higher degree" differentially expressed genes needs more explanation. Although this group of genes is highlighted in the results, they are not featured in any subsequent analyses, leaving their importance unclear.

Thank you for this insight. We have removed this from the methods as it is not relevant to subsequent analyses or conclusions.

- The authors mention maintaining (or at least attempting to maintain) a 1:1 sex ratio for the comparative analysis (Line 525), but it is unclear if this was also done for the S. araneus analysis. If so, was sex included as a covariate (e.g., a random effect) in the differential expression analysis?

We have now incorporated information on sex as described above.

(3) Discussion:

The term "adaptive" is used frequently and liberally throughout the discussion, but the authors should be cautious in interpreting seasonal changes in gene expression as indicators of adaptive evolution. Such changes do not necessarily imply causal or adaptive associations, and this distinction should be clearly stated when discussing the results.

Thank you for this feedback and we agree with your conclusion, while a second expression optimum in the shrew lineage is indicative of adaptive expression, we cannot fully determine whether these are caused by genetic or environmental factors, despite careful attention to experimental design. We have highlighted this as a limitation in the discussion.

(4) Minor Editorial Comment:

Line 105: "... maintenance of an energy budgets..." delete "an"

We have removed this grammatical error.

Author response:

(This author response relates to the first round of peer review by Biophysics Colab. Reviews and responses to both rounds of review are available here: https://sciety.org/articles/activity/10.1101/2023.10.23.563601.)

General Assessment:

Pannexin (Panx) hemichannels are a family of heptameric membrane proteins that form pores in the plasma membrane through which ions and relatively large organic molecules can permeate. ATP release through Panx channels during the process of apoptosis is one established biological role of these proteins in the immune system, but they are widely expressed in many cells throughout the body, including the nervous system, and likely play many interesting and important roles that are yet to be defined. Although several structures have now been solved of different Panx subtypes from different species, their biophysical mechanisms remain poorly understood, including what physiological signals control their activation. Electrophysiological measurements of ionic currents flowing in response to Panx channel activation have shown that some subtypes can be activated by strong membrane depolarization or caspase cleavage of the C-terminus. Here, Henze and colleagues set out to identify endogenous activators of Panx channels, focusing on the Panx1 and Panx2 subtypes, by fractionating mouse liver extracts and screening for activation of Panx channels expressed in mammalian cells using whole-cell patch clamp recordings. The authors present a comprehensive examination with robust methodologies and supporting data that demonstrate that lysophospholipids (LPCs) directly Panx-1 and 2 channels. These methodologies include channel mutagenesis, electrophysiology, ATP release and fluorescence assays, molecular modelling, and cryogenic electron microscopy (cryo-EM). Mouse liver extracts were initially used to identify LPC activators, but the authors go on to individually evaluate many different types of LPCs to determine those that are more specific for Panx channel activation. Importantly, the enzymes that endogenously regulate the production of these LPCs were also assessed along with other by-products that were shown not to promote pannexin channel activation. In addition, the authors used synovial fluid from canine patients, which is enriched in LPCs, to highlight the importance of the findings in pathology. Overall, we think this is likely to be a landmark study because it provides strong evidence that LPCs can function as activators of Panx1 and Panx2 channels, linking two established mediators of inflammatory responses and opening an entirely new area for exploring the biological roles of Panx channels. Although the mechanism of LPC activation of Panx channels remains unresolved, this study provides an excellent foundation for future studies and importantly provides clinical relevance.

We thank the reviewers for their time and effort in reviewing our manuscript. Based on their valuable comments and suggestions, we have made substantial revisions. The updated manuscript now includes two new experiments supporting that lysophospholipid-triggered channel activation promotes the release of signaling molecules critical for immune response and demonstrates that this novel class of agonist activates the inflammasome in human macrophages through endogenously expressed Panx1. To better highlight the significance of our findings, we have excluded the cryo-EM panel from this manuscript. We believe these changes address the main concerns raised by the reviewers and enhance the overall clarity and impact of our findings. Below, we provide a point-by-point response to each of the reviewers’ comments.

Recommendations:

(1) The authors present a tremendous amount of data using different approaches, cells and assays along with a written presentation that is quite abbreviated, which may make comprehension challenging for some readers. We would encourage the authors to expand the written presentation to more fully describe the experiments that were done and how the data were analysed so that the 2 key conclusions can be more fully appreciated by readers. A lot of data is also presented in supplemental figures that could be brought into the main figures and more thoroughly presented and discussed.

We appreciate and agree with the reviewers’ observation. Our initial manuscript may have been challenging to follow due to our use of both wild-type and GS-tagged versions of Panx1 from human and frog origins, combined with different fluorescence techniques across cell types. In this revision, we used only human wild-type Panx1 expressed in HEK293S GnTI- cells, except for activity-guided fractionation experiments, where we used GS-tagged Panx1 expressed in HEK293 cells (Fig. 1). For functional reconstitution studies, we employed YO-PRO-1 uptake assays, as optimizing the Venus-based assay was challenging. We have clarified these exceptions in the main text. We think these adjustments simplify the narrative and ensure an appropriate balance between main and supplemental figures.

(2) It would also be useful to present data on the ion selectivity of Panx channels activated by LPC. How does this compare to data obtained when the channel is activated by depolarization? If the two stimuli activate related open states then the ion selectivity may be quite similar, but perhaps not if the two stimuli activate different open states. The authors earlier work in eLife shows interesting shifts in reversal potentials (Vrev) when substituting external chloride with gluconate but not when substituting external sodium with N-methyl-D-glucamine, and these changed with mutations within the external pore of Panx channels. Related measurements comparing channels activated by LPC with membrane depolarization would be valuable for assessing whether similar or distinct open states are activated by LPC and voltage. It would be ideal to make Vrev measurements using a fixed step depolarization to open the channel and then various steps to more negative voltages to measure tail currents in pinpointing Vrev (a so called instantaneous IV).

We fully agree with the reviewer on the importance of ion selectivity experiments. However, comparing the properties of LPC-activated channels with those activated by membrane depolarization presented technical challenges, as LPC appears to stimulate Panx1 in synergy with voltage. Prolonged LPC exposure destabilizes patches, complicating G-V curve acquisition and kinetic analyses. While such experiments could provide mechanistic insights, we think they are beyond the scope of current study.

(3) Data is presented for expression of Panx channels in different cell types (HEK vs HEKS GnTI-) and different constructs (Panx1 vs Panx1-GS vs other engineered constructs). The authors have tried to be clear about what was done in each experiment, but it can be challenging for the reader to keep everything straight. The labelling in Fig 1E helps a lot, and we encourage the authors to use that approach systematically throughout. It would also help to clearly identify the cell type and channel construct whenever showing traces, like those in Fig 1D. Doing this systematically throughout all the figures would also make it clear where a control is missing. For example, if labelling for the type of cell was included in Fig 1D it would be immediately clear that a GnTI- vector alone control for WT Panx1 is missing as the vector control shown is for HEK cells and formally that is only a control for Panx2 and 3. Can the authors explain why PLC activates Panx1 overexpressed in HEK293 GnTl- cells but not in HEK293 cells? Is this purely a function of expression levels? If so, it would be good to provide that supporting information.

As mentioned above, we believe our revised version is more straightforward to digest. We have improved labeling and provided explanations where necessary to clarify the manuscript. While Panx1 expression levels are indeed higher in GnTI- than in HEK293 cells, we are uncertain whether the absence of detectable currents in HEK293 cells is solely due to expression levels. Some post-translational modifications that inhibit Panx1, such as lysine acetylation, may also impact activity. Future studies are needed to explore these mechanisms further.

(4) The mVenus quenching experiments are somewhat confusing in the way data are presented. In Fig 2B the y axis is labelled fluorescence (%) but when the channel is closed at time = 0 the value of fluorescence is 0 rather than 100 %, and as the channel opens when LPC is added the values grow towards 100 instead of towards 0 as iodide permeates and quenches. It would be helpful if these types of data could be presented more intuitively. Also, how was the initial rate calculated that is plotted in Fig 2C? It would be helpful to show how this is done in a figure panel somewhere. Why was the initial rate expressed as a percent maximum, what is the maximum and why are the values so low? Why is the effect of CBX so weak in these quenching experiments with Panx1 compared to other assays? This assay is used in a lot of experiments so anything that could be done to bolster confidence is what it reports on would be valuable to readers. Bringing in as many control experiments that have been done, including any that are already published, would be helpful.

We modified the Y-axis in Figure 2 to “Quench (%)” for clarity. The data reflects fluorescence reduction over time, starting from LPC addition, normalized to the maximal decrease observed after Triton-X100 addition (3 minutes), enabling consistent quenching value comparisons. Although the quenching value appears small, normalization against complete cell solubilization provides reproducible comparisons. We do not fully understand why CBX effects vary in Venus quenching experiments, but we speculate that its steroid-like pentacyclic structure may influence the lysophospholipid agonistic effects. As noted in prior studies (DOI: 10.1085/jgp.201511505; DOI: 10.7554/eLife.54670), CBX likely acts as an allosteric modulator rather than a simple pore blocker, potentially contributing to these variations.

(5) Could provide more information to help rationalize how Yo-Pro-1, which has a charge of +2, can permeate what are thought to be anion favouring Panx channels? We appreciate that the biophysical properties of Panx channel remain mysterious, but it would help to hear how a bit more about the authors thinking. It might also help to cite other papers that have measured Yo-Pro-1 uptake through Panx channels. Was the Strep-tagged construct of Panx1 expressed in GnTI- cells and shown to be functional using electrophysiology?

Our recent study suggest that the electrostatic landscape along the permeation pathway may influence its ion selectivity (DOI: 10.1101/2024.06.13.598903). However, we have not yet fully elucidated how Panx1 permeates both anions and cations. Based on our findings, ion selectivity may vary with activation stimulus intensity and duration. Cation permeation through Panx1 is often demonstrated with YO-PRO-1, which measures uptake over minutes, unlike electrophysiological measurements conducted over milliseconds to seconds. We referenced two representative studies employing YO-PRO-1 to assess Panx1 activity. Whole-cell current measurements from a similar construct with an intracellular loop insertion indicate that our STREP-tagged construct likely retains functional capacity.

(6) In Fig 5 panel C, data is presented as the ratio of LPC induced current at -60 mV to that measured at +110 mV in the absence of LPC. What is the rationale for analysing the data this way? It would be helpful to also plot the two values separately for all of the constructs presented so the reader can see whether any of the mutants disproportionately alter LPC induced current relative to depolarization activated current. Also, for all currents shown in the figures, the authors should include a dashed coloured line at zero current, both for the LPC activated currents and the voltage steps.

We used the ratio of LPC-induced current to the current measured at +110 mV to determine whether any of the mutants disproportionately affect LPC-induced current relative to depolarization-activated current. Since the mutants that did not respond to LPC also exhibited smaller voltage-stimulated currents than those that did respond, we reasoned that using this ratio would better capture the information the reviewer is suggesting to gauge. Showing the zero current level may be helpful if the goal was to compare basal currents, which in our experience vary significantly from patch to patch. However, since we are comparing LPC- and voltage-induced currents within the same patch, we believe that including basal current measurements would not add useful information to our study.

Given that new experiments included to further highlight the significance of the discovery of Panx1 agonists, we opted to separate structure-based mechanistic studies from this manuscript and removed this experiment along with the docking and cryo-EM studies.

(7) The fragmented NTD density shown in Fig S8 panel A may resemble either lipid density or the average density of both NTD and lipid. For example, Class7 and Class8 in Fig.S8 panel D displayed split densities, which may resemble a phosphate head group and two tails of lipid. A protomer mask may not be the ideal approach to separate different classes of NTD because as shown in Fig S8 panel D, most high-resolution features are located on TM1-4, suggesting that the classification was focused on TM1-4. A more suitable approach would involve using a smaller mask including NTD, TM1, and the neighbouring TM2 region to separate different NTD classes.

We agree with the reviewer and attempted 3D classification using multiple smaller masks including the suggested region. However, the maps remained poorly defined, and we were unable to confidently assign the NTD.

(8) The authors don’t discuss whether the LPC-bound structures display changes in the external part of the pore, which is the anion-selective filter and the narrower part of the pore. If there are no conformational changes there, then the present structures cannot explain permeability to large molecules like ATP. In this context, a plot for the pore dimension will be helpful to see differences along the pore between their different structures. It would also be clearer if the authors overlaid maps of protomers to illustrate differences at the NTD and the "selectivity filter."

Both maps show that the narrowest constriction, formed by W74, has a diameter of approximately 9 Å. Previous steered molecular dynamics simulations suggest that ATP can permeate through such a constriction, implying an ion selection mechanism distinct from a simple steric barrier.

(9) The time between the addition of LPC to the nanodisc-reconstituted protein and grid preparation is not mentioned. Dynamic diffusion of LPC could result in equal probabilities for the bound and unbound forms. This raises the possibility of finding the Primed state in the LPC-bound state as well. Additionally, can the authors rationalize how LPC might reach the pore region when the channel is in the closed state before the application of LPC?

We appreciate the reviewer’s insight. We incubated LPC and nanodisc-reconstituted protein for 30 minutes, speculating that LPC approaches the pore similarly to other lipids in prior structures. In separate studies, we are optimizing conditions to capture more defined conformations.

(10) In the cryo-EM map of the “resting” state (EMDB-21150), a part of the density was interpreted as NTD flipped to the intracellular side. This density, however, is poorly defined, and not connected to the S1 helix, raising concerns about whether this density corresponds to the NTD as seen in the “resting” state structure (PDB-ID: 6VD7). In addition, some residues in the C-terminus (after K333 in frog PANX1) are missing from the atomic model. Some of these residues are predicted by AlphaFold2 to form a short alpha helix and are shown to form a short alpha helix in some published PANX1 structures. Interestingly, in both the AF2 model and 6WBF, this short alpha helix is located approximately in the weak density that the authors suggest represents the “flipped” NTD. We encourage the authors to be cautious in interpreting this part as the “flipped” NTD without further validation or justification.

We agree that the density corresponding the extended NTD into the cytoplasm is relatively weak. In our recent study, we compared two Panx1 structures with or without the mentioned C-terminal helix and found evidence suggesting the likelihood of NTD extension (DOI: 10.1101/2024.06.13.598903). Nevertheless, to prevent potential confusion, we have removed the cryo-EM panel from this manuscript.

(11) Since the authors did not observe densities of bound PLC in the cryo-EM map, it is important to acknowledge in the text the inherent limitations of using docking and mutagenesis methods to locate where PLC binds.

Thank you for the suggestion. We have removed this section to avoid potential confusion.

Optional suggestions:

(1) The authors used MeOH to extract mouse liver for reversed-phase chromatography. Was the study designed to focus on hydrophobic compounds that likely bind to the TMD? Panx1 has both ECD and ICD with substantial sizes that could interact with water soluble compounds? Also, the use of whole-cell recordings to screen fractions would not likely identify polar compounds that interact with the cytoplasmic part of the TMD? It would be useful for the authors to comment on these aspects of their screen and provide their rationale for fractionating liver rather than other tissues.

We have added a rationale in line 90, stating: “The soluble fractions were excluded from this study, as the most polar fraction induced strong channel activities in the absence of exogenously expressed pannexins.” Additionally, we have included a figure to support this rationale (Fig. S1A).

(2) The authors show that LPCs reversibly increase inward currents at a holding voltage of -60 mV (not always specified in legends) in cells expressing Panx1 and 2, and then show families of currents activated by depolarizing voltage steps in the absence of LPC without asking what happens when you depolarize the membrane after LPC activation? If LPCs can be applied for long enough without disrupting recordings, it would be valuable to obtain both I-V relations and G-V relations before and after LPC activation of Panx channels. Does LPC disproportionately increase current at some voltages compared to others? Is the outward rectification reduced by LPC? Does Vrev remain unchanged (see point above)? Its hard to predict what would be observed, but almost any outcome from these experiments would suggest additional experiments to explore the extent to which the open states activated by LPC and depolarization are similar or distinct.

Unfortunately, in our hands, the prolonged application of lysolipids at concentrations necessary to achieve significant currents tends to destabilize the patch. This makes it challenging to obtain G-V curves or perform the previously mentioned kinetic analyses. We believe this destabilization may be due to lysolipids’ surfactant-like qualities, which can disrupt the giga seal. Additionally, prolonged exposure seems to cause channel desensitization, which could be another confounding factor.

(3) From the results presented, the authors cannot rule out that mutagenesis-induced insensitivity of Panx channels to LPCs results from allosteric perturbations in the channels rather than direct binding/gating by LPCs. In Fig 5 panel A-C, the authors introduced double mutants on TM1 and TM2 to interfere with LPC binding, however, the double mutants may also disrupt the interaction network formed within NTD, TM1, and TM2. This disruption could potentially rearrange the conformation of NTD, favouring the resting closed state. Three double Asn mutants, which abolished LPC induced current, also exhibited lower currents through voltage activation in Fig 5S, raising the possibility the mutant channels fail to activate in response to LPC due to an increased energy barrier. One way to gain further insight would be to mutate residues in NTD that interact with those substituted by the three double Asn mutants and to measuring currents from both voltage activation and LPC activation. Such results might help to elucidate whether the three double Asn mutants interfere with LPC binding. It would also be important to show that the voltage-activated currents in Fig. S5 are sensitive to CBX?

Thank you for the comment, with which we agree. Our initial intention was to use the mutagenesis studies to experimentally support the docking study. Due to uncertainties associated with the presented cryo-EM maps, we have decided to remove this study from the current manuscript. We will consider the proposed experiments in a future study.

(4) Could the authors elaborate on how LPC opens Panx1 by altering the conformation of the NTDs in an uncoordinated manner, going from “primed” state to the “active” state. In the “primed” state, the NTDs seem to be ordered by forming interactions with the TMD, thus resulting in the largest (possible?) pore size around the NTDs. In contrast, in the “active” state, the authors suggest that the NTDs are fragmented as a result of uncoordinated rearrangement, which conceivably will lead to a reduction in pore size around NTDs (isn’t it?). It is therefore not intuitive to understand why a conformation with a smaller pore size represents an “active” state.

We believe the uncoordinated arrangement of NTDs is dynamic, allowing for potential variations in pore size during the activated conformation. Alternatively, NTD movement may be coupled with conformational changes in TM1 and the extracellular domain, which in turn could alter the electrostatic properties of the permeation pathway. We believe a functional study exploring this mechanism would be more appropriately presented as a separate study.

(5) Can the authors provide a positive control for these negative results presented in Fig S1B and C?

The positive results are presented in Fig. 1D and E.

(6) Raw images in Fig S6 and Fig S7 should contain units of measurement.

Thank you for pointing this out.

(7) It may be beneficial to show the superposition between primed state and activated state in both protomer and overall structure. In addition, superposition between primed state and PDB 7F8J.

We attempted to superimpose the cryo-EM maps; however, visually highlighting the differences in figure format proved challenging. Higher-resolution maps would allow for model building, which would more effectively convey these distinctions.

(8) Including particles number in each class in Fig S8 panel C and D would help in evaluating the quality of classification.

Noted.

(9) A table for cryo-EM statistics should be included.

Thanks, noted.

(10) n values are often provided as a range within legends but it would be better to provide individual values for each dataset. In many figures you can see most of the data points, which is great, but it would be easy to add n values to the plots themselves, perhaps in parentheses above the data points.

While we agree that transparency is essential, adding n-values to each graph would make some figures less clear and potentially harder to interpret in this case. We believe that the dot plots, n-value range, and statistical analysis provide adequate support for our claims.

(11) The way caspase activation of Panx channels is presented in the introduction could be viewed as dismissive or inflammatory for those who have studied that mechanism. We think the caspase activation literature is quite convincing and there is no need to be dismissive when pointing out that there are good reasons to believe that other mechanisms of activation likely exist. We encourage you to revise the introduction accordingly.

Thank you for this comment. Although we intended to support the caspase activation mechanism in our introduction, we understand that the reviewer’s interpretation indicates a need for clarification. We hope the revised introduction removes any perception of dismissiveness.

(12) Why is the patient data in Fig 4F normalized differently than everything else? Once the above issues with mVenus quenching data are clarified, it would be good to be systematic and use the same approach here.

For Fig. 4F, we used a distinct normalization method to account for substantial day-to-day variation in experiments involving body fluids. Notably, we did not apply this normalization to other experimental panels due to their considerably lower day-to-day variation.

(13) What was the rational for using the structure from ref 35 in the docking task?

The docking task utilized the human orthologue with a flipped-up NTD. We believe that this flipped-up conformation is likely the active form that responds to lysolipids. As our functional experiments primarily use the human orthologue for biological relevance, this structure choice is consistent. Our docking data shows that LPC does not dock at this site when using a construct with the downward-flipped NTD.

(14) Perhaps better to refer to double Asn ‘substitutions’ rather than as ‘mutations’ because that makes one think they are Asn in the wt protein.

Done.

(15) From Fig S1, we gather that Panx2 is much larger than Panx1 and 3. If that is the case, its worth noting that to readers somewhere.

We have added the molecular weight of each subtype in the figure legend.

(16) Please provide holding voltages and zero current levels in all figures presenting currents.

We provided holding voltages. However, the zero current levels vary among the examples presented, making direct comparisons difficult. Since we are comparing currents with and without LPC, we believe that indicating zero current levels is unnecessary for this study.

(17) While the authors successfully establish lysophospholipid-gating of Panx1 and Panx2, Panx3 appears unaffected. It may be advisable to be more specific in the title of the article.

We are uncertain whether Panx3 is unaffected by lysophospholipids, as we have not observed activation of this subtype under any tested conditions.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public Review):

Summary:

This study aims to understand the malaria antigen-specific cTfh profile of children and adults living in a malaria holoendemic area. PBMC samples from children and adults were unstimulated or stimulated with PfSEA-1A or PfGARP in vitro for 6h and analysed by a cTfh-focused panel. Unsupervised clustering and analysis on cTfh were performed.

The main conclusions are:

(1) the cohort of children has more diverse (cTfh1/2/17) recall responses compared to the cohort of adults (mainly cTfh17) and

(2) Pf-GARP stimulates better cTfh17 responses in adults, thus a promising vaccine candidate.

Strengths:

This study is in general well-designed and with excellent data analysis. The use of unsupervised clustering is a nice attempt to understand the heterogeneity of cTfh cells. Figure 9 is a beautiful summary of the findings.

Weaknesses:

(1) Most of my concerns are related to using PfSEA-1A and PfGARP to analyse cTfh in vitro stimulation response. In vitro, stimulation on cTfh cells has been frequently used (e.g. Dan et al, PMID: 27342848), usually by antigen stimulation for 9h and analysed CD69/CD40L expression, or 18h and CD25/OX40. However, the authors use a different strategy that has not been validated to analyse in vitro stimulated cTfh. Also, they excluded CD25+ cells which might be activated cTfh. I am concerned about whether the conclusions based on these results are reliable.

It has been shown that cTfh cells can hardly produce cytokines by Dan et al. However, in this paper, the authors report the significant secretion of IL-4 and IFNg on some cTfh clusters after 6h stimulation. If the stimulation is antigen-specific through TCR, why cTfh1 cells upregulate IL-4 but not IFNg in Figure 6? I believe including the representative FACS plots of IL-4, IFNg, IL21 staining, and using %positive rather than MFI can make the conclusion more convincing. Similarly, the author should validate whether TCR stimulation under their system for 6h can induce robust BCL6/cMAF expression in cTfh cells. Moreover, there is no CD40L expression. Does this mean TCR stimulation mediated BCl6/cMAF upregulation and cytokine secretion precede CD40L expression?

In summary, I am particularly concerned about the method used to analyse PfSEA-1A and PfGARP-specific cTfh responses because it lacks proper validation. I am unsure if the conclusions related to PfSEA-1A/PfGARP-specific responses are reliable.

An unfortunate reality of these types of complex immunologic studies is that it takes time to optimize a multiparameter flow cytometry panel, run this number of samples, and then conduct the analysis (not to mention the time it takes for a manuscript to be accepted for peer-review). An unexpected delay, frankly, was the COVID-19 pandemic when non-essential research lab activities were put on hold. We designed our panel in 2019 and referred to the “T Follicular Helper Cells” Methods and Protocols book from Springer 2015. Obviously the field of human immunology took a huge leap forward during the pandemic as we sought to characterize components of protective immunity, and as a result there are several new markers we will choose for future studies of Tfh subsets. We agree with the reviewer that cytokine expression kinetics differ depending on the in vitro stimulation conditions. Due to small blood volumes obtained from healthy children, we were limited in the number of timepoints we could test. However, since we were most interested in IL21 expression, we found 6 hrs to be the best in combination with the other markers of interest during our optimization experiments. We did find IFNg expression from non-Tfh cells, therefore we believe our stimulation conditions worked.

Dan et al used stimulated tonsils cells to assess the CXCR5^posPD1^posCD45RA^neg Tfh and CXCR5^neg CD45RA^neg non-Tfh whereas in our study, we evaluated CXCR5^posPD1^posCD45RA^neg Tfh from PBMCs. Dan et al PBMCs’ work used EBV/CMV or other pathogen product stimuli and only gated on CD25^posOX40^pos cells which are not the cells we are assessing in our study. This might explain in part the differences in cytokine kinetics, as we evaluated CD25^neg PBMCs only. However, we agree that more recent studies focused on CXCR5^posPD1^pos cells included more Activation-induced marker (AIM) markers, which are missing in our study, inducing a lack of depth in our analysis.

Percentage of positive cells and MFI are complementary data. Indeed, the percentage of positive cells only indicates which cells express the marker of interest without giving a quantitative value of this expression. MFI indicates how much the marker of interest is expressed by cells which is important as it can indicate degree of activation or exhaustion per cell. Meta-cluster analysis is not ideal to assess the percentage of positivity whereas it does provide essential information regarding the intensity of expression. We added supplemental figures 14 (Bcl6 and cMAF), 15 (INFg and IL21) and 16 (IL4 and IL21) where percentage of positive cells were manually gated directly from the total CXCR5^posCD4^posCD45RA^negCD25^neg TfH based on the FMO or negative control, and we overlaid the positive cells on the UMAP of all the CXCR5^posCD4^posCD45RA^negCD25^neg meta-clusters. Results from the manual gating are consistent with the results we show using clustering. However, it helps to better visualize that antigen-specific IL21 expression was statistically significant in children whereas the high background observed for adults did not reveal higher expression after stimulation, perhaps suggesting an upper threshold of cytokine expression (supplemental figure 15). The following sentence has been added in the methods at the end of the “OMIQ analysis” section: “ However, the percentage of positive IFN𝛾, IL-4, IL-21, Bcl6, or cMAF using manual gating can be found in Supplemental Figures 14, 15, and 16 along with the overlay of the gated positive cells on the CD4^posCXCR5^posCD25^neg UMAP and the cytoplots of the gated positive cells for each meta-cluster (Supplemental Figures 14, 15, and 16).”

Indeed cMAF can be induced by TCR signaling, ICOS and IL6 (Imbratta et. al, 2020). However, in our study populations, ICOS was expressed (see Author response image 1, panel A) in absence of any stimulation suggesting that CXCR5^posCD4^posCD25^negCD45RA^neg cells were already capable of expressing cMAF. Indeed, after gating Bcl6 and cMAF positive cells based on their FMOs (Author response image 1, panel B and C, respectively), we overlaid positive cells on the CXCR5^posCD4^posCD25^negCD45RA^neg cells UMAP and we can see that most of our cells already express cMAF alone (Author response image 1, panel D), co-express cMAF and Bcl6 (Author response image 1, panel E), confirming that they are TfH cells, whereas very few cells only expressed Bcl6 alone (Author response image 1, panel F). Because we knew that cT_FH already expresses Bcl6 and cMAF, we focused our analysis on the intensity of their expression to assess if our vaccine candidates were inducing more expression of these transcription factors.

Author response image 1.

(2) The section between lines 246-269 is confusing. Line 249, comparing the abundance after antigen stimulation is improper because 6h stimulation (under Golgi stop) should not induce cell division. I think the major conclusions are contained in Figure 5e, that (A) antigen stimulation will not alter cell number in each cluster and (B) children have more MC03, 06 and fewer MC02, etc.). The authors should consider removing statements between lines 255-259 because the trends are the same regardless of stimulations.

We agree, there is no cell division after 6h and that different meta clusters did not proliferate after this short of in vitro stimulation. The use of the word ‘abundance’ in the context of cluster analysis is in reference to comparing the contribution of events by each group to the concatenated data. After the meta clusters are defined and then deconvoluted by study group, certain meta clusters could be more abundant in one group compared to another - meaning they contributed more events to a particular metacluster.

Dimensionality reduction is more nuanced than manual gating and reveals a continuum of marker expression between the cell subsets, as there is no hard “straight line” threshold, as observed when using in 2D gating. Because of this, differences are revealed in marker expression levels after stimulation making them shift from one cluster to another - thereby changing their abundance.

To clarify how this type of analysis is interpreted, we have modified lines 255-259 as follows:

“In contrast, the quiescent PfSEA-1A- and PfGARP-specific cT_FH2-like cluster (MC02) was significantly more abundant in adults compared to children (Figure 5c and 5d, pf<0.05). Interestingly, following PfGARP stimulation, the activated cT_FH1/17-like subset (MC09) became more abundant in children compared to adults (Figure 5d, pf<0.05 with a False Discovery Rate=0.08), but no additional subsets shifted phenotype after PfSEA-1A stimulation (Figure 5c).”

Reviewer #2 (Public Review):

Summary:

Forconi et al explore the heterogeneity of circulating Tfh cell responses in children and adults from malaria-endemic Kenya, and further compare such differences following stimulation with two malaria antigens. In particular, the authors also raised an important consideration for the study of Tfh cells in general, which is the hidden diversity that may exist within the current 'standard' gating strategies for these cells. The utility of multiparametric flow cytometry as well as unbiased clustering analysis provides a potentially potent methodology for exploring this hidden depth. However, the current state of analysis presented does not aid the understanding of this heterogeneity. This main goal of the study could hopefully be achieved by putting all the parameters used in one context, before dissecting such differences into their specific clinical contexts.

Strengths:

Understanding the full heterogeneity of Tfh cells in the context of infection is an important topic of interest to the community. The study included clinical groupings such as age group differences and differences in response to different malaria antigens to further highlight context-dependent heterogeneity, which offers new knowledge to the field. However, improvements in data analyses and presentation strategies should be made in order to fully utilize the potential of this study.

Weaknesses:

In general, most studies using multiparameter analysis coupled with an unbiased grouping/clustering approach aim to describe differences between all the parameters used for defining groupings, prior to exploring differences between these groupings in specific contexts. However, the authors have opted to separate these into sections using "subset chemokine markers", "surface activation markers" and then "cytokine responses", yet nuances within all three of these major groups were taken into account when defining the various Tfh identities. Thus, it would make sense to show how all of these parameters are associated with one another within one specific context to first logically establish to the readers how can we better define Tfh heterogeneity. When presented this way, some of the identities such as those that are less clear such as "MC03/MC04/ MC05/ MC08" may even be better revealed. once established, all of these clusters can then be subsequently explored in further detail to understand cluster-specific differences in children vs adults, and in the various stimulation conditions. Since the authors also showed that many of the activation markers were not significantly altered post-stimulation thus there is no real obstacle for merging the entire dataset for the first part of this study which is to define Tfh heterogeneity in an unbiased manner regardless of age groups or stimulation conditions. Other studies using similar approaches such as Mathew et al 2020 (doi: 10.1126/science.abc8) or Orecchioni et al 2017 (doi: 10.1038/s41467-017-01015-3) can be referred to for more effective data presentation strategies.

Accordingly, the expression of cytokines and transcription factors can only be reliably detected following stimulation. However, the underlying background responses need to be taken into account for understanding "true" positive signals. The only raw data for this was shown in the form of a heatmap where no proper ordering was given to ensure that readers can easily interpret the expression of these markers following stimulation relative to no stimulation. Thus, it is difficult to reliably interpret any real differences reported without this. Finally, the authors report differences in either cluster abundance or cluster-specific cytokine/ transcription factor expression in Tfh cell subsets when comparing children vs adults, and between the two malaria antigens. The comparisons of cytokine/transcription factor between groups will be more clearly highlighted by appropriately combining groupings rather than keeping them separate as in Figures 6 and 7.

Thank you for sharing these references. Similar to SPADE clustering and ViSNE dimensionality algorithms used in Orecchioni et al, we used all the extracellular markers from our panel in our FlowSOM algorithm with consensus meta-clustering which includes both the chemokine receptors and activation markers even though they are presented separately in our manuscript across the figure 3 and 4. This was explained in the methods section (lines 573 - 587). We then chose the UMAP algorithm as visual dimensionality reduction of the meta-clusters generated by FlowSOM-consensus meta-clustering as explained under the “OMIQ analysis” subpart of our methods (lines 588- 604). Therefore, we believe we have conducted the analysis as this reviewer suggests even if we chose to show the figures that were informative to our story. The heatmap of the results brings the possibility to see which combination of markers respond or not to the different conditions and between groups, all the raw data are present from the supplemental figures 10 to 13 showing, using bar plots, the differences expressed in the heatmaps. We believe it strengthens our interpretation of the results.

Regarding the transcription factor and cytokine background, we added supplemental figures 14, 15 and 16 where we used manual gating to select Bcl6, cMAF, IFNg, IL21 or IL4 positive cells directly from total CXCR5^posCD4^posCD45RA^negCD25^neg TfH cells based on the FMO or negative control, and we overlaid the positive cells on the UMAP of all the CXCR5^posCD4^posCD45RA^negCD25^neg meta-clusters. Moreover, all the dot plots (with their statistics) used for the heatmap figure 6 and 7 can be found in the supplemental figures 10, 11, 12 and 13. These supplemental figures address the concerns above by showing the difference of signals between unstimulated and stimulated conditions.

Reviewer #3 (Public Review):

Summary:

The goal of this study was to carry out an in-depth granular and unbiased phenotyping of peripheral blood circulating Tfh specific to two malaria vaccine candidates, PfSEA-1A and PfGARP, and correlate these with age (children vs adults) and protection from malaria (antibody titers against Plasmodium antigens.). The authors further attempted to identify any specific differences in the Tfh responses to these two distinct malaria antigens.

Strengths:

The authors had access to peripheral blood samples from children and adults living in a malaria-endemic region of Kenya. The authors studied these samples using in vitro restimulation in the presence of specific malaria antigens. The authors generated a very rich data set from these valuable samples using cutting-edge spectral flow cytometry and a 21-plex panel that included a variety of surface markers, cytokines, and transcription factors.

Weaknesses:

- Quantifying antigen-specific T cells by flow cytometry requires the use of either 1- tetramers or 2- in vitro restimulation with specific antigens followed by identification of TCR-activated cells based on de-novo expression of activation markers (e.g. intracellular cytokine staining and/or surface marker staining). Although authors use an in vitro restimulation strategy, they do not focus their study on cells de-novo expressing activation markers as a result of restimulation; therefore, their study is not really on antigen-specific cTfh. Moreover, the authors report no changes in the expression of activation markers commonly used to identify antigen-specific T cells upon in vitro restimulation (including IFNg and CD40L); therefore, it is not clear if their in vitro restimulation with malaria antigens actually worked.

We understand the reviewer’s point of view and apologies for any confusion. IFNg was expressed but not statistically different between groups. Indeed, looking at the CD8 T cells and using manual gating, we were able to show that IFNg was increased but not statistically significant upon stimulation from CD4^posCXCR5^pos cells (supplemental figure 15, panel C), confirming our primary observation using clustering analysis. These results showed that our malaria antigen induced IFNg response in some participants, but not all of them, revealing heterogeneity in this response among individuals within the same group.

Regarding CD40L, in the supplemental figure 7, we can see that some of our meta-clusters expressed more CD40L upon stimulation, but again without leading to statistical differences between groups. Combined with the increased expression of other cytokines and transcription factors, we showed that our stimulation did indeed work. However, because of the high variation within groups, there were no statistical differences across our groups. Because CD40L is not the only marker showing specific T cell activation, and not all T cells respond using this marker alone, a more comprehensive multimarker AIM panel might have highlighted differences between groups. We recognized the limitations of our study and believe that future study will benefit from more activation markers commonly used to identify antigone-specific T cells such as CD69, OX40, 4-1BB (AIM panel), among other markers.

- CXCR5+CD4+ memory T cells have been shown to present multi-potency and plasticity, capable of differentiating to non-Tfh subsets upon re-challenge. Although authors included in their flow panel a good number of markers commonly used in combination to identify Tfh (CXCR5, PD-1, ICOS, Bcl-6, IL-21), they only used one single marker (CXCR5) as their basis to define Tfh, thus providing a weak definition for Tfh cells and follow up downstream analysis.

Sorry for the confusion, even though the subsampled on the CD4^posCXCR5^pos CD25^neg cells to run our FlowSOM, we showed the different levels of expression across meta-clusters (figure 4 panels A and B) of PD1 (Tfh being PD1 positive cells) and ICOS (indicating the activation stage of the Tfh, “T Follicular Helper Cells” Methods and Protocols book from Springer 2015). We also included an overlay of the manually gated double positive Bcl6-cMAF cells on the CXCR5^posCD45RA^negCD25^neg CD4 T cell UMAP plot to show that most of them express Bcl6 (supplemental figure 14). Interestingly, the manually gated IL21 positive cells were less abundant, particularly for children (supplemental figure 15). Because we were not able to include all the markers that are now used to define Tfh cells, we referred to our cell subsets as “TFH-like”. This is an acknowledged limitation of our study. Due to the limited blood volume obtained from children and cost of running multiplex flow cytometry assays, our results showing antigen-specific heterogeneity of Tfh subset will have to be validated in future studies that include these additional defining markers.

- Previous works have used FACS-sorting and in vitro assays for cytokine production and B cell help to study the functional capacity of different cTfh subsets in blood from Plasmodium-infected individuals. In this study, authors do not carry out any such assays to isolate and evaluate the functional capacity of the different Tfh subsets identified. Thus, all the suggestions for the role that these different cTfh subsets may have in vivo in the context of malaria remain highly hypothetical.

Unfortunately, low blood volumes obtained from children prevented us from running in vitro functional assays and the study design did not allow us to correlate them with protection. However, since the function of identified Tfh subsets from malaria-exposed individuals has been evaluated using Pf lysates in other studies, we referenced them when interpreting the differences we reported in Tfh subset recognition between malaria antigens. If either of these antigens move forward into vaccine trials, then evaluating their function would be important.

- The authors have not included malaria unexposed control groups in their study, and experimental groups are relatively small (n=13).

This study design did not include the recruitment of malaria naive negative controls as its goal was to assess malaria antigen-specific responses comparing the quality and abundance between malaria-exposed children to adults to these potential new vaccine targets PfSEA-1A and PfGARP. We did however test 3 malaria-naive adults and found no non-specific activation after stimulation with these two malaria antigens. Since this was done as part of our assay optimization, we did not feel the need to show these negative findings.

And even with our small sample size, we demonstrated significant age-associated differences in malaria antigen-specific responses from cT_FH-like subsets.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Minor points are:

(1) Line 88, cTfh cells are not only from GC-Tfh, they have GC-independent origin (He et al, PMID: 24138884).

The following sentence was added line 88 “Interestingly, cT_FH cells can also come from peripheral cT_FH precursor CCR7^lowPD1^highCXCR5^pos cells; thus, they also have a GC-independent origin (He, Cell, 2013 PMID: 24138884).

(2) I believe all participants were free of blood-stage infection upon enrolment. But can authors clearly state this information between lines 151-159?

We mentioned in the methods, line 495-496 “Participants were eligible if they were healthy and not experiencing any symptoms of malaria at the time venous blood was collected”. However, using qPCR we found 5 children with malaria blood stage. As shown in Author response image 2, comparing malaria free to blood-stage children, no differences were observed without any stimulation. However, MC03 is more abundant upon malaria antigen stimulation in the blood-stage group whereas MC04 is more abundant in the malaria free group upon PfGARP stimulation only confirming that our stimulation worked.

Author response image 2.

Reviewer #3 (Recommendations For The Authors):

(1) The strategy for gating on antigen-specific cTfh cells needs to be revised. The correct approach would be to gate on those cells that respond by de-novo expression of activation markers upon antigen restimulation (also termed activation-induced markers. e.g. CD69, CD40L, CXCL13 and IL-21, Niessl 2020; CD69, CD40L, CD137 and OX40, Lemieux 2023; CD137 and OX40, Grifoni 2020). As it stands, the study is not really on antigen-specific T cells, but rather on the overall CD4 T cell compartment plus or minus antigenic stimulation.

We recognized the limitation in our flow panel design which prevents us from performing this gating. We originally based our panel design on the “T follicular helper cells methods and protocols” book (Springer 2015) which used CD45RA, CD25, CXCR5, CCR6, CXCR3, CCR7, ICOS and PD1 to define cT_FH. We had already optimized our 21-color panel, purchased reagents and started to run our experiments by the time these publications modified how to define TFH cells Niessl, Lemieux and Grifoni’s publication. Indeed we optimized and performed our assay from November 2019 to March 2020, finishing to run the samples during the first quarantine. Because of the urgent needs of research on SARS-CoV-2 that we were involved with from this time and moving forward, the analysis of our TFH work got highly postponed. Moreover, 2020 is also the year where many TFH papers came out with better ways to define cT_FH and responses to antigen stimulations. In our future studies, our panel will include AIM.

(2) It is not clear if the antigenic stimulation actually worked. Does the proportion of IFNg+ or IL-4+ or IL-21+ or CD40L+ or CD25+ CD4 or CD8 T cells increase following in vitro antigen restimulation?

Yes, using manual gating, we are able to show an increase of IL4 (supplemental figure 16 panel B and C), and IL21 (supplemental figure 15 panel J and K) production in both children and adults. However, we did not observe significant production of IFNg (supplemental figure 15, panel C) and changes in CD40L expression (supplemental figure 7) after malaria antigen stimulation, however, our positive control SEB worked. So, yes our stimulation assay worked but these 2 malaria antigens did not significantly induce these cytokines. This could be that they are too low to detect in every participant since they are single antigens and not whole parasite lysates, as other studies have used. It could also be that these antigens don’t stimulate CD40L or IFNg in all our participants. We brought up this limitation as follow in the discussion, line 473: “Although the heterogeneity in the response of CD40L and IFNγ suggests that our tested malaria antigens did not induce significant differences in the expression of these markers in all our participants, our panel did not include other activated induced markers, such as OX40, 4-1BB, and CD69”.

(3) It is not clear what is the proportion of cTfh over the total CD4 T cell compartment among the different groups. Does this vary among different groups? It would be valuable to display this as an old-fashioned combination of contour plots with outliers for illustrating flow cytometry and bar graphs for the cumulative data.

The proportion of CD3^posCD4^posCD25^negCXCR5^pos cTfh cells did not differ within the total number of CD4 T cells between groups (figure 2).

(4) The gating strategy could be refined and become more robust if adding additional markers in combination with CXCR5 for identifying cTfh (e.g. CXCR5+Bcl6+).

Thank you for this suggestion. An overlay of Bcl6 expression can be found in supplemental figure 14 where we confirm that our CXCR5+ cT_FH-like subsets express cMAF and Bcl6.

(5) The protocols for intracellular and intranuclear staining seem to be incomplete in Materials and Methods. In particular, cell permeabilization strategies seem to be missing.

Our apologies for this oversight, we added the following sentences in the methods line 545: “Cells were fixed and permeabilized for 45 mins using the transcription factor buffer set (BD Pharmingen) followed by a wash with the perm-wash buffer. Intracellular staining was performed at 4 °C for 45 more mins followed by two washes using the kit’s perm-wash buffer”.

(6) In Materials and Methods, the authors mention they have used fluorescence minus one control to set their gating strategy. It would be valuable to show these, either on the main body or as part of supplementary figures.

We added the cytoplots of the FMOs and/or negative controls as appropriate in the supplemental figures 14 (cMAF and Bcl6), 15 (IFNg and IL21) and 16 (IL4 and IL21).

(7) Line 194 and Figure 3, it is not clear the criteria that the authors used for down-sampling events before FlowSOM analysis. Was this random? Was this done with unstimulated or stimulated samples?

We chose to down-sample on CD3posCD4^posCD25^negCD45RA^neg and CXCR5^pos cells prior to our FlowSOM to allow more cluster analysis to focus only on the differences among those cells. The down-sampling used 1,000 CD3posCD4^posCD25^neg CD45RA^negCXCR5^pos cells from each fcs file (unstimulated and stimulated samples). If the fcs file had more than 1,000 CXCR5^pos cells, the down-sampling was done randomly by the OMIQ platform algorithm to select only 1,000 CXCR5^pos cells within this specific fcs file. The latest sentence was added to the methods line 593.

(8) Lanes 201, 202, As it stands, the take of the authors on the role of different cTfh subsets during infection remains highly speculative. Are these differences in cTfh phenotypes actually reflected in their in vitro capacity to provide B cell help (e.g. as in the Obeng-Adjei 2015 paper) or to produce IL-21, express co-stimulatory molecules, or any other characteristic that would allow them to better infer their functional roles during infection? Any additional in vitro analysis of the functional capacity of isolated cTfh subsets identified in this research would greatly increase its value.

We agree with the reviewer that this sentence is speculative, and we rephrase it as follow: “First, we found different CXCR5 expression levels between meta-clusters (Figure 3b); CXCR5 is essential for cT_FH cells to migrate to the lymph nodes and interact with B-cells”. We would have liked to perform in vitro functional assays. However, as explained above, we did not have sufficient cells collected from children to do so.

(9) It is not clear why authors omitted IL-17 and did not use IFNg and IL-4 to refine their definition of Th1, Th2 and Th17 cTfh.

We would have liked to include IL-17, however we were constrained by only having access to a 4 lasers cytometer at the time we ran our assay. In light of needing to prioritize markers, when we were designing our flow panel, cTfh1 were shown to be preferentially activated during episodes of acute febrile malaria children (Obeng-Adjei). Therefore, we chose to focus on IFNg and IL4 to differentiate Tfh1 from Tfh2, in addition to other markers as surrogate of functional potential. We did not use IFNg and IL4 to refine our definition of Tfh1, Tfh2 and Tfh17 as recent publications have shown that IL4 is not only expressed in Tfh2 but also in the other Tfh subsets, at lower intensity (Gowthaman among others). Therefore IFNg and IL4 by themselves were not sufficient to properly define the different Tfh subsets. In future studies, we plan to include transcription factor profiles (T-bet, BATF, GATA3) to further refine definitions of Tfh subsets.

(10) Lines, 226, 228, based on the combination of markers that the MC03 subset expresses, it is tempting to think that this is the only "truly" committed Tfh subset from the entire analysis. Please, discuss.

If the reviewer is referring to changes in marker expression levels that indicate they have not reached a level of differentiation that would make them reliable (ie “true) Tfh cells, we agree that this is an important question now that we have technology that can measure and analyse so many phenotypic markers at once. This brings forward the need for the scientific method - to replicate study findings to determine whether they are consistent given the same study design and experimental conditions.

(11) Lines 243 244, Again, is this reflected in functional capacity?

The study described in this manuscript did not include functional assays. However, this did not change the key finding that different malaria antigens behaved differently, demonstrating heterogeneity in Tfh recognition of malaria antigens. Regarding CD40L expression, we did not observe differences between groups, however some individuals had an increase of their CD40L (supplemental figure 7). It is possible that some individuals had responded through other activated induced markers (CD69, ICOS, OX40, 4-1BB among others) and that our stimulation condition was not long enough to assess CD40L expression upon malaria antigen stimulation. This limitation has been addressed by editing the line 243-244 as follows: “we were unable to find statistical differences in the CD40L expression between groups as only few individuals responded through it (supplemental figure 7).”

(12) Lines 243, 244, Are these cTfh subsets exclusively detected in malaria-exposed individuals? This is confounded by the lack of a malaria unexposed control group in this study, which would have been highly valuable.

We agree with the reviewer that having non-naive children would have been valuable as a negative control group. However, this study was conducted in Kenya where all children are suspected to have had at least one malaria infection. We also did not have ethical approval or the means to enroll children in the USA who would not have been exposed to malaria as a negative control group. Since we were also evaluating differences by age group, comparing US adults would not have helped to address this point. Therefore, this remains an open question that might be addressed by another study recruiting children in non-malaria endemic areas.

(13) Line 267, as the authors have not gated on T cells de-novo expressing activation markers in response to antigen restimulation, how do they know these are indeed antigen-specific cTfh?

Omiq analysis accounts for marker expression levels in the resting cells (unstimulated well) for each individual compared to each experimental/stimulated well. The algorithm computationally determines whether that expression level changed without an arbitrary positive threshold, keeping the expression levels as a continuous variable, not dichotomous - which is the power of unbiased cluster analyses. Therefore, we know that these cells are antigen-specific based on the statistical difference in intensity expression between the resting cells and the stimulated ones. Nevertheless, manual gating to show “de-novo” responding cells, produced the same results as assessing the MFI of each meta-cluster (supplemental figures 14, 15 and 16).

(14) Lines, 292-295, it is very surprising that Tfh cells would not produce IL-21 upon restimulation. Have the authors observed upregulation of IL-21 following SEB restimulation?

Yes, we observed IL21 positive cells upon SEB stimulation (supplemental figure 15, panel J and K). However we found unexpectedly high background levels of IL21, specifically within the adult group (supplemental figure 15, panel K and M) making it challenging to find antigen-specific increases above background. Interestingly, an increase in IL21 using manual gating was observed upon PfSEA-1A or PfGARP stimulation in children (supplemental figure 15, panel J and L).

(15) In Figures 3 and 4, it is not clear if there are any significant differences in expression of different markers between different cTfh subsets and/or different conditions. Moreover, the lack of differences in response to antigen stimulation seems to suggest that it did not work adequately.

We intentionally chose 6-hours stimulation to better assess changes in cytokines which we did. However, because it is a short stimulation, we did not expect dramatic changes in the extracellular markers presented in the figure 3 and 4. A longer stimulation, such as 24h, will highlight properly these changes.

(16) Figure 5b would benefit from bar graphs.

Please find below the bar-graphs for the highlighted meta-clusters in figure 5b. We did not include these bar-graphs to our figure 5 as they do not bring new information. They repeat the information already presented through the EdgeR plot.

Author response image 3.

(17) Figures 6 and 7 would greatly benefit from showing individual examples of old-fashioned contour with outliers flow plots to illustrate the different cTfh subsets identified in the study.

The different cT_FH subsets can be found with a contour plot with outliers in the supplemental figure 4.

(18) Figures 3,4, 6, and 7, the authors exclusively focused on the study of MFI to measure the expression of cytokine and transcription factors among different groups/stimulations. Have the authors observed any differences in the percentage or absolute counts of cytokine+ and/or TF+ between different subsets of cTfh and/or different conditions?

Yes. We added the supplemental figures 14 (transcription factors) and 15/16 (cytokines) where cytokines and transcription factors were assessed using manual gating. We found that total CD4^posCXCR5^pos IL4 was significantly increased upon stimulation in both adults and children while IFNg was not. However, we found significantly higher IFNg on total CD8^pos cells showing that the stimulation worked, but the total CD4^posCXCR5^pos did not express IFNg. Finally, we observed a trend of higher IL21^posCD4^posCXCR5^pos in adults, not significant due to high background whereas IL21 was significantly increased upon stimulation in children. Regarding cMAF and Bcl6, both transcription factors were significantly increased upon stimulation within children only.

(19) Figure 8, the definition for high and low PfGARP antibody titers seems rather arbitrary. Are these associations still significant when attempting a regular correlation analysis between Ab values (i.e. Net MFI) and different cTfh subsets?

Yes, the definition for high and low PfGARP antibody levels is arbitrary but when looking at the antibody data (figure 1b), it was naturally bimodal. Therefore as a sub-analysis, we assess the association between PfGARP antibodies levels and cT_FH subsets, see Author response image 4. We checked the correlation between the abundance of the meta-clusters and the level of IgG anti-PfGARP and anti-PfSEA after PfGARP and PfSEA stimulation. We also checked the correlation between the MFI expression of Bcl6 and cMAF after stimulation (PfGARP or PfSEA-1A minus the unstimulated) by the meta-clusters and the level of IgG anti-PfGARP and anti-PfSEA. However, we believe that because of our small sample size, our results are not robust enough and that we risk over-interpreting the data. Therefore, we choose not to include this analysis in the manuscript.

Author response image 4.

(20) The comprehensive 21-plex panel that authors used in this study could generate insights on additional immune cells beyond cTfh (e.g. additional CD4 T cell subsets, CD8 T cells, CD19 B cells). It is not clear why the authors limited their analysis to cTfh only.

The primary goal of the study was to assess the cT_FH response to malaria vaccine candidates. However, we were able to assess the IFNg expression for CD8 T cells upon stimulation using the manual gating as indicated in the supplemental figure 15. Without additional markers to more clearly define other CD4 T cell or B cell subsets, we do not believe this dataset would go deep enough into characterizing antigen-specific responses to malaria antigens that would yield new insight.

(21) Minor point, the punctuation should be revised throughout the manuscript.

Punctuation was revised throughout the manuscript by our departmental scientific writer Dr. Trombly, as per reviewer request.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1:

Weaknesses:

(1) Only Experiment 1 of Rademaker et al (2019) is reanalyzed. The previous study included another experiment (Expt 2) using different types of distractors which did result in distractor-related costs to neural and behavioral measures of working memory. The Rademaker et al (2019) study uses these two results to conclude that neural WM representations are protected from distraction when distraction does not impact behavior, but conditions that do impact behavior also impact neural WM representations. Considering this previous result is critical for relating the present manuscript's results to the previous findings, it seems necessary to address Experiment 2's data in the present work

We thank the reviewer for the proposal to analyze Experiment 2 where subjects completed the same type of visual working memory task, but instead had either a flashing orientation distractor or a naturalistic (gazebo or face) distractor present during two-thirds of the trials. As the reviewer points out, unlike Experiment 1, these two conditions in Experiment 2 had a behavioral impact on recall accuracy, when compared to the blank delay. We have now run the temporal cross-decoding analysis, temporally-stable neural subspace analysis, and condition cross-decoding analysis in Experiment 2. The results from the stable subspace analysis are present in Figure 3, while the results from the temporal cross-decoding analysis and condition cross-decoding analysis are present in the Supplementary Data.

First, we are unable to draw strong conclusions from the temporal cross-decoding analysis, as the decoding accuracies across time in Experiment 2 are much lower compared to Experiment 1. In some ROIs of the naturalistic distractor condition we see that some diagonal elements are not part of the above-chance decoding cluster, making it difficult to draw any conclusions regarding dynamic clusters. We do see some dynamic coding in the naturalistic condition in V3 where the off-diagonals do not show above-chance decoding. Since the temporal cross-decoding provides low accuracies, we do not examine the dynamics of neural subspaces across time.

We do, however, run the stable subspace analysis on the flashing orientation distractor condition. Just like in Experiment 1, we examine temporally stable target and distractor subspaces. When projecting the distractor onto the working memory target subspace, we see a higher overlap between the two as compared to Experiment 1. A similar pattern is seen also when projecting the target onto the distractor subspace. We still see an above-chance principal angle between the target and distractor; however, this angle is qualitatively smaller compared to Experiment 1. This shows that the degree of separation between the two neural subspaces is impacted by behavioral performance during recall.

(2) Primary evidence for 'dynamic coding', especially in the early visual cortex, appears to be related to the transition between encoding/maintenance and maintenance/recall, but the delay period representations seem overall stable, consistent with previous findings

We agree with the reviewer that we primarily see dynamic coding between the encoding/maintenance and at the end of the maintenance periods, implying the WM representations are stable in most ROIs. The only place where we argue that we might see more dynamic coding during the delay itself is in V1 during the noise distractor trials in Experiment 1.

(3) Dynamicism index used in Figure 1f quantifies the proportion of off-diagonal cells with significant differences in decoding performance from the diagonal cell. It's unclear why the proportion of time points is the best metric, rather than something like a change in decoding accuracy. This is addressed in the subsequent analysis considering coding subspaces, but the utility of the Figure 1f analysis remains weakly justified.

We agree that other metrics can also provide a summary of dynamics; here, the dynamicism index just acts as a summary visualizing the dynamic elements. It offers an intuitive way to visualize peaks and troughs of the dynamic code across the extent of the trial.

(4) There is no report of how much total variance is explained by the two PCs defining the subspaces of interest in each condition, and timepoint. It could be the case that the first two principal components in one condition (e.g., sensory distractor) explain less variance than the first two principal components of another condition.

We thank the reviewer for this comment. We have now included the percent variance explained for the two PCs in both the temporally-stable target and distractor subspace and the dynamic subspace analysis. The percent-explained is comparable across analyses; the first PC ranges from 43-50% and the second ranges from 28-37%. The PCs within each analysis (dynamic no-distractor, orientation and noise distractor; temporally-stable target and distractor) are even closer in range (Figure 2c and 3d).

(5) Converting a continuous decoding metric (angular error) to "% decoding accuracy" serves to obfuscate the units of the actual results. Decoding precision (e.g., sd of decoding error histogram) would be more interpretable and better related to both the previous study and behavioral measures of WM performance.

We thank the reviewer for the comments. FCA is a linear function of the angular error that uses the following equation:

We think that the FCA does not obfuscate the results, but instead provides an intuitive scale where 0% accuracy corresponds to a 180° error, 50% to a 90° error and so on. This also makes it easy to reverse-calculate the absolute error if need be. Our lab has previously used this method in other neuroimaging papers with continuous variables (Barbieri et al. 2023, Weber et al. 2024).

We do, however, agree that “% decoding accuracy” does not provide an accurate reflection of the metric used. We have thus now changed “% decoding accuracy” to “Accuracy (% FCA)”.

(6) This report does not make use of behavioral performance data in the Rademaker et al (2019) dataset.

We have now analyzed Experiment 2 which, as previously mentioned by the reviewer and unlike Experiment 1, showed a decrease in recall accuracy during the two distractor conditions. We address the results from Experiment 2 in a previous response (please see Weaknesses 1).

We do not, however, relate single subject behavioral performance to neural measurements, as we do not think there is enough power to do so with a small number of subjects in both Experiment 1 and 2. 

(7) Given there were observed differences between individual retinotopic ROIs in the temporal cross-decoding analyses shown in Figure 1, the lack of data presented for the subspace analyses for the corresponding individual ROIs is a weakness

We have now included an additional supplementary figure that shows individual plots of each ROI for the temporally stable subspace analysis for both Experiment 1 and Experiment 2 (Supplementary Figure 5). 

Reviewer #1 (Recommendations For The Authors):

(1) Is there any relationship between stable/dynamic coding properties and aspects of behavioral performance? This seems like a major missed opportunity to better understand the behavioral relevance or importance of the proposed dynamic and orthogonal coding schemes. For example, is it the case that participants who have more orthogonal coding subspaces between orientation distractor and remembered orientation show less of a behavioral consequence to distracting orientations? Less induced bias? I know these differences weren't significant at the group level in the original study, but maybe individual variability in the metrics of this study can explain differences in performance between participants in the reported dataset

As mentioned in the previous response, we do not run individual correlations between dynamic or orthogonal coding metrics and behavioral performance, because of the small number of subjects in both experiments. We believe that for a brain-behavior correlation between average behavioral error of subjects and an average brain measure, we would need a larger sample size.  

(2) The voxel selection procedure differs from the original study. The authors should add additional detail about the number of voxels included in their analyses, and how this number of voxels compares to that used in the original study.

We have now added a figure summarizing the number of voxels selected across participants. We do select fewer voxels compared to Rademaker et al. 2019 (see their Supplementary Tables 9 and 10 and our Supplementary Figure 8). For example we have ~500 voxels on average in V1 in Experiment 1, while the original study had ~1000. As mentioned in the methods, we aimed to select voxels that reliably responded to both the perception localizer conditions and the working memory trials.

(3) Lines 428-436 specify details about how data is rescaled prior to decoding. The procedure seems to estimate rescaling factors according to some aspect of the training data, and then apply this rescaling to the training and testing data. Is there a possibility of leakage here? That is - do aspects of the training data impact aspects of the testing data, and could a decoder pick up on such leakage to change decoding? It seems this is performed for each training/testing timepoint pair, and so the temporal unfolding of results may depend on this analysis choice.

Thank you for the suggestion. To prevent data leakage, the mean and standard deviation are computed exclusively from the training set. These scaling parameters are then applied to the test set, ensuring that no information from the test set influences the training process. This transformation simply adjusts the test set to the same scale as the training data, without exposing the model to unseen test data during training.

(4) Figure 1d, V1: it looks like the 'dynamics' are a bit non-symmetric - perhaps the authors could comment on this detail of the results? Why would we expect there would be a dynamic cluster on one side of the diagonal, but not the other? Given that this region, condition is the primary evidence for a dynamic code that's not related to the beginning/end of delay (see other comments), figuring this out is of particular importance.

We thank the reviewer for this question. We think that this is just due to small numerical differences in the upper and lower triangles of the matrix, rather than a neuroscientifically interesting effect. However, this is only a speculative observation.

(5) I think it's important to address the issue I raised in "weaknesses" about variance explained by the top N principal components in each condition. What are we supposed to learn from data projected into subspaces fit to different conditions if the subspaces themselves are differently useful?

Thank you, this has now been addressed in a previous comment (please see Weakness 4). 

Reviewer #2:

Weaknesses:

(1) An alternative interpretation of the temporal dynamic pattern is that working memory representations become less reliable over time. As shown by the authors in Figure 1c and Figure 4a, the on-diagonal decoding accuracy generally decreased over time. This implies that the signal-to-noise ratio was decreasing over time. Classifiers trained with data of relatively higher SNR and lower SNR may rely on different features, leading to poor generalization performance. This issue should be addressed in the paper.

We thank the reviewer for raising this issue and we have now run three simulations that aim to address whether a changing SNR across time might create dynamic clusters. 

In the first simulation we created a dataset of 200 voxels that have a sine or cosine response function to orientations between 1° to 180°, the same orientations as the remembered target. A circular shift is applied to each voxel to vary preferred (or maximal) responses of each simulated voxel. We then assess the decoding performance under different SNR conditions during training and testing. For each of the seven iterations we selected 108 responses (out of 180) to train on and 108 to test on. To increase variability the selected trials differed in each iteration. Random white noise was applied to the data and thus the SNR was independently scaled according to the specified levels for train and test data. We then use the same pSVR decoder as in the temporal cross decoding analysis to train and test. 

The second and third simulations more directly address whether increased noise levels  would induce the decoder to rely on different features of the no-distractor and noise distractor data. We use empirical data from the primary visual cortex (V1; where dynamic coding was seen in the noise distractor trials) under the no-distractor and noise distractor conditions for the second and third simulations, respectively. Data from time points 5.6–8.8 seconds after stimulus onset are averaged across five TRs. As in the first simulation, SNR is systematically manipulated by adding white noise. Additionally, to see whether the initial decrease in SNR and subsequent increase would result in dynamic coding clusters, we initially increased and subsequently decreased the amplitude of added noise. The same pSVR decoder was used to train and test on the data with different levels of added noise.

We see an absence of dynamic elements in the SNR cross-decoding matrices, as the decoding accuracy primarily depends on the training data rather than test data. This results in some off-diagonal values in the decoding matrix that are higher, rather than smaller, than corresponding on-diagonal elements.

We have now added a Methods section explaining the simulations in more detail and Supplementary Figure 9 showing the SNR cross-decoding matrices. 

(2) The paper tests against a strong version of stable coding, where neural spaces representing WM contents must remain identical over time. In this version, any changes in the neural space will be evidence of dynamic coding. As the paper acknowledges, there is already ample evidence arguing against this possibility. However, the evidence provided here (dynamic coding cluster, angle between coding spaces) is not as strong as what prior studies have shown for meaningful transformations in neural coding. For instance, the principal angle between coding spaces over time was smaller than 8 degrees, and around 7 degrees between sensory distractors and WM contents. This suggests that the coding space for WM was largely overlapping across time and with that for sensory distractors. Therefore, the major conclusion that working memory contents are dynamically coded is not well-supported by the presented results.

We thank the reviewer for this comment. The principal angles we calculate are above-baseline, meaning that we subtract the within-subspace principal angles from the between-subspace principal angles and take the average. Thus a 7 degree difference does not imply that there are only 7 degrees separating e.g. the sensory distractor from the target; it just indicates that the separation is 7 degrees above chance. 

(3) Relatedly, the main conclusions, such as "VWM code in several visual regions did not generalize well between different time points" and "VWM and feature-matching sensory distractors are encoded in separable coding spaces" are somewhat subjective given that cross-condition generalization analyses consistently showed above chance-level performance. These results could be interpreted as evidence of stable coding. The authors should use more objective descriptions, such as 'temporal generalization decoding showed reduced decoding accuracy in off-diagonals compared to on-diagonals.

Thank you, we agree that our previous claims might have been too strong. We have now toned down our statements in the Abstract and use “did not fully generalize” and “VWM and feature-matching sensory distractors are encoded in coding spaces that do not fully overlap.”

Reviewer #2 (Recommendations For The Authors):

Weakness 1 can potentially be addressed with data simulations that fix the signal pattern, vary the noise pattern, and perform the same temporal generalization analysis to test whether changes in SNR can lead to seemingly dynamic coding formats.

Thank you for the great suggestion. We have now run the suggested simulations. Please see above (response to Weakness 1).

There are mismatches in the statistical symbols shown in Figure 4 and Supplementary Table 2. It seems that there was a swap between the symbols for the noise between-condition and noise within-condition.

Thank you, this has now been fixed.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors investigate ligand and protein-binding processes in GPCRs (including dimerization) by the multiple walker supervised molecular dynamics method. The paper is interesting and it is very well written.

Strengths:

The authors' method is a powerful tool to gain insight on the structural basis for the pharmacology of G protein-coupled receptors.

We thank the Reviewer for the positive comment on the manuscript and the proposed methods.

Reviewer #2 (Public review):

The study by Deganutti and co-workers is a methodological report on an adaptive sampling approach, multiple walker supervised molecular dynamics (mwSuMD), which represents an improved version of the previous SuMD.

Case-studies concern complex conformational transitions in a number of G protein Coupled Receptors (GPCRs) involving long time-scale motions such as binding-unbinding and collective motions of domains or portions. GPCRs are specialized GEFs (guanine nucleotide exchange factors) of heterotrimeric Gα proteins of the Ras GTPase superfamily. They constitute the largest superfamily of membrane proteins and are of central biomedical relevance as privileged targets of currently marketed drugs.

MwSuMD was exploited to address:

a) binding and unbinding of the arginine-vasopressin (AVP) cyclic peptide agonist to the V2 vasopressin receptor (V2R);

b) molecular recognition of the β2-adrenergic receptor (β2-AR) and heterotrimeric GDPbound Gs protein;

c) molecular recognition of the A1-adenosine receptor (A1R) and palmotoylated and geranylgeranylated membrane-anchored heterotrimeric GDP-bound Gi protein;

d) the whole process of GDP release from membrane-anchored heterotrimeric Gs following interaction with the glucagon-like peptide 1 receptor (GLP1R), converted to the active state following interaction with the orthosteric non-peptide agonist danuglipron.

The revised version has improved clarity and rigor compared to the original also thanks to the reduction in the number of complex case studies treated superficially.

The mwSuMD method is solid and valuable, has wide applicability and is compatible with the most world-widely used MD engines. It may be of interest to the computational structural biology community.

The huge amount of high-resolution data on GPCRs makes those systems suitable, although challenging, for method validation and development.

While the approach is less energy-biased than other enhanced sampling methods, knowledge, at the atomic detail, of binding sites/interfaces and conformational states is needed to define the supervised metrics, the higher the resolution of such metrics is the more accurate the outcome is expected to be. Definition of the metrics is a user- and system-dependent process.

We thank the Reviewer for the positive comment on the revised manuscript and mwSuMD. We agree that the choice of supervised metrics is user- and systemdependent. We aim to improve this aspect in the future with the aid of interpretable machine learning.

Reviewer #3 (Public review):

Summary:

In the present work Deganutti et al. report a structural study on GPCR functional dynamics using a computational approach called supervised molecular dynamics.

Strengths:

The study has potential to provide novel insight into GPCR functionality. Example is the interaction between D344 and R385 identified during the Gs coupling by GLP-1R. However, validation of the findings, even computationally through for instance in silico mutagenesis study, is advisable.

Weaknesses:

No significant advance of the existing structural data on GPCR and GPCR/G protein coupling is provided. Most of the results are reproductions of the previously reported structures.

The method focus of our study (mwSuMD) is an enhancement of the supervised molecular dynamics that allows supervising two metrics at the same time and uses a score, rather than a tabù-like algorithm, for handing the simulation. Further changes are the seeding of parallel short replicas (walkers) rather than a series of short simulations, and the software implementation on different MD engines (e.g. Acemd, OpenMM, NAMD, Gromacs).

We agree with the Reviewer that experimental validation of the findings would be advisable, in line with any computational prediction. We are positive that future studies from our group employing mwSuMD will inform mutagenesis and BRET-based experiments.

Reviewer #2 (Recommendations for the authors):

As for GLP1R, I remain convinced that the 7LCI would have been better as a reference for all simulations than 7LCJ, also because 7LCI holds a slightly more complete ECD.

We agree that 7LCJ would have been a better starting point than 7LCI for simulations because it presents the stalk region, contrary to 7LCJ. However, we do not think it might have influenced the output because the stalk is the most flexible segment of GLP1R, and any initial conformation is usually not retained during MD simulations.

Please, correct everywhere the definition of the 6LN2 structure of GPL1R as a ligand-free or apo, because that structure is indeed bound to a negative allosteric modulator docked on the cytosolic end of helix-6

We thank the reviewer for this precision. The text has been modified accordingly.

As for the beta2-AR, the "full-length" AlphaFold model downloaded from the GPCRdb is not an intermediate active state because it is very similar to the receptor in the 3SN6 complex with Gs. Please, eliminate the inappropriate and speculative adjective "intermediate".

We have changed “intermediate” to “not fully active”, which is less speculative since full activation can be achieved only in the presence of the G protein.

Incidentally, in that model, the C-tail, eliminated by the authors, is completely wrong and occupies the G protein binding site. It is not clear to me the reason why the authors preferred to used an AlphaFold model as an input of simulations rather than a high resolution structural model, e.g. 4LDO. Perhaps, the reason is that all ICL regions, including ICL3, were modeled by AlphaFold even if with low confidence. I disagree with that choice.

We understand the reviewer’s point of view. Should we have simulated an “equilibrium” receptor-ligand complex, we would have made the same choice. However, the conformational changes occurring during a G protein binding are so consistent that the starting conformation of the receptor becomes almost irrelevant as long as a sensate structure is used.  

Reviewer #3 (Recommendations for the authors):

The revised version of the manuscript is more concise, focusing only on two systems. However, the authors have responded superficially to the reviewers' comments, merely deleting sections of text, making minor corrections, or adding small additions to the text. In particular, the authors have not addressed the main critical points raised by both Reviewer 2 and Reviewer 3. 

For example, the RMSD values for the binding of PF06882961 to GLP-1R remain high, raising doubts about the predictive capabilities of the method, at least for this type of system.

What is the RMSD of the ligand relative to the experimental pose obtained in the simulations? This value must be included in the text.

We have added this piece of information about PF06882961 RMSD in the text, which on page 6 now reads “We simulated the binding of PF06882961, reaching an RMSD to its bound conformation in 7LCJ of 3.79 +- 0.83 Å (computed on the second half of the merged trajectory, superimposing on GLP-1R Ca atoms of TMD residues 150 to 390), using multistep supervision on different system metrics (Figure 2) to model the structural hallmark of GLP-1R activation (Video S5, Video S6).”

Similarly, the activation mechanism of GLP-1R is only partially simulated.

Furthermore, it is not particularly meaningful to justify the high RMSD values of the SuMD simulations for the binding of Gs to GLP-1R by comparing them with those reported under unbiased MD conditions. "Replica 2, in particular, well reproduced the cryo-EM GLP-1R complex as suggested by RMSDs to 7LCI of 7.59{plus minus}1.58Å, 12.15{plus minus}2.13Å, and 13.73{plus minus}2.24Å for Gα, Gβ, and Gγ respectively. Such values are not far from the RMSDs measured in our previous simulations of GLP-1R in complex with Gs and GLP-149 (Gα = 6.18 {plus minus} 2.40 Å; Gβ = 7.22 {plus minus} 3.12 Å; Gγ = 9.30 {plus minus} 3.65 Å), which indicates overall higher flexibility of Gβ and Gγ compared to Gα, which acts as a sort of fulcrum bound to GLP-1R."

Without delving into the accuracy of the various calculations, the authors should acknowledge that comparing protein structures with such high RMSD values has no meaningful significance in terms of convergence toward the same three-dimensional structure.

The text has been edited to accommodate the reviewer’s suggestion and still give the readers the measure of the high flexibility of Gs bound to GLP-1R. It now reads “Such values do not support convergence with the static experimental structure but are not far from the RMSDs measured in our previous simulations of GLP-1R in complex with G_s and GLP-1 (G_α = 6.18 ± 2.40 Å; G_b = 7.22 ± 3.12 Å; G_g = 9.30 ± 3.65 Å), which indicates overall higher flexibility of G_b and G_g compared to G_α, which acts as a sort of fulcrum bound to GLP-1R.”

Have the authors simulated the binding of the Gs protein using the experimentally active structure of GLP-1R in complex with the ligand PF06882961 (PDB ID 7LCJ)? Such a simulation would be useful to assess the quality of the binding simulation of Gs to the GLP1R/PF06882961 complex obtained from the previous SuMD.

We considered performing the Gs binding simulation to the active structure of GLP-1R.

However, the GLP-1R (and other class B receptors) fully active state, as reported in 7LCJ, depends on the presence of the Gs and can be reached only upon effector coupling. Since it is unlikely that the unbound receptor is already in the fully active state, we reasoned that considering it as a starting point for Gs binding simulations would have been an artifact.

An example of the insufficient depth of the authors' replies can be seen in their response: "We note that among the suggested references, only Mafi et al report about a simulated G protein (in a pre-formed complex) and none of the work sampled TM6 rotation without input of energy."

This statement is inaccurate. For instance, D'Amore et al. (Chem 2024, doi: 10.1016/j.chempr.2024.08.004) simulated Gs coupling to A2A as well as TM6 rotation, as did Maria-Solano and Choi (eLife 2023, doi: 10.7554/eLife.90773.1). The former employed path collective variables metadynamics, which is not cited in the introduction or the discussion, despite its relevance to the methodologies mentioned.

Respectfully, our previous reply is correct, as all of the mentioned articles used enhanced (energy-biased) approaches, so the claim “none of the work sampled TM6 rotation without input of energy” stands. The reference to D’Amore et al. (published after the previous round of reviews of this manuscript) has been added to the introduction; we thank the reviewer for pointing it out. 

Additionally, SuMD employs a tabu algorithm that applies geometric supervision to the simulation, serving as an alternative approach to enhancing sampling compared to the "input of energy" techniques as called by the authors. A fair discussion should clearly acknowledge this aspect of the SuMD methodology.

We have now specified in the Methods that a tabù-like algorithm is part of SuMD, which, despite being the parent technique of mwSuMD, is not the focus of the present work. We provide extended references for readers interested in SuMD. mwSuMD, on the other hand, does not use a tabù-like algorithm but rather a continuative approach based on a score to select the best walker for each batch, as described in the Methods.

Author response:

Reviewer #1 (Evidence, reproducibility and clarity):

Minor comments:

In the results section (lines 498-499), the authors describe free kinetochores in many cells without associated spindle microtubules. However, some nuclei appear to have kinetochores, as presented in Figure 6. Could the authors clarify how this conclusion was derived using transmission electron microscopy (TEM) without serial sectioning, as this is not explicitly mentioned in the materials and methods?

We observed free kinetochores in the ALLAN-KO parasites with no associated spindle microtubules (see Fig. 6Gh), while kinetochores are attached to spindle microtubules in WT-GFP cells (see Fig. 6Gc). To provide further evidence we analysed additional images and found that ALLAN-KO cells have free kinetochores in the centre of nucleus, unattached to spindle microtubules. We provide some more images clearly showing free kinetochores in these cells (new supplementary Fig. S11).

However, in the ALLAN mutant, this difference is not absolute: in a search of over 50 cells, one example of a cell with a “normal” nuclear spindle and attached kinetochores was observed.

The use of serial sectioning has limitations for examining small structures like kinetochores in whole cells. The limitations of the various techniques (for example, SBF-SEM vs tomography) are highlighted in our previous study (Hair et al 2022; PMID: 38092766), and we consider that examining a population of randomly sectioned cells provides a better understanding of the overall incidence of specific features.

Discussion Section:

Could the authors expand on why SUN1 and ALLAN are not required during asexual replication, even though they play essential roles during male gametogenesis?

We observed no phenotype in asexual blood stage parasites associated with the sun1 and allan gene deletions. Several other Plasmodium berghei gene knockout parasites with a phenotype in sexual stages, for example CDPK4 (PMID: 15137943), SRPK (PMID: 20951971), PPKL (PMID: 23028336) and kinesin-5 (PMID: 33154955) have no phenotype in blood stages, so perhaps this is not surprising. One explanation may be the substantial differences in the mode of cell division between these two stages. Asexual blood stages produce new progeny (merozoites) over 24 hours with closed mitosis and asynchronous karyokinesis during schizogony, while male gametogenesis is a rapid process, completed within 15 min to produce eight flagellated gametes. During male gametogenesis the nuclear envelope must expand to accommodate the increased DNA content (from 1N to 8N) before cytokinesis. Furthermore, male gametogenesis is the only stage of the life cycle to make flagella, and axonemes must be assembled in the cytoplasm to produce the flagellated motile male gametes at the end of the process. Thus, these two stages of parasite development have some very different and specific features.

Lines 611-613 states: "These loops serve as structural hubs for spindle assembly and kinetochore attachment at the nuclear MTOC, separating nuclear and cytoplasmic compartments." Could the authors elaborate on the evidence supporting this statement?

We observed the loops/folds in the nuclear envelope (NE) as revealed by SUN1-GFP and 3D TEM images during male gametogenesis. These folds/loops occur mainly in the vicinity of the nuclear MTOC where the spindles are assembled (as visualised by EB1 fluorescence) and attached to kinetochores (as visualised by NDC80 fluorescence). These loops/folds may form due to the contraction of the spindle pole back to the nuclear periphery, inducing distortion of the NE. Since there is no physical segregation of chromosomes during the three rounds of mitosis (DNA increasing from 1N to 8N), we suggest that these folds provide additional space for spindle and kinetochore dynamics within an intact NE to maintain separation from the cytoplasm (as shown by location of kinesin-8B).

In lines 621-622, the authors suggest that ALLAN may have a broader role in NE remodelling across the parasite's lifecycle. Could they reflect on or remind readers of the finding that ALLAN is not essential during the asexual stage?

ALLAN-GFP is expressed throughout the parasite life cycle but as the reviewer points out, a functional role is more pronounced during male gametogenesis. This does not mean that it has no role at other stages of the life cycle even if there is no obvious phenotype following deletion of the gene during the asexual blood stage. The fact that ALLAN is not essential during the asexual blood stage is noted in lines 628-29.

Reviewer #2 (Evidence, reproducibility and clarity):

Introduction

Line 63: The authors stat: "NE is integral to mitosis, supporting spindle formation, kinetochore attachment, and chromosome segregation..". Seemingly at odds, they also say (Line 69) that 'open' "mitosis is "characterized by complete NE disassembly".

The authors could explain better the ideas presented in their quoted review from Dey and Baum, which points out that truly 'open' and 'closed' topologies may not exist and that even in 'open' mitosis, remnants of the NE may help support the mitotic spindle.

We have modified the sentence in which we discuss current opinions about ‘open’ and ‘closed’ mitosis. It is believed that there is no complete disassembly of the NE during open mitosis and no completely intact NE during closed mitosis, respectively. In fact, the NE plays a critical role in the different modes of mitosis during MTOC organisation and spindle dynamics. Please see the modified lines 64-71.

Results

Fig 7 is the final figure; but would be more useful upfront.

We have provided a new introductory figure (Fig 1) showing a schematic of conventional /canonical LINC complexes and evidence of SUN protein functions in model eukaryotes and compare them to what is known in apicomplexans.

Fig 1D. The authors generated a C-terminal GFP-tagged SUN1 transfectants and used ultrastructure expansion microscopy (U-ExM) and structured illumination microscopy (SIM) to examine SUN1-GFP in male gametocytes post-activation. The immuno-labelling of SUN1-GFP in these fixed cells appears very different to the live cell images of SUN1-GFP. The labelling profile comprises distinct punctate structures (particularly in the U-ExM images), suggesting that paraformaldehyde fixation process, followed by the addition of the primary and secondary antibodies has caused coalescing of the SUN1-GFP signal into particular regions within the NE.

We agree with the reviewer. Fixation with paraformaldehyde (PFA) results in a coalescence of the SUN1-GFP signal. We have also tried methanol fixation (see new Fig. S2), but a similar problem was encountered.

Given these fixation issues, the suggestion that the SUN1-GFP signal is concentrated at the BB/ nuclear MTOC and "enriched near spindle poles" needs further support.

These statements seem at odd with the data for live cell imaging where the SUN1-GFP seems evenly distributed around the nuclear periphery. Can the observation be quantitated by calculating the percentage of BB/ nuclear MTOC structures with associated SUN1-GFP puncta? If not, I am not convinced these data help understand the molecular events.

We agree with the reviewer that whilst the live cell imaging showed an even distribution of SUN1-GFP signal, after fixation with either PFA or methanol, then SUN1-GFP puncta are observed in addition to the peripheral location around the stained DNA (Hoechst) (See Fig. S2; puncta are indicated by arrows). These SUN1-GFP labelled puncta were observed at the junction of the nuclear MTOC and the basal body (Fig. 2F). Quantification of the distribution showed that these SUN1-GFP puncta are associated with nuclear MTOC in more than 90 % of cells (18 cells examined). Live cell imaging of the dual labelled parasites; SUN1xkinesin-8B (Fig. 2H) and SUN1x EB1 (Fig. 2I) provides further support for the association of SUN1-GFP puncta with BB (kinesin-8B) /nuclear MTOC (EB1).

The authors then generated dual transfectants and examined the relative locations of different markers in live cells. These data are more informative.

The authors state; " ..SUN1-GFP marked the NE with strong signals located near the nuclear MTOCs situated between the BB tetrads". The nuclear MTOCs are not labelled in this experiment. The SUN1-GFP signal between the kinesin-8B puncta is evident as small puncta on regions of NE distortion. I would prefer to not describe this signal as "strong". The signal is stronger in other regions of the NE.

We have modified the sentence on line 213 to accommodate this suggestion.

Line 219. The authors state; "..SUN1-GFP is partially colocalized with spindle poles as indicated by EB1,.. it shows no overlap with kinetochores (NDC80)." The authors should provide an analysis of the level of overlap at a pixel by pixel level to support this statement.

We now provide the overlap at a pixel-by-pixel level for representative images, and we have quantified more cells (n>30), as documented in the new Fig. S4A. We have also modified the sentence on line 219 to reflect these additions.

The SUN1 construct is C-terminally GFP-tagged. By analogy with human SUN1, the C-terminal SUN domain is expected to be in the NE lumen. That is in a different compartment to EB1, which is located in the nuclear lumen (on the spindle). Thus, the overlap of signal is expected to be minimal.

We agree with the reviewer that the overlap between EB1 and Sun1 signals is expected to be minimal. We have quantified the data and included it in Supplementary Fig. S4A.

Similarly, given that EB1 and NDC80 are known to occupy overlapping locations on the spindle, it seems unlikely that SUN1 can overlap with one and not the other.

We agree with the reviewer’s analysis that EB1 and NDC80 occupy overlapping locations on the spindle, although the length of NDC80 is less at the ends of spindles (see Author response image 1A) as shown in our previous study where we compared the locations of two spindle proteins, ARK2 and EB1, with that of NDC80 (Zeeshan et al, 2022; PMID: 37704606). In the present study we observed that Sun1-GFP partially overlaps with EB1 at the ends of the spindle, but not with NDC80. Please see Author response image 1B.

Author response image 1.

I note on Line 609, the authors state "Our study demonstrates that SUN1 is primarily localized to the nuclear side of the NE.." As per Fig 7D, and as discussed above, the bulk of the protein, including the SUN1 domain, is located in the space between the INM and the ONM.

We appreciate the reviewer’s correction; we have now modified the sentence to indicate that the protein is largely localized in the space between the INM and the ONM on line 617.

Interestingly, as the authors point out, nuclear membrane loops are evident around EB1 and NDC80 focal regions. The data suggests that the contraction of the spindle pole back to the nuclear periphery induces distortion of the NE.

We agree with the reviewer’s suggestion that the data indicate that contraction of spindle poles back to the nuclear periphery may induce distortion of the NE.

The author should discuss further the overlap of findings of this study with that from a recent manuscript (https://doi.org/10.1016/j.cels.2024.10.008). That Sayers et al. study identified a complex of SUN1 and ALLC1 as essential for male fertility in P. berghei. Sayers et al. also provide evidence that this complex particulate in the linkage of the MTOC to the NE and is needed for correct mitotic spindle formation during male gametogenesis.

We thank the reviewer for this suggestion. The study by Sayers et al, (2024) was published while our manuscript was under preparation. It was interesting to see that these complementary studies have similar findings about the role of SUN1 and the novel complex of SUN1-ALLAN. Our study contains a more detailed, in-depth analysis both by Expansion and TEM of SUN1. We include additional studies on the role of ALLAN.  We discuss the overlap in the findings of the two studies in lines 590-605.

While the work is interesting, the conclusions may need to be tempered. The authors suggestion that in the absence of KASH-domain proteins, the SUN1-ALLAN complex forms a non-canonical LINC complex (that is, a connection across the NE), that "achieves precise nuclear and cytoskeletal coordination".

We have toned down the wording of this conclusion in lines 665-677.

In other organisms, KASH interacts with the C-terminal domain on SUN1, which as mentioned above is located between the INM and ONM. By contrast, ALLAN interacts with the N-terminal domain of SUN1, which is located in the nuclear lumen. The SUN1-ALLAN interaction is clearly of interest, and ALLAN might replace some of the roles of lamins. However, the protein that functionally replaces KASH (i.e. links SUN1 to the ONM) remains unidentified.

We agree with reviewer, and future studies will need to focus on identifying the KASH replacement that links SUN1 to the ONM.

It may also be premature to suggest that the SUN1-ALLAN complex is promising target for blocking malaria transmission. How would it be targeted?

We have deleted the sentence that raised this suggestion.

While the above datasets are interesting and internally consistent, there are two other aspects of the manuscript that need further development before they can usefully contribute to the molecular story.

The authors undertook a transcriptomic analysis of Δsun1 and WT gametocytes, at 8 and 30 min post-activation, revealing moderate changes (~2-fold change) in different genes. GO-based analysis suggested up-regulation of genes involved in lipid metabolism. Given the modest changes, it may not be correct to conclude that "lipid metabolism and microtubule function may be critical functions for gametogenesis that can be perturbed by sun1 deletion." These changes may simply be a consequence of the stalled male gametocyte development.

Following the reviewer’s suggestion we have moved these data to the supplementary information (Fig. S5D-I) and toned down their discussion in the results and discussion sections.

The authors have then undertaken a detailed lipid analysis of the Δsun1 and WT gametocytes, before and after activation. Substantial changes in lipid metabolites might not be expected in such a short period of time. And indeed, the changes appear minimal. Similarly, there are only minor changes in a few lipid sub-classes between Δsun1 and WT gametocytes. In my opinion, the data are not sufficient to support the authors conclusion that "SUN1 plays a crucial role, linking lipid metabolism to NE remodelling and gamete formation."

In agreement with the reviewer’s comments we have moved  these data to supplementary information (Fig. S6) and substantially toned down the conclusions based on these findings.

Reviewer #3 (Evidence, reproducibility and clarity):

Major comments:

My main concern with this manuscript is that the authors do conclude not only that SUN1 is important for spindle formation and basal body segregation, but also that it influences for lipid metabolism and NE dynamics. I don't think the data supports this conclusion, for several reasons listed below. I would suggest to remove this claim from the manuscript or at least tone it down unless more supporting data are provided, in particular showing any change in NE dynamics in the SUN1-KO. Instead I would recommend to focus on the more interesting role of SUN1-ALLAN in bipartite MTOC organisation, which likely explains all observed phenotypes (including those in later stages of the parasite life cycle). In addition, some aspects of the knockout phenotype should be quantified to a bit deeper level.

In more detail:

- The lipidomics analysis is clearly the weakest point of the manuscript: The authors state that there are significant changes in some lipid populations between WT and sun1-KO, and between activated and non-activated cells, yet no statistical analysis is shown and the error bars are quite high compared to only minor changes in the means. For some discussed lipids, the result text does not match the graphs, e.g. PA, where the increase upon activation is more pronounced in the SUN1-KO vs WT (contrary to the text), or MAG, which is reduced in the SUN1-KO vs WT (contrary to the text). I don't see the discussed changes in arachidonic acid levels and myristic acid levels in the data either. Even if the authors find after analysis some statistically significant differences between some groups, they should carefully discuss the biological significance of these differences. As it is, I do not think the presented data warrants the conclusion that deletion of SUN1 changes lipid homeostasis, but rather shows that overall lipid homeostasis is not majorly affected by gametogenesis or SUN1 deletion. As a minor comment, if you decide to keep the lipidomics analysis in the manuscript, please state how many replicates were done.

As detailed above we have moved the lipidomics data to supplementary information (Fig. S6) and substantially toned down the discussion of these data in the results and discussion sections.

- I can't quite follow the logic why the authors performed transcriptomic analysis of the SUN1 and how they chose their time points. Their data up to this point indicate that SUN1 has a structural or coordinating role in the bipartite MTOC during male gametogenesis. Based on that it is rather unlikely that SUN1 KO directly leads to transcriptional changes within the 8 min of exflagellation. Isn't it more likely that transcriptional differences are purely a downstream effect of incomplete/failed gametogenesis? This is particularly true for the comparison at 30 min, which compares a mixture of exflagellated/emerged gametes and zygotes in WT to a mixture of aberrant, arrested gametes in the knockout, which will likely not give any meaningful insight. The by far most significant GO-term is then also nuclear-transcribed mRNA catabolic process, which is likely not related at all to SUN1 function (and the authors do not even comment on this in the main text). I would therefore suggest removing the 30 min data set from this manuscript. As a minor point, I would suggest highlighting some of the top de-regulated gene IDs in the volcano plots and stating their function. Also, please state how you prepared the cells for the transcriptomes and in how many replicates this was done.

As suggested by the reviewer we have removed the 30 min post activation data from the manuscript. We have also moved the rest of the transcriptomics data to supplementary information (Fig. S5) and toned down the presentation of this aspect of the work in the results and discussion sections.

- Live-cell imaging of SUN1-GFP does nicely visualise the NE during gametogenesis, showing a highly dynamic NE forming loops and folds, which is very exciting to see. It would be beneficial to also show a video from the life-cell imaging.

We have now added videos to the manuscript as suggested by the reviewer. Please see the supplementary Videos S1 and S2.

In their discussion, the authors state multiple times that NE dynamics are changed upon SUN1 KO. Yet, they do not provide data supporting this claim, i.e. that the extended loops and folds found in the nuclear envelope during gametogenesis are affected in any way by the knockout of SUN1 or ALLAN. What happens to the NE in absence of SUN1? Are there less loops and folds? In absence of a reliable NE marker this may not be entirely easy to address, but at least some SBF-SEM images of the sun1-KO gametocytes could provide insight.

It was difficult to provide SBF-SEM images as that work is beyond the scope of this manuscript. We will consider this approach in our future work. We re-examined many of our TEM images of SUN1-KO and ALLAN-KO parasites and did find some micrographs showing aberrant nuclear membrane folding (<5%) (Please see Author response image 2). However, we also observed similar structures in some of the WT-GFP samples (<5%), so we do not think this is a strong phenotype of the SUN1 or ALLAN mutants.

Author response image 2.

- I think the exciting part of the manuscript is the cell biological role of SUN1 on male gametogenesis, which could be carved out a bit more by a more detailed phenotyping. Specifically it would be good to quantify

(1) If DNA replication to an octoploid state still occurs in SUN1-KO and ALLAN-KO,

DNA replication is not affected in the SUN1-KO and ALLAN-KO mutants: DNA content increases to 8N (data added in Fig. 3J and Fig. S10F).

(2) The proportion of anucleated gametes in WT and the KO lines

We have added these data in Fig. 3K and Fig. S10G

(3) A quantification of the BB clustering phenotype (in which proportion of cells do the authors see this phenotype). This could be addressed by simple fixed immunofluorescence images of the respective WT/KO lines at various time points after activation (or possibly by reanalysis of the already obtained images) and would really improve the manuscript.

We have reanalysed the BB clustering phenotype and added the quantitative data in Fig. 4E and Fig. S7.

Especially the claim that emerged SUN1-KO gametes lack a nucleus is currently only based on single slices of few TEM cells and would benefit from a more thorough quantification in both SUN1- and ALLAN-Kos

We have examined many microgametes (100+ sections). In WT parasites a small proportion of gametes can appear to lack a nucleus if it does not extend all the way to the apical and basal ends (Hair et al. 2022). However, the proportion of microgametes that appear to lack a nucleus (no nucleus seen in any section) was much higher in the SUN1 mutant. In contrast, this difference was not as clear cut in the ALLAN mutant with a small proportion of intact (with axoneme and nucleus) microgametes being observed.

We have done additional analysis of male gametes, looking for the presence of the nucleus by live cell imaging after DNA staining with Hoechst. These data are added in Fig. 3K (for Sun1-KO) and Fig. S10G (for Allan-KO).

- The TEM suggests that in the SUN1-KO, kinetochores are free in the nucleus. Are all kinetochores free or do some still associate to a (minor/incorrectly formed) spindle? The authors could address this by tagging NDC80 in the KO lines.

Our observation and quantification of the data indicated that 100% of kinetochores were attached to spindle microtubules and that 0% were unattached kinetochores in the WT parasites. However, the exact opposite was found for the SUN1 mutant with 100% unattached kinetochores and 0% attached. The result was not quite as clear cut in the ALLAN mutant, with 98% unattached and 2% attached. An important observation was the lack of separation of the nuclear poles and any spindle formation. Spindle formation was never or very rarely observed in the mutants.

- Finally, I think it is curious that in contrast to SUN1, ALLAN seems to be less important, with some KO parasite completing the life cycle. Maybe a more detailed phenotyping as above gives some more hints to where the phenotypic difference between the two proteins lies. I would assume some ALLAN-KO cells can still segregate the basal body. Can the authors speculate/discuss in more detail why these two proteins seems to have slightly different phenotypes?

We agree with the reviewer. Overall, the ALLAN-KO has a less prominent phenotype than that of the Sun1-KO. The main difference is that in the ALLAN-KO mutant some basal body segregation can occur, leading to the production of some fertile microgametocytes, and ookinetes, and oocyst formation (Fig. 8). Approximately 5% of oocysts sporulated to release infective sporozoites that could infect mice in bite back experiments and complete the life cycle. In contrast the Sun1-KO mutant made no healthy oocysts, or infective sporozoites, and could not complete the life cycle in bite back experiments. We have analysed the phenotype in detail and provide quantitative data for gametocyte stages by EM and ExM in Figs. 4 and S8 (SUN1) and Figs. 7 and S11 (ALLAN). We have also performed detailed analysis of oocyst and sporozoite stages and included the data in Fig. 3 (SUN1) and S10 (ALLAN).

Based on the location, and functional and interactome data, we think that SUN1 plays a central role in coordinating nucleoplasm and cytoplasmic events as a key component of the nuclear membrane lumen, whereas ALLAN is located in the nucleoplasm. Deleting the SUN1 gene may disrupt the connection between INM and ONM whereas the deletion of ALLAN may affect only the INM.

Some additional points where the data is not entirely sound yet or could be improved:

- Localisation of SUN1: There seems to be a discrepancy between SUN1-GFP location as observed by live cell microscopy, and by Expansion Microscopy (ExM), similar for ALLAN-GFP. By live-cell microscopy, the SUN1 localisation is much more evenly distributed around the NE, while the localisation in ExM is much more punctuated, and e.g. in Figure 1E seems to be within the nucleus. Do the authors have an explanation for this? Also, in Fig. 1D there are two GFP foci at the cell periphery (bottom left of the image), which I would think are not SUN1-Foci, as they seem to be outside of the cell. Is the antibody specific? Was there a negative control done for the antibody (WT cells stained with GFP antibodies after ExM)?

High resolution SIM and expansion microscopy showed that the SUN1-GFP molecules coalesce to form puncta, in contrast to the more uniform distribution observed by live cell imaging. This apparent difference may be due to a better resolution that could not be achieved by live cell imaging. We agree with the reviewer that the two green foci are outside of the cell. As a negative control we have used WT-ANKA cells (which contain no GFP) and the anti-GFP antibody, which gave no signal. This confirms the specificity of the antibody (please see the new Fig. S3). 

- The authors argue that SIM gave unexpected results due to PFA fixation leading to collapse of the NE loops. However, they also fix their ExM cells and their EM cells with PFA and do not observe a collapse, at least from what I see in the two presented images and in the 3D reconstruction. Is there something else different in the sample preparation?

There was no difference in the fixation process for samples examined by SIM and ExM, but we used an anti-GFP antibody in ExM to visualise the SUN1-GFP, while in SIM the images of GFP signal were collected directly after fixation.  We used both PFA and methanol as fixative, and both methods showed a coalescing of the SUN1-GFP signal (please see the new Fig. S2 and S3).

Can the authors trace their NE in ExM according to the NHS-Ester signal?

We could trace the NE in the ExM by the NHS-ester signal and observed that the SUN1-GFP signal was largely coincident with the NE (Please see the new Fig. S3B).

- Fig 2D: It would be good to not just show images of oocysts but actually quantify their size from images. Also, have the authors determined the sporozoite numbers in SUN1-KO?

We have measured oocyst size (data added in new Fig. 3) and added the sporozoite quantification data in Fig. 3D.

- Line 481-483: the authors state that oocyst size is reduced in ALLAN-KO but do not show the data. Please quantify oocyst size or at least show representative images. Also the drastic decrease in sporozoite numbers (Fig. 6D, E) is not mentioned in the text. Please add reference to Fig S7D when talking about the bite back data.

We have added the oocyst size data in Fig. S10. We mention the changes in sporozoite numbers (now  shown in Fig. 7D, E), and refer to  the bite back data shown in current Fig. 7E.

- Fig S1C, 6C: Both WB images are stitched, but this is not clearly indicated e.g. by leaving a small gap between the lanes. Also please show a loading control along with the western blots. Also there seems to be a (unspecific?) band in the control, running at the same height as Allan-GFP WB. What exactly is the control?

We have provided the original blot showing the bands of ALLAN-GFP and SUN1-GFP. As a positive control, we used an RNA associated protein (RAP-GFP) that is highly expressed in Plasmodium and regularly used in our lab for this purpose.

- Regarding the crossing experiment: The authors conclude from this cross that SUN1 is only needed in males, yet for this conclusion they would need to also show that a cross with a female line does not rescue the phenotype. The authors should repeat the cross with a male-deficient line to really test if the phenotype is an exclusively male phenotype. In addition, line 270-272 states that no oocysts/sporozoites were detected in sun1-ko and nek4-ko parasites. However, the figure 2E shows only oocysts, not sporozoites, and shows also that sun1-ko does form oocysts, albeit dead ones.

We have now performed the experiment of crossing the Sun1-KO parasite line with a male deficient line (Hap2-KO) and added the data in Fig. 3I. We have added images showing sporozoites in oocysts.

- In Fig S1 the authors show that they also generated a SUN1-mCherry line, yet they do not use it in any of the presented experiments (unless I missed it). Would it be beneficial to cross the SUN1-mCherry line with the Allan1-GFP line to test colocalisation (possibly also by expansion microscopy)?

We did generate a SUN1-mCherry line, with the intent to cross ALLAN-GFP and SUN1-mCherry lines and observe the co-location of the proteins. Despite multiple attempts this cross was unsuccessful. This may have been due to their close proximity such that the addition of both GFP and mCherry was difficult to facilitate a proper protein-protein interaction between either of the proteins.

- Line 498: "In a significant proportion of cells" - What was the proportion of cells, and what does significant mean in this context?

Approximately 67% of cells showed the clumping of BBs. We have now added the numbers in Figs. 6H and S11I.

- The authors should discuss a bit more how their work relates to the work of Sayers et al. 2024, which also identified the SUN1-ALLAN complex. The paper is cited, but only very briefly commented on.

We have extended this discussion now in lines 590-605.

Suggestions how to improve the writing and data presentation.

- General presentation of microscopy images: Considering that large parts of the manuscript are based on microscopy data, their presentation could be improved. Single-channel microscopy images would benefit from being depicted in gray scale instead of color, which would make it easier to see the structures and intensities (especially for blue channels).

Whilst we agree with the reviewer, sometimes it is difficult to see the features in the merged images. Therefore, we would like to request to be allowed to retain the colours, which can be easily followed in both individual and merged images.

Also, it would be good to harmonize in which panels arrows are shown (e.g. Fig 1G, where some white arrows are in the SUN1-GFP panel, while others are in the merge panel, but they presumably indicate the same thing.). At the same time, Fig 1H doesn't have any with arrows, even though the figure legend states so.

We apologise for this lack of consistency, and we have now added arrows wherever they are missing to harmonise in the presentations.

Fig 3A and S4 show the same experiment but are coloured in different colours (NHS-Eester in green vs grey scale).

- Are the scale bars of all expansion microscopy images adjusted for the expansion factor?

Yes, the scale bars are adjusted accordingly.

- The figure legends would benefit from streamlining, as they have very different style between figures (eg Fig. 6 which has a concise figure legend vs microscopy figures where figure legends are very long and describe not only the figure but the results)

The figure legends have been streamlined, with removal of the description of results.

- Line 155-156: The text makes it sound like the expression only happens after activation. is that the case? Are these images activated or non-activated gametocytes?

They are expressed before activation, but the signal intensifies after activation. Images from before and after activation of gametocytes have been added in Fig. S1F.

- Line 267: Reference to the original nek4-KO paper missing

This reference is now included.

- Line 301: The reference to Figure 2J seems to be a bit arbitrarily placed. Also, this schematic of lipid metabolism is never discussed in relation to the transcriptomic or lipidomic data.

We have moved these data to supplementary information and modified the text.

- Line 347-349 states that gametes emerged, but the referenced figure shows activated gametocytes before exflagellation.

We have corrected the text to the start of exflagellation.

- Line 588: Spelling mistake in SUN1-domain

Corrected.

- Line 726/731: i missing in anti-GFP

Corrected.

- Line 787-789: statement of scale bar and number of cells imaged is not at the right position in the figure legend.

Moved to right place

- Line 779, 783: "shades of green" should be just "green". Same goes for line 986, 989 with "shades of grey"

Changed.

- Line 974, 976: please correct to WT-GFP and dsun1

Corrected.

- Line 1041, 1044: WT-GFP instead of WTGFP.

Corrected to WT-GFP.

- Fig 1B, D, E, Fig S1G, H: What are the time points of imaging?

We have added the time points to the images in these figures.

- Fig 1D/Line 727: the scale of the scale bar on the inset is missing.

We have added the scale bar.

- Fig 3 E-G and 6H-J: Please indicate total number of cells/images analysed per quantification, either in the graphs themselves or in the figure legend.

We indicate now the number of cells analysed in individual figures and also in Fig. S5C and S8C, respectively.

- Fig 5B: What is NP

Nuclear Pole (NP), also known as the nuclear/acentriolar MTOC (Zeeshan et al 2022; PMID: 35550346).

- Fig S1B/D: The legend states that there is an arrow indicating the band, but there is none.

We have added the arrow.

- Fig S2C: Is the scale bar really the same for the zygote and the ookinete?

We have checked this and used the same for both zygote and ookinete.

- Fig S3C, S7C: which stages was qRT-PCR done on?

Gametocytes activated for 8 min.

- Fig. S3D, S7D: According to the figure legend, three independent experiments were performed. How many mice were used per experiment? It would be good to depict the individual data points instead of the bar graph. For S7D, 3 data points are depicted (one in WT, two in allan-KO), what do they mean?

The bite back experiment was performed using 15-20 mosquitoes infected with WT-GFP and gene knockout lines to feed on one naïve mouse each, in three different experiments. We have now included the data points in the bar diagrams.

- Fig S3: Panel letters E and G are missing

We have updated the lettering in current Fig. S5

- Fig 3D: Please indicate what those boxes are. I presume that these are the insets show in b, e and j, but it is never mentioned. J is not even larger than i. Also, f is quite cropped, it would be good to see the large-scale image it comes from to see where in the nucleus these kinetochores are placed. Were there unbound kinetochores found in WT?

We mention the boxes in the figure legends. It is rare to find unbound kinetochores in WT parasite. We provide large scale and zoomed-in images of free kinetochores in Fig. S8.

- Fig S4: Insets are not mentioned in the figure legend. Please add scale bar to zoom-ins

We now describe the insets in the figure legends and have added scale bars to the zoomed-in images.

- Fig S5A, B: Please indicate which inset belongs to which sub-panel. Where does Ac stem from?

We have now included the full image showing the inset (new Fig. S8).

- Fig S5C and S8C: Change "DNA" to "Nucleus".

We have changed “DNA” to “Nucleus”. Now they are Fig. S8K and S11I.

Reviewer #3 (Significance):

Yet, the statement that SUN1 is also important for lipid homoeostasis and NE dynamics is currently not backed up by sufficient data. I believe that the manuscript would benefit from removing the less convincing transcriptomic and lipidomic datasets and rather focus on more deeply characterising the cell biology of the knockouts. This way, the results would be interesting not only for parasitologists, but also for more general cell biologists.

We have moved the lipidomics and transcriptomics data to supplementary information and toned down the emphasis on these data to make the manuscript more focused on the cell biology and analysis of the genetic KO data.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

This paper contains what could be described as a "classic" approach towards evaluating a novel taste stimuli in an animal model, including standard behavioral tests (some with nerve transections), taste nerve physiology, and immunocytochemistry of taste cells of the tongue. The stimulus being tested is ornithine, from a class of stimuli called "kokumi" (in terms of human taste); these kokumi stimuli appear to enhance other canonical tastes, increasing what are essentially hedonic attributes of other stimuli. The mechanism for ornithine detection is thought to be GPRC6A receptors expressed in taste cells. The authors showed evidence for this in an earlier paper with mice; this paper evaluates ornithine taste in a rat model, and comes to a similar conclusion, albeit with some small differences between the two rodent species.

Strengths:

The data show effects of ornithine on taste/intake in laboratory rats: In two-bottle and briefer intake tests, adding ornithine results in higher intake of most, but all not all stimuli tested. Bilateral chorda tympani (CT) nerve cuts or the addition of GPRC6A antagonists decreased or eliminated these effects. Ornithine also evoked responses by itself in the CT nerve, but mainly at higher concentrations; at lower concentrations it potentiated the response to monosodium glutamate. Finally, immunocytochemistry of taste cell expression indicated that GPRC6A was expressed predominantly in the anterior tongue, and co-localized (to a small extent) with only IP3R3, indicative of expression in a subset of type II taste receptor cells.

Weaknesses:

As the authors are aware, it is difficult to assess a complex human taste with complex attributes, such as kokumi, in an animal model. In these experiments they attempt to uncover mechanistic insights about how ornithine potentiates other stimuli by using a variety of established experimental approaches in rats. They partially succeed by finding evidence that GPRC6A may mediate effects of ornithine when it is used at lower concentrations. In the revision they have scaled back their interpretations accordingly. A supplementary experiment measuring certain aspects of the effects of ornithine added to Miso soup in human subjects is included for the express purpose of establishing that the kokumi sensation of a complex solution is enhanced by ornithine; however, they do not use any such complex solutions in the rat studies. Moreover, the sample size of the human experiment is (still) small - it really doesn't belong in the same manuscript with the rat studies.

Despite the reviewer’s suggestion, we would like to include the human sensory experiment. Our rationale is that we must first demonstrate that the kokumi of miso soup is enhanced by the addition of ornithine, which is then followed by basic animal experiments to investigate the underlying mechanisms of kokumi in humans.

We did not present the additive effects of ornithine on miso soup in the present rat study because our previous companion paper (Fig. 1B in Mizuta et al., 2021, Ref. #26) already confirmed that miso soup supplemented with 3 mM L-ornithine (but not D-ornithine) was statistically significantly (P < 0.001) preferred to plain miso soup by mice.

Furthermore, we believe that our sample size (n = 22) is comparable to those employed in other studies. For example, the representative kokumi studies by Ohsu et al. (Ref. #9), Ueda et al. (Ref. #10), Shibata et al. (Ref. #20), Dunkel et al. (Ref. #37), and Yang et al. (Ref. #44) used sample sizes of 20, 19, 17, 9, and 15, respectively.

Reviewer #2 (Public review):

Summary:

The authors used rats to determine the receptor for a food-related perception (kokumi) that has been characterized in humans. They employ a combination of behavioral, electrophysiological, and immunohistochemical results to support their conclusion that ornithine-mediated kokumi effects are mediated by the GPRC6A receptor. They complemented the rat data with some human psychophysical data. I find the results intriguing, but believe that the authors overinterpret their data.

Strengths:

The authors provide compelling evidence that ornithine enhances the palatability of several chemical stimuli (i.e., IMP, MSG, MPG, Intralipos, sucrose, NaCl, quinine). Ornithine also increases CT nerve responses to MSG. Additionally, the authors provide evidence that the effects of ornithine are mediated by GPRC6A, a G-protein-coupled receptor family C group 6 subtype A, and that this receptor is expressed primarily in fungiform taste buds. Taken together, these results indicate that ornithine enhances the palatability of multiple taste stimuli in rats and that the enhancement is mediated, at least in part, within fungiform taste buds. This is an important finding that could stand on its own. The question of whether ornithine produces these effects by eliciting kokumi-like perceptions (see below) should be presented as speculation in the Discussion section.

Weaknesses:

I am still unconvinced that the measurements in rats reflect the "kokumi" taste percept described in humans. The authors conducted long-term preference tests, 10-min avidity tests and whole chorda tympani (CT) nerve recordings. None of these procedures specifically model features of "kokumi" perception in humans, which (according to the authors) include increasing "intensity of whole complex tastes (rich flavor with complex tastes), mouthfulness (spread of taste and flavor throughout the oral cavity), and persistence of taste (lingering flavor)." While it may be possible to develop behavioral assays in rats (or mice) that effectively model kokumi taste perception in humans, the authors have not made any effort to do so. As a result, I do not think that the rat data provide support for the main conclusion of the study--that "ornithine is a kokumi substance and GPRC6A is a novel kokumi receptor."

Kokumi can be assessed in humans, as demonstrated by the enhanced kokumi perception observed when miso soup is supplemented with ornithine (Fig. S1). Currently, we do not have a method to measure the same kokumi perception in animals. However, in the two-bottle preference test, our previous companion paper (Fig. 1B in Mizuta et al. 2021, Ref. #26) confirmed that miso soup supplemented with 3 mM L-ornithine (but not D-ornithine) was statistically significantly (P < 0.001) preferred over plain miso soup by mice.

Of the three attributes of kokumi perception in humans, the “intensity of whole complex tastes (rich flavor with complex tastes)” was partly demonstrated in the present rat study. In contrast, “mouthfulness (the spread of taste and flavor throughout the oral cavity)” could not be directly detected in animals and had to be inferred in the Discussion. “Persistence of taste (lingering flavor)” was evident at least in the chorda tympani responses; however, because the tongue was rinsed 30 seconds after the onset of stimulation, the duration of the response was not fully recorded.

It is well accepted in sensory physiology that the stronger the stimulus, the larger the tonic response—and consequently, the longer it takes for the response to return to baseline. For example, Kawasaki et al. (2016, Ref. #45) clearly showed that the duration of sensation increased proportionally with the concentration of MSG, lactic acid, and NaCl in human sensory tests. The essence of this explanation has been incorporated into the Discussion (p. 12).

Why are the authors hypothesizing that the primary impacts of ornithine are on the peripheral taste system? While the CT recordings provide support for peripheral taste enhancement, they do not rule out the possibility of additional central enhancement. Indeed, based on the definition of human kokumi described above, it is likely that the effects of kokumi stimuli in humans are mediated at least in part by the central flavor system.

We agree with the reviewer’s comment. Our CT recordings indicate that the effects of kokumi stimuli on taste enhancement occur primarily at the peripheral taste organs. The resulting sensory signals are then transmitted to the brain, where they are processed by the central gustatory and flavor systems, ultimately giving rise to kokumi attributes. This central involvement in kokumi perception is discussed on page 12. Although kokumi substances exert their effects at low concentrations—levels at which the substance itself (e.g., ornithine) does not become more favorable or (in the case of γ-Glu-Val-Gly) exhibits no distinct taste—we cannot rule out the possibility that even faint taste signals from these substances are transmitted to the brain and interact with other taste modalities.

The authors include (in the supplemental data section) a pilot study that examined the impact of ornithine on variety of subjective measures of flavor perception in humans. The presence of this pilot study within the larger rat study does not really mice sense. While I agree with the authors that there is value in conducting parallel tests in both humans and rodents, I think that this can only be done effectively when the measurements in both species are the same. For this reason, I recommend that the human data be published in a separate article.

Despite the reviewer’s suggestion, we intend to include the human sensory experiment. Our rationale is that we must first demonstrate that the kokumi of miso soup is enhanced by the addition of ornithine, and then follow up with basic animal experiments to investigate the potential underlying mechanisms of kokumi in humans.

In our previous companion paper (Fig. 1B in Mizuta et al., 2021, Ref. #26), we confirmed with statistical significance (P < 0.001) that mice preferred miso soup supplemented with 3 mM L-ornithine (but not D-ornithine) over plain miso soup. However, as explained in our response to Reviewer #2’s first concern (in the Public review), it is difficult to measure two of the three kokumi attributes—aside from the “intensity of whole complex tastes (rich flavor with complex tastes)”—in animal models.

The authors indicated on several occasions (e.g., see Abstract) that ornithine produced "synergistic" effects on the CT nerve response to chemical stimuli. "Synergy" is used to describe a situation where two stimuli produce an effect that is greater than the sum of the response to each stimulus alone (i.e., 2 + 2 = 5). As far as I can tell, the CT recordings in Fig. 3 do not reflect a synergism.

We appreciate your comments regarding the definition of synergy. In Fig. 5 (not Fig. 3), please note the difference in the scaling of the ordinate between Fig. 5D (ornithine responses) and Fig. 5E (MSG responses). When both responses are presented on the same scale, it becomes evident that the response to 1 mM ornithine is negligibly small compared to the MSG response, which clearly indicates that the response to the mixture of MSG and 1 mM ornithine exceeds the sum of the individual responses to MSG and 1 mM ornithine. Therefore, we have described the effect as “synergistic” rather than “additive.” The same observation applies to the mice experiments in our previous companion paper (Fig. 8 in Mizuta et al. 2021, Ref. #26), where synergistic effects are similarly demonstrated by graphical representation. We have also added the following sentence to the legend of Fig. 5:

“Note the different scaling of the ordinate in (D) and (E).”

Reviewer #3 (Public review):

Summary:

In this study the authors set out to investigate whether GPRC6A mediates kokumi taste initiated by the amino acid L-ornithine. They used Wistar rats, a standard laboratory strain, as the primary model and also performed an informative taste test in humans, in which miso soup was supplemented with various concentrations of L-ornithine. The findings are valuable and overall the evidence is solid. L-Ornithine should be considered to be a useful test substance in future studies of kokumi taste and the class C G protein coupled receptor known as GPRC6A (C6A) along with its homolog, the calcium-sensing receptor (CaSR) should be considered candidate mediators of kokumi taste. The researchers confirmed in rats their previous work on Ornithine and C6A in mice (Mizuta et al Nutrients 2021).

Strengths:

The overall experimental design is solid based on two bottle preference tests in rats. After determining the optimal concentration for L-Ornithine (1 mM) in the presence of MSG, it was added to various tastants including: inosine 5'-monophosphate; monosodium glutamate (MSG); mono-potassium glutamate (MPG); intralipos (a soybean oil emulsion); sucrose; sodium chloride (NaCl; salt); citric acid (sour) and quinine hydrochloride (bitter). Robust effects of ornithine were observed in the cases of IMP, MSG, MPG and sucrose; and little or no effects were observed in the cases of sodium chloride, citric acid; quinine HCl. The researchers then focused on the preference for Ornithine-containing MSG solutions. Inclusion of the C6A inhibitors Calindol (0.3 mM but not 0.06 mM) or the gallate derivative EGCG (0.1 mM but not 0.03 mM) eliminated the preference for solutions that contained Ornithine in addition to MSG. The researchers next performed transections of the chord tympani nerves (with sham operation controls) in anesthetized rats to identify a role of the chorda tympani branches of the facial nerves (cranial nerve VII) in the preference for Ornithine-containing MSG solutions. This finding implicates the anterior half-two thirds of the tongue in ornithine-induced kokumi taste. They then used electrical recordings from intact chorda tympani nerves in anesthetized rats to demonstrate that ornithine enhanced MSG-induced responses following the application of tastants to the anterior surface of the tongue. They went on to show that this enhanced response was insensitive to amiloride, selected to inhibit 'salt tastant' responses mediated by the epithelial Na+ channel, but eliminated by Calindol. Finally they performed immunohistochemistry on sections of rat tongue demonstrating C6A positive spindle-shaped cells in fungiform papillae that partially overlapped in its distribution with the IP3 type-3 receptor, used as a marker of Type-II cells, but not with (i) gustducin, the G protein partner of Tas1 receptors (T1Rs), used as a marker of a subset of type-II cells; or (ii) 5-HT (serotonin) and Synaptosome-associated protein 25 kDa (SNAP-25) used as markers of Type-III cells.

At least two other receptors in addition to C6A might mediate taste responses to ornithine: (i) the CaSR, which binds and responds to multiple L-amino acids (Conigrave et al, PNAS 2000), and which has been previously reported to mediate kokumi taste (Ohsu et al., JBC 2010) as well as responses to Ornithine (Shin et al., Cell Signaling 2020); and (ii) T1R1/T1R3 heterodimers which also respond to L-amino acids and exhibit enhanced responses to IMP (Nelson et al., Nature 2001). These alternatives are appropriately discussed and, taken together, the experimental results favor the authors' interpretation that C6A mediates the Ornithine responses. The authors provide preliminary data in Suppl. 3 for the possibility of co-expression of C6A with the CaSR.

Weaknesses:

The authors point out that animal models pose some difficulties of interpretation in studies of taste and raise the possibility in the Discussion that umami substances may enhance the taste response to ornithine (Line 271, Page 9).

Ornithine and umami substances interact to produce synergistic effects in both directions—ornithine enhances responses to umami substances, and vice versa. These effects may depend on the concentrations used, as described in the Discussion (pp. 9–10). Further studies are required to clarify the precise nature of this interaction.

One issue that is not addressed, and could be usefully addressed in the Discussion, relates to the potential effects of kokumi substances on the threshold concentrations of key tastants such as glutamate. Thus, an extension of taste distribution to additional areas of the mouth (previously referred to as 'mouthfulness') and persistence of taste/flavor responses (previously referred to as 'continuity') could arise from a reduction in the threshold concentrations of umami and other substances that evoke taste responses.

Thank you for this important suggestion. If ornithine reduces the threshold concentrations of tastants—including glutamate—and enhances their suprathreshold responses, then adding ornithine may activate additional taste cells. This effect could explain kokumi attributes such as an “extension of taste distribution” and possibly the “persistence of responses.” As shown in Fig. 2, the lowest concentrations used for each taste stimulus are near or below the thresholds, which indicates that threshold concentrations are reduced—especially for MSG and MPG. We have incorporated this possibility into the Discussion as follows (p.12):

“Kokumi substances may reduce the threshold concentrations as well as they increase the suprathreshold responses of tastants. Once the threshold concentrations are lowered, additional taste cells in the oral cavity become activated, and this information is transmitted to the brain. As a result, the brain perceives this input as coming from a wider area of the mouth.”

The status of one of the compounds used as an inhibitor of C6A, the gallate derivative EGCG, as a potential inhibitor of the CaSR or T1R1/T1R3 is unknown. It would have been helpful to show that a specific inhibitor of the CaSR failed to block the ornithine response.

Thank you for this important comment. We attempted to identify a specific inhibitor of CaSR. Although we considered using NPS-2143—a commonly used CaSR inhibitor—it is known to also inhibit GPRC6A. We agree that using a specific CaSR inhibitor would be beneficial and plan to pursue this in future studies.

It would have been helpful to include a positive control kokumi substance in the two bottle preference experiment (e.g., one of the known gamma glutamyl peptides such as gamma-glu-Val-Gly or glutathione), to compare the relative potencies of the control kokumi compound and Ornithine, and to compare the sensitivities of the two responses to C6A and CaSR inhibitors.

We agree with this comment. In retrospect, it may have been advantageous to directly compare the potencies of CaSR and GPRC6A agonists in enhancing taste preferences—and to evaluate the sensitivity of these preferences to CaSR and GPRC6A antagonists. However, we did not include γ-Glu-Val-Gly in the present study because we have already reported its supplementation effects on the ingestion of basic taste solutions in rats using the same methodology in a separate paper (Yamamoto and Mizuta, 2022, Ref. #25). The results from both studies are compared in the Discussion (p. 11).

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Major:

I am not convinced by the Author's arguments for including the human data. I appreciate their efforts in adding a few (5) subjects and improving the description, but it still feels like it is shoehorned into this paper, and would be better published as a different manuscript.

This human study is short, but it is complete rather than preliminary. The rationale for us to include the human data as supplementary information is shown in responses to the reviewer’s Public review.

Minor concerns:

Page 3 paragraph 1: Suggest "contributing to palatability".

Thank you for this suggestion. We have rewritten the text as follows:

“…, the brain further processes these sensations to evoke emotional responses, contributing to palatability or unpleasantness”.

Page 4 paragraph 2: The text still assumes that "kokumi" is a meaningful descriptor for what rodents experience. Re-wording the following sentence like this could help:

"Neuroscientific studies in mice and rats provide evidence that gluthione and y-Glu-Val-Gly activate CaSRs, and modify behavioral responses to other tastants in a way that may correspond to kokumi taste as experienced by humans. However, to our..."

Or something similar.

Thank you for this suggestion. We have rewritten the sentence according to your suggestion as follows:

"Neuroscientific studies (23,25,30) in mice and rats provide evidence that glutathione and y-Glu-Val-Gly activate CaSRs, and modify behavioral responses to other tastants in a way that may correspond to kokumi as experienced by humans”.

Page 7 paragraph 1 - put the concentrations of Calindol and EGCG used (in the physiology exps) in the text.

We have added the concentrations: “300 µM calindol and 100 µM EGCG”.

Reviewer #2 (Recommendations for the authors):

I have included all of my recommendations in the public review section.

Reviewer #3 (Recommendations for the authors):

Although the definitions of 'thickness', 'mouthfulness' and 'continuity' have been revised very helpfully in the Introduction, 'mouthfulness' reappears at other points in the MS e.g., Page 4, Results, Line 3; Page 9, Line 3. It is best replaced by the new definition in these other locations too.

We wish to clarify that our revised text stated, “…to clarify that kokumi attributes are inherently gustatory, in the present study we use the terms ‘intensity of whole complex tastes (rich flavor with complex tastes)’ instead of ‘thickness,’ ‘mouthfulness (spread of taste and flavor throughout the oral cavity)’ instead of ‘continuity,’ and ‘persistence of taste (lingering flavor)’ instead of ‘continuity.’” The term “mouthfulness” was retained in our text, though we provided a more specific explanation. In the re-revised version, we have added “(spread of taste in the oral cavity)” immediately after “mouthfulness.”

I doubt that many scientific readers will be familliar with the term 'intragemmal nerve fibres' (Page 8, Line 4). It is used appropriately but it would be helpful to briefly define/explain it.

We have added an explanation as follows:

“… intragemmal nerve fibers, which are nerve processes that extend directly into the structure of the taste bud to transmit taste signals from taste cells to the brain.”

I previously pointed out the overlap between the CaSR's amino acid (AA) and gamma-glutamyl-peptide binding site. I was surprised by the authors' response which appeared to miss the point being made. It was based on the impacts of selected mutations in the receptor's Venus FlyTrap domain (Broadhead JBC 2011) on the responses to AAs and glutathione analogs. The significantly more active analog, S-methylglutathione is of additional interest because, like glutathione itself, it is present in mammalian body fluids. My apologies to the authors for not more carefully explaining this point.

Thank you for this comment. Both CaSR and GPRC6A are recognized as broad-spectrum amino acid sensors; however, their agonist profiles differ. Aromatic amino acids preferentially activate CaSR, whereas basic amino acids tend to activate GPRC6A. For instance, among basic amino acids, ornithine is a potent and specific activator of GPRC6A, while γ-Glu-Val-Gly in addition to amino acids is a high-potency activator of CaSR. It remains unclear how effectively ornithine activates CaSR and whether γ-glutamyl peptides also activate GPRC6A. These questions should be addressed in future studies.

Author response:

We thank the reviewers for their evaluation, for helpful suggestions to improve clarity and accuracy, and for their positive reception of the manuscript. We will incorporate their suggestions in a revised manuscript. Here, we respond to their major comments. 

The reviewers suggest that a molecular study of Hofstenia’s reproductive systems would be beneficial, as would mechanistic explanations for its unusual reproductive behavior. We agree with the reviewers that both of these would be interesting avenues, although we think this is outside the scope of this current manuscript. This manuscript studies growth and reproductive dynamics in acoels, and establishes a foundation to study its underlying molecular, developmental, and physiological machinery. 

Our previous molecular work, using scRNAseq and FISH, identified several germline markers. Here, we show that two of them are specific markers of testes and ovaries, respectively. This, together, with our new anatomical data, allows us to identify the expression domains of most of these other markers more clearly. Some markers may be expressed in a presumptive common germline that eventually splits into an anterior male germline and posterior female germline. We agree with the reviewers that understanding the dynamics of germline differentiation and its molecular genetic underpinnings would be very interesting, and we hope to address this in future work. 

As the reviewers note, we do not understand how sperm is stored, how the worm’s own sperm can travel to its ovaries to enable selfing, or how eggs in the ovaries travel within the body. We agree with the reviewers that understanding these processes would be very interesting. Our histological and molecular work so far has been unable to find tube-like structures or other cavities for storage and transport. Potentially, cells could move within the parenchyma. Explaining these events will require substantial effort (including mechanistic studies of cell behavior and ultrastructural studies that the reviewers suggest), and we hope to do this in future work. 

We agree with Reviewer 1 that it is interesting that Piwi-1 expression is only observed in the ovaries and not in the testes - unusual given its broad germline expression in many taxa. Although there are several possible explanations for this finding (for eg. Piwi-1 could be expressed at low levels in male germline, perhaps other Piwi proteins are expressed in male germline, or Piwi may play roles in male germline progenitors that are not co-located with maturing sperm, etc), we do not currently know why this is so, and we will discuss these possibilities in our revised manuscript.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors demonstrate impairments induced by a high cholesterol diet on GLP-1R dependent glucoregulation in vivo as well as an improvement after reduction in cholesterol synthesis with simvastatin in pancreatic islets. They also map sites of cholesterol high occupancy and residence time on active versus inactive GLP-1Rs using coarse-grained molecular dynamics (cgMD) simulations and screened for key residues selected from these sites and performed detailed analyses of the effects of mutating one of these residues, Val229, to alanine on GLP-1R interactions with cholesterol, plasma membrane behaviour, clustering, trafficking and signalling in pancreatic beta cells and primary islets, and describe an improved insulin secretion profile for the V229A mutant receptor.

These are extensive and very impressive studies indeed. I am impressed with the tireless effort exerted to understand the details of molecular mechanisms involved in the effects of cholesterol for GLP-1 activation of its receptor. In general, the study is convincing, the manuscript well written and the data well presented.

Some of the changes are small and insignificant which makes one wonder how important the observations are. For instance, in figure 2 E (which is difficult to interpret anyway because the data are presented in percent, conveniently hiding the absolute results) does not show a significant result of the cyclodextrin except for insignificant increases in basal secretion. That is not identical to impairment of GLP-1 receptor signaling!

We assume that the reviewer refers to Figure 1E, where we show the percentage of insulin secretion in response to 11 mM glucose +/- exendin-4 stimulation in mouse islets pretreated with vehicle or MβCD loaded with 20 mM cholesterol. While we concur with the reviewer that the effect in this case is triggered by increased basal insulin secretion at 11 mM glucose, exendin-4 appears to no longer compensate for this increase by proportionally amplifying insulin responses in cholesterol-loaded islets, leading to a significantly decreased exendin-4induced insulin secretion fold increase under these circumstances, as shown in Figure 1F. We interpret these results as a defect in the GLP-1R capacity to amplify insulin secretion beyond the basal level to the same extent as in vehicle conditions. An alternative explanation is that there is a maximum level of insulin secretion in our cells, and 11 mM glucose + exendin-4 stimulation gets close to that value. With the increasing effect of cholesterol-loaded MβCD on basal secretion at 11 mM glucose, exendin-4 stimulation would then appear to work less well.

We have performed a simple experiment to investigate this possibility: insulin secretion following stimulation with a secretagogue cocktail (20 mM glucose, 30 mM KCl, 10 µM FSK and 100 µM IBMX) in islets +/- MβCD/cholesterol loading to determine if maximal stimulation had been reached or not in our original experiment. This experiment, now included in Supplementary Figure 1C, demonstrates that insulin secretion can increase up to ~4% (from ~2%) in our islets, supporting our initial conclusion. We have also included absolute insulin concentrations as well as percentages of secretion for all the experiments included in the study in the new Supplementary File 1 to improve the completeness of the report.

To me the most important experiment of them all is the simvastatin experiment, but the results rest on very few numbers and there is a large variation. Apparently, in a previous study using more extensive reduction in cholesterol the opposite response was detected casting doubt on the significance of the current observation. I agree with the authors that the use of cyclodextrin may have been associated with other changes in plasma membrane structure than cholesterol depletion at the GLP-1 receptor.

We agree with the reviewer that the insulin secretion results in vehicle versus LPDS/simvastatin treated mouse islets (Figure 1H, I) are relatively variable. We have therefore performed 2 extra biological repeats of this experiment (for a total n of 7). Results now show a significant increase in exendin-4-stimulated secretion with no change in basal secretion in islets pre-incubated with LPDS/simvastatin.  

The entire discussion regarding the importance of cholesterol would benefit tremendously from studies of GLP-1 induced insulin secretion in people with different cholesterol levels before and after treatment with cholesterol-lowering agents. I suspect that such a study would not reveal major differences.

We agree with the reviewer that such study would be highly relevant. While this falls outside the scope of the present paper, we encourage other researchers with access to clinical data on GLP-1R agonist responses in individuals taking cholesterol lowering agents to share their results with the scientific community. We have highlighted this point in the paper discussion to emphasise the importance of more research in this area.

Reviewer #2 (Public review):

Summary:

In this manuscript the authors provided a proof of concept that they can identify and mutate a cholesterol-binding site of a high-interest class B receptor, the GLP-1R, and functionally characterize the impact of this mutation on receptor behavior in the membrane and downstream signaling with the intent that similar methods can be useful to optimize small molecules that as ligands or allosteric modulators of GLP-1R can improve the therapeutic tools targeting this signaling system.

Strengths:

The majority of results on receptor behavior are elucidated in INS-1 cells expressing the wt or mutant GLP-1R, with one experiment translating the findings to primary mouse beta-cells. I think this paper lays a very strong foundation to characterize this mutation and does a good job discussing how complex cholesterol-receptor interactions can be (ie lower cholesterol binding to V229A GLP-1R, yet increased segregation to lipid rafts). Table 1 and Figure 9 are very beneficial to summarize the findings. The lower interaction with cholesterol and lower membrane diffusion in V229A GLP-1R resembles the reduced diffusion of wt GLP-1R with simv-induced cholesterol reductions, although by presumably decreasing the cholesterol available to interact with wt GLP-1R. This could be interesting to see if lowering cholesterol alters other behaviors of wt GLP-1R that look similar to V229A GLP-1R. I further wonder if the authors expect that increased cholesterol content of islets (with loading of MβCD saturated with cholesterol or high-cholesterol diets) would elevate baseline GLP-1R membrane diffusion, and if a more broad relationship can be drawn between GLP-1R membrane movement and downstream signaling.

Membrane diffusion experiments are difficult to perform in intact islets as our method requires cell monolayers for RICS analysis. We however agree that it is of interest to investigate if cholesterol loading affects GLP-1R diffusion. To this end, we have performed further RICS analysis in INS-1 832/3 SNAP/FLAG-hGLP-1R cells pretreated with vehicle or MβCD loaded with 20 mM cholesterol (new Supplementary Figures 1D and 1E). Interestingly, results show significantly increased plasma membrane diffusion of exendin-4-stimulated receptors, with no change in basal diffusion, following MβCD/cholesterol loading. This behaviour differs from that of the V229A mutant receptor which shows reduced diffusion under basal conditions, a pattern that mimics that of the WT receptor under low cholesterol conditions (by pre-treatment with LPDS/simvastatin).

Weaknesses:

I think there are no obvious weaknesses in this manuscript and overall, I believe the authors achieved their aims and have demonstrated the importance of cholesterol interactions on GLP-1R functioning in beta-cells. I think this paper will be of interest to many physiologists who may not be familiar with many of the techniques used in this paper and the authors largely do a good job explaining the goals of using each method in the results section.

The intent of some methods, for example the Laurdan probe studies, are better expanded in the discussion.

We have expanded on the rationale behind the use of Laurdan to assess behaviours of lipid packed membrane nanodomains in the methods, results and discussion of the revised manuscript.

I found it unclear what exactly was being measured to assess 'receptor activity' in Fig 7E and F.

Figures 7E and F refer to bystander complementation assays measuring the recruitment of nanobody 37 (Nb37)-SmBiT, which binds to active Gas, to either the plasma membrane (labelled with KRAS CAAX motif-LgBiT), or to endosomes (labelled with Endofin FYVE domain-LgBiT) in response to GLP-1R stimulation with exendin-4. This assay therefore measures GLP-1R activation specifically at each of these two subcellular locations. We have included a schematic of this assay in the new Supplementary Figure 3 to clarify the aim of these experiments.

Certainly many follow-up experiments are possible from these initial findings and of primary interest is how this mutation affects insulin homeostasis in vivo under different physiological conditions. One of the biggest pathologies in insulin homeostasis in obesity/t2d is an elevation of baseline insulin release (as modeled in Fig 1E) that renders the fold-change in glucose stimulated insulin levels lower and physiologically less effective. No difference in primary mouse islet baseline insulin secretion was seen here but I wonder if this mutation would ameliorate diet-induced baseline hyperinsulinemia.

We concur with the reviewer that it would be interesting to determine the effects of the GLP1R V229A mutation on insulin secretion responses under diet-induced metabolic stress conditions. While performing in vivo experiments on glucoregulation in mice harbouring the V229A mutation falls outside the scope of the present study, we have included ex vivo insulin secretion experiments in islets from GLP-1R KO mice transduced with adenoviruses expressing SNAP/FLAG-hGLP-1R WT or V229A and subsequently treated with vehicle versus MβCD loaded with 20 mM cholesterol to replicate the conditions of Figure 1E in the new Supplementary Figure 4.

I would have liked to see the actual islet cholesterol content after 5wks high-cholesterol diet measured to correlate increased cholesterol load with diminished glucose-stimulated inulin. While not necessary for this paper, a comparison of islet cholesterol content after this cholesterol diet vs the more typical 60% HFD used in obesity research would be beneficial for GLP-1 physiology research broadly to take these findings into consideration with model choice.

We have included these data in Supplementary Figure 1A.

Another area to further investigate is does this mutation alter ex4 interaction/affinity/time of binding to GLP-1 or are all of the described findings due to changes in behavior and function of the receptor?

To answer this question, have performed binding affinity experiments, which show no differences, in INS-1 832/3 SNAP/FLAG-hGLP-1R WT versus V229A cells (new Supplementary Figure 2D).

Lastly, I wonder if V229A would have the same impact in a different cell type, especially in neurons? How similar are the cholesterol profiles of beta-cells and neurons? How this mutation (and future developed small molecules) may affect satiation, gut motility, and especially nausea, are of high translational interest. The comparison is drawn in the discussion between this mutation and ex4-phe1 to have biased agonism towards Gs over beta-arrestin signaling. Ex4-phe1 lowered pica behavior (a proxy for nausea) in the authors previously co-authored paper on ex4-phe1 (PMID 29686402) and I think drawing a parallel for this mutation or modification of cholesterol binding to potentially mitigate nausea is worth highlighting.

While experiments in neurons are outside the scope of the present study, we have added this worthy point to the discussion and hypothesise on possible effects of GLP-1R mutants with modified cholesterol interactions on central GLP-1R actions in the revised manuscript.

Reviewer #1 (Recommendations for the authors):

There are no line numbers

These have now been added.

Abstract: "Cholesterol is a plasma membrane enriched lipid" - sorry for being finicky, but shouldn't this read; "a lipid often enriched in plasma membranes"

We have modified the abstract to state that: “Cholesterol is a lipid enriched at the plasma membrane”.

p. 4 "Moreover, islets extracted from high cholesterol-fed mice". How do you "extract islets"?

We have exchanged the term “extracted” by “isolated”. Islet isolation is described in the paper methods section.

p. 4 The sentence "These effects were accompanied by decreased GLP-1R plasma membrane diffusion under vehicle conditions, measured by Raster Image Correlation Spectroscopy (RICS) in rat insulinoma INS-1 832/3 cells with endogenous GLP-1R deleted [INS-1 832/3 GLP-1R KO cells (27)] stably expressing SNAP/FLAG-tagged human GLP-1R (SNAP/FLAG-hGLP-1R), an effect that is normally triggered by agonist binding (28), as also observed here (Supplementary Figure 1C, D)" is a masterpiece of complexity. Perhaps breaking up would facilitate reading?

This paragraph has now been modified in the revised manuscript.

p. 5. I cannot evaluate the "coarse grain molecular dynamics" studies.

Reviewer #2 (Recommendations for the authors):

I view this as an excellent manuscript with very comprehensive work and clear translational relevance. I don't think any further experiments are needed for the scope outlined in this manuscript. The discussion is already long but a short postulation on how this may translate to GLP-1R-cholesterol interactions in other cell types, specifically neurons with the intent on manipulating satiation and nausea, could be worthwhile.

This has now been added.

The only thing for readability I would suggest is a sentence in the results mentioning why you're doing the Laurdan analysis, and what is the output for assessing 'receptor activity' in the membrane and endosomes.

Both points have now been added.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1 (Public review):

Summary:

The authors examine CD8 T cell selective pressure in early HCV infection using. They propose that after initial CD8-T mediated loss of virus fitness, in some participants around 3 months after infection, HCV acquires compensatory mutations and improved fitness leading to virus progression.

Strengths:

Throughout the paper, the authors apply well-established approaches in studies of acute to chronic HIV infection for studies of HCV infection. This lends rigor the to the authors' work.

Weaknesses:

(1) The Discussion could be strengthened by a direct discussion of the parallels/differences in results between HIV and HCV infections in terms of T cell selection, entropy, and fitness.

We have added a direct discussion of the parallels/differences between HIV and HCV throughout the discussion including at lines 308 – 310 and 315 -327.

Lines 308-310: “In fact, many parallels can be drawn between HIV infections and HCV infections in the context of emerging viral species that escape T cell immune responses.”

Lines: 315-327: “One major difference between HCV and HIV infection is the event where patients infected with HCV have an approximately 25% chance to naturally clear the infection as opposed to just achieving viral control in HIV infections. Here, we probed the underlying mechanism, and questioned how the host immune response and HCV mutational landscape can allow the virus to escape the immune system. To understand this process, taking inspiration from HIV studies (24), a quantitative analysis of viral fitness relative to viral haplotypes was conducted using longitudinal samples to investigate whether a similar phenomenon was identified in HCV infections for our cohort for patients who progress to chronic infection. We observed a decrease in population average relative fitness in the period of <90DPI with respect to the T/F virus in chronic subjects infected with HCV. The decrease in fitness correlated positively with IFN-γ ELISPOT responses and negatively with SE indicating that CD8+ T-cell responses drove the rapid emergence of immune escape variants, which initially reduced viral fitness. This is similarly reflected in HIV infected patients where strong CD8+ T-cell responses drove quicker emergence of immune escape variants, often accompanied by compensatory mutations (24).”

(2) In the Results, please describe the Barton model functionality and why the fitness landscape model was most applicable for studies of HCV viral diversity.

This has been added to the introduction section rather than Results as we feel that it is more appropriate to show why it is most applicable to HCV viral diversity in the background section of the manuscript. We write at lines 77-90:

“Barton et al.’s [23] approach to understand HIV mutational landscape resulting in immune escape had two fundamental points: 1) replicative fitness depends on the virus sequence and the requirement to consider the effect of co-occurring mutations, and 2) evolutionary dynamics (e.g. host immune pressure). Together they pave the way to predict the mutational space in which viral strains can change given the unique immune pressure exerted by individuals infected with HIV. This model fits well with the pathology of HCV infection. For instance, HIV and HCV are both RNA viruses with rapid rate of mutation. Additionally, like HIV, chronic infection is an outcome for HCV infected individuals, however, unlike HIV, there is a 25% probability that individuals infected with HCV will naturally clear the virus. Previously published studies [9] have shown that HIV also goes through a genetic bottleneck which results in the T/F virus losing dominance and replaced by a chronic subtype, identified by the immune escape mutations. The concepts in Barton’s model and its functionality to assess the fitness based on the complex interaction between viral sequence composition and host immune response is also applicable to early HCV infection.”

(3) Recognize the caveats of the HCV mapping data presented.

We have now recognized the caveats of the HCV mapping data at lines 354-256 “While our findings here are promising, it should be recognized that although the bioinformatics tool (iedb_tool.py) proved useful for identifying potential epitopes, there could be epitopes that are not predicted or false-positive from the output which could lead to missing real epitopes”

(4) The authors should provide more data or cite publications to support the authors' statement that HCV-specific CD8 T cell responses decline following infection.

We have now clarified at lines 352-353 that the decline was toward “selected epitopes that showed evidence of escape”.

Furthermore, we have cited two publications at line 352 that support our statement.

(5) Similarly, as the authors' measurements of HCV T and humoral responses were not exhaustive, the text describing the decline of T cells with the onset of humoral immunity needs caveats or more rigorous discussion with citations (Discussion lines 319-321).

We have now added a caveat in the discussion at lines 357-360 which reads

“In conclusion, this study provides initial insights into the evolutionary dynamics of HCV, showing that an early, robust CD8+ T-cell response without nAbs strongly selects against the T/F virus, enabling it to escape and establish chronic infection. However, these findings are preliminary and not exhaustive, warranting further investigation to fully understand these dynamics. “

(6) What role does antigen drive play in these data -for both T can and antibody induction?

It is possible that HLA-adapted mutations could limit CD8 T cell induction if the HLAs were matched between transmission pairs, as has been shown previously for HIV (https://doi.org/10.1371/journal.ppat.1008177) with some data for HCV (https://journals.asm.org/doi/10.1128/jvi.00912-06). However, we apologise as we are not entirely sure that this is what the reviewer is asking for in this instance.

(7) Figure 3 - are the X and Y axes wrongly labelled? The Divergent ranges of population fitness do not make sense.

Our apologies, there was an error with the plot in Figure 3 and the X and Y axis were wrongly labelled. This has now been resolved.

(8) Figure S3 - is the green line, average virus fitness?

This has now been clarified in Figure S3.

(9) Use the term antibody epitopes, not B cell epitopes.

We now use the term antibody epitopes throughout the manuscript.

Reviewer #1 (Recommendations for the authors):

Recommendations for improving the writing and presentation:

(1) Introduction:

Line 52: 'carry mutations B/T cell epitopes'. Two points

i) These are antibody epitopes (and antibody selection) not B cell epitopes

We have corrected this sentence at line 55 which now reads: “carry mutations within epitopes targeted by B cells and CD8+ T cells”.

ii) To avoid confusion, add text that mutations were generated following selection in the donor.

For HCV, it is unclear if mutations are generated following selection or have been occurring in low frequencies outside detection range. Only when selection by host immune pressure arises do the potentially low-frequency variants become dominant. However, we do acknowledge it is potentially misleading to only mention new variants replacing the transmitted/founder population. We have modified the sentence at line 52 to read:

“At this stage either an existing variant that was occurring in low-frequency outside detection range or an existing variant with novel mutations generated following immune selection is observed in those who progress to chronic infection”

- Lines 51-56: Human studies of escape and progression are associative, not causative as implied.

Correct, evidence suggesting that escape and progression are currently associative. We have now corrected these lines to no longer suggest causation.

- Line 65: Suggest you clarify your meaning of 'easier'?

This sentence, now at line 72, has been modified to: “subtype 1b viruses have a higher probability to evade immune responses”

(2) Results:

- Line 147: Barton model (ref'd in Intro) is directly referred to here but not referenced.

The reference has been added.

- The authors should cite previous HIV literature describing associations between the rate of escape and Shannon Entropy e.g. the interaction between immunodominance, entropy, and rate of escape in acute HIV infection was described in Liu et al JCI 2013 but is not cited.

We have now cited previous HIV research at line 147-151, adding Liu et al:

“Additionally, the interaction between immunodominance, entropy, and escape rate in acute HIV infection has been described, where immunodominance during acute infection was the most significant factor influencing CD8+ T cell pressure, with higher immunodominance linked to faster escape (27). In contrast, lower epitope entropy slowed escape, and together, immunodominance and entropy explained half of the variability in escape timing (27).”

- Line 319: The authors suggest that HCV-specific CD8 T cell response declines following early infection. On what are they basing this statement? The authors show their measured T cell responses decline but their approach uses selected epitopes and they are therefore unable to assess total HCV T cell response in participants (Where there is no escape, are T cell magnitudes maintained or do they still decline?). Can the authors cite other studies to support their statement?

We have now clarified that the decline was toward “selected epitopes that showed evidence of escape”. Furthermore, we also cite two studies to support our findings.

- Throughout the authors talk in terms of CD8 T cells but the ELISpot detects both CD4 and CD8 T cell responses. I suggest the authors be more explicit that their peptide design (9-10mers) is strongly biased to only the detection of CD8 T cells.

To make this clearer and more explicit we have now added to the methods section at line 433-435:

“While the ELISpot assay detects responses from both CD4 and CD8 T cells, our peptide design (9-10mers) is strongly biased toward CD8 T-cell detection. We have therefore interpreted ELISpot responses primarily in terms of CD8 T-cell activity.”

- The points made in lines 307-321 could be more succinct

We have now edited the discussion (lines 307 – 321) to make the points more succinct (now lines 307-323).

Minor corrections to text, figures:

- Figure 2: suggest making the Key bigger and more obvious.

We have now made the key bigger and more obvious

- Figure 3 A & D....is there an error on the X-axis...are you really reporting ELISpot data of < 1 spot/10^6? Perhaps the X and Y axes are wrongly labelled?

Our apologies, there was an error with the plot in Figure 3 and the X and Y axis were wrongly labelled. This has now been resolved.

- Figure 5: As this is PBMC, remove CD8 from the description of ELISpot. 

We have now removed CD8 from the description of ELISpot in both Figure 5 and Figure S3

Reviewer #2 (Public review):

Summary:

In this work, Walker and collaborators study the evolution of hepatitis C virus (HCV) in a cohort of 14 subjects with recent HCV infections. They focus in particular on the interplay between HCV and the immune system, including the accumulation of mutations in CD8+ T cell epitopes to evade immunity. Using a computational method to estimate the fitness effects of HCV mutations, they find that viral fitness declines as the virus mutates to escape T-cell responses. In long-term infections, they found that viral fitness can rebound later in infection as HCV accumulates additional mutations.

Strengths:

This work is especially interesting for several reasons. Individuals who developed chronic infections were followed over fairly long times and, in most cases, samples of the viral population were obtained frequently. At the same time, the authors also measured CD8+ T cell and antibody responses to infection. The analysis of HCV evolution focused not only on variation within particular CD8+ T cell epitopes but also on the surrounding proteins. Overall, this work is notable for integrating information about HCV sequence evolution, host immune responses, and computational metrics of fitness and sequence variation. The evidence presented by the authors supports the main conclusions of the paper described above.

Weaknesses:

One notable weakness of the present version of the manuscript is a lack of clarity in the description of the method of fitness estimation. In the previous studies of HIV and HCV cited by the authors, fitness models were derived by fitting the model (equation between lines 435 and 436) to viral sequence data collected from many different individuals. In the section "Estimating survival fitness of viral variants," it is not entirely clear if Walker and collaborators have used the same approach (i.e., fitting the model to viral sequences from many individuals), or whether they have used the sequence data from each individual to produce models that are specific to each subject. If it is the former, then the authors should describe where these sequences were obtained and the statistics of the data.

If the fitness models were inferred based on the data from each subject, then more explanation is needed. In prior work, the use of these models to estimate fitness was justified by arguing that sequence variants common to many individuals are likely to be well-tolerated by the virus, while ones that are rare are likely to have high fitness costs. This justification is less clear for sequence variation within a single individual, where the viral population has had much less time to "explore" the sequence landscape. Nonetheless, there is precedent for this kind of analysis (see, e.g., Asti et al., PLoS Comput Biol 2016). If the authors took this approach, then this point should be discussed clearly and contrasted with the prior HIV and HCV studies.

We thank the reviewer for pointing out the weakness in our explanation and description of the fitness model. The model has been generated using publicly released viral sequences and this has been described in a previous publication by Hart et al. 2015. T/F virus from each of the subjects chronically infected with HCV in our cohort were given to the model by Hart et al. to estimate the initial viral fitness of the T/F variant. Subsequent time points of each subject containing the subvariants of the viral population were also estimated using the same model (each subtype). For each subject, these subvariant viral fitness values were divided by the fitness value of the initial T/F virus (hence relative fitness of the earliest time points with no mutations in the epitope regions were a value of 1.000). All other fitness values are therefore relative fitness to the T/F variant.

We have further clarified this point in the methods section “Estimating survival fitness of viral variant” to better describe how the data of the model was sourced (Lines 465-499).

To add to the reviewer’s point, we agree that sequence variants common to many individuals are likely to be well-tolerated by the virus and this event was observed in our findings as our data suggested that immune escape variants tended to revert to variants that were closer the global consensus strain. Our previous publications have indicated that T/F viruses during transmission were variants that were “fit” for transmission between hosts, especially in cases where the donor was a chronic progressor, a single T/F is often observed. Progression to immune escape and adaptation to chronic infection in the new host has an in-between process of genetic expansion via replication followed by a bottleneck event under immune pressure where overall fitness (overall survivability including replication and exploring immune escape pathways) can change. Under this assumption we questioned whether the observation reported in HIV studies (i.e. mutation landscapes that allow HIV adaptation to host) also happens in HCV infections. Furthermore, cohort used in this study is a rare cohort where patients were tracked from uninfected, to HCV RNA+, to seroconversion and finally either clearing the virus or progression to chronic infection. Thus, it is of importance to understand the difference between clearance and chronic progression.

Another important point for clarification is the definition of fitness. In the abstract, the authors note that multiple studies have shown that viral escape variants can have reduced fitness, "diminishing the survival of the viral strain within the host, and the capacity of the variant to survive future transmission events." It would be helpful to distinguish between this notion of fitness, which has sometimes been referred to as "intrinsic fitness," and a definition of fitness that describes the success of different viral strains within a particular individual, including the potential benefits of immune escape. In many cases, escape variants displace variants without escape mutations, showing that their ability to survive and replicate within a specific host is actually improved relative to variants without escape mutations. However, escape mutations may harm the virus's ability to replicate in other contexts. Given the major role that fitness plays in this paper, it would be helpful for readers to clearly discuss how fitness is defined and to distinguish between fitness within and between hosts (potentially also mentioning relevant concepts such as "transmission fitness," i.e., the relative ability of a particular variant to establish new infections).

Thank you for pointing out the weakness of our definition of fitness. We have now clarified this at multiple sections of the paper: In the abstract at lines 18-21 and in the introduction at lines 64-69.

These read:

Lines 18-21: “However, this generic definition can be further divided into two categories where intrinsic fitness describes the viral fitness without the influence of any immune pressure and effective fitness considers both intrinsic fitness with the influence of host immune pressure.”

Lines 64-69: “This generic definition of fitness can be further divided into intrinsic fitness (also referred to as replicative fitness), where the fitness of sequence composition of the variant is estimated without the influence of host immune pressure. On the other hand, effective fitness (from here on referred to as viral fitness) considers fundamental intrinsic fitness with host immune pressure acting as a selective force to direct mutational landscape (19)[REF], which subsequently influences future transmission events as it dictates which subvariants remain in the quasispecies.”

One concern about the analysis is in the test of Shannon entropy as a way to quantify the rate of escape. The authors describe computing the entropy at multiple time points preceding the time when escape mutations were observed to fix in a particular epitope. Which entropy values were used to compare with the escape rate? If just the time point directly preceding the fixation of escape mutations, could escape mutations have already been present in the population at that time, increasing the entropy and thus drawing an association with the rate of escape? It would also be helpful for readers to include a definition of entropy in the methods, in addition to a reference to prior work. For example, it is not clear what is being averaged when "average SE" is described.

We thank the reviewer to point out the ambiguity in describing average SE. This has been rectified by adding more information in the methods section (Lines 397 to 400):

“Briefly, SE was calculated using the frequency of occurrence of SNPs based on per codon position, this was further normalized by the length of the number of codons in the sequence which made up respective protein. An average SE value was calculated for each time point in each protein region for all subjects until the fixation event.”

To answer the reviewer’s question, we computed entropy at multiple time points preceding the observation in the escape mutation. The escape rate was calculated for the epitopes targeted by immune response. We compared the average SE based on change of each codon position and then normalised by protein length, where the region contained the epitope and the time it took to reach fixation. We observed that if the protein region had a higher rate of variation (i.e. higher average SE) then we also see a quicker emergence of an immune escape epitope. Since we took SE from the very first time point and all subsequent time points until fixation, we do not think that escape mutations already been present at the population would alter the findings of the association with rate of escape. Especially, these escape mutations were rarely observed at early time points. It is likely that due to host immune pressure that the escape variant could be observed, the SE therefore suggest the liberty of exploration in the mutation landscape. If the region was highly restrictive where any mutations would result in a failed variant, then we should observe relatively lower values of average SE. In other words, the higher variability that is allowed in the region, the greater the probability that it will find a solution to achieve immune escape.

Reviewer #2 (Recommendations for the authors):

In addition to the main points above, there are a few minor comments and suggestions about the presentation of the data.

(1) It's not clear how, precisely, the model-based fitness has been calculated and normalized. It would be helpful for the authors to describe this explicitly. Especially in Figure 3, the plotted fitness values lie in dramatically different ranges, which should be explained (maybe this is just an error with the plot?).

We have now clarified how the model-based fitness has been calculated and normalized in the method section “Estimating survival fitness of viral variants” at line 465-472.

“The model used for estimating viral fitness has been previously described by Hart et al. (19). Briefly, the original approach used HCV subtype 1a sequences to generate the model for the NS5B protein region. To update the model for other regions (NS3 and NS2) as well as other HCV subtypes in this study, subtype 1b and subtype 3a sequences were extracted from the Los Almos National Laboratory HCV database. An intrinsic fitness model was first generated for each subtype for NS5B, NS3 and NS2 region of the HCV polyprotein. Then using, longitudinally sequenced data from patients chronically infected with HCV as well as clinically documented immune escape to describe high viral fitness variants, we generated estimates of the viral fitness for subjects chronically infected with HCV in our cohort.”

Our apologies, there was an error with the plot in Figure 3. This has now been resolved.

(2) In different plots, the authors show every pairwise comparison of ELISPOT values, population fitness, average SE, and rate of escape. It may be helpful to make one large matrix of plots that shows all of these pairwise comparisons at the same time. This could make it clear how all the variables are associated with one another. To be clear, this is a suggestion that the authors can consider at their discretion.

Thank you for the suggestion to create a matrix of plots for pairwise comparisons. While this approach could indeed clarify variable associations, implementing it is outside the scope of this project. We appreciate the idea and may consider it in future studies as we continue to expand on this work.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

The crystal structure of the Sld3CBD-Cdc45 complex presented by Li et al. is a novel contribution that significantly advances our understanding of CMG formation during the rate-limiting step of DNA replication initiation. This structure provides insights into the intermediate steps of CMG formation. The study builds upon previously known structures of Sld3 and Cdc45 and offers new perspectives into how Cdc45 is loaded onto MCM DH through Sld3-Sld7. The most notable finding is the structural difference in Sld3CBD when bound to Cdc45, particularly the arrangement of the α8-helix, which is essential for Cdc45 binding and may also pertain to its metazoan counterpart, Treslin. Additionally, the conformational shift in the DHHA1 domain of Cdc45 suggests a possible mechanism for its binding to MCM2NTD.

Strengths:

The manuscript is generally well-written, with a precise structural analysis and a solid methodological section that will significantly advance future studies in the field. The predictions based on structural alignments are intriguing and provide a new direction for exploring CMG formation, potentially shaping the future of DNA replication research.

Weaknesses:

The main weakness of the manuscript lies in the lack of experimental validation for the proposed Sld3-Sld7-Cdc45 model. Specifically, the claim that Sld3 binding to Cdc45-MCM does not inhibit GINS binding, a finding that contradicts previous research, is not sufficiently substantiated with experimental evidence. To strengthen their model, the authors must provide additional experimental data to support this mechanism. Also, the authors have not compared the recently published Cryo-EM structures of the metazoan CMG helicases with their predicted models to see if Sld3/Treslin does not cause any clash with the GINS when bound to the CMG. Still, the work holds great potential in its current form but requires further experiments to confirm the authors' conclusions.

We appreciate the reviewers’ careful reading and the comments.

Our structural analysis of Sld3CBD-Cdc45 showed the detailed interaction map between Sld3CBD and Cdc45 at 2.6 Å resolution. The Sld3, MCM and GINS binding sites of Cdc45 completely differed, suggesting that the Sld3CBD, Cdc45 and GINS could bind to MCM together. The SCMG-DNA model confirmed such a binding manner, although our study does not show how this binding manner affects the GINS loading by other initiation factors (Dpb11, Sld2, et. al). Regarding the previous studies, competition of Sld3 and GINS for binding to Cdc45 or Cdc45-MCM (Bruck et. al), which may be caused by the conformation change of Cdc45 DHHA1 between Sld3CBD-Cdc45 and CMG. We modified our manuscript and discussed (P7/L168-173, and P10/L282-286). Following the comment, we checked the recently published Cryo-EM structure (PDBID:8Q6O) with their predicted models of the metazoan CMG helicases (P7/L198-P8/L202) and added the Cdc45 mutation experiments to confirm our conclusion ([Recommendations for the authors] Q18).

Reviewer #2 (Public review):

Summary

The manuscript presents valuable findings, particularly in the crystal structure of the Sld3CBD-Cdc45 interaction and the identification of additional sequences involved in their binding. The modeling of the Sld7-Sld3CBD-CDC45 subcomplex is novel, and the results provide insights into potential conformational changes that occur upon interaction. However, the work remains incomplete as several main claims are only partially supported by experimental data, particularly the proposed model for Sld3 interaction with GINS on the CMG. Additionally, the single-stranded DNA binding data from different species do not convincingly advance the manuscript's central arguments.

Strengths

(1) The Sld3CBD-Cdc45 structure is a novel contribution, revealing critical residues involved in the interaction.

(2) The model structures generated from the crystal data are well presented and provide valuable insights into the interaction sequences between Sld3 and Cdc45.

(3) The experiments testing the requirements for interaction sequences are thorough and conducted well, with clear figures supporting the conclusions.

(4) The conformational changes observed in Sld3 and Cdc45 upon binding are interesting and enhance our understanding of the interaction.

(5) The modeling of the Sld7-Sld3CBD-CDC45 subcomplex is a new and valuable addition to the field.

Weaknesses

(1) The proposed model for Sld3 interacting with GINS on the CMG needs more experimental validation and conflicts with published findings. These discrepancies need more detailed discussion and exploration.

Our structural analysis experiment of Sld3CBD-Cdc45 showed the detailed interaction information between Sld3CBD and Cdc45 at 2.6 Å resolution. The Sld3CBD-binding site of Cdc45 is completely different from that of GINS and MCM binding to Cdc45, suggesting that the Sld3CBD, Cdc45, and GINS could bind to MCM together. The SCMG-DNA model confirmed such a binding manner. Following the comment, we added a Cdc45 mutant analysis, disrupting the binding to MCM and GINS but not affecting the Sld3CBD binding (Supplementary Figure 9). Our model is consistent with the GINS-loading requirement (the phosphorylation of Sld3 on Cdc45-MCM) and has no discrepancies with the stepwise loading fashion (Please see the responses to [Recommendations for the authors] Reviewer#1-Q14-15]). Regarding the previous studies, competition of Sld3 and GINS for binding to Cdc45 or Cdc45-MCM (Bruck et. al), by in vitro binding experiments, please see the responses to [Recommendations for the authors] Q6.

(2) The section on the binding of Sld3 complexes to origin single-stranded DNA needs significant improvement. The comparisons between Sld3-CBD, Sld3CBD-Cdc45, and Sld7-Sld3CBD-Cdc45 involve complexes from different species, limiting the comparisons' value.

As suggested, we tried to improve the ssDNA-binding section (Please see the responses to [Recommendations for the authors]: Q4 and Q5). We used Sld7-Sld3CBD-Cdc45 from different sources due to limitations in protein expression. These two sources belong to the same family and the proteins Sld7, Sld3 and Cdc45 have sequence conservation with similar structures predicted by the alphafold3 (RMSD = 0.356, 1.392, and 0.891 for Ca atoms of Sld7CTD, Sld7NTD-Sld3NTD, and Sld3CBD-Cdc45). Such similarity in source and protein lever allows us to do the comparison.

(3) The authors' model proposing the release of Sld3 from CMG based on its binding to single-stranded DNA is unclear and needs more elaboration.

Considering that ssDNA (ssARS1) is produced by CMG, the ssDNA-binding of Sld3 should happen after forming an active CMG. Therefore, the results of ssDNA binding experiments implied that the Sld3 release could be with the binding to ssDNA produced by CMG. We tried to present more elaborations in the revised version. (Please see the responses to [Recommendations for the authors] Q4, Q5).

Reviewer #3 (Public review):

Summary:

The paper by Li et al. describes the crystal structure of a complex of Sld3-Cdc45-binding domain (CBD) with Cdc45 and a model of the dimer of an Sld3-binding protein, Sld7, with two Sld3-CBD-Cdc45 for the tethering. In addition, the authors showed the genetic analysis of the amino acid substitution of residues of Sld3 in the interface with Cdc45 and biochemical analysis of the protein interaction between Sld3 and Cdc45 as well as DNA binding activity of Sld3 to the single-strand DNAs of the ARS sequence.

Strengths:

The authors provided a nice model of an intermediate step in the assembly of an active Cdc45-MCM-GINS (CMG) double hexamers at the replication origin, which is mediated by the Sld3-Sld7 complex. The dimer of the Sld3-Sld7 complexes tethers two MCM hexamers together for the recruitment of GINS-Pol epsilon on the replication origin.

Weaknesses:

The biochemical analysis should be carefully evaluated with more quantitative ways to strengthen the authors' conclusion.

We thank your positive assessment. We provided more quantitative information and tried to quantify the experiments as suggested (Please see the responses to [Recommendations for the authors]).

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

I have several concerns that I will outline below, accompanied by my suggestions.

(1) "The title of the paper- "Structural and functional insights into Cdc45 recruitment by Sld7-Sld3 for CMG complex Formation," appears misleading because it appears that authors present a structure of Sld3-Sld7 in complex with Cdc45, which is not the case here. If authors can provide additional structures proving the function of this complex, then this title justifies it. Otherwise, I recommend making a title that justifies the presented work in its current form.

Following the comment, we change the title to “Sld3CBD-Cdc45 structural insights into Cdc45 recruitment for CMG complex formation”.

(2) In lines 70-72, where the authors mention the known structures of different proteins, intermediates, and complexes, I recommend including PDB IDs of the described structures and reference citations. This will help the readers to analyze what is missing in the pathway and why this structure is essential.

Following the comment, we added PBDIDs and references (P3/L72-74).

(3) The representation of Figure 1A is unclear and looks clumsy. If the structure were rotated in another orientation, where α8 and α9 would be displayed on the forward side, it would be more helpful to understand the complex forming regions by looking at the structure. Also, I recommend highlighting the α8 and α9 in a contrasting color to be easily visible and attract readers' attention. Similarly, it would also be helpful if DHAA1 would be shown in a different color.

Following the comment, we modified the Figure1 to show α8 and α9 of Sld3CBD and DHAA1 of Cdc45 clearly in revised version.

(4) Can authors add a supplementary figure showing the probability of disorderness of the α8 helix region in the Sld3? Also, highlight what region became ordered in their structure.

Yes, we have showed the disordered α8 helix region and highlight ordered α8 in the Sld3 in Figure S4 A.

(5) Can you compare the Cdc45 long distorted helix (Supplementary Figure 3B) in the Sld3-Cdc45 complex with the Xenoupus and drosophila Cdc45 from their CMG structures? Also, can the authors explain why this helix is destabilized in their structure but is relatively stable in another Cdc45 structure (in CMG and HuCdc45)?

We have checked all Cdc45 from published cryo-EM CMG structures, including Xenopus CMG-donson (8Q6O) and Drosophila CMG (6RAW), and all of them ordered the long helix in the CMG complex, whereas this long helix was disordered in the crystal structure of Sld3CBD-Cdc45 and Entamoeba histolytica Cdc45. The crystal packing around the long helix showed that it looks to be stabilized by crystal packing only in huCdc45, therefore we suggested that this long helix is detestable for crystallization.

(6) I recommend adding the following parameters to Supplementary Table 2: 1. Rmerge values, 2. Wilson B factor, 3. Average B factor, and 4. Total number of molecules in ASU.

We are sorry to make a mistake about Rmerge in Table 2. We correct it. We added the Wilson B factor, the average B factor, and the total number of Sld3CBD-Cde45 in ASU.

(7) Can authors provide the B factor values of the α8 helix of Sld3?

We checked the B factor values of the helix α8CTP of Sld3 in Sld3CBD-Cdc45. Since this helix binds to Cdc45 stably, the average B factor of the main chain is 45 Ų less than that of the whole structure. We added the average B factor of helix α8CTP into the Supplementary Figure 4A legend.

(8) Can authors explain why higher Ramachandran outliers exist in their structure? Can it be reduced below 1% during refinement?

There are 13 outliers (1.67%) in different places: four are close to the disorder regions (poor electron map), four are in a loop with poor map and the remains are turn parts or a loop. For the residues with poor electron maps, we could not modify them to the allow Ramachandran region with low Rfree value, so we could not reduce them to below 1% during refinement while keeping the current Rfree value.

(9) In Supplementary Figure 8, please show the CD spectra of the Sld3WT. Why is the Sld3-3S peak relatively flat? Was the sample precipitating while doing the measurements, or does it have less concentration than others?

To check the folding of the mutants, we did CD experiments with the estimated secondary structure elements. Because WT Sld3CBD was prepared in a complex with Cdc45, while the mutants of Sld3CBD existed along, we calculated the elements of secondary structure from the crystal structure of Sld3CBD-Cdc45. The concentration of samples was controlled to the same level for CD measurement. The relative plat of the Sld3-3S peak may be caused by precipitating while doing the measurement.

(10) Can authors generate the alpha fold three models of the Sld3CBD-Cdc45-MCM-dsDNA and SCMG-dsDNA and compare them with the models they have generated?

We tried to predict the Sld3CBD-Cdc45-MCM-dsDNA and SCMG-dsDNA using Alphafold3. Although the results showed similar structures to our models, many parts were disordered. So, we did not use the predicted structures.

(11) The authors say that the overall molecular mass of the Sld7-Sld3ΔC-Cdc45 was >400kDa on the SEC column. However, the column used for purifying this complex and the standards that were run on it for molecular weight calculations have not been written anywhere. If the Superdex 200 column was used, then the sample of more than 400kDa should not elute at the position shown in Supplementary Figure 2B. I recommend showing the standard MW plot and where the elution volume of the Sld7-Sld3ΔC-Cdc45 lies on the standard curve. Also, add how molecular weight calculations were done and the calculated molecular mass.

Following the comment, we added a measurement of Superdex 200 16/60 column (SEC) using a standard sample kit into Supplementary Figure 2 to show that the molecular weight of the peak at the position was estimated to be > 400 k Da.

(12) I also recommend using at least one of the techniques, either SEC-MALS or AUC, to calculate the actual molecular mass of the Sld7-Sld3ΔC-Cdc45 complex and to find its oligomeric state. If the authors want to prove their hypothesis that a dimer of this complex binds to MCMDH, it is essential to show that it exists as a dimer. Based on the current SEC profile, it appears as a monomer peak if the S200 SEC column is being used.

As the response to (11), we added the standard MW plot (measurement using Superdex 200 16/60 column) using a standard sample kit. The molecular weight at the peak elution position of Sld7-Sld3ΔC-Cdc45 was estimated to be 429k Da. Considering that the Sld7-Sld3ΔC-Cdc45 dimer should be a flexible long-shaped molecule, the elution position could be at a larger molecular weight position than the real one (158 x 2 k Da). We also tried to confirm the particle size using SEC-SAXS, as the response to the next question (13).

(13) Dynamic light scattering is not the most accurate method for calculating intermolecular distance. I recommend using another technique that calculates the accurate molecular distances between two Cdc45 if Sld7-Sld3ΔC-Cdc45 is forming a dimer. Techniques such as FRET could be used. Otherwise, some complementary methods, such as SAXS, could also be used to generate a low-resolution envelope and fit the speculated dimer model inside, or authors could try negative staining the purified Sld7-Sld3ΔC-Cdc45 and generate 2D class averages and low-resolution ab initio models to see how the structure of this complex appears and whether it satisfies the speculated model of the dimeric complex.

We have tried both negative staining TEM and SEC-SAXS experiments. We could not obtain images good enough of negative staining of TEM to generate 2D class averages and low-resolution ab initio models. The results of SEC-SAXS provided a molecular weight of 370 - 420 kDa, and an Rg > 85 Å, which are consistent with our conclusion from SEC and DLS results but with large error due to the measurement temperature at 10-15°C (measuring equipment limitation). The peak of SCE-SAXS under measurement conditions was not as sharp as purification at 4°C and SAXS data is not good enough to make a molecular model, so we did not add them to our manuscript.

(14) Authors mentioned in the introduction section (lines 72-73) that based on the single-molecule experiments, Cdc45 is recruited in a stepwise manner to MCMDH. If this is true and if Sld7-Sld3ΔC-Cdc45 forms a dimer, this is also true, then for stepwise recruitment, the dimer will have to break into monomers, and this will be an energy-expensive process for the cell. So, would such a process occur physiologically? Can the authors explain how this would physiologically happen inside the cell?

Sld7-Sld3-Cdc45 consists of domains linked by long loops, so the dimer Cdc45-Sld3-[Sld7]2-Sld3-Cdc45 is flexible long-sharp. Such a flexible dimer does not mean that two Cdc45 molecules must bind to MCM DH simultaneously and may bind to MCM DH by stepwise manner. The dimer formation of Sld7-Sld3-Cdc45 is advantageous for recruiting efficiently and saving energy. Moreover, our proposal of Cdc45-Sld3-[Sld7]2-Sld3-Cdc45 on MCM DH could be a stage during CMG formation in the cell. Following the comment, we added such descriptions (P7/L194, and P10/L276-279).

(15) Can authors show experimentally that a dimer of Sld7-Sld3ΔC-Cdc45 is binding to MCMDH and not a monomer in a stepwise fashion?

In our study, we provided experiments of particle size to show the dimer of Sld7-Sld3-Cdc45 off MCM DH and a model of SCMG to indicate the dimer of Sld7-Sld3ΔC-Cdc45 on MCM DH. This question should be addressed future by the Cryo-EM of Sld7-Sld3-Cdc45-MCM DH or Sld7-Sld3-CMG. As the response to Q14, the flexible dimer of Sld7-Sld3ΔC-Cdc45 binding on MCMDH does not contradict the stepwise-loading fashion. The dimer of Sld7-Sld3ΔC-Cdc45 binding on MCM DH shows a stage.

(16) Can authors highlight where Sld7 will lie on their model shown in Figures 3A and 3C, considering their model shown in 3B is true?

We predict that the Sld7-Sld3-Cdc45 should be in a dimer form of Cdc45-Sld3-[Sld7]2-Sld3-Cdc45 based on the structures and the particle size analysis. The Sld7 dimer could be across MCM DH on the top of Figure 3A right and 3C right. However, we could not add the Sld7 molecule to the models because there is no interaction data between Sld7 and MCM.

(17) In Supplementary Figure 10, can authors show the residues between the loop region highlighted in the dotted circle to show that there is no steric clash between the residues in that region of their predicted model?

Following the comment, we added the residues in Supplementary Figure 10 (Supplementary Figure 11 in the revised version) to show no steric clash in our predicted model.

(18) It is essential to show experimentally that Sld3CBD neighbors MCM2 and binds Cdc45 on the opposite side of the GINS binding site. I recommend that the authors design an experiment that proves this statement. Mutagenesis experiments for the predicted residues that could be involved in interaction with proper controls might help to prove this point. Since this is the overall crux of the paper, it has to be demonstrated experimentally.

We thank the reviewer’s recommendation. Our structural analysis experiment shows the interaction information between Sld3CBD and Cdc45 at 2.6 Å resolution. The Sld3CBD-binding site, GINS-binding site, and MCM-binding site of Cdc45 are completely different, indicating that the Sld3CBD, Cdc45 and GINS could bind to MCM together. The SCMG model confirmed such a binding manner. Following the recommendation, we added mutant analysis of Cdc45 G367D and W481R, which was reported to disrupt the binding to MCM and GINS, respectively. Both mutants do not affect the binging to Sld3CBD as we predicted (Supplementary Figure 9B). We modified our manuscript and discussed this point more clearly (P7/L170-173).

(19) I recommend rewriting the sentence in lines 208-210. During EMSA experiments, new bands do not appear; instead, there is no shift at lower ratios, so you see a band similar to the control for Sld3CBD-Cdc45. So, re-write the sentence correctly to avoid confusion when interpreting the result.

Following the comment, we rewrote this sentence to "The ssDNA band remained (Figure 4B) and new bands corresponding to the ssDNA–protein complex appeared in CBB staining PAGE (Supplementary Figures 13) when the Sld3CBD–Cdc45 complex was mixed with ssDNA at the same ratio, indicating that the binding affinity of Sld3CBD–Cdc45 for ssDNA was lower than that of Sld3CBD alone” (P8/L226-229)

(20) Since CDK-mediated phosphorylation of Sld3 is known to be required for GINS loading, the ssDNA binding affinity of phosphorylated Sld3 remains the same. I wonder what would happen if phosphorylated Sld3 were used for the experiment shown in Figure 4B.

The CDK phosphorylation site is located at Sld3CTD and our ssDNA-binding experiment did not include the Sld3CTD, so phosphorylated Sld3 does not affect the results shown in Figure 4B.

(21) Sld3CBD-Cdc45 has a reduced binding affinity for ss DNA, and Sld7-Sld3ΔC-Cdc45 and Sl7-Sld3ΔC have a similar binding affinity to Sld3CBD based on figure 4B. It appears that Sld3CBD reduces the DNA binding affinity of CDC45 or vice versa. Is it correct to say so?

Our opinion is “vice versa”. Cdc45 reduces the ssDNA-binding affinity of Sld3CBD. Although we could not point out the ssDNA-binding sites of Sld3CBD, the surface charge of Sld3CBD implies that α8CTP could contribute to ssDNA-binding (Supplementary Figures 15).

(22) Cdc45 binds to the ssDNA by itself, but in the case of Sld3CBD-Cdc45, the binding affinity is reduced for Sld3CBD and Cdc45. Based on their structure, can authors explain what leads to this complex's reduced binding affinity to the ssDNA? Including a figure showing how Sld7-Sld3CBD-Cdc45 interacts with the DNA would be a nice idea.

Previous studies showed that Cdc45 binds tighter to long ssDNA (> 60 bases) and the C-terminus of Cdc45 is responsible for the ssDNA binding activity. The structure of Sld3CBD-Cdc45 shows the C-terminal domain DHHA1 of Cdc45 binds to Sld3CBD, which may lead to Sld3CBD-Cdc45 complex reduced ssDNA-binding affinity of Cdc45. We agree that showing a figure of how Sld7-Sld3CBD-Cdc45 interacts with ssDNA is a nice idea. However, there is no detailed interaction information between Sld7-Sld3Δ-Cdc45 and ssDNA, so we could not give a figure to show the ssDNA-binding manner. We added a figure to show the surface charges of Sld3CBD of Sld3CBD-Cdc45, and Sld3NTD-Sld7NTD, respectively (Supplemental Figure 15).

(23) Based on the predicted model of Sld7-Sld3 and Cdc45 complex, can authors explain how Sld7 would restore the DNA binding ability of the Sld3CBD?

It can be considered that Sld7 and Sld3NTD could bind ssDNA. Although we did not perform the ssDNA-binding assay of Sld7, the Sld3NTD-Sld7NTD surface shows a large positive charge area which may contribute to ssDNA-binding (Supplemental Figure 15). We added the explanation (P9/L245-248).

(24) It would be important to show binding measurements and Kd values of all the different complexes shown in Figure 4B with ssDNA to explain the dissociation of Cdc45 from Sld7-Sld3 after the CMG formation. I also recommend describing the statement from lines 224-227 more clearly how Sld7-Sld3-Cdc45 is loading Cdc45 on CMG.

As the reviewer mentioned, the binding measurements and Kd of values of all the different complexes are important to explain the dissociation of Sld7-Sld3 from CMG. The pull-down assay using chromatography may be affected by balancing the binding affinity and chromatography conditions. Therefore, we used EMSA with native-PAGE, which is closest to the natural state. However, the disadvantage is that the Kd values could not be estimated. For lines 224-227, the ssARS1-binding affinity of Sld3 and its complex should relate to the dissociation of Sld7–Sld3 from the CMG complex but not Cdc45 loading, because ssARS1 is unwound from dsDNA by the CMG complex after Cdc45 and GINS loading. We modified the description (P9/L248-251).

(25) Can authors explain why SDS-PAGE was used to assess the ssDNA (See line 420)?

We are sorry for making this mistake and corrected it to “polyacrylamide gel electrophoresis”.

(26) In line 421, can the authors elaborate on a TMK buffer?

We are sorry for this omission and added the content of the TMK buffer (P16/L453).

(27) I am curious to know if the authors also attempted to Crystallize the Sld7-Sld3CBD-Cdc45 complex. This complex structure would support the authors' hypothesis in this article.

We tried to crystallize Sld7-Sld3Δ-Cdc45 but could not get crystals. We also tried using cryo-EM but failed to obtain data.

Reviewer #2 (Recommendations for the authors):

(1) The manuscript would be strengthened if the authors acknowledged in greater detail how their work agrees with or disagrees with Itou et al. (PMID: 25126958 DOI: 10.1016/j.str.2014.07.001). The introduction insufficiently described the findings of that previous work in lines 63-64.

We compared Sld3CBD in Sld3CBD-Cdc45 to the monomer reported by Itou et al. (PMID: 25126958 DOI: 10.1016/j.str.2014.07.001) in the section of [The overall structure of Sld3CBD-Cdc45] and point out the structural similarity and difference (P5/L105-106), especially, conformation change of Sld3CBD α8 for binding to Cdcd45, which agrees to the mutant experiments of Itou et al., (P3/L126-127). Another Cdc45-binding site of Sld3CBD in the Sld3CBD-Cdc45 complex is α9 not residues predicted in previous studies.

(2) Figure 2. Could you please perform and present data from multiple biological replicates (e.g., at least two independent experiments) for each mutant strain? This would help ensure that the observed pull-downs (2A-B) and growth patterns (2C) are consistent and reproducible.

We have done pull-downs three times from co-expression to purification and pull-down assay. We added descriptions to the method of [Mutant analysis of Sld3 and Cdc45]. The growth patterns are two times in Figure 2C.

(3) Figure 3B. The match between the predicted complex length and particle size measured by dynamic light scattering (DLS) is striking. Did the authors run the analysis with vehicle controls and particle size standards? There is no mention of these controls.

Following the comment, we added the control data of buffer and standard protein lysozyme, and the descriptions to the method of [Dynamic light scattering].

(4) Figure 4. In lines 216-217, the authors write that the binding of the K. marxianus complex "demonstrates that the presence of Sld7 could restore the single-stranded DNA binding capacity of Sld3." Another explanation is that complexes from each species bind differently. If the authors want to make a strong claim, they should compare the binding of complexes containing the same proteins.

Agree with the comment, to make a strong claim using samples from the same source is better. Due to limitations in protein overexpression, we used Sld7-Sld3ΔC-Cdc45 from different sources two sources belong to the identical family (Saccharomycetaceae) and the proteins Sld7, Sld3 and Cdc45 have sequence conservation with similar structures (RMSD = 0.356, 1.392, and 0.891 for Ca atoms of Sld7CTD, Sld7NTD-Sld3NTD, and Sld3CBD-Cdc45) predicted by the alphafold3. Such similarity in source and protein level allows us to do the comparison. Moreover, we modified the description to “indicates that the presence of Sld7 and Sld3NTD could increase the ssDNA-binding affinity to a level comparable to that of Sld3CBD.

(5) The logic of the following is unclear: "Considering that ssDNA is unwound from dsDNA by the helicase CMG complex, Sld7-Sld3ΔC-Cdc45, and Sld7-Sld3C having a stronger ssDNA-binding capacity than Sld3CBD-Cdc45 may imply a relationship between the dissociation of Sld7-Sld3 from the CMG complex and binding to ssDNA unwound by CMG." (Lines 224-227). How do the authors imagine that the binding affinity difference due to Sld7 contributes to the release of Sld3? Please explain.

Considering that ssARS1 is unwound from dsARS1 by the activated helicase CMG complex formed after loading Cdc45 and GINS, Sld3–Sld7 having a stronger ssARS1-binding affinity may provide an advantage for the dissociation of Sld7–Sld3 from the CMG complex. We modified the sentence of Lines 224-227 (P9/L248-251).

(6) The authors suggest that the release of Sld3 from the helicase is related to its association with single-stranded ARS1 DNA. They refer to the work of Bruck et al. (doi: 10.1074/jbc.M111.226332), which demonstrates that single-stranded origin DNA inhibits the interaction between Sld3 and MCM2-7 in vitro. The authors selectively choose data from this previous work, only including data that supports their model while disregarding other data. This approach hinders progress in the field. Specifically, Bruck proposed a model in which the association of Sld3 and GINS with MCM2-7 is mutually exclusive, explaining how Sld3 is released upon CMG assembly. In Figure 3 of the authors' model, they suggest that Sld3 can associate with MCM2-7 through CDC45, even when GINS is bound. Furthermore, Bruck's work showed that ssARS1-2 does not disrupt the Sld3-Cdc45 interaction. Instead, Bruck's data demonstrated that ssARS1-2 disrupts the interaction between MCM2-7 and Sld3 without Cdc45. While we do not expect the authors to consider all data in the literature when formulating a model, we urge them to acknowledge and discuss other critical data that challenges their model. Additionally, it would be beneficial for the field if the authors include both modes of Sld3 interaction with MCM2-7 (i.e., directly with MCM or through CDC45) when proposing a model for how CMG assembly and Sld3 release occurs.

In our discussion, we referred to the studies of Bruck’s data (doi: 10.1074/jbc.M111.226332) but did not discuss more because we didn’t perform similar experiments in vitro, and we do not think that no discussion hinders progress in the field. Promoting research progress, the new experiment should provide a new proposal and updated knowledge. Although we do not know exactly the positional relationship between Sld3 and Dpb11-Sld2 on MCM during GINS recruiting, the Sld3CBD-Cdc45 structure shows clearly that the Sld3CBD-binding site of Cdc45 is completely different from that of GINS and MCM binding to Cdc45. The model SCMG confirmed such a binding manner, Sld3, Cdc45 and GINS could bind together. The competition of Sld3 and GINS for binding to Cdc45 or Cdc45-MCM reported by Bruck et. al, may be caused by the conformation change of Cdc45 DHHA1 between Sld3CBD-Cdc45 and CMG, or without other initiation factors (CMG formation is regulated by the initial factors). We modified the discussion (P10/L282-286). Regarding ssARS1-binding, we did not discuss with Bruck's data that ARS1-2 does not disrupt the Sld3-Cdc45 interaction, because the data does not conflict with our proposal, although the data does not have an advantage. We propose that the release of Sld3 and Sld7 from CMG could be associated with the binding of ssARS1 unwound by CMG, but the dissociation event of Sl3-Sld7 doesn’t only ssARS1-binding. The exploration of unwound-ssARS1 causes the conformation change of CMG, which may be another event for Sld3-Sld7 dissociation. However, we do not have more experiments to confirm this and Bruck’s ssDNA-binding experiment did not use all of Sld3, Cdc45 and MCM, so we do not discuss more with Bruck’ data in the revised version (P11/L303-305).,

Reviewer #3 (Recommendations for the authors):

Major points:

(1) Figure 1, Sld3CBD-Cdc45 complex: Please indicate the number of critical residues and those of alpha-helixes and beta-sheets in this Figure or Supplemental Figure to confirm the authors' claim.

Following the comment, we added the number of alpha-helixes and beta-sheets with residue numbers in Figure 1, and Supplemental Figures 4 and 5. We also added a topology diagram (Supplemental Figure 3).

(2) Figure 2A and B: Please quantify the interaction here with a proper statistical comparison.

In the experiments of Figures 2A and 2B, we used a co-expression system to co-purify the complexes and check their binding. For quantifying, we added the concentrations of the samples used in the Method of [Mutant analysis of Sld3 and Cdc45].

(3) Figure 3B, EMSA: If these are from the EMSA assay, at least free DNAs and protein-bound DNAs are present on the gel. However, the authors showed one band, which seems to be free DNA in Figure 3B and separately the smear band of the protein complex in Supplementary Figure 12, and judged the DNA binding by the disappearance of the band (line 207). Interestingly, in the case of Sld3CBD, there are few smear bands (Supplementary Figure 12). Where is DNA in this case? The disappearance could be due to the contaminated nucleases (need a control non-specific DNA). Without showing the Sld3CBD-DNA complex in the gel, the conclusion that the DNA binding activity of Sld3CBD-Cdc45 to DNA is lower than Sld3CBD alone (line 210) is very much speculative. The same is true for Sld7-Sld3dC-Cdc45.

Please explain the method (EMSA) briefly in the main text and show a whole gel in both Figures. If the authors insist that the Sld3 DNA-binding activity is altered with Cdc43 (and MCM), it is better to perform a more quantitative DNA binding assay such as BIAcore (surface plasmon), etc.

In the EMSA, we use SYBR (Figure 4B) and CBB (Supplementary Figure 13) staining to show bands of ssDNA and protein, respectively. As the reviewer mentioned, the disappearance of the bands could be due to the contaminated nucleases, we did experiments with non-specific ssDNA-binding as a control using the same proteins shown in Supplementary Figure 14. So, we are convinced that the disappearance of the ssDNA bands or not disappearance could occur when binding to protein or not. We added such explanations in the text (P9/L242-244). As we mentioned in the legend of Supplementary Figure 13, the Sld3CBD could not enter the gel, even when bound to ssDNA, because the pI values exceeded the pH of the running buffer.

Following the reviewer's comments, we attempted a pull-down experiment using Histag (C-terminal histag of Sld3CBD/Sld3ΔC). Unfortunately, we encountered difficulties in achieving the balance between binding and chromatography conditions.

(4) Figure 3B: Please quantify the DNA binding here with a proper statistical comparison with triplicate.

For EMSA (Figure 3B), we used samples of ssDNA:protein= 1:0. 1:1, 1:2, 1:4 and 0:1 molecular ratios with 10 pM as a 1 unit. We added concentrations of the samples in the Method of [Electrophoretic mobility shift assay for ssDNA binding].

Following the comment, we tried to quantify the binding strength by integrating the grayscale of the bands in gel photos. However, we are concerned because this quantitative calculation through grayscale could not provide an accurate representation of results. Many sample groups cannot be run on one gel. Therefore, the gel differences in parameters cause large errors in the calculation as shown in Author response image 1. Although the calculated integral grayscale chart is consistent with our conclusion, we do not want to add this to our manuscript.

Author response image 1.

(5) Because of poor writing, the authors need to ask for English editing.

We are very sorry for the language. We asked a company (Editag, https:www.editage.jp) to do a native speaker revision and used AI to recheck English.

Minor points:

(1) Lines 47-58, Supplementary Figure 1: Although the sentences describe well how CMG assembles on the replication origin, the figure does not reflect what is written, but rather shows a simple schematic figure related to the work. However, for the general readers, it is very useful to see a general model of the CMG assembly. Then, the authors need to emphasize the steps focused in this study.

Thank you for your thoughtful comments. We optimized Figure 1 and hope it will be more understandable to general readers.

(2) Line 50, DDK[6F0L](superscript): what is 5F0L?

We are sorry for this mistake, that is a PDBID of the DDK structure. we deleted 6F0L.

(3) Lines 68 and 69, ssDNA and dsDNA: should be "single-stranded DNA (ssDNA)" and double-stranded DNA (dsDNA) when these words appear for the first time.

Following the comment, we modified it to “single-stranded DNA (ssDNA)” and “double-stranded DNA (dsDNA)” (P3/L68,70).

(4) Line 84, Cdc45s: What "s" means here?

We are sorry for this mistake, we modified it to “Cdc45”.

(5) Line 87, Sld3deltaC: What is Sld3deltaC? This is the deletion of either the Cdc45-binding domain or the C-terminal domain.

Sld3ΔC is a deletion of the C-terminal domain of Sld3. We added the residue range and explanation (P4/L91).

(6) Line 103: Although the authors mentioned beta-sheets 1-14 in the text, there is no indication in Figures. It is impossible to see the authors' conclusion.

The secondary structure elements of Sld3CBD-Cdc45 are shown in Supplementary Figures 4 and 5. Following the comment, we added a topology diagram of Sld3CBD and Cdc45 in the Sld3CBD-Cdc45 complex as Supplementary Figure 3 and added citations when describing structural elements.

(7) Line 106, huCdc45: Does this mean human Cdc45? If so, it should be "human CDC45 (huCDC45). CMG form is from budding yeast? Please specify the species.

Yes, huCdc45 is human Cdc45. We modified it into “human CDC45 (huCdc45)”.

(8) Line 107, Supplemental Figure 3B, black ovals: Please add "alpha7" in the Figure.

Following the comment, we added a label of Cdc45 α7 to Supplemental Figure 3B and 3C (Supplemental Figure 4B and 4C in revised version).

(9) Line 128, DHHA1: What is this? Please explain it in the text.

Following the comment, we added the information on DHHA1 (P3/L75-77).

(10) Line 130, beta13, and beta14: If the authors would like to point out these structures, please indicate where these sheets are in Figures.

We added a topology diagram as Supplementary Figure 3 to show the β-sheet in DHH and added a citation in the text.

(11) Line 133: Please add (Figure 1B) after the a8CTP.

Following the comment, we added “(Figure 1C)” (1B is 1C in revised version) after the α8CTP (P6/L133).

(12) Line 140: After DHHA1, please add (Figure 1C).

Following the comment, we added the figure citation after the DHHA1 (P6/L140).

(13) Line 142: After DHHA1, please add (Figure 1D).

Following the comment, we added the figure citation after the DHHA1 (P6/L142).

(14) Line 149, Sld3-Y seemed to retain a faint interaction with Cdc45. The Cdc45 band is too faint here. Moreover, as shown above, without the quantification with proper statistics, it is hard to draw this kind of conclusion.

We agree that the Cdc45 band corresponding to Sld3-Y in the pull-down assay was very faint, so we performed an in vivo experiment (Fig2C) to confirm this result.

(15) Line 149, Figure 2A and B: What kind of interaction assay was used here? Simple pull-down. It seems to eluate from the column. If so, how do the authors evaluate the presence of the proteins in different fractions? Please explain the method briefly in the main text.

Figure 2 shows a co-express pull-down binding assay. To describe the co-express pull-down experiments clearly, we added more explanations in the Methods [Mutation analysis of Sld3 and Cdc45].

(16) Line 154-155: Please show the quantification to see if the reduced binding is statistically significant.

Here, we explain why Cdc45-A remained Sld3CBD-bind ability. Although mutant Cdc45-A has reduced three hydrogen bonds with D344 of Sld3CBD, the remaining hydrogen-bond network keeps contact between Sld3CBD and Cdc45.

(17) Line 158, cell death: "No growth" does not mean cell death. Please rephrase here.

Following the comment, we modified it to “no growth” (P6/L158).

(18) Line 166: After CMG dimer, please add "respectively".

Following the comment, we added the word “, respectively” after CMG dimer (P7/L178).

(19) Line 194-195: I can not catch the meaning. Please rephrase here to clarify the claim. What are ssARS1-2 and ARS1-5?

Following the comment, we added more information about ssDNA fragments at the beginning of this section (P8/L210-214).

(20) Figure 4A and Supplemental Figure 12 top, schematic figure of ARS region. It is hard to catch. More explanation of the nature of the DNA substrates and much better schematic presentations would be appreciated.

Following the comment, we added more information about ARS1 to the figure legend.

(21) Figure 1A, dotted ovals should be dotted squares as shown in the enlarged images on the bottom.

Following the comment, we modified Figure 1A and the legend to change the dotted ovals into dotted squares.

These paragraphs tell us that it is important to collect comprehensive data, think about the impact of relevant parties, and considering the groups that are easily overlooked before making decisions. This reminds me of cyberbullying in the society today. Lots of people only listen to one side of the story. They get emotionally stirred up by comments on a popular influencer’s social post and end up participating in online bullying against the other group. This kind of behavior stems from a lack of critical thinking and the unwillingness to investigate the truth from multiple perspectives, which can have serious consequences.

An example in social media is internet violence. Most people cite common shaming as upholding justice but are remiss in forgetting the psychological as well as emotional harm inflicted on the target. By focusing on the enjoyment of calling someone out at any cost while forgetting the long-term impact on the target's well-being, users forget the harm that their actions may result in eventually

In this NPR transcript we learn that basically Elon Musk has broken a contract with twitter as he "secretly stopped taking action to buy twitter" this idea shows which the two people in this transcript mention of changing mind when he feels and trashing the company, it is interesting that even billionaires think that their wish-washy thinking may not harm others or the reputation of others when it actually does.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Revision Plan (Response to Reviewers)

1. General Statements [optional]

Response: We are pleased the reviewers appreciate the power of this novel proteomics methodology that allowed us to uncover new depths on the complexity of the ribosome ubiquitination code in response to stress. We also appreciate that the reviewers think that this is a "very timely" study and "interesting to a broad audience" that can change the models of translation control currently adopted in the field. Characterizing complex cellular processes is critical to advance scientific knowledge and our work is the first of its kind using targeted proteomics methods to unveil the integrated complexity of ribosome ubiquitin signals in eukaryotic systems. We also appreciate the fairness of the comments received and below we offer a comprehensive revision plan substantially addressing the main points raised by the reviewers. According to the reviewers' suggestions, we will also expand our studies to two additional E3 ligases (Mag2 and Not4) known to ubiquitinate ribosomes, which will create an even more complete perspective of ubiquitin roles in translation regulation.

2. Description of the planned revisions

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

The authors present a potentially powerful proteomics platform using parallel reaction monitoring (PRM) to quantitatively profile ribosomal protein (RP) ubiquitylation, with a focus on yeast under hydrogen peroxide (H₂O₂) stress. This approach robustly identifies both known and novel RP modifications, including basal ubiquitylation events previously undetected, and identifies Hel2-dependent mechanisms. The data support the conclusion that RPs are regulated by a multifaceted ubiquitin code, establishing a good foundation for the study.

However, the study's focus shifts in a manner that introduces several limitations. Following the rigorous PRM-based analyses, the reliance on Western blotting without replication or quantification (e.g., single-experiment data in Figs. 3-5) significantly weakens the evidence. Experimental design becomes inconsistent, with variable combinations of stressors (H₂O₂, MMS, 4-NQO) and genetic backgrounds (WT, hel2Δ, rad6Δ) that preclude systematic comparisons. For instance, Fig. 3C/E and Fig. 4 omit critical controls (e.g., MMS in Fig. 4, rad6Δ in Fig. 3E), while Fig. 5 conflates distinct variables by comparing H₂O₂-treated rad6Δ with MMS-treated hel2Δ-a design that obscures causal relationships. Furthermore, Fig. 3F highlights that 4-NQO and MMS elicit divergent responses in hel2Δ, undermining the rationale for using these stressors interchangeably. These inconsistencies culminate in a fragmented narrative; attempts to link ISR activation or ribosome stalling to RP ubiquitylation become impossible, leaving the primary takeaway as "stress responses are complex" rather than advancing mechanistic insight.

        __Response: __We appreciate the evaluation of our work and that the power of our proteomics method established a good foundation for the study. We also understand the reviewer's concerns and we will detail below a plan to enhance quantification and increase systematic comparisons. The experiments presented here were conducted with biological replicates, but in several instances, we focused on presence and absence of bands, or their pattern (mono vs poly-ub) because of the semi-quantitative nature of immunoblots. We will revise the figures and present their quantification and statistical analyses. In additional, we did not intend to use these stressors interchangeably, but instead, to use select conditions to highlight the complexity the stress response. In particular, we followed up with H2O2 *versus* 4-NQO because both chemicals are considered sources of oxidative stress. Even though it is unfeasible to compare every single stress condition in every strain background, in the revised version, we will include additional controls to increase the cohesion of the narrative, and expand the comparison between MMS, H2O2, and 4-NQO, as suggested. Details below.

To strengthen the work, the following revisions are essential:

R1.1. Repeat and quantify immunoblots: All Western blotting data require biological replicates and statistical analysis to support claims.

        __Response: __As requested, we will display quantification and statistical analysis of the suggested and new immunoblots that will be conducted during the revision period.

R1.3. Remove non-parallel comparisons: The mRNA expression analysis in Fig. 5, which compares dissimilar conditions (e.g., rad6Δ + H₂O₂ vs. hel2Δ + MMS), should be omitted or redesigned to enable direct, strain- and stressor-matched contrasts.

        __Response: __We will follow the reviewers' suggestion and redesign the analysis to increase consistency and prioritize data under identical conditions. To increase confidence in the mRNA data analysis, we intend to perform follow up experiments and analyze protein abundance of *ARG proteins* and *CTT1 *under different conditions. The remaining data using non-parallel comparisons will be moved to supplemental material and de-emphasized in the final version of the manuscript.

R1.4. Standardize experimental variables: Restructure the study to maintain identical genetic backgrounds and stressors across all figures, enabling systematic interrogation of enzyme- or stress-specific effects on the ubiquitin code.

        __Response: __To ensure a better comparison across strains and conditions, we will re-run several experiments and focus on our main stress conditions. Specifically:

• 3D: We plan to re-run this experiment and include MMS

• 3E: We plan to perform the same panel of experiments in rad6D ,and display WT data as main figure.

• 4A-B: We plan to perform translation output (HPG incorporation) experiments with MMS as suggested

• 4C: We plan to re-run blots for p-eIF2a under MMS for improved comparison.

Reviewer #1 (Significance (Required)):

The authors present a potentially powerful proteomics platform using parallel reaction monitoring (PRM) to quantitatively profile ribosomal protein (RP) ubiquitylation, with a focus on yeast under hydrogen peroxide (H₂O₂) stress. This approach robustly identifies both known and novel RP modifications, including basal ubiquitylation events previously undetected, and identifies Hel2-dependent mechanisms. The data support the conclusion that RPs are regulated by a multifaceted ubiquitin code, establishing a good foundation for the study.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

In this manuscript the authors use a new target proteomics approach to quantify site-specific ubiquitin modification across the ribosome before and after oxidative stress. Then they validate their findings following in particular ubiquitination of Rps20 and Rps3 and extend their analysis to different forms of oxidative stress. Finally they question the relevance of two known actors of ribosome ubiquitination, Hel2 and Rad6. It is not easy to summarize the observations because in fact the major finding is that the patterns of ribosome ubiquitination occur in a stresser and enyzme specific manner (even when considering only oxidative stress). However, the complexity revealed by this study is very relevant for the field, because it underlies that the ubiquitination code of ribosomes is not easy to interpret with regard to translation dynamics and responses to stress or players involved. It suggests that some of the models that have generally been adopted probably need to be amended or completed. I am not a proteomics expert, so I cannot comment on the validity of the new proteomics approach, of whether the methods are appropriately described to reproduce the experiments. However, for the follow up experiments, the results following Rps20 and Rps3 ubiquitination are well performed, nicely controlled and are appropriately interpreted.

Maybe what one can regret is that the authors have limited their analysis to the study of Hel2 and Rad6, and not included other enyzmes that have already been associated with regulation of ribosome ubiquitination, to get a more complete picture. It may not take that much time to test more mutants, but of course there is the risk that rather than enable to make a working model it might make things even more complex.

        __Response: __We value the positive evaluation of our work. We also appreciate the notion that it meaningfully expands the knowledge on the complexity of the ribosome ubiquitination code, challenges the current models of translation control, and conducted well-performed, and nicely controlled experiments. To address the main concern of the reviewer, we will expand our work by studying two additional enzymes involved in ribosome ubiquitination (Mag2 and Not4) and provide a more comprehensive picture of this integrated system. Specifically, we will generate yeast strains deleted for *MAG2* and *NOT4*, and evaluate their impact in ribosome ubiquitination under our main conditions of stress. We will investigate the role of these additional E3s in translation output (HPG incorporation), and in inducing the integrated stress response via phosphorylated eIF2α and Gcn4 expression. Additional follow up experiments will be performed according to our initial results.

Reviewer #2 (Significance (Required)):

In recent years, regulation of translation elongation dynamics has emerged as a much more relevant site of control of gene expression that previously envisonned. The ribosome has emerged as a hub for control of stress responses. Therefore this study is certainly very timely and interesting for a broad audience. However, it does fall short of giving any simple picture, and maybe the only point one can question is whether it is interesting to publish a manuscript that concludes that regulation is complicated, without really being able to provide any kind of suggestive model.

My feeling is nevertheless that it will impact how scientists in the field design their experiments and what they will conclude. It will certainly also drive new experiments and approaches, and lead to investigations on how all the different players in regulation of ribosome modification talk to each other and signal to signaling pathways.

        __Response: __We appreciate the comments and the balanced view that studies like ours will still be impactful and contribute to a number of fields in multiple and meaningful ways. With the new experiments proposed here, and used of additional mutants and strains, we intend to propose and provide a more unified model that explain this complex and dynamic relationship.

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Recent studies have shown that the ubiquitination of uS3 (Rps3) is crucial for the quality control of nonfunctional rRNA, specifically in the process known as 18S noncoding RNA degradation (NRD). Additionally, the ubiquitination of uS10 (Rps20) plays a significant role in ribosome-associated quality control (RQC). However, the dynamics of ribosome ubiquitination in response to oxidative stress are not yet fully understood.

In this study, the authors developed a targeted proteomics method to quantify the dynamics of ribosome ubiquitination in response to oxidative stress, both relatively and stoichiometrically. They identified 11 ribosomal sites that exhibited increased ubiquitin modification after exposure to hydrogen peroxide (H2O2). This included two known targets: uS10 and uS3 (of Hel2), which recognize collided ribosomes and initiate the processes of 18S NRD and translation quality control (RQC). Using isotope-labeled peptides, the researchers demonstrated that these modifications are non-stoichiometric and display significant variability among different peptides.

Furthermore, the authors explored how specific enzymes in the ubiquitin system affect these modifications and their impact on global translation regulation. They found that uS3 (Rps3) and uS10 (Rps20) were modified differently by various stressors, which in turn influenced the Integrated Stress Response (ISR). The authors suggest that different types of stressors alter the pattern of ubiquitinated ribosomes, with Rad6 and Hel2 potentially competing for specific subpopulations of ribosomes.

Overall, this study emphasizes the complexity of the ubiquitin ribosomal code. However, further experiments are necessary to validate these findings before publication.

Major Comments:

I consider the additional experiments essential to support the claims of the paper.

R3.1. To understand the roles of ribosome ubiquitination at the specific sites, the authors must perform stressor-specific suppression of global translation, as demonstrated in Figures 4 and 5. This should include the uS10-K6R/K8R and uS3-K212R mutants.

        __Response: __We understand the importance of the suggested experiment. We have already requested and kindly received strains expressing these mutations, which will reduce the time required to successfully address this point. We will perform our translation and ISR assays such as the one referred by the reviewer in Figs. 4A-C and 5E, and results will determine the role of individual ribosome ubiquitination sites in translation control.

R3.2. It is crucial to ensure that experiments are adequately replicated and that statistical analysis is thorough, with precise quantification. For a more accurate comparison between wild-type (WT) and Hel2 deletion mutants regarding ribosome ubiquitination, the authors should quantify the ubiquitinated ribosomes in both WT and Hel2 mutants under stress conditions. This quantification should be conducted on the same blot, using diluted control samples. Similarly, in Figures 3F and 4C, for an accurate comparison between WT and Hel2 or Rad6 deletion mutants, the authors should quantify the ubiquitinated ribosomes across these conditions. Again, this quantification should be performed on the same blot with the dilution of control samples.

        __Response: __As was also requested by reviewer 1 and discussed above (point R1.1), we will conduct quantification and display statistical analyses for our immunoblots. In addition, we will re-run the aforementioned experiments to improve quantification following the reviewers' request (same gel & diluted control samples).

Reviewer #3 (Significance (Required)):

• General assessment:

Recent studies reveal that the ubiquitination of uS3 (Rps3) is essential for the quality control of nonfunctional rRNA (18S NRD), while the ubiquitination of uS10 (Rps20) plays a crucial role in ribosome-associated quality control (RQC). However, the dynamics of ribosome ubiquitination in response to oxidative stress remain unclear.

• Advance:

In this study, the authors developed a targeted proteomics method to quantify ribosome ubiquitination dynamics in response to oxidative stress, both relatively and stoichiometrically. By utilizing isotope-labeled peptides, they demonstrated that these modifications are non-stoichiometric and exhibit significant variability across different peptides. They identified 11 ribosomal sites that showed increased ubiquitin modification following H2O2 exposure, including two known targets of Hel2, which recognize collided ribosomes and induce translation quality control (RQC).

• Audience: This information will be of interest to a specialized audience in the fields of translation, ribosome function, quality control, ubiquitination, and proteostasis.

• The field: Translation, ribosome function, quality control, ubiquitination, and proteostasis.

__ Response:__ We appreciate that our work will be valuable to a number of fields in protein dynamics and that our method advances the field by measuring ribosome ubiquitination relatively and stoichiometrically in response to stress.

3. Description of the revisions that have already been incorporated in the transferred manuscript

Response: All requested changes require experiments and data analyses, and a complete revision plan is delineated above in section #2.

• *

4. Description of analyses that authors prefer not to carry out

• *

R1.2. Leverage the PRM platform: Apply the established quantitative proteomics approach to validate or extend findings in Fig. 3 (e.g., RAD6-dependent ubiquitylation), ensuring methodological consistency.

        __Response: __Although we understand the interest on the proposed result for consistency, this is the only requested experiment that we do not intend to conduct. Because of the lack of overall ubiquitination of ribosomal proteins in *rad6**D* in response to H2O2 (e.g., Silva et al., 2015, Simoes et al., 2022), we believe that this PRM experiment in unlikely to produce meaningful insight on the ubiquitination code. In this context, we expected that sites regulated by Hel2 will be the ones largely modified in rad6*D *and we followed up on them via immunoblot. Moreover, this experiment would not be time or cost-effective, and resources and efforts could be used to strengthen other important areas of the manuscript, such as including the E3's Mag2 and Not4 into our work.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #2

Evidence, reproducibility and clarity

This is a very interesting paper investigating the fitness and cellular effects of mutations that drive dihedral protein complex into forming filaments. The Levy group have previously shown that this can happen relatively easily in such complexes and this paper now investigates the cellular consequences of this phenomenon. The study is very rigorous biophysically and very surprisingly comes up empty in terms of an effect: apparently this kind of self-assembly can easily be tolerated in yeast, which was certainly not my expectation. This is a very interesting result, because it implies that such assemblies may evolve neutrally because they fulfill the two key requirements for such a trajectory: They are genetically easily accessible (in as little as a single mutation), and they have perhaps no detrimental effect on fitness. This immediately poses two very interesting questions: Are some natural proteins that are known to form filaments in the cell perhaps examples of such neutral trajectories? And if this trait is truly neutral (as long as it doesn't affect the base biochemical function of the protein in question), why don't we observe more proteins form these kinds of ordered assemblies.

I have no major comments about the experiments as I find that in general very carefully carried out. I have two more general comments:

1. The fitness effect of these assemblies, if one exists, seems very small. I think it's worth remembering that even very small fitness effects beyond even what competition experiments can reveal could in principle be enough to keep assembly-inducing alleles at very low frequencies in natural populations. Perhaps this could be acknowledged in the paper somewhere.
2. The proteins used in this study I think were chosen such that they do not have an important function in yeast that could be disrupted by assembly This allows the effect of the large scale assemblies to be measured in isolation. If I deduced this correctly, this should probably be pointed out agin in this paper (I apologise if I missed this).
3. The model system in which these effects were tested for is yeast. This organism has a rigid cell wall and I was wondering if this makes it more tolerant to large scale assemblages than wall-less eukaryotes. Could the authors comment on this?

Minor points:

In Figure 2D, what are the fits? And is there any analysis that rules out expression effects on the mutant caused by higher levels of the wild-type? The error bars in Figure 2E are not defined.

Significance

This is a remarkably rigours paper that investigates whether self-assembly into large structures has any fitness effect on a single celled organism. This is very relevant, because a landmark paper from the Levy group showed that many proteins are very close in genetic terms to forming such assemblies. The general expectation I think would have been that this phenomenon is pretty harmful. This would have explained why such filaments are relatively rare as far as we know. This paper now does a large number of highly rigours experiments to first prove beyond doubt that a range of model proteins really can be coaxed into forming such filaments in yeast cells through a very small number of mutations. Its perhaps most surprising result is that this does not negatively affect yeast cells.

From an evolutionary perspective, this is a very interesting and highly surprising result. It forces us to rethink why such filaments are not more common in Nature. Two possible answers come to mind: First, it's possible that filamentation is not directly harmful to the cell, but that assembling proteins into filaments can interfere with their basic biochemical function (which was not tested for here).

Second, perhaps assembly does cause a fitness defect, but one so small that it is hard to measure experimentally. Natural selection is very powerful, and even fitness coefficients we struggle to measure in the laboratory can have significant effects in the wild. If this is true, we might expect such filaments to be more common in organisms with small effective population sizes, in which selection is less effective.

A third possibility is of course that the prevalence of such self-assembly is under-appreciated. Perhaps more proteins than we currently know assemble into these structures under some conditions without any benefit or detriment to the organism.

These are all fascinating implications of this work that straddle the fields of evolutionary genetics and biochemistry and are therefore relevant to a very wide audience. My own expertise is in these two fields. I also think that this work will be exciting for synthetic biologists, because it proves that these kinds of assemblies are well tolerated inside cells.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1 (Public Review):

The article emphasizes vocal social behavior but none of the experiments involve a social element. Marmosets are recorded in isolation which could be sufficient for examining the development of vocal behavior in that particular context. However, the early-life maturation of vocal behavior is strongly influenced by social interactions with conspecifics. For example, the transition of cries and subharmonic phees which are high-entropy calls to more low-entropy mature phees is affected by social reinforcement from the parents. And this effect extends cross context where differences in these interaction patterns extend to vocal behavior when the marmosets are alone. From the chord diagrams, cries still consist of a significant proportion of call types in lesioned animals. Additionally, though it is an intriguing finding that the infants' phee calls have acoustic differences being 'blunted of variation, less diverse and more regular,' the suggestion that the social message conveyed by these infants was 'deficient, limited, and/or indiscriminate' is not but can be tested with, for example, playback experiments.

We recognize that our definition of vocal social behavior is not within the normal realm of direct social interactions. We were particularly interested in marmoset vocalizations as a social signal, such as phees, cries and twitter, even when their family members or conspecifics are not visibly present. Generally speaking, in the laboratory, infant marmosets make few calls when in the presence of another conspecific, but when isolated they naturally make phee calls to reach out to their distantly located relatives. In this context, while we did not assess the animals interacting directly, we assessed what are normally referred to as ‘social contact calls,’ hence the term ‘social vocalizations.’ Playback recordings might provide potential evidence of antiphonal calling as a means of social interaction and might reveal the poor quality of the social message conveyed by the infant, but even here, the vocalizing marmoset would be calling to a non-visible conspecific. Thus, although our experiment lacked a direct social element, our data suggest that in the absence of a functioning ACC in early life, infant calls that convey social information, and which would elicit feedback from parents and other family members, may be compromised, and this could potentially influence how that infant develops its social interactive skills. We have now commented on the significance of social vocalizations in the introductory text (page 3) and discussion (page 15).

The manuscript would benefit from the addition of more details to be able to better determine if the conclusions are well supported by the data. Understanding that this is very difficult data to get, the number of marmosets and some variability in the collection of the data would allow for the plotting of each individual across figures. For example, in the behavioral figures, which is the marmoset that is in the behavioral data that has a sparing of the ACC lesion in one hemisphere? Certain figures, described below in the recommendations for the authors, could also do with additional description.

Thanks for these suggestions. We have plotted the individual animals in the relevant figures and addressed the comments and recommendations listed below.

Reviewer #1 (Recommendations For The Authors):

Given the number of marmosets, variability in the collected data, lesion extent, and different controls, I would like to see more plots with individuals indicated (perhaps with different symbols). More details could also be added for several plots.

Figure 2D (new) and 2E now have plots that represent the individual animals, each represented by a different symbol.

Figure 2A) Since lesions are bilateral, could you also show the extent of the lesions on the other side for completeness?

Our intention was to process one hemisphere of each brain for Golgi staining to examine changes in cell morphology in the ACC and associated brain regions following the lesion. Unfortunately, the Golgi stain was unsuccessful. Consequently, we were unable to use the tissue to reconstruct the bilateral extent of the lesion. We did, however, first establish the bilateral nature of the lesion through coronal slices of the animals MRI scan before processing the intact hemisphere to confirm the bilateral extent of the lesion. The MRI scans (every 5th section) for each control and lesioned animal is compiled in a figure in the supplementary materials (Fig. S1). These scans show that the ACC-lesioned animals have bilateral lesions with one animal (ACC1) showing some sparing in one hemisphere, as we noted in the text. We have now made reference to this supplemental figure in the text (page 5).

Figure 2B/C) In Figure 2B, control and ACC lesions are in the columns while right next to it in 2C, ACC lesion and control are in the rows. Could these figures be adjusted so that they are consistent?

We have now adjusted these figures and updated the figure legends accordingly.

Figure 2C) Is there quantification of the 'loss of neurons and respective increase in glial cells at the lesioned site especially at the interface between gray and white matter'? There are multiple slices for each animal.

Thanks for suggesting this. We have now quantified these data which are presented as a new graph as Fig. 2D. These data revealed a significant loss of neurons (NeuN) in the ACC group as well as an increase in glial cells (GFAP and Iba1) relative to the controls. The figure legend and results have also been updated.

Figure 2C) It is difficult for me to distinguish between white and purple - could you show color channels independently since images were split into separate channels for each fluorophore?

Fig. 2C has been revised to better visualize the neurons and glia at the gray and white matter interface. We found that grayscale images for each channel offered a better contrast than separating the channels for each fluorophore.

Figure 2C/D) I like how there are individual dots here for the individual marmosets. Since there are four in each group, could they be represented throughout with symbols (with a key indicating the pair and also the control condition)? For example, were there changes in the histology for control animals that got saline injections as opposed to those that didn't get any surgery?

We have highlighted the individual animals with different symbols in the figures. Although some animals were twin pairs, it was not possible to have twins in all cases. Only two sets were twins. We have indicated the symbols that represent the twin pair in Fig. 2 as well as the MRI scans of the twin pairs in Fig. S1. There were no observed changes in histology for the sham animals relative to the other non-sham controls. The MRI scan for one sham CON2 shows herniated tissue in the right hemisphere which is a normal consequence of brain exposure caused by a craniotomy.

Figure 3D-E) Here, individual data points could be informative especially given that some animals are missing data past the third week.

To prevent cluttering the figure with too many data points, we have added the sample size for each group in the figure legend (pages 33).

Figure 3D/F) What exactly is the period that goes into this analysis? In the text, 'Further analysis showed that the ACC lesion had minimal effects on the rate of most call types during this period'. Is this period from weeks 3 to 6 relative to the proportions in week 2? I think I also don't quite understand the chord diagram. The legend says 'the numbers around each chord diagram represents relative probability value for each call type transition' so how does that relate to the proportion of these call types? It looks like there is a wider slice for cries for ACC-lesioned animals each week. I also don't see in the week 4 chord diagram, the text description of 'elevation in the rate of 'other' calls, which comprised tsik, egg, eck, chatter and seep calls. These calls were significantly elevated in animals after the ACC lesion."

We apologize for the confusion. Fig 3D and Fig 3F are not directly related. Fig. 3D shows the different types of emitted calls. The figure shows the averaged data per group pooled from post-surgery weeks (week 3 – week 6). It represents the proportion of individual call types relative to the total number of calls during each recording period. The only major finding here was the increased rate of ‘other’ calls comprising tsik, egg, ock, chatter and seep calls. These calls were significantly elevated in animals after the ACC lesion.

While Fig. 3D represents the differences in the proportion of calls, the chord diagrams in Fig. 3F represents the probability of call-to-call transition obtained from a probability matrix. At postnatal week 6, marmosets with ACC lesions showed a higher likelihood of transitions between all call types, but less frequent transitions between social contact calls relative to sham controls. The chord diagrams visualize the weighted probabilities and directionality of these transitions between the different call types. Weighted probabilities were used to account for variations in call counts. The thickness of the arrows or links indicates the probability of a call transition, while the numbers surrounding each chord diagram represent the relative probability value for each specific transition. We have now reworded the text and clarified these details in the figure legend (pages 32-33).

Figure 3E) How is the ratio on the y-axis calculated here?

The y-axis represents the averaged value of the ratios of the number of social contact calls relative to non-social contact calls in each recording per subject per group (i.e., (x̄ (# social calls / # non-social calls). This is now included in the figure legend and the axis is updated (page 32).

Also, cries could be considered a 'social contact call' since they are produced by infants to elicit responses from the parents. There is also the hypothesis in the literature that cries transition into phees.

The reviewer is correct. Cries are often considered a social contact call because they elicit parental feedback. We decided to separate cry-calls from other social contact calls for two reasons. First, in our sample, we found cry behavior to be highly variable across the animals. For example, one control infant cried incessantly whereas another control infant cried less than normal. This extreme variability in animals of the same group masked the features between animals that reliably differentiated between them. Second, cry-calls elicit feedback from parents who are normally within the vicinity of the infant whereas phee calls elicit antiphonal phee calls from any distantly located conspecific. In other words, the context in which these calls are often elicited are very different.

The use of 'syntactical' is a bit jarring to me because outside of linguistics, its use in animal communication generally refers to meaning-bearing units that can be combined into well-formed complexes such as pod-specific whale songs or predator alarm calls with concatenated syllable types in some species of monkeys. To my knowledge, individual phee syllables have not been currently shown to carry information on their own and may be better described as 'sequential' rather than 'syntactical'.

We agree. We have made this change accordingly.

Figure 4B) How many phee calls with differing numbers of syllables are present each week? How equal is the distribution given that later analyses go up to 5 syllables?

The total number of phee calls with differing number of syllables ranged between 20-40 phees. This number varied between subjects, per week. The most common were 3- and 4-syllable phee calls which ranged from 7-15. Due to this variability, Fig. 4B presents the average syllable count. The axis is now updated.

Figure 4C-E) How is the data combined here? Is there a 2nd syllable, the combined data from the 2nd syllable from phee calls of all lengths (1 - 5?). If so, are there differences based on how long the total sequence is?

The combined data represents the specific syllable (e.g., the 1st syllable in a 2-syllable phee, in a 3-syllable phee and in a 4-syllable phee) irrespective of the length of the sequence in a sequence. No differences were observed between 2nd syllable in a 2 syllable phee and 2nd syllable in a 3 or a 4 syllable phee. We have included this detail in the figure legend (page 33-34).

So duration is a vocal parameter that is highly dependent on physical factors such as body size and lung volume, where there differences in physical growth between the pairs of ACC-lesioned marmosets and their twins? Entropy is less closely tied to these physical factors but has previously been shown to decrease as phee calls mature, which we can also see in the negative relationship of the control animals. Do you know of experiments that show that lower entropy calls are more 'blunted'?

Thank you for raising the important issue of physical growth factors. For twin pairs, it is not uncommon for one infant to be slightly bigger, heavier or stronger than the other presumably because one gets more access to food. With increasing age, we did not observe significant changes in bodyweight between the groups. We examined grip strength in all infants as a means of assessing how well the infant was able to access food during nursing. Poor grip strength would indicate a lower propensity to ‘hang on’ to the mother for nursing which could lead to lower weight gain and reduced physical growth. We found that both grip strength and body weight increased as the infants got older and both parameters were equivalent. We have included an additional figure to show the normal increase in both weight and grip strength to the supplemental materials (Fig. S3) and have made reference to this in the text (page 8).

As for entropy, it’s impact on the emotional quality of vocalizations has not been systematically explored. Generally speaking, high entropy relates to high randomness and distortion in the signal. Accordingly, one view posits low-entropy phee calls represent mature sounding calls relative to noisy and immature high-entropy calls (e.g., Takahasi et al 2017). In the current study, the reduction in syllable entropy observed for both groups of animals with increasing age is consistent with this view. At the same time entropy can relate to vocal complexity; high entropy refers to complex and variable sound patterns whereas low entropy sounds are predictable, less diverse and simple vocal sequences (Kershenbaum, A. 2013. Entropy rate as a measure of animal vocal complexity. Bioacoustics, 23(3), 195–208). One possibility is that call maturity does not equate directly to emotional quality. In other words, a low-entropy mature call can also be lacking in emotion as observed in humans with ACC damage; these patients show mature speech, but they lack the variations in rhythms, patterns and intonation (i.e., prosody) that would normally convey emotional salience and meaning. Our observation of a reduction in phee syllable entropy in the ACC group in the context of being short and loud with reduced peak frequency is consistent with this view. Our use of the word ‘blunt’ was to convey how the calls exhibited by the ACC group were potentially lacking emotional meaning. Beyond this speculation, we are not aware of any papers that have examined the relationship between entropy and blunted calls directly. We have now included this speculation in the discussion (pages 12-13).

Reviewer #2 (Public Review):

The authors state that the integrity of white matter tracts at the injection site was impacted but do not show data.

We have added representative micrographs of a control and ACC-lesioned animal in a new supplementary figure which shows the neurotoxin impacted the integrity of white matter tracts local to the site of the lesion (Fig. S2).

The study only provides data up to the 6th week after birth. Given the plasticity of the cortex, it would be interesting to see if these impairments in vocal behavior persist throughout adulthood or if the lesioned marmosets will recover their social-vocal behavior compared to the control animals.

We agree. Our original intention was to examine behavior into adulthood. Unfortunately, the COVID-19 pandemic compromised the continuation of the study. We were limited by the data that we were allowed to acquire due to imposed restrictions. Some non-vocalization data collected when the animals were young adults is currently being prepared for another paper.

Even though this study focuses entirely on the development of social vocalizations, providing data about altered social non-vocal behaviors that accompany ACC lesions is missing. This data can provide further insights and generate new hypotheses about the exact role of ACC in social vocal development. For example, do these marmosets behave differently towards their conspecifics or family members and vice versa, and is this an alternate cause for the observed changes in social-vocal development?

We agree. At the time however, apparatus for assessing behavior between the infant’s family and non-family members was not available. Assessing such behaviors in the animals holding room posed some difficulty since marmosets are easily distracted by other animals as well as the presence of an experimenter, amongst other things. This is an area of investigation we are currently pursuing.

Reviewer #3 (Public Review):

It is striking to find that the vocal repertoire of infant marmosets was not significantly affected by ACC lesions. During development, the neural circuits are still maturing and the role of different brain regions may evolve over time. While the ACC likely contributes to vocalizations across the lifespan, its relative importance may vary depending on the developmental stage. In neonates, vocalizations may be more reflexive or driven by physiological needs. At this stage, the ACC may play a role in basic socioemotional regulation but may not be as critical for vocal production. Since the animals lived for two years, further analysis might be helpful to elucidate the precise role of ACC in the vocal behavior of marmosets.

Figure 3D. According to the Introduction "...infant ACC lesions abolish the characteristic cries that infants normally issue when separated from its mother". Are the present results in marmosets showing the opposite effect? Please discuss.

To date, the work of Maclean (1985) is the only publication that describes the effect of early cingulate ablation on the spontaneous production of ‘separation calls’ largely construed as cries, coos and whimpers in response to maternal separation. All of this work was largely performed in rhesus macaques or squirrel monkeys. In addition to ablating the cingulate cortex, Maclean found that it was necessary to ablate the subcallosal (areas 25) and preseptal cingulate cortex (presumably referring to prelimbic area 32) to permanently eliminate the spontaneous production of separation cry calls. Our ablation of the ACC was more circumscribed to area 24 and is therefore consistent with MacLean’s earlier work that removal of ACC alone does not eliminate cry behavior. In adults, ACC ablation is insufficient at eliminating vocalization as well. We make reference to this on pages 13-14 of the discussion.

Figure 3E and Discussion. Phees are mature contact calls and cries immature contact calls (Zhang et al, 2019, Nat Commun). Therefore, I would rather say that the proportion of immature (cries) contact calls increases vs the mature (phee, trill, twitters) contact calls in the ACC group. Cries are also "isolated-induced contact calls" to attract the attention of the caregivers.

The reviewer is correct in that cries are directed towards caregivers but in our sample, cry behavior was highly variable between the infants. Consequently, in Fig. 3E social contact calls include phee, twitter and trill calls but does not include cries which were separated (see also response to reviewer #1). Many of the calls made during babbling were immature in their spectral pattern (compare phee calls between Fig. 3A and 3B). Cries typically transitioned into phees, twitters or trills before they fully matured. Fig 3E shows that the controls made more isolation-induced social contact calls at postnatal week 6 which were presumably maturing at this time point. Thus, if anything, there was an increase in the proportion of mature contact calls vs immature contact calls with increasing age.

Figure 4D. Animal location and head direction within the recording incubator can have significant effects on the perceived amplitude of a call. Were these factors taken into account?

The reviewer makes an excellent observation. Unfortunately, we did not account for location and head direction because the infants were quite mobile in the incubator. The directional microphone was hidden from view because the infants were distracted by it, and positioned ~12 cm from the marmoset, and placed in the exact same location for every recording. In addition, calls with phantom frequencies were eliminated during visual inspection of spectrograms. Beyond these details, location and head direction were not taken into account.

Figure 4E. When a phee call has a higher amplitude, as is the case for the ACC group (Figure 4D), the energy of the signal will be concentrated more strongly at the phee call frequency ~8KHz. This concentration of the energy reduces the variability in the frequency distribution, leading to lower entropy. The interpretation of the results should be reconsidered. A faint call (control group) can exhibit more variability in the frequency content since the energy is distributed across a wider range of frequencies contributing to higher entropy. It can still be "fixed, regular, and stereotyped" if the behavior is consistent or predictable with little variation. Also, to define ACC calls as "monotonic" I would rather search for the lack of frequency modulation, amplitude variation, or narrower bandwidth.

We very much appreciate this explanation. We were able to identify the maximum frequency that closely matched pitch of a sound for each syllable in a multisyllabic phee. New Fig. 4E shows that the peak frequency for each phee syllable was lower in the ACC-lesioned monkeys which may directly translate to the low entropy observed in this group. The term “monotonic” was used to relate our data to the classical and long-standing evidence of human ACC lesions causing monotonous intonation of speech. When all factors are taken into account, it is evident that the vocal phee signature of the ACC-lesioned animal was structurally different to the controls implicating a less complex and stereotyped ACC signal. Further studies are needed to systematically explore the relationship between entropy and emotional quality of vocalizations

Apart from the changes in the vocal behavior, did the AAC lesions manifest in any other observable cognitive, emotional, or social behavior? ACC plays a role in processing pain and modulating pain perception. Could that be the reason for the observed increase in the proportion of cries in the ACC group and the increase in the phee call amplitude? Did the cries in the ACC group also display a higher amplitude than the cries in the control group?

It was our intention to acquire as much data as possible from these infants as they matured from a cognitive, social and emotional perspective. Unfortunately, our study was hampered by variety of reasons including the COVID-19 pandemic which imposed major restrictions on our ability to continue with the experiment in a time sensitive manner. In addition, the development and construction of the custom apparatus to measure these behaviors was stalled during this period further preventing us from collecting behavioral data at regular time intervals. As for the cry behavior, the number of cries, in the ACC group were very low especially at postnatal week 5 and 6. Consequently, there were very few data points to work with.

Discussion. Louder calls have the potential to travel longer distances compared to fainter calls, possess higher energy levels, and can propagate through the environment more effectively. If the ACC group produced louder phee syllables, how could be the message conveyed over long distances "deficient, limited, and/or indiscriminate"?

Thanks for raising this interesting concept. Not all calls emitted by the animals were loud. We specifically examined the long-distance phee call in this regard. The phee syllables emitted by the ACC group were high amplitude with low frequencies, short duration and low entropy. Taking these factors into account, it is conceivable that the phee calls produced by the ACC group could not effectively convey their message over long distances despite their propagation through the environment. We have made reference to this in the discussion where we focus is specifically on the phee calls only (pages 12).

Abstract: Do marmosets have syntax? Consider replacing "syntactical" with a more appropriate term (maybe "syntax-like").

Thanks for this suggestion. We have replaced the term syntactical with ‘sequential’ as per the recommendation of reviewer #1.

Introduction: "...cries that infants normally issue when separated from its mother". Please replace "its" with "their".

This has been corrected.

Results: Is the reference to Fig 1B related to the text?

We have included and referred to Fig. 1B in the text (results and methods) to show other researchers how they can use this technique as a reliable and safe means of monitoring tidal volume under anesthesia in small infant marmoset without intubation.

I understand that both "spectrograph" and "spectrogram" are used to analyze the frequency content of a signal. Nevertheless, "spectrogram" refers to the visual representation of the frequency content of a signal over time, and this term is commonly used in audio signal processing and specifically in the vocal communication field. I would recommend replacing "spectrograph" with "spectrogram".

Thanks for this suggestion. We have corrected this throughout the manuscript.

(Concerning the previous comment in the public review). Cries are uttered to attract the attention of the caregivers. The increase in the proportion of cries in the ACC group does not match the sentence: "...these infants appeared to make little effort in using vocalizations to solicit social contact when socially isolated".

We apologize for the confusion. It is not the case that the ACC animals make more cries. Cry calls were highly variable amongst the animals. Consequently, although Fig 3D gives the impression that the proportion of cries in higher in ACC animals they did not differ significantly from the controls. Due to their high variability, cries were removed in the measurement of social contact. Accordingly, Fig. 3E does not include cry behavior; it shows that the ACC animals engage less in social contact calls.

Related to Figure 3. What is the difference between "egg" and "eck" calls? Do you mean "ock"?

We apologize. This was a typo. It should be ock calls.

Figure 4B. Is the sample size five animals per group and per week? Overlapping data points seem to be placed next to each other. Why in some groups (e.g. ACC 6 weeks) less than five dots are visible?

The sample size differed per week because of the lack of recording during the COVID restrictions. In Fig 4b, we have now separated the overlapping dots. We have also added the sample size of the groups in the figure legends.

Would the authors expect to see stronger differences between the lesioned and the control groups when comparing a later developmental stage? The animals were euthanized at the age of

These speculation is certainly feasible and yes, we were hoping to establish this level of detail with testing at later developmental stages. This is an aspect of development we are currently pursuing.

Could these experiments be conducted?

I’m afraid these animals are longer available, but we are currently conducting experiments in other animals with early life neurochemical manipulations who show behavioral changes into early adulthood.

ACC lesion: It is reported that the lesions extended past 24b into motor area 6M. Did the animal display any motor control disability?

Surprisingly, despite the lesion encroaching into 6M, these animals showed no observable motor impairment. We assessed the animals grip strength and body weight and discovered normal strength and growth in weight in both controls and the lesioned group. We have added this data as supplemental information (Fig. S3).

Author response:

The following is the authors’ response to the original reviews

Reviewer #1 (Public review):

Summary:

This study investigates what happens to the stimulus-driven responses of V4 neurons when an item is held in working memory. Monkeys are trained to perform memory-guided saccades: they must remember the location of a visual cue and then, after a delay, make an eye movement to the remembered location. In addition, a background stimulus (a grating) is presented that varies in contrast and orientation across trials. This stimulus serves to probe the V4 responses, is present throughout the trial, and is task-irrelevant. Using this design, the authors report memory-driven changes in the LFP power spectrum, changes in synchronization between the V4 spikes and the ongoing LFP, and no significant changes in firing rate.

Strengths:

(1) The logic of the experiment is nicely laid out.

(2) The presentation is clear and concise.

(3) The analyses are thorough, careful, and yield unambiguous results.

(4) Together, the recording and inactivation data demonstrate quite convincingly that the signal stored in FEF is communicated to V4 and that, under the current experimental conditions, the impact from FEF manifests as variations in the timing of the stimulus-evoked V4 spikes and not in the intensity of the evoked activity (i.e., firing rate).

Weaknesses:

I think there are two limitations of the study that are important for evaluating the potential functional implications of the data. If these were acknowledged and discussed, it would be easier to situate these results in the broader context of the topic, and their importance would be conveyed more fairly and transparently.

(1) While it may be true that no firing rate modulations were observed in this case, this may have been because the probe stimuli in the task were behaviorally irrelevant; if anything, they might have served as distracters to the monkey's actual task (the MGS). From this perspective, the lack of rate modulation could simply mean that the monkeys were successful in attending the relevant cue and shielding their performance from the potentially distracting effect of the background gratings. Had the visual probes been in some way behaviorally relevant and/or spatially localized (instead of full field), the data might have looked very different.

Any task design involves tradeoffs; if the visual stimulus was behaviorally relevant, then any observed neurophysiological changes would be more confounded by possible attentional effects. We cannot exclude the possibility that a different task or different stimuli would produce different results; we ourselves have reported firing rate enhancements for other types of visual probes during an MGS task (Merrikhi et al. 2017). We have added an acknowledgement of these limitations in the discussion section (lines 323-330 in untracked version). At minimum, our results show a dissociation between the top-down modulation of phase coding, which is enhanced during WM even for these task-irrelevant stimuli, and rate coding. Establishing whether and how this phase coding is related to perception and behavior will be an important direction for future work.

With this in mind, it would be prudent to dial down the tone of the conclusions, which stretch well beyond the current experimental conditions (see recommendations).

We have edited the title (removing the word ‘primarily’) and key sentences throughout to tone down the conclusions, generally to state that the importance of a phase code in WM modulations is *possible* given the observed results, rather than certain (see abstract lines 26-27, introduction lines 59-62, conclusion lines 310-311).

(2) Another point worth discussing is that although the FEF delay-period activity corresponds to a remembered location, it can also be interpreted as an attended location, or as a motor plan for the upcoming eye movement. These are overlapping constructs that are difficult to disentangle, but it would be important to mention them given prior studies of attentional or saccade-related modulation in V4. The firing rate modulations reported in some of those cases provide a stark contrast with the findings here, and I again suspect that the differences may be due at least in part to the differing experimental conditions, rather than a drastically different encoding mode or functional linkage between FEF and V4.

We have added a paragraph to the discussion section addressing links to attention and motor planning (lines 315-333), and specifically acknowledging the inherent difficulties of fully dissociating these effects when interpreting our results (lines 323-330).

Reviewer #2 (Public review):

Summary:

It is generally believed that higher-order areas in the prefrontal cortex guide selection during working memory and attention through signals that selectively recruit neuronal populations in sensory areas that encode the relevant feature. In this work, Parto-Dezfouli and colleagues tested how these prefrontal signals influence activity in visual area V4 using a spatial working memory task. They recorded neuronal activity from visual area V4 and found that information about visual features at the behaviorally relevant part of space during the memory period is carried in a spatially selective manner in the timing of spikes relative to a beta oscillation (phase coding) rather than in the average firing rate (rate code). The authors further tested whether there is a causal link between prefrontal input and the phase encoding of visual information during the memory period. They found that indeed inactivation of the frontal eye fields, a prefrontal area known to send spatial signals to V4, decreased beta oscillatory activity in V4 and information about the visual features. The authors went one step further to develop a neural model that replicated the experimental findings and suggested that changes in the average firing rate of individual neurons might be a result of small changes in the exact beta oscillation frequency within V4. These data provide important new insights into the possible mechanisms through which top-down signals can influence activity in hierarchically lower sensory areas and can therefore have a significant impact on the Systems, Cognitive, and Computational Neuroscience fields.

Strengths:

This is a well-written paper with a well-thought-out experimental design. The authors used a smart variation of the memory-guided saccade task to assess how information about the visual features of stimuli is encoded during the memory period. By using a grating of various contrasts and orientations as the background the authors ensured that bottom-up visual input would drive responses in visual area V4 in the delay period, something that is not commonly done in experimental settings in the same task. Moreover, one of the major strengths of the study is the use of different approaches including analysis of electrophysiological data using advanced computational methods of analysis, manipulation of activity through inactivation of the prefrontal cortex to establish causality of top-down signals on local activity signatures (beta oscillations, spike locking and information carried) as well as computational neuronal modeling. This has helped extend an observation into a possible mechanism well supported by the results.

Weaknesses:

Although the authors provide support for their conclusions from different approaches, I found that the selection of some of the analyses and statistical assessments made it harder for the reader to follow the comparison between a rate code and a phase code. Specifically, the authors wish to assess whether stimulus information is carried selectively for the relevant position through a firing rate or a phase code. Results for the rate code are shown in Figures 1B-G and for the phase code are shown in Figure 2. Whereas an F-statistic is shown over time in Figure 1F (and Figure S1) no such analysis is shown for LFP power. Similarly, following FEF inactivation there is no data on how that influences V4 firing rates and information carried by firing rates in the two conditions (for positions inside and outside the V4 RF). In the same vein, no data are shown on how the inactivation affects beta phase coding in the OUT condition.

Per the reviewer’s suggestion, we have added several new supplementary figures. We now show the F-statistic for discriminability over time for the LFP timecourse (Fig. S2), and as a function of power in various frequencies (Fig. S4). We have added before/after inactivation comparisons of the LFP and spiking activity, and their respective F-statistics for discrimination between contrasts and orientations in Fig. S9. Lastly, we added a supplementary figure evaluating the impact of FEF inactivation on beta phase coding in the OUT condition, showing no significant change (Fig. S11).

Moreover, some of the statistical assessments could be carried out differently including all conditions to provide more insight into mechanisms. For example, a two-way ANOVA followed by post hoc tests could be employed to include comparisons across both spatial (IN, OUT) and visual feature conditions (see results in Figures 2D, S4, etc.). Figure 2D suggests that the absence of selectivity in the OUT condition (no significant difference between high and low contrast stimuli) is mainly due to an increase in slope in the OUT condition for the low contrast stimulus compared to that for the same stimulus in the IN condition. If this turns out to be true it would provide important information that the authors should address.

We have updated the STA slope measurement, excluding the low contrast condition which lacks a clear peak in the STA. Additionally, we equalized the bin widths and aligned the x-axes for better visual comparability. Then, we performed a two-way ANOVA, analyzing the effects of spatial features (IN vs. OUT) and visual conditions (contrast and orientation). The results showed a significant effect of the visual feature on both orientation (F = 3.96, p=0.046) and contrast (F = 14.26, p<10^-3). However, neither the spatial feature nor the spatial-visual interaction exhibited significant effects for orientation (F = 0.52, p=0.473, F=1.56, p=0.212) or contrast (F = 2.19, p=0.139, F=1.15, p=0.283).

There are also a few conceptual gaps that leave the reader wondering whether the results and conclusion are general enough. Specifically,

(1) The authors used microstimulation in the FEF to determine RFs. It is thus possible that the FEF sites that were inactivated were largely more motor-related. Given that beta oscillations and motor preparatory activity have been found to be correlated and motor sites show increased beta oscillatory activity in the delay period, it is possible that the effect of FEF inactivation on V4 beta oscillations is due to inactivation of the main source of beta activity. Had the authors inactivated sites with a preponderance of visual neurons in the FEF would the results be different?

We do not believe this to be likely based on what is known anatomically and functionally about this circuitry. Anatomically, the projections from FEF to V4 arise primarily from the supragranular layers, not layers which contain the highest proportion of motor activity (Barone et al. 2000, Pouget et al. 2009, Markov et al. 2013). Functionally, based on electrical identification of V4-projecting FEF neurons, we know that FEF to V4 projections are predominantly characterized by delay rather than motor activity (Merrikhi et al. 2017). We have now tried to emphasize these points when we introduce the inactivation experiments (lines 185-186).

Experimentally, the spread of the pharmacological effect with our infusion system is quite large relative to any clustering of visual vs. motor neurons within the FEF, with behavioral consequences of inactivation spreading to cover a substantial portion of the visual hemifield (e.g., Noudoost et al. 2014, Clark et al. 2014), and so our manipulation lacks the spatial resolution to selectively target motor vs. other FEF neurons.

(2) Somewhat related to this point and given the prominence of low-frequency activity in deeper layers of the visual cortex according to some previous studies, it is not clear where the authors' V4 recordings were located. The authors report that they do have data from linear arrays, so it should be possible to address this.

Unfortunately, our chamber placement for V4 has produced linear array penetration angles which do not reliably allow identification of cortical layers. We are aware of previous results showing layer-specific effects of attention in V4 (e.g., Pettine et al. 2019, Buffalo et al. 2011), and it would indeed be interesting to determine whether our observed WM-driven changes follow similar patterns. We may be able to analyze a subset of the data with current source density analysis to look for layer-specific effects in the future, but are not able to provide any information at this time.

(3) The authors suggest that a change in the exact frequency of oscillation underlies the increase in firing rate for different stimulus features. However, the shift in frequency is prominent for contrast but not for orientation, something that raises questions about the general applicability of this observation for different visual features.

While the shift in peak frequency across contrasts is more prominent than that across orientations (Fig. S3A-B), the relationship between orientation and peak frequency is also significant (one-way ANOVA for peak frequency across contrasts, F_Contrast=10.72, p<10^-4; or across orientations, F_Orientation=3, p=0.030; stats have been added to Fig. S3 caption). This finding also aligns with previous studies, which reported slight peak frequency shifts (~1–2 Hz) in the context of attention (Fries, 2015). To address the question of whether the frequency-firing rate correlation generalizes to orientation-driven changes, we now examine the relationship between peak frequency and firing rate separately for each contrast level (Fig. S14). The average normalized response as a function of peak frequency, pooled across subsamples of trials from each of 145 V4 neurons (100 subsamples/neuron), IN vs. OUT conditions, shows a significant correlation during the delay period for each contrast (contrast low (F_Condition=0.03, p=0.867; F_Frequency=141.86, p<10^-18; F_Interaction=10.70, p=0.002, ANCOVA), contrast middle (F_Condition=7.18, p=0.009; F_Frequency=96.76, p<10^-14; F_Interaction=0.13, p=0.716, ANCOVA), contrast high (F_Condition=12.51, p=0.001; F_Frequency=333.74, p<10^-29; F_Interaction=7.91, p=0.006, ANCOVA).

(4) One of the major points of the study is the primacy of the phase code over the rate code during the delay period. Specifically, here it is shown that information about the visual features of a stimulus carried by the rate code is similar for relevant and irrelevant locations during the delay period. This contrasts with what several studies have shown for attention in which case information carried in firing rates about stimuli in the attended location is enhanced relative to that for stimuli in the unattended location. If we are to understand how top-down signals work in cognitive functions it is inevitable to compare working memory with attention. The possible source of this difference is not clear and is not discussed. The reader is left wondering whether perhaps a different measure or analysis (e.g. a percent explained variance analysis) might reveal differences during the delay period for different visual features across the two spatial conditions.

We have added discussion regarding the relationship of these results to previous findings during attention in the discussion section (lines 315-333).

The use of the memory-guided saccade task has certain disadvantages in the context of this study. Although delay activity is interpreted as memory activity by the authors, it is in principle possible that it reflects preparation for the upcoming saccade, spatial attention (particularly since there is a stimulus in the RF), etc. This could potentially change the conclusion and perspective.

We have added a new discussion paragraph addressing the relationship to attention and motor planning (lines 315-333). We have also moderated the language used to describe our conclusions throughout the manuscript in light of this ambiguity.

For the position outside the V4 RF, there is a decrease in both beta oscillations and the clustering of spikes at a specific phase. It is therefore possible that the decrease in information about the stimuli features is a byproduct of the decrease in beta power and phase locking. Decreased oscillatory activity and phase locking can result in less reliable estimates of phase, which could decrease the mutual information estimates.

Looking at the SNR as a ratio of power in the beta band to all other bands, there is no significant drop in SNR between conditions (SNRIN = 4.074+-984, SNROUT = 4.333+-0.834 OUT, p=0.341, Wilcoxon signed-rank). Therefore, we do not think that the change in phase coding is merely a result of less reliable phase estimates.

The authors propose that coherent oscillations could be the mechanism through which the prefrontal cortex influences beta activity in V4. I assume they mean coherent oscillations between the prefrontal cortex and V4. Given that they do have simultaneous recordings from the two areas they could test this hypothesis on their own data, however, they do not provide any results on that.

This paper only includes inactivation data. We are working on analyzing the simultaneous recording data for a future publication.

The authors make a strong point about the relevance of changes in the oscillation frequency and how this may result in an increase in firing rate although it could also be the reverse - an increase in firing rate leading to an increase in the frequency peak. It is not clear at all how these changes in frequency could come about. A more nuanced discussion based on both experimental and modeling data is necessary to appreciate the source and role (if any) of this observation.

As the reviewer notes, it is difficult to determine whether the frequency changes drive the rate changes, vice versa, or whether both are generated in parallel by a common source. We have adjusted our language to reflect this (lines 291-293). Future modeling work may be able to shed more light on the causal relationships between various neural signatures.

Reviewer #3 (Public review):

Summary:

In this report, the authors test the necessity of prefrontal cortex (specifically, FEF) activity in driving changes in oscillatory power, spike rate, and spike timing of extrastriate visual cortex neurons during a visual-spatial working memory (WM) task. The authors recorded LFP and spikes in V4 while macaques remembered a single spatial location over a delay period during which task-irrelevant background gratings were displayed on the screen with varying orientation and contrast. V4 oscillations (in the beta range) scaled with WM maintenance, and the information encoded by spike timing relative to beta band LFP about the task-irrelevant background orientation depended on remembered location. They also compared recorded signals in V4 with and without muscimol inactivation of FEF, demonstrating the importance of FEF input for WM-induced changes in oscillatory amplitude, phase coding, and information encoded about background orientations. Finally, they built a network model that can account for some of these results. Together, these results show that FEF provides meaningful input to the visual cortex that is used to alter neural activity and that these signals can impact information coding of task-irrelevant information during a WM delay.

Strengths:

(1) Elegant and robust experiment that allows for clear tests for the necessity of FEF activity in WM-induced changes in V4 activity.

(2) Comprehensive and broad analyses of interactions between LFP and spike timing provide compelling evidence for FEF-modulated phase coding of task-irrelevant stimuli at remembered location.

(3) Convincing modeling efforts.

Weaknesses:

(1) 0% contrast background data (standard memory-guided saccade task) are not reported in the manuscript. While these data cannot be used to consider information content of spike rate/time about task-irrelevant background stimuli, this condition is still informative as a 'baseline' (and a more typical example of a WM task).

We have added a new supplementary figure to show the effect of WM on V4 LFP power and SPL in 0% contrast trials (Fig. S6). These results (increases in beta LFP power and SPL when remembering the V4 RF location) match our previous report for the effect of spatial WM on LFP power and SPL within extrastriate area MT (Bahmani et al. 2018).

(2) Throughout the manuscript, the primary measurements of neural coding pertain to task-irrelevant stimuli (the orientation/contrast of the background, which is unrelated to the animal's task to remember a spatial location). The remembered location impacts the coding of these stimulus variables, but it's unclear how this relates to WM representations themselves.

Indeed, here we have focused on how maintaining spatial WM impacts visual processing of incoming sensory information, rather than on how the spatial WM signal itself is represented and maintained. Behaviorally, this impact on visual signals could be related to the effects of the content of WM on perception and reaction times (e.g., Soto et al. 2008, Awh et al. 1998, Teng et al. 2019), but no such link to behavior is shown in our data.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

As mentioned above, the two points I raised in the public review merit a bit of development in the Discussion. In addition, the authors should revise some of their conclusions.

For instance (L217):

"The finding that WM mainly modulates phase coded information within extrastriate areas fundamentally shifts our understanding of how the top-down influence of prefrontal cortex shapes the neural representation, suggesting that inducing oscillations is the main way WM recruits sensory areas."

In my opinion, this one is over-the-top on various counts.

Here is another exaggerated instance (L298):

"...leading us to conclude that representations based on the average firing rate of neurons are not the primary way that top-down signals enhance sensory processing."

Again, as noted above, the problem is that one could make the case that the top-down signals are, in fact, highly effective, since they are completely quashing any distracter-related modulation in firing rate across RFs. There is only so much that one can conclude from responses to stimuli that are task-irrelevant, uniform across space, and constant over the course of a trial.

I think even the title goes too far. What the work shows, by all accounts, is that the sustained activity in FEF has a definitive impact on V4 *even* with respect to a sustained, irrelevant background stimulus. The result is very robust in this sense. However, this is quite different from saying that the *primary* means of functional control for FEF is via phase coding. Establishing that would require ruling out other forms of control (i.e., rate coding) in all or a wide range of experimental conditions. That is far from the restricted set of conditions tested here and is also at variance with many other experiments demonstrating effects of attention or even FEF microstimulation on V4 firing activity.

To reiterate, in my opinion, the work is carefully executed and the data are interesting and largely unambiguous. I simply take issue with what can be reliably concluded, and how the results fit with the rest of the literature. Revisions along these lines would improve the readability of the paper considerably.

We have edited the title (removing the word ‘primarily’) and key sentences throughout to tone down the conclusions, generally to state that the importance of a phase code in WM modulations is *possible* given the observed results, rather than certain (see abstract lines 26-27, introduction lines 59-62, conclusion lines 310-311).

Reviewer #3 (Recommendations for the authors):

(1) My primary comment that came up multiple times as I read the manuscript (and which is summarized above) is that I wasn't ever sure why the authors are focused on analyzing neural coding of task-irrelevant sensory information during a WM task as a function of WM contents (remembered location). Most studies of neural codes supporting WM often focus on coding the remembered information - not other information. Conceptually, it seems that the brain would want to suppress - or at least not enhance - representations of task-irrelevant information when performing a demanding task, especially when there is no search requirement, and when there is no feature correspondence between the remembered and viewed stimuli. (i.e., the interaction between WM and visual input is more obvious for visual search for a remembered target). Why, in theory, would a visual region need to improve its coding of non-remembered information as a function of WM? This isn't meant to detract from the results, which are indeed very interesting and I think quite informative. The authors are correct that this is certainly relevant for sensory recruitment models of WM - there's clear evidence for a role of feedback from PFC to extrastriate cortex - but what role, specifically, each region plays in this task is critical to describe clearly, especially given the task-irrelevance of the input. Put another way: what if the animal was remembering an oriented grating? In that case, MI between spike-based measures and orientation would be directly relevant to questions of neural WM representations, as the remembered feature is itself being modeled. But here, the focus seems to be on incidental coding.

Indeed, here we have focused on how maintaining spatial WM impacts visual processing of incoming sensory information, rather than on how the spatial WM signal itself is represented and maintained. Behaviorally, this impact on visual signals could be related to the effects of the content of WM on perception and reaction times (e.g., Soto et al. 2008, Awh et al. 1998, Teng et al. 2019), but no such link to behavior is shown in our data.

Whether similar phase coding is also used to represent the content of object WM (for example, if the animal was remembering an oriented grating), or whether phase coding is only observed for WM’s modulation of the representation of incoming sensory signals, is an important question to be addressed in future work.

(2) Related to the above, the phrasing of the second sentence of the Discussion (lines 291-292) is ambiguous - do the authors mean that the FEF sends signals that carry WM content to V4, or that FEF sends projections to V4, and V4 has the WM content? As presently phrased, either of these are reasonable interpretations, yet they're directly opposing one another (the next sentence clarifies, but I imagine the authors want to minimize any confusion).

We have edited this sentence to read, “Within prefrontal areas, FEF sends direct projections to extrastriate visual areas, and activity in these projections reflects the content of WM.”

(3) I'm curious about how the authors consider the spatial WM task here different from a cued spatial attention task. Indeed, both require sustained use of a location for further task performance. The section of the Discussion addressing similar results with attention (lines 307-311) presently just summarizes the similarities of results but doesn't offer a theoretical perspective for how/why these different types of tasks would be expected to show similar neural mechanisms.

We have added discussion regarding the relationship of these results to previous findings during attention in the discussion section (lines 315-333).

(4) As far as I can tell, there is no consideration of behavioral performance on the memory-guided saccade task (RT, precision) across the different stimulus background conditions. This should be reported for completeness, and to determine whether there is an impact of the (likely) task-irrelevant background on task performance. This analysis should also be reported for Figure 3's results characterizing how FEF inactivation disrupts behavior (if background conditions were varied, see point 7 below).

We have added the effect of inactivation on behavioral RT and % correct across the different stimulus background conditions (Fig. S8). Background contrast and orientation did not impact either RT or % correct.

(5) Results from Figure 2 (especially Figures 2A-B) concerning phase-locked spiking in V4 should be shown for 0%-contrast trials as well, as these trials better align with 'typical' WM tasks.

We have added a new supplementary figure to show the effect of WM on V4 LFP power and SPL in 0% contrast trials (Fig. S6). These results (increases in beta LFP power and SPL) match our previous report for the effect of spatial WM on LFP power and SPL within extrastriate area MT (Bahmani et al. 2018).

(6) The magnitude of SPL difference in aggregate (Figure 2B) is much, much smaller than that of the example site shown (Figure 2A), such that Figure 2A's neuron doesn't appear to be visible on Figure 2B's scatterplot. Perhaps a more representative sample could be shown? Or, the full range of x/y axes in Figure 2B could be plotted to illustrate the full distribution.

We have updated Fig. 2A with a more representative sample neuron.

(7) I'm a bit confused about the FEF inactivation experiments. In the Methods (lines 512-513), the authors mention there was no background stimulus presented during the inactivation experiment, and instead, a typical 8-location MGS task was employed. However, in the results on pg 8 (Lines 201-214), and Figure 3G, the authors quantify a phase code MI. The previous phase code MI analysis was looking at MI between each spike's phase and the background stimulus - but if there's no background, what's used to compute phase code MI? Perhaps what they meant to write was that, in addition to the primary task with a manipulation of background properties, an 8-location MGS task was additionally employed.

The reviewer is correct that both tasks were used after inactivation (the 8-location task to assess the spread of the behavioral effect of inactivation, and the MGS-background task for measuring MI). We have edited the methods text to clarify.

(8) How is % Correct defined for the MGS task? (what is the error threshold? Especially for the results described in lines 192-193).

The % correct is defined as correct completed trials divided by the total number of trials; the target window was a circle with radius of 2 or 4 dva (depending on cue eccentricity). These details have been added to the Methods.

(9) The paragraph from lines 183-200 describes a number of behavioral results concerning "scatter" and "RT" - the RT shown seems extremely high, and perhaps is normalized. Details of this normalization should be included in the Methods. The "scatter" is listed as dva, but it's not clear how scatter is quantified (std dev of endpoint distribution? Mean absolute error), nor how target eccentricity is incorporated (as scatter is likely higher for greater target eccentricity).

We have renamed ‘scatter’ to ‘saccade error’ in the text to match the figure, and now provide details in the Methods section. Both RT and saccade error are normalized for each session, details are now provided in the Methods. Since error was normalized for each session before performing population statistics, no other adjustment for eccentricity was made.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1:

(1) Line numbers are missing.

Added

(2) VR classroom. Was this a completely custom design based on Unity, or was this developed on top of some pre-existing code? Many aspects of the VR classroom scenario are only introduced (e.g., how was the lip-speech synchronisation done exactly?). Additional detail is required. Also, is or will the experiment code be shared publicly with appropriate documentation? It would also be useful to share brief example video-clips.

We have added details about the VR classroom programming to the methods section (p. 6-7), and we have now included a video-example as supplementary material.

“Development and programming of the VR classroom were done primarily in-house, using assets (avatars and environment) were sourced from pre-existing databases. The classroom environment was adapted from assets provided by Tirgames on TurboSquid (https://www.turbosquid.com/Search/Artists/Tirgames) and modified to meet the experimental needs. The avatars and their basic animations were sourced from the Mixamo library, which at the time of development supported legacy avatars with facial blendshapes (this functionality is no longer available in current versions of Mixamo). A brief video example of the VR classroom is available at: https://osf.io/rf6t8.

“To achieve realistic lip-speech synchronization, the teacher’s lip movements were controlled by the temporal envelope of the speech, adjusting both timing and mouth size dynamically. His body motions were animated using natural talking gestures.”

While we do intent to make the dataset publicly available for other researchers, at this point we are not making the code for the VR classroom public. However, we are happy to share it on an individual-basis with other researchers who might find it useful for their own research in the future.

(3) "normalized to the same loudness level using the software Audacity". Please specify the Audacity function and parameters.

We have added these details (p.7)

“All sound-events were normalized to the same loudness level using the Normalize function in the audio-editing software Audacity (theaudacityteam.org, ver 3.4), with the peak amplitude parameter set to -5 dB, and trimmed to a duration of 300 milliseconds.“

(4) Did the authors check if the participants were already familiar with some of the content in the mini-lectures?

This is a good point. Since the mini-lectures spanned many different topics, we did not pre-screen participants for familiarity with the topics, and it is possible that some of the participants had some pre-existing knowledge.

In hindsight, it would have been good to have added some reflective questions regarding participants prior knowledge as well as other questions such as level of interest in the topic and/or how well they understood the content. These are elements that we hope to include in future versions of the VR classroom.

(5) "Independent Component Analysis (ICA) was then used to further remove components associated with horizontal or vertical eye movements and heartbeats". Please specify how this selection was carried out.

Selection of ICA components was done manually based on visual inspection of their time-course patterns and topographical distributions, to identify components characteristic of blinks, horizontal eye-movements and heartbeats). Examples of these distinct components are provided in Author response image 1 below. These is now specified in the methods section.

Author response image 1.

(6) "EEG data was further bandpass filtered between 0.8 and 20 Hz". If I understand correctly, the data was filtered a second time. If that's the case, please do not do that, as that will introduce additional and unnecessary filtering artifacts. Instead, the authors should replace the original filter with this one (so, filtering the data only once). Please see de Cheveigne and Nelkn, Neuron, 2019 for an explanation. Also, please provide an explanation of the rationale for further restricting the cut-off bands in the methods section. Finally, further details on the filters should be included (filter type and order, for example).

Yes, the data was indeed filtered twice. The first filter is done as part of the preprocessing procedure, in order to remove extremely high- and low- frequency noise but retain most activity within the range of “neural” activity. This broad range is mostly important for the ICA procedure, so as to adequately separate between ocular and neural contribution to the recorded signal.

However, since both the speech tracking responses and ERPs are typically less broadband and are comprised mostly of lower frequencies (e.g., those that make up the speech-envelope), a second narrower filter was applied to improve TRF model-fit and make ERPs more interpretable.

In both cases we used a fourth order zero-phase Butterworth IIR filter with 1-seconds of padding, as implemented in the Fieldtrip toolbox. We have added these details to the manuscript.

(7) "(~ 5 minutes of data in total), which is insufficient for deriving reliable TRFs". That is a bit pessimistic and vague. What does "reliable" mean? I would tend to agree when talking about individual subject TRFs, which 5 min per participant can be enough at the group level. Also, this depends on the specific speech material. If the features are univariate or multivariate. Etc. Please narrow down and clarify this statement.

We determined that the data in the Quiet condition (~5 min) was insufficient for performing reliable TRF analysis, by assessing whether its predictive-power was significantly better than chance. As shown in Author response image 2 below, the predictive power achieved using this data was not higher than values obtained in permuted data (p = 0.43). Therefore, we did not feel that it was appropriate to include TRF analysis of the Quiet condition in this manuscript. We have now clarified this in the manuscript (p. 10)

Author response image 2.

(8) "Based on previous research in by our group (Kaufman & Zion Golumbic 2023), we chose to use a constant regularization ridge parameter (λ= 100) for all participants and conditions". This is an insufficient explanation. I understand that there is a previous paper involved. However, such an unconventional choice that goes against the original definition and typical use of these methods should be clearly reported in this manuscript.

We apologize for not clarifying this point sufficiently, and have added an explanation of this methodological choice (p.11):

“The mTRF toolbox uses a ridge-regression approach for L2 regularization of the model to ensure better generalization to new data. We tested a range of ridge parameter values (λ's) and used a leave-one-out cross-validation procedure to assess the model’s predictive power, whereby in each iteration, all but one trials are used to train the model, and it is then applied to the left-out trial. The predictive power of the model (for each λ) is estimated as the Pearson’s correlation between the predicted neural responses and the actual neural responses, separately for each electrode, averages across all iterations. We report results of the model with the λ the yielded the highest predictive power at the group-level (rather than selecting a different λ for each participant which can lead to incomparable TRF models across participants; see discussion in Kaufman & Zion Golumbic 2023).”

Assuming that the explanation will be sufficiently convincing, which is not a trivial case to make, the next issue that I will bring up is that the lambda value depends on the magnitude of input and output vectors. While the input features are normalised, I don't see that described for the EEG signals. So I assume they are not normalized. In that case, the lambda would have at least to be adapted between subjects to account for their different magnitude.

We apologize for omitting this detail – yes, the EEG signals were normalized prior to conducting the TRF analysis. We have updated the methods section to explicitly state this pre-processing step (p.10).

Another clarification, is that value (i.e., 100) would not be comparable either across subjects or across studies. But maybe the authors have a simple explanation for that choice? (note that this point is very important as this could lead others to use TRF methods in an inappropriate way - but I understand that the authors might have specific reasons to do so here). Note that, if the issue is finding a reliable lambda per subject, a more reasonable choice would be to use a fixed lambda selected on a generic (i.e., group-level) model. However selecting an arbitrary lambda could be problematic (e.g., would the results replicate with another lambda; and similarly, what if a different EEG system was used, with different overall magnitude, hence the different impact of the regularisation).

We fully agree that selecting an arbitrary lambda is problematic (esp across studies). As clarified above, the group-level lambda chosen here for the encoding more was data-driven, optimized based on group-level predictive power.

(9) "L2 regularization of the model, to reduce its complexity". Could the authors explain what "reduce its complexity" refers to?

Our intension here was to state that the L2 regularization constrains the model’s weights so that it can better generalize between to left-out data. However, for clarity we have now removed this statement.

(10) The same lambda value was used for the decoding model. From personal experience, that is very unlikely to be the optimal selection. Decoding models typically require a different (usually larger) lambda than forward models, which can be due to different reasons (different SNR of "input" of the model and, crucially, very different dimensionality).

We agree with the reviewer that treatment of regularization parameters might not be identical for encoding and decoding models. Our initial search of lambda parameters was limited to λ= 0.01 - 100, with λ= 100 showing the best reconstruction correlations. However, following the reviewer’s suggestion we have now broadened the range and found that, in fact reconstruction correlations are further improved and the best lambda is λ= 1000 (see Author response image 3 below, left panel). Importantly, the difference in decoding reconstruction correlations between the groups is maintained regardless of the choice of lambda (although the effect-size varies; see Author response image 3, right panel). We have now updated the text to reflect results of the model with λ= 1000.

Author response image 3.

(11) Skin conductance analysis. Additional details are required. For example, how was the linear interpolation done exactly? The raw data was downsampled, sure. But was an anti-aliasing filter applied? What filter exactly? What implementation for the CDA was run exactly?

We have added the following details to the methods section (p. 14):

“The Skin Conductance (SC) signal was analyzed using the Ledalab MATLAB toolbox (version 3.4.9; Benedek and Kaernbach, 2010; http://www.ledalab.de/) and custom-written scripts. The raw data was downsampled to 16Hz using FieldTrip's ft_resampledata function, which applies a built-in anti-aliasing low-pass filter to prevent aliasing artifacts. Data were inspected manually for any noticeable artifacts (large ‘jumps’), and if present were corrected using linear interpolation in Ledalab. A continuous decomposition analysis (CDA) was employed to separate the tonic and phasic SC responses for each participant. The CDA was conducted using the 'sdeco' mode (signal decomposition), which iteratively optimizes the separation of tonic and phasic components using the default regularization settings.”

(12) "N1- and P2 peaks of the speech tracking response". Have the authors considered using the N1-P2 complex rather than the two peaks separately? Just a thought.

This is an interesting suggestion, and we know that this has been used sometimes in more traditional ERP literature. In this case, since neither peak was modulated across groups, we did not think this would yield different results. However, it is a good point to keep in mind for future work.

(13) Figure 4B. The ticks are missing. From what I can see (but it's hard without the ticks), the N1 seems later than in other speech-EEG tracking experiments (where is closer to ~80ms). Could the authors comment on that? Or maybe this looks similar to some of the authors' previous work?

We apologize for this and have added ticks to the figure.

In terms of time-course, a N1 peak at around 100ms is compatible with many of our previous studies, as well as those from other groups.

(14) Figure 4C. Strange thin vertical grey bar to remove.

Fixed.

(15) Figure 4B: What about the topographies for the TRF weights? Could the authors show that for the main components?

Yes. The topographies of the main TRF components are similar to those of the predictive power and are compatible with auditory responses. We have added them to Figure 4B.

(16) Figure 4B: I just noticed that this is a grand average TRF. That is ok (but not ideal) only because the referencing is to the mastoids. The more appropriate way of doing this is to look at the GFP, instead, which estimates the presence of dipoles. And then look at topographies of the components. Averaging across channels makes the plotted TRF weaker and noisier. I suggest adding the GFP to the plot. Also, the colour scale in Figure 4A is deceiving, as blue is usually used for +/- in plots of the weights. While that is a heatmap, where using a single colour or even yellow to red would be less deceiving at first look. Only cosmetics, indeed. The result is interesting nonetheless!

We apologize for this, and agree with the reviewer that it is better not to average across EEG channels. In the revised Figure, we now show the TRFs based on the average of electrodes FC1, FC2, and FCz, which exhibited the strongest activity for the two main components.

Following the previous comment, we have also included the topographical representation of the TRF main components, to give readers a whole-head perspective of the TRF.

We have also fixed the color-scales.

We are glad that the reviewer finds this result interesting!

(17) Figure 4C. This looks like a missed opportunity. That metric shows a significant difference overall. But is that underpinned but a generally lower envelope reconstruction correlation, or by a larger deviation in those correlations (so, that metric is as for the control in some moments, but it drops more frequently due to distractibility)?

We understand the reviewer’s point here, and ideally would like to be able to address this in a more fine-grained analysis, for example on a trial-by-trial basis. However, the design of the current experiment was not optimized for this, in terms of (for example) number of trials, the distribution of sound-events and behavioral outcomes. We hope to be able to address this issue in our future research.

(18) I am not a fan of the term "accuracy" for indicating envelope reconstruction correlations. Accuracy is a term typically associated with classification. Regression models are typically measured through errors, loss, and sometimes correlations. 'Accuracy' is inaccurate (no joke intended).

We accept this comment and now used the term “reconstruction correlation”.

(19) Discussion. "The most robust finding in". I suggest using more precise terminology. For example, "largest effect-size".

We agree and have changed the terminology (p. 31).

(20) "individuals who exhibited higher alpha-power [...]". I probably missed this. But could the authors clarify this result? From what I can see, alpha did not show an effect on the group. Is this referring to Table 2? Could the authors elaborate on that? How does that reconcile with the non-significant effect of the group? In that same sentence, do you mean "and were more likely"? If that's the case, and they were more likely to report attentional difficulties, how is it that there is no group-effect when studying alpha?

Yes, this sentence refers to the linear regression models described in Figure 10 and in Table 2. As the reviewer correctly points out, this is one place where there is a discrepancy between the results of the between-group analysis (ADHD diagnosis yes/no) and the regression analysis, which treats ADHD symptoms as a continuum, across both groups. The same is true for the gaze-shift data, which also did not show a significance between-group effect but was identified in the regression analysis as contributing to explaining the variance in ADHD symptoms.

We discuss this point on pages 30-31, noting that “although the two groups are clearly separable from each other, they are far from uniform in the severity of symptoms experienced”, which motivated the inclusion of both analyses in this paper.

At the bottom of p. 31 we specifically address the similarities and differences between the between-group and regression-based results. In our opinion, this pattern emphasizes that while neither approach is ‘conclusive’, looking at the data through both lenses contributes to an overall better understanding of the contributing factors, as well as highlighting that “no single neurophysiological measure alone is sufficient for explaining differences between the individuals – whether through the lens of clinical diagnosis or through report of symptoms”.

(21) "why in the latter case the neural speech-decoding accuracy did not contribute to explaining ASRS scores [...]". My previous point 1 on separating overall envelope decoding from its deviation could help there. The envelope decoding correlation might go up and down due to SNR, while you might be more interested in the dynamics over time (i.e., looking at the reconstructions over time).

Again, we appreciate this comment, but believe that this additional analysis is outside the scope of what would be reliably-feasible with the current dataset. However, since the data will be made publicly available, perhaps other researchers will have better ideas as to how to do this.

(22) Data and code sharing should be discussed. Also, specific links/names and version numbers should be included for the various libraries used.

We are currently working on organizing the data to make it publicly available on the Open Science Project.

We have updated links and version numbers for the various toolboxes/software used, throughout the manuscript.

Reviewer #2:

(1) While it is highly appreciated to study selective attention in a naturalistic context, the readers would expect to see whether there are any potential similarities or differences in the cognitive and neural mechanisms between contexts. Whether the classic findings about selective attention would be challenged, rebutted, or confirmed? Whether we should expect any novel findings in such a novel context? Moreover, there are some studies on selective attention in the naturalistic context though not in the classroom, it would be better to formulate specific hypotheses based on previous findings both in the strictly controlled and naturalistic contexts.

Yes, we fully agree that comparing results across different contexts would be extremely beneficial and important.

The current paper serves as an important proof-first-concept demonstrating the plausibility and scientific potential of using combined EEG-VR-eyetracking to study neurophysiological aspects of attention and distractibility, but is also the basis for formulating specific hypothesis that will be tested in follow-up studies.

If fact, a follow up study is already ongoing in our lab, where we are looking into this point, by testing users in different VR scenarios (e.g., classroom, café, office etc.), and assessing whether similar neurophysiological patterns are observed across contexts and to what degree they are replicable within and across individuals. We hope to share these data with the community in the near future.

(2) Previous studies suggest handedness and hemispheric dominance might impact the processing of information in each hemisphere. Whether these issues have been taken into consideration and appropriately addressed?

This is an interesting point. In this study we did not specifically control for handedness/hemispheric dominance, since most of the neurophysiological measured used here are sensory/auditory in their nature, and therefore potentially invariant to handedness. Moreover, the EEG signal is typically not very sensitive to hemispheric dominance, at least for the measures used here. However, this might be something to consider more explicitly in future studies. Nonetheless, we have added handedness information to the Methods section (p. 5): “46 right-handed, 3 left-handed”

(3) It would be interesting to know how students felt about the Virtual Classroom context, whether it is indeed close to the real classroom or to some extent different.

Yes, we agree. Obviously, the VR classroom differs in many ways from a real classroom, in terms of the perceptual experience, social aspects and interactive possibilities. We did ask participants about their VR experience after the experiment, and most reported feeling highly immersed in the VR environment and engaged in the task, with a strong sense of presence in the virtual-classroom.

We note that, in parallel to the VR studies in our lab, we are also conducting experiments in real classrooms, and we hope that the cross-study comparison will be able to shed more light on these similarities/differences.

(4) One intriguing issue is whether neural tracking of the teacher's speech can index students' attention, as the tracking of speech may be relevant to various factors such as sound processing without semantic access.

Another excellent point. While separating the ‘acoustic’ and ‘semantic’ contributions to the speech tracking response is non-trivial, we are currently working on methodological approaches to do this (again, in future studies) following, for example, the hierarchical TRF approach used by Brodbeck et al. and others.

(5) There are many results associated with various metrics, and many results did not show a significant difference between the ADHD group and the control group. It is difficult to find the crucial information that supports the conclusion. I suggest the authors reorganize the results section and report the significant results first, and to which comparison(s) the readers should pay attention.

We apologize if the organization of the results section was difficult to follow. This is indeed a challenge when collecting so many different neurophysiological metrics.

To facilitate this, we have now added a paragraph at the beginning of the result section, clarifying its structure (p.16):

The current dataset is extremely rich, consisting of many different behavioral, neural and physiological responses. In reporting these results, we have separated between metrics that are associated with paying attention to the teacher (behavioral performance, neural tracking of the teacher’s speech, and looking at the teacher), those capturing responses to the irrelevant sound-events (ERPs and event-related changes in SC and gaze); as well as more global neurophysiological measures that may be associated with the listeners’ overall ‘state’ of attention or arousal (alpha- and beta-power and tonic SC).

Moreover, within each section we have ordered the analysis such that the ones with significant effects are first. We hope that this contributes to the clarity of the results section.

(6) The difference between artificial and non-verbal humans should be introduced earlier in the introduction and let the readers know what should be expected and why.

We have added this to the Introduction (p. 4)

(7) It would be better to discuss the results against a theoretical background rather than majorly focusing on technical aspects.

We appreciate this comment. In our opinion, the discussion does contain a substantial theoretical component, both regarding theories of attention and attention-deficits, and also regarding their potential neural correlates. However, we agree that there is always room for more in depth discussion.

Reviewer #3:

Major:

(1) While the study introduced a well-designed experiment with comprehensive physiological measures and thorough analyses, the key insights derived from the experiment are unclear. For example, does the high ecological validity provide a more sensitive biomarker or a new physiological measure of attention deficit compared to previous studies? Or does the study shed light on new mechanisms of attention deficit, such as the simultaneous presence of inattention and distraction (as mentioned in the Conclusion)? The authors should clearly articulate their contributions.

Thanks for this comment.

We would not say that this paper is able to provide a ‘more sensitive biomarker’ or a ‘new physiological measure of attention’ – in order to make those type of grand statements we would need to have much more converging evidence from multiple studies and using both replication and generalization approaches.

Rather, from our perspective, the key contribution of this work is in broadening the scope of research regarding the neurophysiological mechanisms involved in attention and distraction.

Specifically, this work:

(1) Offers a significant methodological advancement of the field – demonstrating the plausibility and scientific potential of using combined EEG-VR-eyetracking to study neurophysiological aspects of attention and distractibility in contexts that ‘mimic’ real-life situations (rather than highly controlled computerized tasks).

(2) Provides a solid basis formulating specific mechanistic hypothesis regarding the neurophysiological metrics associated with attention and distraction, the interplay between them, and their potential relation to ADHD-symptoms. Rather than being an end-point, we see these results as a start-point for future studies that emphasize ecological validity and generalizability across contexts, that will hopefully lead to improved mechanisms understanding and potential biomarkers of real-life attentional capabilities (see also response to Rev #2 comment #1 above).

(3) Highlights differences and similarities between the current results and those obtained in traditional ‘highly controlled’ studies of attention (e.g., in the way ERPs to sound-events differ between ADHD and controls; variability in gaze and alpha-power; and more broadly about whether ADHD symptoms do or don’t map onto specific neurophysiological metrics). Again, we do not claim to give a definitive ’answer’ to these issues, but rather to provide a new type of data that can expands the conversation and address the ecological validity gap in attention research.

(2) Based on the multivariate analyses, ASRS scores correlate better with the physiological measures rather than the binary deficit category. It may be worthwhile to report the correlation between physiological measures and ASRS scores for the univariate analyses. Additionally, the correlation between physiological measures and behavioral accuracy might also be interesting.

Thanks for this. The beta-values reported for the regression analysis reflect the correlations between the different physiological measures and the ASRS scores (p. 30). From a statistical perspective, it is better to report these values rather than the univariate correlation-coefficients, since these represent the ‘unique’ relationship with each factor, after controlling for all the others.

The univariate correlations between the physiological measures themselves, as well as with behavioral accuracy, are reported in Figure 10

(3) For the TRF and decoding analysis, the authors used a constant regularization parameter per a previous study. However, the optimal regularization parameter is data-dependent and may differ between encoding and decoding analyses. Furthermore, the authors did not conduct TRF analysis for the quiet condition due to the limited ~5 minutes of data. However, such a data duration is generally sufficient to derive a stable TRF with significant predictive power (Mesik and Wojtczak, 2023).

The reviewer raises two important points, also raised by Rev #1 (see above).

Regarding the choice of regularization parameters, we have now clarified that although we used a common lambda value for all participants, it was selected in a data-driven manner, so as to achieve an optimal predictive power at the group-level.

See revised methods section:

“The mTRF toolbox uses a ridge-regression approach for L2 regularization of the model to ensure better generalization to new data. We tested a range of ridge parameter values (λ's) and used a leave-one-out cross-validation procedure to assess the model’s predictive power, whereby in each iteration, all but one trials are used to train the model, and it is then applied to the left-out trial. The predictive power of the model (for each λ) is estimated as the Pearson’s correlation between the predicted neural responses and the actual neural responses, separately for each electrode, averages across all iterations. We report results of the model with the λ the yielded the highest predictive power at the group-level (rather than selecting a different λ for each participant which can lead to incomparable TRF models across participants; see discussion in Kaufman & Zion Golumbic 2023).”

Regarding whether data was sufficient in the Quiet condition for performing TRF analysis – we are aware of the important work by Mesik & Wojtczak, and had initially used this estimate when designing our study. However, when assessing the predictive-power of the TRF model trained on data from the Quiet condition, we found that it was not significantly better than chance (see Author response image 2, ‘real’ predictive power vs. permuted data). Therefore, we ultimately did not feel that it was appropriate to include TRF analysis of the Quiet condition in this manuscript. We have now clarified this in the manuscript (p. 10)

(4) As shown in Figure 4, for ADHD participants, decoding accuracy appears to be lower than the predictive power of TRF. This result is surprising because more data (i.e., data from all electrodes) is used in the decoding analysis.

This is an interesting point – however, in our experience it is not necessarily the case that decoding accuracy (i.e., reconstruction correlation with the stimulus) is higher than encoding predictive-power. While both metrics use Pearson’s’ correlations, they quantify the similarity between two different types of signals (the EEG and the speech-envelope). Although the decoding procedure does use data from all electrodes, many of them don’t actually contain meaningful information regarding the stimulus, and thus could just as well hinder the overall performance of the decoding.

(5) Beyond the current analyses, the authors may consider analyzing inter-subject correlation, especially for the gaze signal analysis. Given that the area of interest during the lesson changes dynamically, the teacher might not always be the focal point. Therefore, the correlation of gaze locations between subjects might be better than the percentage of gaze duration on the teacher.

Thanks for this suggestion. We have tried to look into this, however working with eye-gaze in a 3-D space is extremely complex and we are not able to calculate reliable correlations between participants.

(6) Some preprocessing steps relied on visual and subjective inspection. For instance, " Visual inspection was performed to identify and remove gross artifacts (excluding eye movements) " (P9); " The raw data was downsampled to 16Hz and inspected for any noticeable artifacts " (P13). Please consider using objective processes or provide standards for subjective inspections.

We are aware of the possible differences between objective methods of artifact rejection vs. use of manual visual inspection, however we still prefer the manual (subjective) approach. As noted, in this case only very large artifacts were removed, exceeding ~ 4 SD of the amplitude variability, so as to preserve as many full-length trials as possible.

(7) Numerous significance testing methods were employed in the manuscript. While I appreciate the detailed information provided, describing these methods in a separate section within the Methods would be more general and clearer. Additionally, the authors may consider using a linear mixed-effects model, which is more widely adopted in current neuroscience studies and can account for random subject effects.

Indeed, there are many statistical tests in the paper, given the diverse types of neurophysiological data collected here. We actually thought that describing the statistics per method rather than in a separate “general” section would be easier to follow, but we understand that readers might diverge in their preferences.

Regarding the use of mixed-effect models – this is indeed a great approach. However, it requires deriving reliable metrics on a per-trial basis, and while this might be plausible for some of our metrics, the EEG and GSR metrics are less reliable at this level. This is why we ultimately chose to aggregate across trials and use a regular regression model rather than mixed-effects.

(8) Some participant information is missing, such as their academic majors. Given that only two lesson topics were used, the participants' majors may be a relevant factor.

To clarify – the mini-lectures presented here actually covered a large variety of topics, broadly falling within the domains of history, science and social-science and technology. Regarding participants’ academic majors, these were relatively diverse, as can be seen in Author response table 1 and Author response image 4.

Author response table 1.

Author response image 4.

(9) Did the multiple regression model include cross-validation? Please provide details regarding this.

Yes, we used a leave-one-out cross validation procedure. We have now clarified this in the methods section which now reads:

“The mTRF toolbox uses a ridge-regression approach for L2 regularization of the model to ensure better generalization to new data. We tested a range of ridge parameter values (λ's) and used a leave-one-out cross-validation procedure to assess the model’s predictive power, whereby in each iteration, all but one trials are used to train the model, and it is then applied to the left-out trial. The predictive power of the model (for each λ) is estimated as the Pearson’s correlation between the predicted neural responses and the actual neural responses, separately for each electrode, averages across all iterations. We report results of the model with the λ the yielded the highest predictive power at the group-level (rather than selecting a different λ for each participant which can lead to incomparable TRF models across participants; see discussion in Kaufman & Zion Golumbic 2023).”

Minor:

(10) Typographical errors: P5, "forty-nine 49 participants"; P21, "$ref"; P26, "Table X"; P4, please provide the full name for "SC" when first mentioned.

Thanks! corrected

Although not disagreeing with the use of personas to understand user problems, is it not concerning that through their creation we may be including our own personal biases and backgrounds? I found it hard to truly understand and empathize with the scenario, most likely because my own scenario and lifestyle is very different to Ko's. My positionality as a young, healthy college aged person would impact my ability to define issues with this user persona. I think that the designer's background and situation should be heavily considered before the creation of a user persona or scenario in order to minimize the implications of bias.

This is an important statement that stands out to me. It connects to what I learned in INFO380 where we discussed how when companies are making designs and adding features, they have to consider which ones are the most beneficial and desirable. I think that this shows the importance of designers doing research and learning more about the stakeholders involved to help them make these decisions. If research isn't done properly, the company may waste a lot of resources and time. I also think it is important to consider underrepresented demographics that may be users of the product and see how they can possibly be considered in these decision-making processes. This makes me appreciate designers even more as this process is not easy and can be difficult having to make decisions that don't please some users but this might be something they learn as they develop their skills since you can't please everyone.

This part brings up a very interesting point about who is responsible for a bots actions. I think it is who ever used it, the creator might have different intentions for its use, and the bot may get used differently.

Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

Manuscript number: RC-2024-02465

Corresponding author(s): Saravanan, Palani

1. General Statements

We would like to thank the Review Commons Team for handling our manuscript and the Reviewers for their constructive feedback and suggestions. In our revised manuscript, we have addressed and incorporated all the major suggestions of the reviewers, and we have also added new significant data on the role of Tropomyosin in regulation of endocytosis through its control over actin monomer pool maintenance and actin network homeostasis. We believe that with all these additions, our study has significantly gained in quality, strength of conclusions made, and scope for future work.

2. Point-by-point description of the revisions

Reviewer #1

Evidence, reproducibility and clarity

There are 2 Major issues -

Having an -ala-ser- linker between the GFP and tropomyosin mimics acetylation. This is not the case, and more likely the this linker acts as a spacer that allows tropomyosin polymers to form on the actin, and without it there is steric hindrance. A similar result would be seen with a simple flexible uncharged linker. It has been shown in a number of labs that the GFP itself masks the effect of the charge on the amino terminal methionine. This is consistent with NMR, crystallographic and cryo structural studies. Biochemical studies should be presented to demonstrate that the impact of a linker for the conclusions stated to be made, which provide the basis of a major part of this study.

Response: We would like to clarify that all mNG-Tpm constructs used in our study contain a 40 amino-acid (aa) flexible linker between the N-terminal mNG fluorescent protein and the Tpm protein as per our earlier published study (Hatano et al., 2022). During initial optimization, we have also experimented with linker length and the 40aa-linker length works optimally for clear visualization of Tpm onto actin cable structures in budding yeast, fission yeast (both S. pombe and S. japonicus), and mammalian cells (Hatano et al., 2022). These constructs have also been used since in other studies (Wirshing et al., 2023; Wirshing and Goode, 2024) and currently represents the best possible strategy to visualize Tpm isoforms in live cells. In our study, we characterized these proteins for functionality and found that both mNG-Tpm1 and mNG-Tpm2 were functional and can rescue the synthetic lethality observed in Dtpm1Dtpm2 cells. During our study, we observed that mNG-Tpm1 expression from a single-copy integration vector did not restore full length actin cables in Dtpm1 cells (Fig. 1B, 1C). We hypothesized that this could be a result of reduced binding affinity of the tagged tropomyosin due to lack of normal N-terminal acetylation which stabilizes the N-terminus. The 40aa linker is unstructured and may not be able to neutralize the charge on the N-terminal Methionine, thus, we tried to insert -Ala-Ser- dipeptide which has been routinely used in vitro biochemical studies to stabilize the N-terminal helix and impart a similar effect as the N-terminal acetylation (Alioto et al., 2016; Palani et al., 2019; Christensen et al., 2017) by restoring normal binding affinity of Tpm to F-actin (Monteiro et al., 1994; Greenfield et al., 1994). We observed that addition of the -Ala-Ser- dipeptide to mNG-Tpm fusion, indeed, restored full length actin cables when expressed in Dtpm1 cells, performing significantly better in our in vivo experiments (Fig. 1B, 1C). We agree with the reviewer that the -AS- dipeptide addition may not mimic N-terminal acetylation structurally but as per previous studies, it may stabilize the N-terminus of Tpm and allow normal head-to-tail dimer formation (Greenfield et al., 1994; Monteiro et al., 1994; Frye et al., 2010). We have discussed this in our new Discussion section (Lines 350-372). Since, the addition of -AS- dipeptide was referred to as "acetyl-mimic (am)" in a previous study (Alioto et al., 2016), we continued to use the same nomenclature in our study. Now as per your suggestions and to be more accurate, we have renamed "mNG-amTpm" constructs as "mNG-ASTpm" throughout the study to not confuse or claim that -AS- addition mimics acetylation. In any case, we have not seen any other ill effect of -AS- dipeptide introduction in addition to our 40 amino acid linker suggesting that it can also be considered part of the linker. Although, we agree with the reviewer that biochemical characterization of the effect of linker would be important to determine, we strongly believe that it is currently outside the scope of this study and should be taken up for future work with these proteins. Our study has majorly aimed to understand the functionality and utility of these mNG-Tpm fusion proteins for cell biological experiments in vivo, which was not done earlier in any other model system.

My major issue however is making the conclusions stated here, using an amino-terminal fluorescent protein tag that s likely to impact any type of isoform selection at the end of the actin polymer. Carboxyl terminal tagging may have a reduced effect, but modifying the ends of the tropomyosin, which are integral in stabilising end to end interactions with itself on the actin filament, never mind any section systems that may/maynot be present in the cell, is not appropriate.

Response: We agree with the reviewer that N-terminal tagging of tropomyosin may have effects on its function, but these constructs represent the only fluorescently tagged functional tropomyosin constructs available currently while C-terminal fusions are either non-functional (we were unable to construct strains with endogenous Tpm1 gene fused C-terminally to GFP) or do not localize clearly to actin structures (See Figure R1 showing endogenous C-terminally tagged Tpm2-yeGFP that shows almost no localization to actin cables). To our knowledge, our study represents a first effort to understand the question of spatial sorting of Tpm isoforms, Tpm1 and Tpm2, in S. cerevisiae and any future developments with better visualization strategies for Tpm isoforms without compromising native N-terminal modifications and function will help improve our understanding of these proteins in vivo. We have also discussed these possibilities in our new Discussion section (Lines 391-396).

Significance

This paper explores the role of formin in determining the localisation of different tropomyosins to different actin polymers and cellular locations within budding yeast. Previous studies have indicated a role for the actin nucleating proteins in recruiting different forms of tropomyosin within fission yeast. In mammalian cells there is variation in the role of formins in affiecting tropomyosin localisation - variation between cell type. There is also evidence that other actin binding proteins, and tropomyosin abundance play roles in regulating the tropomyosin-actin association according to cell type. Biochemical studies have previously been undertaken using budding yeast and fission yeast that the core actin polymerisation domain of formins do not interact with tropomyosin directly. The significance of this study, given the above, and the concerns raised is not clear to this reviewer.

Response: __Our study explores multiple facets of Tropomyosin (Tpm) biology. The lack of functional tagged Tpm has been a major bottleneck in understanding Tpm isoform diversity and function across eukaryotes. In our study, we characterize the first functional tagged Tpm proteins (Fig. 1, Fig. S1) and use them to answer long-standing questions about localization and spatial sorting of Tpm isoforms in the model organism S. cerevisiae (Fig. 2, Fig. 3, Fig. S2, Fig. S3). We also discover that the dual Tpm isoforms, Tpm1 and Tpm2, are functionally redundant for actin cable organization and function, while having gained divergent functions in Retrograde Actin Cable Flow (RACF) (Fig. 4, Fig. 5A-D, Fig. S4, Fig. S5, Fig. S6). We have now added new data on role of global Tpm levels controlling endocytosis via maintenance of normal linear-to-branched actin network homeostasis in S. cerevisiae (Fig. 5E-G)__. We respectfully differ with the reviewer on their assessment of our study and request the reviewer to read our revised manuscript which discusses the significance, limitations, and future perspectives of our study in detail.

Reviewer #2

Evidence, reproducibility and clarity

This manuscript by Dhar, Bagyashree, Palani and colleagues examines the function of the two tropomyosins, Tpm1 and Tpm2, in the budding yeast S. cerevisiae. Previous work had shown that deletion of tpm1 and tpm2 causes synthetic lethality, indicating overlapping function, but also proposed that the two tropomyosins have distinct functions, based on the observation that strong overexpression of Tpm2 causes defects in bud placement and fails to rescue tpm1∆ phenotypes (Drees et al, JCB 1995). The manuscript first describes very functional mNeonGreen tagged version of Tpm1 and Tpm2, where an alanine-serine dipeptide is inserted before the first methionine to mimic acetylation. It then proposes that the Tpm1 and Tpm2 exhibit indistinguishable localization and that low level overexpression (?) of Tpm2 can replace Tpm1 for stabilization of actin cables and cell polarization, suggesting almost completely redundant functions. They also propose on specific function of Tpm2 in regulating retrograde actin cable flow.

Overall, the data are very clean, well presented and quantified, but in several places are not fully convincing of the claims. Because the claims that Tpm1 and Tpm2 have largely overlapping function and localization are in contradiction to previous publication in S. cerevisiae and also different from data published in other organisms, it is important to consolidate them. There are fairly simple experiments that should be done to consolidate the claims of indistinguishable localization, and levels of expression, for which the authors have excellent reagents at their disposal.

1. Functionality of the acetyl-mimic tagged tropomyosin constructs: The overall very good functionality of the tagged Tpm constructs is convincing, but the authors should be more accurate in their description, as their data show that they are not perfectly functional. For instance, the use of "completely functional" in the discussion is excessive. In the results, the statement that mNG-Tpm1 expression restores normal growth (page 3, line 69) is inaccurate. Fig S1C shows that tpm1∆ cells expressing mNG-Tpm1 grow more slowly than WT cells. (The next part of the same sentence, stating it only partially restores length of actin cables should cite only Fig S1E, not S1F.) Similarly, the growth curve in Fig S1C suggests that mNG-amTpm1, while better than mNG-Tpm1 does not fully restore the growth defect observed in tpm1∆ (in contrast to what is stated on p. 4 line 81). A more stringent test of functionality would be to probe whether mNG-amTpm1 can rescue the synthetic lethality of the tpm1∆ tpm2∆ double mutant, which would also allow to test the functionality of mNG-amTpm2.

__Response: __We would like to thank the reviewer for his feedback and suggestions. Based on the suggestions, we have now more accurately described the growth rescue observed by expression of mNG-ASTpm1 in Dtpm1 cells in the revised text. We have also removed the use of "completely functional" to describe mNG-Tpm functionality and corrected any errors in Figure citations in the revised manuscript.

As per reviewers' suggestion, we have now tested rescue of synthetic lethality of Dtpm1Dtpm2 cells by expression of all mNG-Tpm variants and we find that all of them are capable of restoring the viability of Dtpm1Dtpm2 cells when expressed under their native promoters via a high-copy plasmid (pRS425) (Fig. S1E) but only mNG-Tpm1 and mNG-ASTpm1 restored viability of Dtpm1Dtpm2 cells when expressed under their native promoters via an integration plasmid (pRS305) (Fig. S1F). These results clearly suggest that while both mNG-Tpm1 and mNG-Tpm2 constructs are functional, Tpm1 tolerates the presence of the N-terminal fluorescent tag better than Tpm2. These observations now enhance our understanding of the functionality of these mNG-Tpm fusion proteins and will be a useful resource for their usage and experimental design in future studies in vivo.

It would also be nice to comment on whether the mNG-amTpm constructs really mimicking acetylation. Given the Ala-Ser peptide ahead of the starting Met is linked N-terminally to mNG, it is not immediately clear it will have the same effect as a free acetyl group decorating the N-terminal Met.

Response: __We agree with the reviewer's observation and for the sake of clarity and accuracy, we have now renamed "mNG-amTpm" with "mNG-ASTpm". The use of -AS- dipeptide is very routine in studies with Tpm (Alioto et al., 2016; Palani et al., 2019; Christensen et al., 2017) and its addition restores normal binding affinities to Tpm proteins purified from E. coli (Monteiro et al., 1994). We agree with the reviewer that the -AS- dipeptide addition may not mimic N-terminal acetylation structurally but as per previous studies, it may help neutralize the impact of a freely protonated Met on the alpha-helical structure and stabilize the N-terminus helix of Tpm and allow normal head-to-tail dimer formation (Monteiro et al., 1994; Frye et al., 2010; Greenfield et al., 1994). Consistent with this, we also observe a highly significant improvement in actin cable length when expressing mNG-ASTpm as compared to mNG-Tpm in Dtpm1 cells, suggesting an improvement in function probably due to increased binding affinity (Fig. 1B, 1C). We have also discussed this in our answer to Question 1 of Reviewer 1 and the revised manuscript (Lines 350-372)__.

__ Localization of Tpm1 and Tpm2:__Given the claimed full functionality of mNG-amTpm constructs and the conclusion from this section of the paper that relative local concentrations may be the major factor in determining tropomyosin localization to actin filament networks, I am concerned that the analysis of localization was done in strains expressing the mNG-amTpm construct in addition to the endogenous untagged genes. (This is not expressly stated in the manuscript, but it is my understanding from reading the strain list.) This means that there is a roughly two-fold overexpression of either tropomyosin, which may affect localization. A comparison of localization in strains where the tagged copy is the sole Tpm1 (respectively Tpm2) source would be much more conclusive. This is important as the results are making a claim in opposition to previous work and observation in other organisms.

Response: __We thank the reviewer for this observation and their suggestions. We agree that relative concentrations of functional Tpm1 and Tpm2 in cells may influence the extent of their localizations. As per the reviewer's suggestion, we have now conducted our quantitative analysis in cells lacking endogenous Tpm1 and only expressing mNG-ASTpm1 from an integrated plasmid copy at the leu2 locus and the data is presented in new __Figure S3. We compared Tpm-bound cable length (Fig. S3A, S3B) __and Tpm-bound cable number (Fig. S3A, S3C) along with actin cable length (Fig. S3D, S3E) and actin cable number (Fig. S3D, S3F) in wildtype, Dbnr1, and Dbni1 cells. Our analysis revealed that mNG-ASTpm1 localized to actin cable structures in wildtype, Dbnr1, and Dbni1 cells and the decrease observed in Tpm-bound cable length and number upon loss of either Bnr1 or Bni1, was accompanied by a corresponding decrease in actin cable length and number upon loss of either Bnr1 or Bni1. Thus, this analysis reached the same conclusion as our earlier analysis (Fig. 2) that mNG-ASTpm1 does not show preference between Bnr1 and Bni1-made actin cables. mNG-ASTpm2 did not restore functionality, when expressed as single integrated copy, in Dtpm1Dtpm2 cells (new results in __Fig. S1E, S1F, S5A) thus, we could not conduct a similar analysis for mNG-ASTpm2. This suggests that use of mNG-ASTpm2 would be more meaningful in the presence of endogenous Tpm2 as previously done in Fig. 2D-F.

We have now also performed additional yeast mating experiments with cells lacking bnr1 gene and expressing either mNG-ASTpm1 or mNG-ASTpm2 and the data is shown in new Figure 3. From these observations, we observe that both mNG-ASTpm1 and mNG-ASTpm2 localize to the mating fusion focus in a Bnr1-independent manner (Fig. 3B, 3D) and suggests that they bind to Bni1-made actin cables that are involved in polarized growth of the mating projection. These results also add strength to our conclusion that Tpm1 and Tpm2 localize to actin cables irrespective of which formin nucleates them. Overall, these new results highlight and reiterate our model of formin-isoform independent binding of Tpm1 and Tpm2 in S. cerevisiae.

In fact, although the authors conclude that the tropomyosins do not exhibit preference for certain actin structures, in the images shown in Fig 2A and 2D, there seems to be a clear bias for Tpm1 to decorate cables preferentially in the bud, while Tpm2 appears to decorate them more in the mother cell. Is that a bias of these chosen images, or does this reflect a more general trend? A quantification of relative fluorescence levels in bud/mother may be indicative.

Response: __We thank the reviewer for pointing this out. Our data and analysis do not suggest that Tpm1 and Tpm2 show any preference for decoration of cables in either mother or bud compartment. As per the reviewer's suggestion, we have now quantified the ratio of mean mNG fluorescence in the bud to the mother (Bud/Mother) and the data is shown in __Figure. S2G. The bud-to-mother ratio was similar for mNG-ASTpm1 and mNG-ASTpm2 in wildtype cells, and the ratio increased in Dbnr1 cells and decreased in Dbni1 cells for both mNG-ASTpm1 and mNG-ASTpm2 (Fig. S2G). __This is consistent with the decreased actin cable signal in the mother compartment in Dbnr1 cells and decreased actin cable signal in the bud compartment in Dbni1 cells (Fig. S2A-D). Thus, our new analysis shows that both mNG-ASTpm1 and mNG-ASTpm2 have similar changes in their concentration (mean fluorescence) upon loss of either formins Bnr1 and Bni1 and show similar ratios in wildtype cells as well, suggesting no preference for binding to actin cables in either bud or mother compartment. The preference inferred by the reviewer seems to be a bias of the current representative images and thus, we have replaced the images in __Fig. 2A, 2D to more accurately represent the population.

The difficulty in preserving mNG-amTpm after fixation means that authors could not quantify relative Tpm/actin cable directly in single fixed cells. Did they try to label actin cables with Lifeact instead of using phalloidin, and thus perform the analysis in live cells?

__Response: __We did not use LifeAct for our analysis as LifeAct is known to cause expression-dependent artefacts in cells (Courtemanche et al., 2016; Flores et al., 2019; Xu and Du, 2021) and it also competes with proteins that regulate normal cable organization like cofilin. Use of LifeAct would necessitate standardization of expression to avoid such artefacts in vivo. Also, phalloidin staining provides the best staining of actin cables and allows for better quantitative results in our experiments. The use of LifeAct along with mNG-Tpm would also require optimization with a red fluorescent protein which usually tend to have lower brightness and photostability. However, during the revision of our study, a new study from Prof. Goode's lab has developed and optimized expression of new LifeAct-3xmNeonGreen constructs for use in S. cerevisiae (Wirshing and Goode, 2024). Thus, a similar strategy of using tandem copies of bright and photostable red fluorescent proteins can be explored for use in combination with mNG-Tpm in the future studies.

__ Complementation of tpm1∆ by Tpm2:__

I am confused about the quantification of Tpm2 expression by RT-PCR shown in Fig S3F. This figure shows that tpm2 mRNA expression levels are identical in cells with an empty plasmid or with a tpm2-encoding plasmid. In both strains (which lack tpm1), as well as in the WT control, one tpm2 copy is in the genome, but only one strain has a second tpm2 copy expressed from a centromeric plasmid, yet the results of the RT-PCR are not significantly different. (If anything, the levels are lower in the tpm2 plasmid-containing strain.) The methods state that the primers were chosen in the gene, so likely do not distinguish the genomic from the plasmid allele. However, the text claims a 1-fold increase in expression, and functional experiments show a near-complete rescue of the tpm1∆ phenotype. This is surprising and confusing and should be resolved to understand whether higher levels of Tpm2 are really the cause of the observed phenotypic rescue.

The authors could for instance probe for protein levels. I believe they have specific nanobodies against tropomyosin. If not, they could use expression of functional mNG-amTpm2 to rescue tpm1∆. Here, the expression of the protein can be directly visualized.

Response: __We thank the reviewer for pointing this out. We would like to clarify that in our RT-qPCR experiments, the primers were chosen within the Tpm1 and Tpm2 gene and do not distinguish between transcripts from endogenous or plasmid copy. We have now mentioned this in the Materials and Methods section of the revised manuscript. So, they represent a relative estimate of the total mRNA of these genes present in cells. We were consistently able to detect ~19 fold increase in Tpm2 total mRNA levels as compared to wildtype and ∆tpm1 cells (Fig. S4D) when tpm2 was expressed from a high-copy plasmid (pRS425). This increase in Tpm2 mRNA levels was accompanied by a rescue in growth (Fig. S4A) and actin cable organization (Fig. S4B) of ∆tpm1 cells containing pRS425-ptpm2TPM2. When tpm2 was expressed from a low-copy number centromeric plasmid (pRS316), we detected a ~2 fold increase in Tpm2 transcript levels when using the tpm1 promoter and no significant change was detected when using tpm2 promoter (Fig. S4E)__. We have made sure that these results are accurately described in the revised manuscript.

As per the reviewer's suggestion, we have now conducted a more extensive analysis to ascertain the expression levels of Tpm2 in our experiments and the data is now presented in new Figure S5. We used mNG-ASTpm1 and mNG-ASTpm2 to rescue growth of ∆tpm1 (Fig. S5A) and correlated growth rescue with protein levels using quantified fluorescence intensity (Fig. S5B, S5C) and western blotting (anti-mNG) (Fig. S5D, S5E). We find that ∆tpm1 cells containing pRS425-ptpm1mNG-ASTpm1 had the highest protein level followed by pRS425-ptpm2 mNG-ASTpm2, pRS305-ptpm1mNG-ASTpm1, and the least protein levels were found in pRS305-ptpm2 mNG-ASTpm2 containing ∆tpm1 cells in both fluorescence intensity and western blotting quantifications (Fig. S5C, S5E). Surprisingly, we were not able to detect any protein levels in ∆tpm1 cells containing pRS305-ptpm2 mNG-ASTpm2 with western blotting (Fig. S5D) which was also accompanied by a lack of growth rescue (Fig. S5A). This most likely due to weak expression from the native Tpm2 promoter which is consistent with previous literature (Drees et al., 1995). Taken together, this data clearly shows that the rescue observed in ∆tpm1 cells is caused due to increased expression of mNG-ASTpm2 in cells and supports our conclusion that increase in Tpm2 expression leads to restoration of normal growth and actin cables in ∆tpm1 cells.

__ Specific function of Tpm2:__

The data about the retrograde actin flow is interpreted as a specific function of Tpm2, but there is no evidence that Tpm1 does not also share this function. To reach this conclusion one would have to investigate retrograde actin flow in tpm1∆ (difficult as cables are weak) or for instance test whether Tpm1 expression restores normal retrograde flow to tpm2∆ cells.

Response: __We agree with the reviewer and as per the reviewer's suggestion, we have performed another experiment which include wildtype, ∆tpm2 cells containing empty pRS316 vector or pRS316-ptpm2TPM1 or pRS316-ptpm1TPM1. We find that RACF rate increased in ∆tpm2 cells as compared to wildtype and was restored to wildtype levels by exogenous expression of Tpm2 but not Tpm1 (Fig. S6E, S6F). Since, actin cables were not detectable in ∆tpm1 cells, we measured RACF rates in ∆tpm1 cells expressing Tpm1 or Tpm2 from a plasmid copy, which restored actin cables as shown previously in __Fig. 5A-C. We observed that RACF rates were similar to wildtype in ∆tpm1 cells expressing either Tpm1 or Tpm2 (Fig. S6E, S6F), suggesting that Tpm1 is not involved in RACF regulation. Taken together, these results suggest a specific role for Tpm2, but not Tpm1, in RACF regulation in S. cerevisiae, consistent with previous literature (Huckaba et al., 2006).

Minor comments: __1.__The growth of tpm1∆ with empty plasmid in Fig S3A is strangely strong (different from other figures).

Response: We thank the reviewer for pointing this out. We have now repeated the drop test multiple times (Fig. R2), but we see similar growth rates as the drop test already presented in Fig. S4A. __At this point, it would be difficult to ascertain the basis of this difference observed at 23{degree sign}C and 30{degree sign}C, but a recent study that links leucine levels to actin cable stability (Sing et al., 2022) might explain the faster growth of these ∆tpm1 cells containing a leu2 gene carrying high-copy plasmid. However, there is no effect on growth rate at 37{degree sign}C which is consistent with other spot assays shown in __Fig. S1D, S4F, S5A.

Significance

I am a cell biologist with expertise in both yeast and actin cytoskeleton.

The question of how tropomyosin localizes to specific actin networks is still open and a current avenue of study. Studies in other organisms have shown that different tropomyosin isoforms, or their acetylated vs non-acetylated versions, localize to distinct actin structures. Proposed mechanisms include competition with other ABPs and preference imposed by the formin nucleator. The current study re-examines the function and localization of the two tropomyosin proteins from the budding yeast and reaches the conclusion that they co-decorate all formin-assembled structures and also share most functions, leading to the simple conclusion that the more important contribution of Tpm1 is simply linked to its higher expression. Once consolidated, the study will appeal to researchers working on the actin cytoskeleton.

We thank the reviewer for their positive assessment of our work and the constructive feedback that has greatly improved the quality of our study. After addressing the points raised by the reviewer, we believe that our study has significantly gained in consolidating the major conclusions of our work.

**Referees cross-commenting**

Having read the other reviewers' comments, I do agree with reviewer 1 that it is not clear whether the Ala-Ser linker really mimics acetylation. I am less convinced than reviewer 3 that the key conclusions of the study are well supported, notably the issue of Tpm2 expression levels is not convincing to me.

Response: __We acknowledge the reviewer's point about the effect of Ala-Ser dipeptide and would request the reviewer to refer to our response to Reviewer 1 (Question 1) for a more detailed discussion on this. We have also extensively addressed the question of Tpm2 expression levels as suggested by the reviewer (new data in __Figure S5) which has further strengthened the conclusions of our study.

__Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Summary:__ The study presents the first fully functional fluorescently tagged Tpm proteins, enabling detailed probing of Tpm isoform localization and functions in live cells. The authors created a modified fusion protein, mNG-amTpm, which mimicked native N-terminal acetylation and restored both normal growth and full-length actin cables in yeast cells lacking native Tpm proteins, demonstrating the constructs' full functionality. They also show that Tpm1 and Tpm2 do not have a preference for actin cables nucleated by different formins (Bnr1 and Bni1). Contrary to previous reports, the study found that overexpressing Tpm2 in Δtpm1 cells could restore growth rates and actin cable formation. Furthermore, it is shown that despite its evolutionary divergence, Tpm2 retains actin-protective functions and can compensate for the loss of Tpm1, contributing to cellular robustness.

Major and Minor Comments: 1. The key conclusions of this paper are convincing. However, I suggest that more detail be provided regarding the image analysis used in this study. Specifically, since threshold settings can impact the quality of the generated data and, therefore, its interpretation, it would be useful to see a representative example of the quantification methods used for actin cable length/number (as in refs. 80 and 81) and mitochondria morphology. These could be presented as Supplemental Figures. Additionally, it would help to interpret the results if the authors could be more specific about the statistical tests that were used.

Response: __We agree with the reviewer's suggestions and have now updated our Materials and Methods section to describe the image analysis pipelines used in more detail. We have also added examples of quantification procedure for actin cable length/number and mitochondrial morphology as an additional Supplementary __Figure S7. Briefly, the following pipelines were used:

• Actin cable length and number analysis: This was done exactly as mentioned in McInally et al., 2021, McInally et al., 2022. Actin cables were manually traced in Fiji as shown in __ S7A__, and then the traces files for each cell were run through a Python script (adapted from McInally et al., 2022) that outputs mean actin cable length and number per cell.
• Mitochondria morphology: Mitochondria Analyzer plug-in in Fiji was used to segment out the mitochondrial fragments. The parameters used for 2D segmentation of mitochondria were first optimized using "2D Threshold Optimize" to find the most accurate segmentation and then the same parameters were run on all images. After segmentation of the mitochondrial network, measurements of fragment number were done using "Analyze Particles" function in Fiji. An example of the overall process is shown in __ S7B.__ As per the reviewer's suggestion, we have now included the description of the statistical test used in the Figure Legends of each Figure in the revised manuscript. We have used One-Way Anova with Tukey's Multiple Comparison test, Kruskal-Wallis test with Dunn's Multiple Comparisons, and Unpaired Two-tailed t-test using the in-built functions in GraphPad Prism (v.6.04).

**Referees cross-commenting**

I agree with both reviewers 1 and 2 regarding the issues with the Ala-Ser acetylation mimic and Tpm2 expression levels, respectively. I think the authors should be more careful in how they frame the results, but I consider that these issues do not invalidate the main conclusions of this study.

Response: __We acknowledge the reviewer's concern about the Ala-Ser dipeptide and would request them to refer our earlier discussion on this in response to Reviewer 1 (Question 1) and Reviewer 2 (Question 2). We would also request the reviewer to refer to our answer to Reviewer 2 (Question 6) where we have extensively addressed the question of Tpm2 expression levels and their effect on rescue of Dtpm1 cells. This data is now presented as new __Figure S5 in our revised manuscript.

Reviewer#3 (Significance (Required)):

The finding that Tpm2 can compensate for the loss of Tpm1, restoring actin cable organization and normal growth rates, challenges previous assumptions about the non-redundant functions of these isoforms in Saccharomyces cerevisiae (ref. 16). It also supports a concentration-dependent and formin-independent localization of Tpm isoforms to actin cables in this species. The development of fully functional fluorescently tagged Tpm proteins is a significant methodological advancement. This advancement overcomes previous visualization challenges and allows for accurate in vivo studies of Tpm function and regulation in S. cerevisiae.

The findings will be of particular interest to researchers in the field of cellular and molecular biology who study actin cytoskeleton dynamics. Additionally, it will be relevant for those utilizing advanced microscopy and live-cell imaging techniques.

As a researcher, my experience lies in cytoskeleton dynamics and protein interactions, though I do not have specific experience related to tropomyosin. I use different yeast species as models and routinely employ live-cell imaging as a tool.

We thank the reviewer for their positive outlook and assessment of our study. We have incorporated all their suggestions, and we are confident that the revised manuscript has significantly improved in quality due to these additions.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #2 (Public review):

Summary

In this extensive comparative study, Moreno-Borrallo and colleagues examine the relationships between plasma glucose levels, albumin glycation levels, diet and lifehistory traits across birds. Their results confirmed the expected positive relationship between plasma blood glucose level and albumin glycation rate but also provided findings that are somewhat surprising or contrast with findings of some previous studies (positive relationships between blood glucose and lifespan, or absent relationships between blood glucose and clutch mass or diet). This is the first extensive comparative analysis of glycation rates and their relationships to plasma glucose levels and life history traits in birds that is based on data collected in a single study, with blood glucose and glycation measured using unified analytical methods (except for blood glucose data for 13 species collected from a database).

Strengths

This is an emerging topic gaining momentum in evolutionary physiology, which makes this study a timely, novel and important contribution. The study is based on a novel data set collected by the authors from 88 bird species (67 in captivity, 21 in the wild) of 22 orders, except for 13 species, for which data were collected from a database of veterinary and animal care records of zoo animals (ZIMS). This novel data set itself greatly contributes to the pool of available data on avian glycemia, as previous comparative studies either extracted data from various studies or a ZIMS database (therefore potentially containing much more noise due to different methodologies or other unstandardised factors), or only collected data from a single order, namely Passeriformes. The data further represents the first comparative avian data set on albumin glycation obtained using a unified methodology. The authors used LC-MS to determine glycation levels, which does not have problems with specificity and sensitivity that may occur with assays used in previous studies. The data analysis is thorough, and the conclusions are substantiated. Overall, this is an important study representing a substantial contribution to the emerging field evolutionary physiology focused on ecology and evolution of blood/plasma glucose levels and resistance to glycation.

Weaknesses

Unfortunately, the authors did not record handling time (i.e., time elapsed between capture and blood sampling), which may be an important source of noise because handling-stress-induced increase in blood glucose has previously been reported. Moreover, the authors themselves demonstrate that handling stress increases variance in blood glucose levels. Both effects (elevated mean and variance) are evident in Figure ESM1.2. However, this likely makes their significant findings regarding glucose levels and their associations with lifespan or glycation rate more conservative, as highlighted by the authors.

Recommendations for the authors:

Reviewer #2 (Recommendations for the authors):

I understand that your main objective regarding glycation rate and lifespan, was to analyse the species resistance to glycation with respect to lifespan, while factoring out the species-specific variation in blood glucose level. However, I still believe that the absolute glycation level (i.e., not controlled for blood glucose level) may also be important for the evolution of lifespan. Given that blood glucose is positively related to both glycation and lifespan (although with a plateau in the latter case), lifespan could possibly be positively correlated with absolute glycation levels. If significant, that would be an interesting and counterintuitive finding, which would call for an explanation, thereby potentially stimulating further research. If not significant, it would show that long-lived species do not have higher glycation levels, despite having higher blood glucose levels, thereby strengthening your argument about higher resistance of longlived species to glycation. So, in my opinion, the inclusion of an additional model of glycation level on life-history traits, without controlling for blood glucose, is worth considering.

We include now this model as supplementary material, indicating it in several parts of the text, including some of these issues we discussed here.

Lines 230-231: Please, provide a citation for these GVIF thresholds

We include it now.

Figure 3: I think that showing both glucose and glycation rate on the linear scale, rather than log scale, would better illustrate your conclusion - the slowing rise of glycation rate with increasing glucose levels.

That is a good point, although it may also be confusing for readers to see a graph that represents the data in a different way as the models. Maybe showing both graphs (as 3.A and 3.B) can solve it?

Figure 4. I recommend stating in the caption that the whiskers do not represent interquartile ranges (a standard option in box plots) but credible intervals as mentioned in the current version of the public author response.

Sorry about that, it was missed. Now it is included. Nevertheless, interquartile ranges from the posterior distributions can still be observed here represented with the boxes. Then the whiskers are the credible intervals.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this manuscript, Guo and colleagues used a cell rounding assay to screen a library of compounds for inhibition of TcdB, an important toxin produced by Clostridioides difficile. Caffeic acid and derivatives were identified as promising leads, and caffeic acid phenethyl ester (CAPE) was further investigated.

Strengths:

Considering the high morbidity rate associated with C. difficile infections (CDI), this manuscript presents valuable research in the investigation of novel therapeutics to combat this pressing issue. Given the rising antibiotic resistance in CDI, the significance of this work is particularly noteworthy. The authors employed a robust set of methods and confirmatory tests, which strengthened the validity of the findings. The explanations provided are clear, and the scientific rationale behind the results is well-articulated. The manuscript is extremely well-written and organized. There is a clear flow in the description of the experiments performed. Also, the authors have investigated the effects of CAPE on TcdB in careful detail and reported compelling evidence that this is a meaningful and potentially useful metabolite for further studies.

Weaknesses:

This is really a manuscript about CAPE, not caffeic acid, and the title should reflect that. Also, a few details are missing from the description of the experiments. The authors should carefully revise the manuscript to ascertain that all details that could affect the interpretation of their results are presented clearly. Just as an example, the authors state in the results section that TcdB was incubated with compounds and then added to cells. Was there a wash step in between? Could compound carryover affect how the cells reacted independently from TcdB? This is just an example of how the authors should be careful with descriptions of their experimental procedures. Lastly, authors should be careful when drawing conclusions from the analysis of microbiota composition data. Ascribing causality to correlational relationships is a recurring issue in the microbiome field. Therefore, I suggest authors carefully revise the manuscript and tone down some statements about the impact of CAPE treatment on the gut microbiota.

Thanks for your constructive suggestion. We have carefully revised the manuscript, including the description of title, results and methods sections.

Reviewer #2 (Public review):

Summary:

This work is towards the development of nonantibiotic treatment for C. difficile. The authors screened a chemical library for activity against the C. difficile toxin TcdB, and found a group of compounds with antitoxin activity. Caffeic acid derivatives were highly represented within this group of antitoxin compounds, and the remaining portion of this work involves defining the mechanism of action of caffeic acid phenethyl ester (CAPE) and testing CAPE in mouse C. difficile infection model. The authors conclude CAPE attenuates C. difficile disease by limiting toxin activity and increasing microbial diversity during C. difficile infection.

Strengths/ Weaknesses:

The strategy employed by the authors is sound although not necessarily novel. A compound that can target multiple steps in the pathogenies of C. difficile would be an exciting finding. However, the data presented does not convincingly demonstrate that CAPE attenuates C. difficile disease and the mechanism of action of CAPE is not convincingly defined. The following points highlight the rationale for my evaluation.

(1) The toxin exposure in tissue culture seems brief (Figure 1). Do longer incubation times between the toxin and cells still show CAPE prevents toxin activity?

Thanks for your comments. The cytotoxicity assay was employed to directly assess the protective capacity of CAPE against cell death induced by TcdB. Our observations at 1 and 12 h post-TcdB exposure revealed that CAPE effectively mitigated the toxic effects of the TcdB at both time points, demonstrating its potent protective role. Please see Figure S1.

(2) The conclusion that CAPE has antitoxin activity during infection would be strengthened if the mouse was pretreated with CAPE before toxin injections (Figure 1D).

Thanks for your constructive comments. According to your suggestion, we administered TcdB 2 h after pretreatment with CAPE. The outcomes demonstrated that CAPE pretreatment significantly enhanced the survival rate of the intoxicated mice, confirming that CAPE retains its antitoxin efficacy during the infection process. Please see Figure S2.

(3) CAPE does not bind to TcdB with high affinity as shown by SPR (Figure 4). A higher affinity may be necessary to inhibit TcdB during infection. The GTD binds with millimolar affinity and does not show saturable binding. Is the GTD the binding site for CAPE? Auto processing is also affected by CAPE indicating CAPE is binding non-GTD sites on TcdB.

Thanks for your comments. Our findings indicate that the GTD domain is a critical binding site for CAPE. CAPE exerts its protective effects at multiple stages of TcdB-mediated cell death, including inhibiting TcdB's self-cleavage and blocking the activity of GTD, thereby preventing the glycosylation modification of Rac1 by TcdB.

(4) In the infection model, CAPE does not statistically significantly attenuate weight loss during C. difficile infection (Figure 6). I recognize that weight loss is an indirect measure of C. difficile disease but histopathology also does not show substantial disease alleviation (see below).

Thanks for your comments. Our comparative analysis revealed a notable distinction in the body weight of mice on the third day post-infection (Figure 6B). Similarly, the dry/wet stool ratio exhibited a comparable pattern, suggesting that treatment with phenethyl caffeic acid ameliorated Clostridium difficile-induced diarrhea to a significant degree (Figure 6C).

(5) In the infection model (Figure 6), the histopathology analysis shows substantial improvement in edema but limited improvement in cellular infiltration and epithelial damage. Histopathology is probably the most critical parameter in this model and a compound with disease-modifying effects should provide substantial improvements.

Thanks for your comments. Edema, inflammatory factor infiltration, and epithelial damage served as key evaluation metrics. Statistical analysis revealed that the pathological scores of mice treated with CAPE were markedly reduced compared to those in the model group (Figure 6F).

(6) The reduction in C. difficile colonization is interesting. It is unclear if this is due to antitoxin activity and/or due to CAPE modifying the gut microbiota and metabolites (Figure 6). To interpret these data, a control is needed that has CAPE treatment without C. difficile infection or infection with an atoxicogenic strain.

The observed reduction in C. difficile fecal colonization following drug treatment may be attributed to the CAPE's antitoxin properties or its capacity to modify the intestinal microbiota and metabolites. These two mechanisms likely work in tandem to combat CDI. CDI is primarily triggered by the toxins A (TcdA) and B (TcdB) secreted by the bacterium. Certain therapies, including monoclonal antibodies like bezlotoxumab, target CDI by neutralizing these toxins, thereby mitigating gut damage and subsequent C. difficile colonization(1,2). The establishment of C. difficile in the gut is intricately linked to the equilibrium of the intestinal microbiota. Although antibiotic treatments can inhibit C. difficile growth, they may also disrupt the microbial balance, potentially facilitating the overgrowth of other pathogens. Consequently, interventions such as fecal microbiota transplantation (FMT) are designed to reestablish gut flora balance and consequently decrease C. difficile colonization(3,4). Moreover, the administration of probiotics and prebiotics is considered to reduce C. difficile colonization by modifying the gut environment(5,6).

(7) Similar to the CAPE data, the melatonin data does not display potent antitoxin activity and the mouse model experiment shows marginal improvement in the histopathological analysis (Figure 9). Using 100 µg/ml of melatonin (~ 400 micromolar) to inactivate TcdB in cell culture seems high. Can that level be achieved in the gut?

The uptake and dissemination of melatonin within the body varies with the dose administered. For instance, in rats, the bioavailability of melatonin following administration was found to be 53.5%, whereas in dogs, bioavailability was nearly complete (100%) at a dose of 10 mg/kg, yet it decreased to 16.9% at a lower dose of 1 mg/kg(7). This data suggests that the absorption of melatonin differs across various animal species and is influenced by the dose administered. Moreover, it underscores the higher potential bioavailability of melatonin, implying that a dose of 200 mg/kg should be adequate to achieve the desired concentration in the body post-administration.

(8) The following parameters should be considered and would aid in the interpretation of this work. Does CAPE directly affect the growth of C. difficile? Does CAPE affect the secretion of TcdB from C. difficile? Does CAPE alter the sporulation and germination of C. diffcile?

We incorporated CAPE into the MIC assay for detecting C. difficile, as well as for assessing the sporulation capacity of C. difficile and evaluating the secretion level of TcdB. The findings revealed that CAPE markedly repressed tcdB transcription at a concentration of 16 μg/mL and effectively suppressed the growth and sporulation of C. difficile BAA-1870 at a concentration of 32 μg/mL. Please see Figure S3.

References:

(1) Skinner AM, et al. Efficacy of bezlotoxumab to prevent recurrent Clostridioides difficile infection (CDI) in patients with multiple prior recurrent CDI. Anaerobe. 2023 Dec; 84: 102788.

(2) Wilcox MH, et al. Bezlotoxumab for Prevention of Recurrent Clostridium difficile Infection. N Engl J Med. 2017 Jan 26;376(4):305-317.

(3) Khoruts A, Sadowsky MJ. Understanding the mechanisms of faecal microbiota transplantation. Nat Rev Gastroenterol Hepatol. 2016 Sep;13(9):508-16.

(4) Khoruts A, Staley C, Sadowsky MJ. Faecal microbiota transplantation for Clostridioides difficile: mechanisms and pharmacology. Nat Rev Gastroenterol Hepatol. 2021 Jan;18(1):67-80.

(5) Mills JP, Rao K, Young VB. Probiotics for prevention of Clostridium difficile infection. Curr Opin Gastroenterol. 2018 Jan;34(1):3-10.

(6) Lau CS, Chamberlain RS. Probiotics are effective at preventing Clostridium difficile-associated diarrhea: a systematic review and meta-analysis. Int J Gen Med. 2016 Feb 22; 9:27-37.

(7) Yeleswaram K, et al. Pharmacokinetics and oral bioavailability of exogenous melatonin in preclinical animal models and clinical implications. J Pineal Res. 1997 Jan;22(1):45-51.

Reviewer #3 (Public review):

Summary:

The study is well written, and the results are solid and well demonstrated. It shows a field that can be explored for the treatment of CDI.

Strengths:

The results are really good, and the CAPE shows a good and promising alternative for treating CDI. The methodology and results are well presented, with tables and figures that corroborate them. It is solid work and very promising.

Weaknesses:

Some references are too old or missing.

Thanks for your constructive suggestion. We have included and refreshed several references to enhance the manuscript.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

While the manuscript convincingly demonstrates that CAPE affects the TcdB toxin and reduces its toxicity in vitro, it would be beneficial to include data on the effect of CAPE on the growth of C. difficile. This would help ensure that the observed in vivo effects are not merely due to reduced bacterial growth but rather due to the specific action of CAPE on the toxin.

Thanks for your constructive suggestion. We have augmented our findings with the impact of CAPE on the bacteria themselves, revealing that CAPE not only hampers the growth of the bacterial cells but also suppresses their capacity to produce spores. Please see Figure S3.

(1) Line 41, line 115 - authors should clarify what they mean when mentioning Bacteroides within parentheses.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(2) Line 71 - Is C. difficile really found "in the environment"?

Thanks for your comments. C. difficile is prevalent across various natural settings, including soil and water ecosystems. A study has identified highly diverse strains of this bacterium within environmental samples(1). Moreover, the significant presence of C. difficile in soil and lawn specimens collected near Australian hospitals indicates that the organism is indeed a common inhabitant in the environment(2).

(3) Lines 128-130 - Was there a wash step here? What could be the impact of compound carryover in this experiment?

Thanks for your comments. Following pre-incubation of TcdB with CAPE, remove the compounds that have not bound to TcdB through centrifugation. The persistence of the compound in the culture post-washing could result in an inflated assessment of its efficacy, particularly if it continues to engage with TcdB or the cells beyond the initial 1-hour pre-incubation window. The carryover of the compound might also give rise to misleading positive results, where the compound seems to confer protection or inhibition against TcdB-mediated cell rounding, whereas such effects are actually due to the lingering activity of the compound. This carryover could skew the determination of the compound's minimum effective concentration, as the effective concentration interacting with the cells might be inadvertently elevated. Furthermore, if the compounds possess cytotoxic properties or impact cell viability, carryover could generate artifacts in cell morphology that are unrelated to the direct interaction between TcdB and the compounds.

(4) Lines 133-134 - I suggest authors mention how many caffeic acid derivatives there were in the entire library so that the suggested "enrichment" of them in the group of bioactive compounds can be better judged.

Thanks for your comments. The natural compound library contained eight caffeic acid derivatives, of which methyl caffeic acid and ferulic acid displayed no efficacy. This information has been incorporated into the manuscript.

(5) Line 135 - I recommend the authors add the molarity of the compound solutions used.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(6) Line 247 - I think the term "CAPE mice" is confusing. Please use a full description.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(7) Line 248 - I also think the terms "model mice" and "model group" are confusing. Maybe call them "control mice"?

Thanks for your comments. The terms "model mice" and "model group" are indeed synonymous, and we have subsequently clarified that control mice refer to those that have not been infected with C. difficile.

(8) Line 273 - "most abundant species at the genus level" is incorrect. I think what you mean is "most abundant TAXA".

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(9) Line 278 - Please include your p-value cut-off together with the LDA score.

Thanks for your comments. We have revised the above description to “LDA score > 3.5, p < 0.05”.

(10) Line 292 - Details on how metabolomics was performed should be included here.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(11) Line 299 - 1.5 is a fairly low cut-off. The authors should at a minimum also include the p-value cut-off used.

Response: Thanks for your comments. We have revised the above description to “fold change > 1.5, p < 0.05”.

(12) Line 307 - Purine "degradation" would be better here.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(13) Line 328 onward - The melatonin experiment is a weird one. Although I fully understand the rationale behind testing the effect of melatonin in the mouse model, the idea that just because melatonin levels changed in the gut it would act as a direct inhibitor of TcdB was very far-fetched, even though it ended up working. Authors should explain this in the manuscript.

Thanks for your comments. Furthermore, beyond our murine studies, we have confirmed that melatonin significantly diminishes TcdB-induced cytotoxicity at the cellular level (Figure 9A). Additionally, it has been documented that melatonin, acting as an antimicrobial adjuvant and anti-inflammatory agent, can decrease the recurrence of CDI(3). Consequently, we contend that the aforementioned statement is substantiated.

(14) Lines 429-435 - There are seemingly contradictory pieces of information here. The authors state that adenosine is released from cells upon inflammation and that CAPE treatment caused an increase in adenosine levels. Later in this section, the authors state that adenosine prevents TcdA-mediated damage and inflammation. This should be clarified and better discussed.

Thanks for your comments. Adenosine modulates immune responses and inflammatory cascades by interacting with its receptors, including its capacity to suppress the secretion of specific pro-inflammatory mediators. We have updated this depiction in the manuscript.

(15) Lines 513-514 - How was this phenotype quantified?

Thanks for your comments. Initially, we introduced TcdB at a final concentration of 0.2 ng/mL along with various concentrations of compounds into 1 mL of medium for a 1-h pre-incubation period. Subsequently, unbound compounds were removed through centrifugation, and the resulting mixture was then applied to the cells.

(16) Figure 3 - panels are labeled incorrectly.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(17) Figure 5C - it is unclear what the different colors and labels represent.

Thanks for your comments. In the depicted graph, blue denotes the total binding energy, red signifies the electrostatic interactions, green corresponds to the van der Waals forces, and orange indicates solvation or hydration effects. The horizontal axis represents the mutation of the amino acid residue at the respective position to alanine. As illustrated in Figure 5C, the mutations W520A and GTD exhibit the highest binding energies.

References:

(1) Janezic S, et al. Highly Divergent Clostridium difficile Strains Isolated from the Environment. PLoS One. 2016 Nov 23;11(11): e0167101.

(2) Perumalsamy S, Putsathit P, Riley TV. High prevalence of Clostridium difficile in soil, mulch and lawn samples from the grounds of Western Australian hospitals. Anaerobe. 2019 Dec; 60:102065.

(3) Sutton SS, et al. Melatonin as an Antimicrobial Adjuvant and Anti-Inflammatory for the Management of Recurrent Clostridioides difficile Infection. Antibiotics (Basel). 2022 Oct 25;11(11):1472.

Reviewer #2 (Recommendations for the authors):

Minor comments and questions.

(1) Which form of TcdB is being used in these experiments?

Thanks for your comments. The TcdB proteins used in this study are TcdB1 subtypes.

(2) Why are THP-1 cells being used in these assays?

Thanks for your comments. For the purposes of this study, we employed a diverse array of cell lines, including Vero, HeLa, THP-1, Caco-2, and HEK293T. Each cell line was selected to serve a specific experimental objective. The inclusion of the THP-1 cell line was necessitated by the need to incorporate a macrophage cell line to ensure the comprehensive nature of our experiments, allowing for the testing of both epithelial cells and macrophages. C. difficile is a kind of intestinal pathogenic bacteria, and immune clearance plays a vital role in the process of pathogen infection, so THP-1 cells are used as important immune cells.

(3) Please improve the quality of the microscopy images in Figure 1.

Thanks for your comments. We have improved the quality of the microscopy images in Figure 1.

(4) Does the flow cytometry experiment in Figure 2B show internalization? Surface-bound toxins would provide the same histogram.

Thanks for your comments. Figure 2B was employed to assess the internalization of TcdB, and the findings indicate that CAPE does not influence the internalization process of TcdB.

(5) The sensogram in Figure 4A does not look typical and should be clarified.

Thanks for your comments. Typically, small molecules and proteins engage in a rapid binding and dissociation dynamic. However, as depicted in Figure 4A, the interaction between CAPE and TcdB demonstrates a gradual progression towards equilibrium. This behavior can be primarily explained by the swift occupation of the protein's primary binding sites by the small molecule in the initial stages. Subsequently, CAPE binds to secondary or lower affinity sites, extending the time needed to reach equilibrium. Additionally, the likelihood of CAPE binding to multiple sites on TcdB requires time for the exploration and occupation of these diverse locations before equilibrium is attained, we have incorporated an analysis of this potential scenario into the manuscript.

Reviewer #3 (Recommendations for the authors):

These are my suggestions for the text:

(1) Line 29: high recurrent rates.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(2) Line 32: Where is the caffeic acid identified? I think a line should be included.

Thanks for your comments. Caffeic acid was identified from natural compounds library and we have completed the corresponding modifications according to the suggestions.

(3) Line 39: C. difficile is not italic.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(4) Line 41: Bacteroides spp.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

(5) Line 56: This number of casualties 56.000 is still happening or it was in the past?

Thanks for your comments. The mortality rates reported in the manuscript reflect a downturn in the incidence and fatality of CDI around 2017(1), as the infection gained broader recognition. Nonetheless, a recent study reveals that the mortality rate for CDI cases in Germany can soar to 45.7% within a year, with the overall economic burden amounting to approximately 1.6 billion euros. This underscores the ongoing significance of CDI as a global public health challenge(2).

(6) Line 104: Where did the idea of testing caffeic acid come from? Any previous study of the authors? Any studies with the inhibition of other pathogens?

Thanks for your comments. Initially, we conducted a screen of a compound library comprising 2,076 compounds and identified several potent inhibitors, which, upon structural analysis, were revealed to be caffeic acid derivatives. Prior to our investigation, no studies had explored the potential of CAPE in this context.

(7) Line 115: Bacteroides spp.

Thanks for your comments. We have completed the corresponding modifications according to the suggestions.

Results section

(8) Did the authors try the caffeic acid with the TcdA or binary toxin? I know this is not the purpose of the study, but TcdA toxin has a high identity structure with TcdB and generates inflammation in the gut via neutrophils. Negative strains for the major toxins and positive for the binary toxin also cause severe cases of CDI.

Thanks for your comments. Although we acknowledge the significance of TcdA and binary toxins in CDI, we did not investigate the impact of CAPE on these toxins. Our focus was exclusively on the effect of CAPE against TcdB, as it is the primary virulence factor in C. difficile pathogenesis. Since TcdA and TcdB are highly similar in structure, we will analyze the neutralization effect of CAPE on TcdA in later studies.

(9) Does caffeic acid have any effect on C. difficle? Or does it only gain the toxins? That would be ideal.

Thanks for your comments. We have included additional related assays in our study. Beyond directly neutralizing TcdB, CAPE also demonstrates the capacity to inhibit the growth and spore formation of C. difficile.

(10) Line 230: C. difficile BAA-1870 is a clinical strain? There are no details about it in the paper.

Thanks for your comments. C. difficile BAA-1870 (RT027/ST1), a highly virulent isolate frequently employed in research(3-6), was kindly donated by Professor Aiwu Wu. We have meticulously noted the PCR ribotype in our manuscript.

(11) Line 236: Did the mice fully recover from CDI after the administration of the CAPE? Was one dose enough?

Thanks for your comments. CAPE was administered orally at 24 h intervals, commencing with the initial dose on Day 0. By the time a significant difference was observed on Day 3, the treatment had been administered a total of three times.

Methodology

(12) Most of the methods do not have a reference.

Thanks for your comments. We have added several references to the methods.

Discussion section

(13) The first two paragraphs of the discussion should be summarized. Those details were already explained in the introduction.

Thanks for your comments. The discussion section and the introduction address slightly different focal points; therefore, we aim to retain the first two paragraphs to maintain continuity and context.

(14) Line 382: Bezolotoxumab was approved by the FDA in 2016. It is not recent.

Thanks for your comments. We have revised the above description.

(15) Line 410: "Despite the high 410 cure rate and increasing popularity of FMT, its safety remains controversial. Although this is true, recently (2022) the FDA approved the Rebyota, which was later cited by the authors.

Thanks for your comments. We have revised the above description.

(16) Lines 415-416: "the abundance of Bacteroides, a critical gut microbiota component that is required for C. difficile resistance". There is only one reference cited by the authors. I suppose that if it is true, more studies should be mentioned. Why are probiotics with Bacteroides spp. not available in the market?

Thanks for your comments. We have supplemented additional references. The scarcity of probiotic products containing Bacteroides spp. on the market is primarily attributable to the stringent requirements of their survival conditions. As most Bacteroides spp. are anaerobic, they thrive in oxygen-deprived environments. This unique survival trait poses challenges in maintaining their viability during product preservation and distribution, which in turn escalates production costs and complexity. Furthermore, despite the significant role of Bacteroides in gut health, research into its potential probiotic benefits and safety is comparatively underexplored.

References:

(1) Guh AY, et al. Emerging Infections Program Clostridioides difficile Infection Working Group. Trends in U.S. Burden of Clostridioides difficile Infection and Outcomes. N Engl J Med. 2020 Apr 2;382(14):1320-1330.

(2) Schley K, et al. Costs and Outcomes of Clostridioides difficile Infections in Germany: A Retrospective Health Claims Data Analysis. Infect Dis Ther. 2024 Nov 20.

(3) Saito R, et al. Hypervirulent clade 2, ribotype 019/sequence type 67 Clostridioides difficile strain from Japan. Gut Pathog. 2019 Nov 4; 11:54.

(4) Pellissery AJ, Vinayamohan PG, Venkitanarayanan K. In vitro antivirulence activity of baicalin against Clostridioides difficile. J Med Microbiol. 2020 Apr;69(4):631-639.

(5) Shao X, et al. Chemical Space Exploration around Thieno[3,2-d]pyrimidin-4(3H)-one Scaffold Led to a Novel Class of Highly Active Clostridium difficile Inhibitors. J Med Chem. 2019 Nov 14;62(21):9772-9791.

(6) Mooyottu S, Flock G, Venkitanarayanan K. Carvacrol reduces Clostridium difficile sporulation and spore outgrowth in vitro. J Med Microbiol. 2017 Aug;66(8):1229-1234.

This makes a lot of sense because I think students who come from low-income backgrounds have always had to work extra hard to end up at the same place as their wealthier counterparts who have more resources and thus more opportunities. It may make them feel "weak" to ask for help even though it is totally normal for us to reach out to professors when we are struggling. I think there is a big psychological effect that is going on here. No one wants to feel like they cannot handle a class or exam, especially if they have put a lot of pressure on themselves to overcome their situation. It is totally understandable when lower-income students have trouble reaching our, but we should work on creating a safe space where students feel comfortable reaching out regardless of their situations.

Author response:

The following is the authors’ response to the original reviews

Public Reviews: 

Reviewer #1 (Public review): 

Summary: 

Govindan and Conrad use a genome-wide CRISPR screen to identify genes regulating retention of intron 4 in OGT, leveraging an intron retention reporter system previously described (PMID: 35895270). Their OGT intron 4 reporter reliably responds to O-GlcNAc levels, mirroring the endogenous splicing event. Through a genome-wide CRISPR knockout library, they uncover a range of splicing-related genes, including multiple core spliceosome components, acting as negative regulators of OGT intron 4 retention. They choose to follow up on SFSWAP, a largely understudied splicing regulator shown to undergo rapid phosphorylation in response to O-GlcNAc level changes (PMID: 32329777). RNA-sequencing reveals that SFSWAP depletion not only promotes OGT intron 4 splicing but also broadly induces exon inclusion and intron splicing, affecting decoy exon usage. While this study offers interesting insights into intron retention and O-GlcNAc signaling regulation, the RNA sequencing experiments lack the essential controls needed to provide full confidence to the authors' conclusions. 

Strengths: 

(1) This study presents an elegant genetic screening approach to identify regulators of intron retention, uncovering core spliceosome genes as unexpected positive regulators of intron retention. 

(2) The work proposes a novel functional role for SFSWAP in splicing regulation, suggesting that it acts as a negative regulator of splicing and cassette exon inclusion, which contrasts with expected SR-related protein functions. 

(3) The authors suggest an intriguing model where SFSWAP, along with other spliceosome proteins, promotes intron retention by associating with decoy exons. 

We thank the reviewer for recognizing and detailing the strengths of our manuscript. 

Weaknesses: 

(1) The conclusions on SFSWAP impact on alternative splicing are based on cells treated with two pooled siRNAs for five days. This extended incubation time without independent siRNA treatments raises concerns about off-target effects and indirect effects from secondary gene expression changes, potentially limiting confidence in direct SFSWAP-dependent splicing regulation. Rescue experiments and shorter siRNA-treatment incubation times could address these issues. 

We repeated our SFSWAP knockdown analysis and analyzed both OGT e4-e5 junction splicing and SFSWAP transcript levels by RT-qPCR (now included in Sup. Fig. S4) from day 2 to day 5 post siRNA treatment. We observed that the time point at which OGT intron 4 removal increases (day 2) coincides with the time at which SFSWAP transcript levels start decrease, consistent with a direct effect of SFSWAP knockdown on OGT intron 4 splicing. Moreover, the effect of SFSWAP knockdown on OGT intron 4 splicing peaks between day 4-5, supporting our use of these longer time points to cast a wide net for SFSWAP targets.

(2) The mechanistic role of SFSWAP in splicing would benefit from further exploration. Key questions remain, such as whether SFSWAP directly binds RNA, specifically the introns and exons (including the decoy exons) it appears to regulate. Furthermore, given that SFSWAP phosphorylation is influenced by changes in O-GlcNAc signaling, it would be interesting to investigate this relationship further. While generating specific phosphomutants may not yield definitive insights due to redundancy and also beyond the scope of the study, the authors could examine whether distinct SFSWAP domains, such as the SR and SURP domains, which likely overlap with phosphorylation sites, are necessary for regulating OGT intron 4 splicing. 

We absolutely agree with the reviewer that the current work stops short of a detailed mechanistic study, and we have made every attempt to be circumspect in our interpretations to reflect that limitation. In addition, we are very interested in delving more deeply into the mechanistic aspects of this regulation. In fact, we have initiated many of the experiments suggested by the reviewer (and more), but in each case, rigorous interpretable results will require a minimum another year’s time. 

For example, we have used crosslinking and biotin labeling techniques (using previously available reagents from Eclipsebio) to test whether SFSWAP binds RNA. The results were negative, but the lack of strong SFSWAP antibodies required that we use a transiently expressed myc-tagged SFSWAP. Therefore, this negative result could be an artifact of the exogenous expression and/or tagging. Given the difficulties of “proving the negative”, considerably more work will be required to substantiate this finding. As another example, we intend to develop a complementation assay as suggested. For an essential gene, the ideal complementation system employs a degron system, and we have spent months attempting to generate a homozygous AID-tagged SFSWAP. Unfortunately, we so far have only found heterozygotes. Of course, this could be because the tag interferes with function, the insert was not efficiently incorporated by homologous repair, or that we simply haven’t yet screened a sufficient number of clones. We’re confident that these technical issues that can be addressed, but they will take a significant amount of time to resolve. While we would ideally define a mechanism, we think that the data reported here outlining functions for SFSWAP in splicing represent a body of work sufficient for publication. 

(3) Data presentation could be improved (specific suggestions are included in the recommendations section). Furthermore, Excel tables with gene expression and splicing analysis results should be provided as supplementary datasheets. Finally, a more detailed explanation of statistical analyses is necessary in certain sections. 

We have addressed all specific suggestions as detailed in the recommendations below.

Reviewer #2 (Public review): 

Summary: 

The paper describes an effort to identify the factors responsible for intron retention and alternate exon splicing in a complex system known to be regulated by the O-GlcNAc cycling system. The CRISPR/Cas9 system was used to identify potential factors. The bioinformatic analysis is sophisticated and compelling. The conclusions are of general interest and advance the field significantly. 

Strengths: 

(1) Exhaustive analysis of potential splicing factors in an unbiased screen. 

(2) Extensive genome wide bioinformatic analysis. 

(3) Thoughtful discussion and literature survey. 

We thank the reviewer for recognizing and detailing the strengths of our manuscript. 

Weaknesses: 

(1) No firm evidence linking SFSWAP to an O-GlcNAc specific mechanism. 

We couldn’t agree more with this critique. Indeed, our intention at the outset for the screen was to find an O-GlcNAc sensor linking OGT splicing with O-GlcNAc levels. As often occurs with high-throughput screens, we didn’t find exactly what we were looking for, but the screen nonetheless pointed us to interesting biology. Prompted by our screen, we describe new insights into the function of SFSWAP a relatively uncharacterized essential gene. Currently, we are testing other candidates from our screen, and we are performing additional studies to identify potential O-GlcNAc sensors.  

(2) Resulting model leaves many unanswered questions. 

We agree (see Reviewer 1, point 2 response).  

Reviewer #3 (Public review): 

Summary: 

The major novel finding in this study is that SFSWAP, a splicing factor containing an RS domain but no canonical RNA binding domain, functions as a negative regulator of splicing. More specifically, it promotes retention of specific introns in a wide variety of transcripts including transcripts from the OGT gene previously studied by the Conrad lab. The balance between OGT intron retention and OGT complete splicing is an important regulator of O-GlcNAc expression levels in cells. 

Strengths: 

An elegant CRISPR knockout screen employed a GFP reporter, in which GFP is efficiently expressed only when the OGT retained intron is removed (so that the transcript will be exported from the nucleus to allow for translation of GFP). Factors whose CRISPR knockdown causes decreased intron retention therefore increase GFP, and can be identified by sequencing RNA of GFP-sorted cells. SFSWAP was thus convincingly identified as a negative regulator of OGT retained intron splicing. More focused studies of OGT intron retention indicate that it may function by regulating a decoy exon previously identified in the intron, and that this may extend to other transcripts with decoy exons. 

We thank the reviewer for recognizing the strengths of our manuscript. 

Weaknesses: 

The mechanism by which SFSWAP represses retained introns is unclear, although some data suggests it can operate (in OGT) at the level of a recently reported decoy exon within that intron.

Interesting/appropriate speculation about possible mechanisms are provided and will likely be the subject of future studies. 

We completely agree that this is a limitation of the current study (see above). Now that we have a better understanding of SFSWAP functions, we will continue to explore SFSWAP mechanisms as suggested. 

Overall the study is well done and carefully described but some figures and some experiments should be described in more detail. 

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors): 

(1) Clarify and add missing statistical details across the figures. For example, Figure S2 lacks statistical comparisons, and in Figures 4A and 4C the tests applied should be specified in the legend. 

We have added appropriate statistical analysis wherever missing and edited figure legends to specify the tests used.

(2) The authors are strongly encouraged to provide detailed tables of gene expression and alternative splicing analyses from RNA-Seq experiments (e.g., edgeR, rMATS, Whippet, and MAJIQ), as this would enhance transparency and facilitate data interpretation. 

We have added tables for gene expression and alternate splicing analysis as suggested (Suppl. tables 3-

6).

(3) Although the legend sometimes indicates differently (e.g., Figure 3b, 5a, 5c, etc), the volcano plots showing the splicing changes do not contain a cutoff for marginally differential percent spliced in or intron retention values. 

The legends have been edited to reflect the correct statistical and/or PSI cutoffs.

(4) For consistency, use a consistent volcano plot format across all relevant figures (Figures 3b, 5a-c, S3, S4, S7, and S8), including cutoffs for differential splicing and the total count of up- and down-regulated events. 

Due to different statistical frameworks and calculations employed by different alternate splicing pipelines, we could not use the same cutoffs for different pipelines.  However, we have now indicated the number of up- and down-regulated events for consistency among the volcano plots.

(5) What is the overlap of differentially regulated events between the different analytical methodologies applied? 

We analyzed the degree of overlap between the three pipelines used in the paper using a Venn diagram (added to Suppl. Fig. S7). However, as widely reported in literature (e.g., Olofsson et al., 2023; Biochem Biophys Res Commun. 2023; doi: 10.1016/j.bbrc.2023.02.053.), the degree of overlap between pipelines is quite low.

(6) To further substantiate your conclusions, additional validations of RNA-Seq splicing data, ideally visualized on an agarose gel, would be valuable, especially for exons and introns regulated by SFSWAP, and particularly for OGT decoy exons in Figure 4c. 

We have not included these experiments as we focused on other critiques for this resubmission. Because the RNA-seq, RT-PCR and RT-qPCR data all align, we are confident that the products we are seeing are correctly identified and orthogonally validated (Figs 2d, 4a, 4b, and 4c).  

(7) It would be more informative if the CRISPR screen data were presented in a format where both the adjusted p-value and LFC values of the hits are presented. Perhaps a volcano plot? 

We have now included these graphs in revised Supplementary Figure S2. 

(8) In Figure 2d, a cartoon showing primer binding sites for each panel could aid interpretation, particularly in explaining the unexpected simultaneous increase in OGT mRNA and intron retention upon SFSWAP knockdown. 

We have added a cartoon showing primer binding sites similar to that shown in Fig. 4a.

(9) Page 9, line 1, states that SFSWAP autoregulates its expression by controlling intron retention. Including a Sashimi plot would provide visual support for this claim. 

The data suggesting that SFSWAP autoregulates its own transcript abundance were reported in Zachar et al. (1994), not from our own studies. Validation of those data with our RNA-seq data is confounded by the fact that we are using siRNAs to knockdown the SFSWAP RNA at the transcript level (Fig. S15). 

(10) In the legend of Figure S2 the authors state that negative results are inconclusive because RNA knockdowns are not verified by western blotting or qRT-PCR. This is correct, but the reviewer would also argue that the positive results are also inconclusive as they are not supported by a rescue experiment to confirm that the effect is not due to off-target effects. 

This is a fair point with respect to the siRNA experiments on their own. However, the CRISPR screen was performed with sgRNAs, and MAGeCK RRA scores are high only for those genes that have multiple sgRNAs that up-regulate the gene. Examination of the SFSWAP sgRNAs individually shows that three of four SFSWAP sgRNAs had false discovery rates ≤10^-42 for GFP upregulation. Thus, the siRNAs provide an additional orthogonal approach. It seems unlikely that the siRNAs, and three independent sgRNAs will have the same off-target results. Thus, these combined observations support the conclusion that SFSWAP loss leads to decreased OGT intron retention.  

(11) For clarity in Figure 3a, consider using differential % spliced in or intron retention bar plots with directionality (positive and negative axis) and labeling siSFSWAP as the primary condition. 

(12) Consider presenting Figure 5D as a box plot with a Wilcoxon test for statistical comparison. 

For both points 11 and 12, we have tried the graphs as the reviewer suggested. While these were good suggestions, in both cases we felt that the original plots ended up presenting a clearer presentation of the data (see Author response image 1).

Author response image 1.

(13) Please expand the Methods section to detail the Whippet and MAJIQ analyses. 

We have expanded the methods section to include additional details of the alternate splicing analysis.

(14) Include coordinates for the four possible OGT decoy exon combinations analyzed in the Methods section. 

We have added the coordinates of all four decoy forms in the methods section.  

(15) A section on SFSWAP mass spectrometry is listed in Methods but is missing from the manuscript. 

This section has now been removed.

Reviewer #2 (Recommendations for the authors): 

This is an excellent contribution. The paper describes an effort to identify the factors responsible for intron retention and alternate exon splicing in a complex system known to be regulated by the O-GlcNAc cycling system. The CRISPR/Cas9 system was used to identify potential factors. The bioinformatic analysis is sophisticated and compelling. The conclusions are of general interest and advance the field significantly. 

Some specific recommendations. 

(1) The plots in Figure 3 describing SI and ES events are confusing to this reader. Perhaps the violin plot is not the best way to visualize these events. The same holds true for the histograms in the lower panel of Figure 3. Not sure what to make of these plots. 

For Figure 3b, we include both scatter and violin plots to represent the same data in two distinct ways. For Figure 3d, we agree that these are not the simplest plots to understand, and we have spent significant time trying to come up with a better way of displaying these trends in GC content as they relate to SE and RI events. Unfortunately, we were unable to identify a clearer way to present these data. 

(2) The model (Figure 6) is very useful but confusing. The legend and the Figure itself are somewhat inconsistent. The bottom line of the figure is apparent but I fear that the authors are trying to convey a more complete model than is apparent from this figure. Please revise. 

We have simplified the figure from the previous submission. As mentioned above, we admit that mechanistic details remain unknown. However, we have tried to generate a model that reflects our data, adds some speculative elements to be tested in the future, but remains as simple as possible. We are not quite sure what the reviewer was referring to as “somewhat inconsistent”, but we have attempted to clarify the model in the revised Discussion and Figure legend.  

(3) It is unclear how normalization of the RNA seq experiments was performed (eg. Figure S5 and 6).  

The normalization differences in Fig. S5 and S6 (now Fig S8 and S9) were due to scaling differences during the use of rmats2sashimiplot software. We have now replaced Fig. S5 to reflect correctly scaled images.

I am enthusiastic about the manuscript and feel that with some clarification it will be an important contribution. 

Thank you for these positive comments about our study!

Reviewer #3 (Recommendations for the authors): 

(1) In Figure 1f, it is clear that siRNA-mediated knockdown of OGT greatly increases spliced RNA as the cells attempt to compensate by more efficient intron removal (three left lanes). However, there is no discussion of the various treatments with TG or OSMI. Might quantitation of these lanes not also show the desired effects of TG and OSMI on spliced transcript levels? 

The strong effect of OGT knockdown masks the (comparatively modest) effects of subsequent inhibitor treatments on the reporter RNA. We have edited the results section to clarify this.

(2) In Figure 2c, why is the size difference between spliced RNA and intron-retained RNA so different in the GFP-probed gel (right) compared with the OGT-probed gel (left)? Even recognizing that the GFP probe is directed against reporter transcripts, and the OGT probe (I think) is directed against endogenous OGT transcripts, shouldn't the difference between spliced and unspliced bands be the same, i.e., +/- the intron 4 sequence. Also, why does the GFP probe detect the unspliced transcript so poorly? 

The fully spliced endogenous OGT mRNA is ~5.5 kb while the fully spliced reporter is only ~1.6kb, so the difference in size (the apparent shift relative to the mRNA) is quite different. Moreover, the two panels in Fig 2c are not precisely scaled to one another, so direct comparisons cannot be made. 

The intron retained isoform does not accumulate to high levels in this reporter, a phenotype that we also observed with our GFP reporter designed to probe the regulation of the MAT2A retained intron (Scarborough et al., 2021). We are not certain about the reason for these observations, but suspect that the reporter RNA’s retained intron isoforms are less stable in the nucleus than their endogenous counterparts. Alternatively, the lack of splicing may affect 3´ processing of the transcripts so that they do not accumulate to the high levels observed for the wild-type genes. 

(3) Please provide more information about the RNA-seq experiments. How many replicates were performed under each of the various conditions? The methods section says three replicates were performed for the UPF1/TG experiments; was this also true for the SFSWAP experiments?  

All RNA-seq experiments were performed in biological triplicates. We have edited the methods section to clarify this.

(4) Relatedly, the several IGV screenshots shown in Figure 3C presumably represent the triplicate RNA seq experiments. In part D, how many experiments does the data represent? Is it a compilation of three experiments? 

Fig. 3d is derived from alternate splicing analysis performed on three biological replicates. We have added the number of replicates (n=3) on the figure to clarify this. We have also noted that the three IGV tracks represent biological replicates in the Figure legend for 3c.  

(5) Please provide more details regarding the qRT-PCR experiments. 

We have provided the positions of primer sets used for RT-qPCR analysis and cartoon depictions of target sites below the data wherever appropriate.

(6) In the discussion of decoy exon function (in the Discussion section), several relevant observations are cited to support a model in which decoy exons promote assembly of splicing factors. One might also cite the finding that eCLIP profiling has found enriched binding of U2AF1 and U2AF2 at the 5' splice site region of decoy exons (reference 16). 

Excellent point. This has now been added to the Discussion. 

Minor corrections / clarifications: 

(1) In the Figure 2A legend, CRISPR is misspelled. 

Corrected.

(2) In the discussion, the phrase "indirectly inhibits splicing of exons 4 and 5, but promoting stable unproductive assembly of the spliceosome", the word "but" should probably be "by". 

Corrected.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #2 (Public review):

The authors have constructively responded to previous referee comments and I believe that the manuscript is a useful addition to the literature. I particularly appreciate the quantitative approach to social behavior, but have two cautionary comments.

(1) Conceptually it is important to further justify why this particular maximum entropy model is appropriate. Maximum entropy models have been applied across a dizzying array of biological systems, including genes, neurons, the immune system, as well as animal behavior, so would seem quite beneficial to explain the particular benefits here, for mouse social behavior as coarse-grained through the eco-hab chamber occupancy. This would be an excellent chance to amplify what the models can offer for biological understanding, particularly in the realm of social behavior

We thank the reviewer for this comment. Maximum entropy models, along with other statistical inference methods that learn interaction patterns from simultaneously-measured degrees of freedom, help distinguish various types of interactions, e.g. direct vs. indirect interactions among animals, individual preference to food vs. social interaction with pairs. As research on social behavior expands from focusing on pairs of animals to studying groups in (semi-)naturalistic environments, maximum entropy models serve as a crucial link between high-throughput data and the need to identify and distinguish interaction rules. Specifically, among all possible maximum entropy models, the pairwise maximum entropy model is one of the simplest that can describe interactions among individuals, which serves as an excellent starting point to understand collective and social behavior in animals.

Although the Eco-HAB setup currently records spatially coarse-grained data, it still provides more spatial information compared to the traditional three-chamber tests used to assess sociability for rodents. By showing that the maximum entropy model can effectively analyze Eco-HAB data, we hope to highlight its potential in research of social behavior in animals.

To amplify what the models can offer for biological understanding particularly in the realm of social behavior, We have updated the Introduction to add a more logical structure to the need of using maximum entropy models to identify interactions among mice. Additionally, we updated the first paragraph of the Discussion to make it specific that it is the use of maximum entropy models that identifies interaction patterns from the high-throughput data. Finally, we have also added in the Discussion (line 422-425) arguments supporting the specific use of pairwise maximum entropy models to study social behaviors.

(2) Maximum entropy models of even intermediate size systems involve a large number of parameters. The authors are transparent about that limitation here, but I still worry that the conclusion of the sufficiency of pairwise interactions is simply not general, and this may also relate to the differences from previous work. If, as the authors suggest in the discussion, this difference is one of a choice of variables, then that point could be emphasized. The suggestion of a follow up study with a smaller number of mice is excellent.

We thank the reviewer for raising the issue and agree that the caveat of how general pairwise interactions can describe social behavior of animals needs to be discussed. We have added a sentence in the Discussion to point out this important caveat. “More generally, this discrepancy when looking at different choices of variables raises the issue that when studying social behavior of animals in a group, it is important to test and compare interaction models with different complexity (e.g. pairwise or with higher-order interactions).” We have also toned down our conclusion to limit our results of pairwise interactions describing mice co-localization patterns to the data collected in Eco-HAB (also see Reviewer 3 Major Point 2).

Reviewer #3 (Public review):

Summary:

Chen et al. present a thorough statistical analysis of social interactions, more precisely, co-occupying the same chamber in the Eco-HAB measurement system. They also test the effect of manipulating the prelimbic cortex by using TIMP-1 that inhibits the MMP-9 matrix metalloproteinase. They conclude that altering neural plasticity in the prelimbic cortex does not eliminate social interactions, but it strongly impacts social information transmission.

Strengths:

The quantitative approach to analyzing social interactions is laudable and the study is interesting. It demonstrates that the Eco-HAB can be used for high throughput, standardized and automated tests of the effects of brain manipulations on social structure in large groups of mice.

Weaknesses:

A demonstration of TIMP-1 impairing neural plasticity specifically in the prelimbic cortex of the treated animals would greatly strengthen the biological conclusions. The Eco-HAB provides coarser spatial information compared to some other approaches, which may influence the conclusions.

Recommendations for the authors:  

Reviewer #3 (Recommendations for the authors):

Major points

(1) Do the Authors have evidence that TIMP-1 was effective, as well as specific to the prelimbic cortex?

We refer to the literature for the effectiveness and specificity of TIMP-1 to the prelimbic cortex.

Specifically, the study by Okulski et al. (Biol. Psychiatry 2007) provides clear evidence that TIMP1 plays a role in synaptic plasticity in the prefrontal cortex. They showed that TIMP-1 is induced in the medial prefrontal cortex (mPFC) following stimulation that triggers late long-term potentiation (LTP), a key model of synaptic plasticity. Overexpression of TIMP-1 in the mPFC blocked the activity of matrix metalloproteinases (MMPs) and prevented the induction of late LTP in vivo. Similar effects were observed with pharmacological inhibition of MMP-9 in vitro, reinforcing the idea that TIMP-1 regulates extracellular proteolysis as part of the plasticity mechanism in the prefrontal cortex. These findings confirm that TIMP-1 is both effective and active in this specific brain region.

Further evidence comes from Puścian et al. (Mol. Psychiatry 2022), who used TIMP-1-loaded nanoparticles to influence neuronal plasticity in the amygdala. They found that TIMP-1 affected MMP expression, LTP, and dendritic morphology, showing its impact on synaptic modifications. More directly relevant, Winiarski et al. (Sci. Adv. 2025) demonstrated that injecting TIMP-1-loaded nanoparticles into the prelimbic cortex altered responses to social stimuli, further supporting the idea that TIMP-1 has region-specific effects on behavioral processes.

We have also updated the main text (page 8, 1st paragraph of “Effect of impairing neuronal plasticity in the PL on subterritory preferences and sociability”) of the manuscript to include the above references.

(2) The Authors seem to suggest that one main reason for the different results compared to Shemesh et al. 2013 was the coarseness of the Eco-HAB data. In this case, I think this conclusion should be toned down because of this significant caveat.

We thank the reviewer for pointing this out, and agree that this caveat and difference should be emphasized. To tone down the conclusion, we have

(1) added details about the Eco-HAB (it being coarse-grained, etc.) in the abstract to tone down the conclusion.

(2) added to the results summary in the Discussion (top of page 12) that the results are “within in the setup of the semi-naturalistic Eco-HAB experiments”

(3) added to the Discussion (page 13) that the different results compared to Shemesh et al 2013 means that general studies of social behavior need to compare models with different levels of complexity (e.g. pairwise vs. higher-order interactions). (Also see Reviewer 2 Comment 2.)

Minor points

(1) Please explain what is measured in Fig. 1C (what is on the y axis?).

Figure 1C shows the activity of the mice as measured by the rate of transitions, i.e. the number of times the mice switch boxes during each hour of the day, averaged over all N = 15 mice and T = 10 days (cohort M1). The error bars represent variability of activities across individuals or across days. For mouse-to-mouse variability (blue), we first compute for each mouse its number of transitions averaged over the same hour for all 10 days, then we compute its standard deviation across all 15 mice and plot it as error bars. For day-to-day variability (orange), we first compute for each day the number of transitions for each hour averaged over all mice, then compute its standard deviation across all 10 days as the errorbar. We have added the detailed explanation in the caption of Figure 1C.

(2) In Fig. 3, it would be better to present the control group also in the main figure instead of the supplementary.

We have merged Figure 3 and Figure 3 Supplementary 1 to present the control group also in the main figure.

(3) In Fig. 3 and corresponding supplements, there seems to be a large difference between males and females. I think this would deserve some more discussion.

While not being the main focus of this paper, we agree with the reviewer that the difference between male and female is important and deserves attention in the discussion and also future study. Thus we have added a paragraph in the Discussion (line 394-399, bottom of page 12).

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

The authors set out to analyse the roles of the teichoic acids of Streptococcus pneumoniae in supporting the maintenance of the periplasmic region. Previous work has proposed the periplasm to be present in Gram positive bacteria and here advanced electron microscopy approach was used. This also showed a likely role for both wall and lipo-teichoic acids in maintaining the periplasm. Next, the authors use a metabolic labelling approach to analyse the teichoic acids. This is a clear strength as this method cannot be used for most other well studied organisms. The labelling was coupled with super-resolution microscopy to be able to map the teichoic acids at the subcellular level and a series of gel separation experiments to unravel the nature of the teichoic acids and the contribution of genes previously proposed to be required for their display. The manuscript could be an important addition to the field but there are a number of technical issues which somewhat undermine the conclusions drawn at the moment. These are shown below and should be addressed. More minor points are covered in the private Recommendations for Authors.

Weaknesses to be addressed:

(1) l. 144 Was there really only one sample that gave this resolution? Biological repeats of all experiments are required.

CEMOVIS is a very challenging method that is not amenable to numerous repeats. However, multiple images were recorded from at least two independent samples for each strain. Additional sample images are shown in a new Fig. S3.

CETOVIS is even more challenging (only two publications in Pubmed since 2015) and was performed on a single ultrathin section that, exceptionally, laid perfectly flat on the EM grid, allowing tomography data acquisition on ∆tacL cells. The reconstructed tomogram confirmed the absence of a granular layer in the depth of the section. Additionally, the numbering of Fig. S4A-B (previously misidentified as Fig. S2A-B) has been corrected in the text of V2.

(2) Fig. 4A. Is the pellet recovered at "low" speeds not just some of the membrane that would sediment at this speed with or without LTA? Can a control be done using an integral membrane protein and Western Blot? Using the tacL mutant would show the behaviour of membranes alone.

We think that the pellet is not just some of the membrane but most of it. In support of this view, the “low” speed pellets after enzymatic cell lysis contain not just some membrane lipids, but most of them (Fig. S10A). We therefore expect membrane proteins to be also present in this fraction. We performed a Western blot using antibodies against the membrane protein PBP2x (new Fig. S7C). Unfortunately, no signal was detected most likely due to protein degradation from contaminant proteases that we could trace to the purchased mutanolysin. The same sedimentation properties were observed with the ∆tacL strain as shown in Fig. 6A. However, in the ∆tacL strain the membrane pellet still contains membrane-bound TA precursors. It is therefore impossible to test definitely if pneumococcal membranes totally devoid of TA would sediment in the same way.

(3) Fig. 4A. Using enzymatic digestion of the cell wall and then sedimentation will allow cell wall associated proteins (and other material) to become bound to the membranes and potentially effect sedimentation properties. This is what is in fact suggested by the authors (l. 1000, Fig. S6). In order to determine if the sedimentation properties observed are due to an artefact of the lysis conditions a physical breakage of the cells, using a French Press, should be carried out and then membranes purified by differential centrifugation. This is a standard, and well-established method (low-speed to remove debris and high-speed to sediment membranes) that has been used for S. pneumoniae over many years but would seem counter to the results in the current manuscript (for instance Hakenbeck, R. and Kohiyama, M. (1982), Purification of Penicillin-Binding Protein 3 from Streptococcus pneumoniae. European Journal of Biochemistry, 127: 231-236).

Thank you for this suggestion. We have tested this hypothesis by breaking cells with a Microfluidizer followed by differential centrifugation. This experiment, which requires an important minimal volume, was performed with unlabeled cells (due to the cost of reagents) and assessed by Western blot using antibodies against the membrane protein PBP2x (new Fig. S7C). In this case, the majority of the membrane material was found in the high-speed pellet, as expected.

We also applied the spheroplast lysis procedure of Flores-Kim et al. to the labeled cells, and found that most of the labeled material sedimented at low speed (new Fig. S7B), as observed with our own procedure.

With these new results, the section on membrane density has been removed from the Supplementary Information. Instead, the fractionation is further discussed in terms of size of membrane fragments and presence of intact spheroplasts in the notes in Supplementary Information preceding Fig. S7.

(4) l. 303-305. The authors suggest that the observed LTA-like bands disappear in a pulse chase experiment (Fig. 6B). What is the difference between this and Fig. 5B, where the bands do not disappear? Fig. 5C is the WT and was only pulse labelled for 5 min and so would one not expect the LTA-like bands to disappear as in 6B?

Fig. 6B shows a pulse-chase experiment with strain ∆tacL, whereas Fig. 5C shows a similar experiment with the parental WT strain. The disappearance of the LTA-like band pattern with the ∆tacL strain (Fig. 6B), and their persistence in the WT strain (Fig. 5C), indicate that these bands are the undecaprenyl-linked TA in ∆tacL and proper LTA in the WT. A sentence has been added to better explain this point in V2.

Note that we have exchanged the previous Fig. 5C and Fig. S13B, so that the experiments of Fig. 5A and 5C are in the same medium, as suggested by Reviewer #2.

(5) Fig. 6B, l. 243-269 and l. 398-410. If, as stated, most of the LTA-like bands are actually precursor then how can the quantification of LTA stand as stated in the text? The "Titration of Cellular TA" section should be re-evaluated or removed? If you compare Fig. 6C WT extract incubated at RT and 110oC it seems like a large decrease in amount of material at the higher temperature. Thus, the WT has a lot of precursors in the membrane? This needs to be quantified.

Indeed, the quantification of the ratio of LTA and WTA in the WT strain rests on the assumption that the amount of membrane-linked polymerized TA precursors is negligible in this strain. This assumption is now stated in the Titration section. We think it is the case. The true LTA and TA precursors do not have exactly the same electrophoretic mobility, being shifted relative to each other by about half a ladder “step”. This difference is visible when samples are run in adjacent lanes on the same gel, as in the new Fig. 6C. The difference of migration was well documented in the original paper about the deletion of tacL, although tacL was known as rafX at that time, and the ladders were misidentified as WTA (Wu et al. 2014. A novel protein, RafX, is important for common cell wall polysaccharide biosynthesis in Streptococcus pneumoniae: implications for bacterial virulence. J Bacteriol. 196, 3324-34. doi: 10.1128/JB.01696-14). This reference was added in V2. The experiment in the new Fig. 6C was repeated to have all samples on the same gel and treated at a lower temperature. The minor effect on the amount of LTA when WT cells are heated at pH 4.2 may be due to the removal of some labeled phosphocholine. We have NMR evidence that the phosphocholine in position D is labile to acidic treatment of LTA, which may lack in some cases, as reported by Hess et al. (Nat Commun. 2017 Dec 12;8(1):2093. doi: 10.1038/s41467-017-01720-z).

(6) L. 339-351, Fig. 6A. A single lane on a gel is not very convincing as to the role of LytR. Here, and throughout the manuscript, wherever statements concerning levels of material are made, quantification needs to be done over appropriate numbers of repeats and with densitometry data shown in SI.

Yes indeed. Apart from the titration of TA in the WT strain, we haven’t yet carried out a thorough quantification of TA or LTA/WTA ratio in different strains and conditions, although we intend to do so in a follow-up study, using the novel opportunities offered by the method presented here.

However, to better substantiate our statement regarding the ∆lytR strain, we have quantified two experiments performed in C-medium with azido-choline, and two experiments of pulse labeling in BHI medium. The results are presented in the additional supplementary Fig. S14. The value of 51% was a calculation error, and was corrected to 41%. Likewise, the decrease in the WTA/LTA ratio was corrected to 5 to 7-fold.

(7) 14. l. 385-391. Contrary to the statement in the text, the zwitterionic TA will have associated counterions that result in net neutrality. It will just have both -ve and +ve counterions in equal amounts (dependent on their valency), which doesn't matter if it is doing the job of balancing osmolarity (rather than charge).

Thank you for pointing out this point. The paragraph has been corrected in V2.

Reviewer #2 (Public review):

The Gram-positive cell wall contains for a large part of TAs, and is essential for most bacteria. However, TA biosynthesis and regulation is highly understudied because of the difficulties in working with these molecules. This study closes some of our important knowledge gaps related to this and provides new and improved methods to study TAs. It also shows an interesting role for TAs in maintaining a 'periplasmic space' in Gram positives. Overall, this is an important piece of work. It would have been more satisfying if the possible causal link between TAs and periplasmic space would have been more deeply investigated with complemented mutants and CEMOVIS. For the moment, there is clearly something happening but it is not clear if this only happens in TA mutants or also in strains with capsules/without capsules and in PG mutants, or in lafB (essential for production of another glycolipid) mutants. Finally, some very strong statements are made suggesting several papers in the literature are incorrect, without actually providing any substantiation/evidence supporting these claims. Nevertheless, I support the publication of this work as it pioneers some new methods that will definitively move the field forward.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) l. 55 It is stated that TA are generally not essential. This needs to be introduced in a little more detail as in several species they are collectively. Need some more references here to give context.

We have expended the paragraph and added a selection of references in V2.

(2) l. 63 and Fig. 1A. Is the model based on the images from this paper? Is the periplasm as thick as the peptidoglycan layer? Would you not expect the density of WTA to be the same throughout the wall, rather than less inside? Do the authors think that the TA are present as rods in the cell envelope and because of this the periplasm looks a little like a bilayer, is this so? Is the relative thickness of the layers based on the data in the paper (Table 1)?

The model proposed in Fig. 1A is not based on our data. It is a representation of the model proposed by Harold Erickson, and the appropriate reference has been added to the figure legend in V2. We do not speculate on the relative density of WTA inside the peptidoglycan layer, at the surface or in the periplasm. The only constraint from the model is that the density of WTA in the periplasm should be sufficient for self-exclusion and allow the brush polymer theory to apply. The legend has been amended in V2.

We indeed think that the bilayer appearance of the periplasmic space in the wild type strain, and the single layer periplasmic space in the ∆tacL and ∆lytR support the Erickson’s model. Although the model was drawn arbitrarily, it turns out that the relative thickness of the peptidoglycan and periplasmic scale is in rough agreement with the measurements reported in Table 1.

(3) Fig. 2. It is hard to orient oneself to see the layers. The use of the term periplasmic space (l. 132) and throughout is probably not wise as it is not a space.

We prefer to retain this nomenclature since the term periplasmic space has been used in all the cell envelope CEMOVIS publications and is at the core of Erickson’s hypothesis about these observations and teichoic acids.

(4) L. 147. This is not referring to Fig. S2A-B as suggested but Fig. S3A-B.

This has been corrected.

(5) l. 148. How do you know the densities observed are due to PG or certainly PG alone? Perhaps it is better to call this the cell wall.

Yes. Cell wall is a better nomenclature and the text and Table 1 have been corrected in V2, in accordance with Fig. 2.

(6) l. 165. It is also worth noting that peripheral cell wall synthesis also happens at the same site so this may well not be just division.

Yes. We have replaced “division site” by “mid-cell” in V2.

(7) l. 214 What is the debris? If PG digestion has been successful then there will be marginal debris. Is this pellet translucent (like membranes)? If you use fluorescently labelled PG in the preparation has it all disappeared, as would be expected by fully digested and solubilised material?

In traditional protocols of bacterial membrane preparation, a low-speed centrifugation is first performed to discard “debris” that to our knowledge have not been well characterized but are thought to consist of unbroken cells and large fragments of cell wall. After enzymatic degradation of the pneumococcal cell wall, the low-speed pellet is not translucent as in typical membrane pellets after ultracentrifugation, but is rather loose, unlike a dense pellet of unbroken cells. A description of the pellet appearance was added in V2.

It is a good idea to check if some labeled PG is also pelleted at low-speed after digestion. In a double labeling experiment using azido-choline and a novel unpublished metabolic probe of the PG, we found that the PG was fully digested and labeled fragments migrated as a couple of fuzzy bands likely corresponding to different labeled peptides. These species were not pelleted at low speed.

(8) l. 219. Can you give a reference to certify that the low mobility material is WTA? Why does it migrate differently than LTA? Or is the PG digestion not efficient?

WTA released from sacculi by alkaline lysis were found to migrate as a smear at the top of native gels revealed by alcian-blue silver staining, which is incompatible with SDS (Flores-Kim, 2019, 2022). The references have be added in V2. It could be argued in this case that the smearing was due to partial degradation of the WTA by the alkaline treatment.

Bui et al. (2012) reported the preparation of WTA by enzymatic digestion of sacculi, but the resulting WTA were without muropeptide, presumably due to a step of boiling at pH 5 used to deactivate the enzymes.

To our knowledge, this is the first report of pneumococcal WTA prepared by digestion of sacculi and analyzed by SDS-PAGE. Since the migration of WTA in native and SDS-PAGE is similar, we hypothesize that they do not interact significantly with the dodecyl sulphate, in contrast to the LTA, which bear a lipidic moiety. The fuzziness of the WTA migration pattern may also result from the greater heterogeneity due to the attached muropeptide, such as different lengths (di-, tetra-saccharide…), different peptides despite the action of LytA (tri-, tetra-peptide…), different O-acetylation status, etc.

(9) L. 226-227, Fig S8. Presumably several of the major bands on the Coomassie stained gel are the lysozyme, mutanolysin, recombinant LytA, DNase and RNase used to digest the cell wall etc.? Can the sizes of these proteins be marked on the gel. Do any of them come down with the material at low-speed centrifugation?

We have provided a gel showing the different enzymes individually and mixed (new Fig. S9G). While performing several experiments of this type, we found that the mutanolysin might be contaminated with proteases. The enzymes do not appear to sediment at low speed.

(10) Fig. S9B. It is difficult to interpret what is in the image as there appear to be 2 populations of material (grey and sometimes more raised). Does the 20,000 g material look the same?

Fig. S10B is a 20,000 × g pellet. We agree that there appears to be two types of membrane vesicles, but we do not know their nature.

(11) l. 277 and Fig. 5A. Why is it "remarkable" that there are apparently more longer LTA molecules as the cell reach stationary phase?

This is the first time that a change of TA length is documented. Such a change could conceivably have consequences in the binding and activity of CBPs and the physiology of the cell envelope in general. These questions should be adressed in future studies.

(12) l. 280. How do you know which is the 6-repeat unit?

It is an assumption based on previous analyses by Gisch et al.( J Biol Chem 2013, 288(22):15654-67. doi: 10.1074/jbc.M112.446963). The reference was added.

(13) Fig. 5A and C. Panel C, the cells were grown in a different medium and so are not comparable to Panel A. Why is Fig. S12B not substituted for 5B? Presumably these are exponential phase cells.

We have interverted the Fig. S13B and 5C in V2, as suggested, and changed the text and legends accordingly.

Reviewer #2 (Recommendations for the authors):

L30: vitreous sections?

Corrected in V2.

L32: as their main universal function --> as a universal function. To show it's the main universal function, you will need to look at this across various bacterial species.

Changed to “possible universal function” in V2.

L35: enabled the titration the actual --> titration of the actual?

Corrected in V2.

L34: consider breaking up this very long sentence.

Done in V2.

L37: may compensate the absence--> may compensate for the absence.

Corrected in V2.

L45: Using metabolic labeling and electrophoresis showed --> Metabolic labeling and...

Corrected in V2.

L46: This finding casts doubts on previous results, since most LTA were likely unknowingly discarded in these studies. This needs to be rephrased and is unnecessarily callous. While the current work casts doubts on any quantitative assessments of actual LTA levels measured in previous studies, it does not mean any qualitative assessments or conclusions drawn from these experiments are wrong. Better would be to say: These findings suggest that previously reported quantitative assessments of LTA levels are likely underestimating actual LTA levels, since much of the LTA would have been unknowingly discarded.

If the authors do think that actual conclusions are wrong in previous work, then they need to be more explicit and explain why they were wrong.

Yes indeed. The statement was toned down in V2.

L55: Although generally non-essential. I would remove or rephrase this statement. I don't think any TA mutant will survive out in the wild and will be essential under a certain condition. So perhaps not essential for growth under ideal conditions, but for the rest pretty essential.

The paragraph was amended by qualifying the essentiality to laboratory conditions and including selected references.

L95: Note that the prevailing model until reference 20 (Gibson and Veening) was that the TA is polymerized intracellularly (see e.g. Figure 2 of PMID: 22432701, DOI: 10.1089/mdr.2012.0026). This intracellular polymerisation model seemed unlikely according to Gibson and Veening ('As TarP is classified by PFAM as a Wzy-type polymerase with predicted active site outside the cell, we speculate that TarP and TarQ polymerize the TA extracellularly in contrast to previous reports.'), but there is no experimental evidence as far as this referee knows of either model being correct.

Despite the lack of experimental evidence, we think that Gibson and Veening are very likely correct, based on their argument, and also by analogy with the synthesis of other surface polysaccharides from undecaprenyl- or dolichol-linked precursors. It is unfortunate that Figure 2 of PMID: 22432701, DOI: 10.1089/mdr.2012.0026 was published in this way, since there was no evidence for a cytoplasmic polymerization, to our knowledge.

L97: It is commonly believed, although I'm not sure it has ever been shown, that the capsule is covalently attached at the same position on the PG as WTA. Therefore, there must be some sort of regulation/competition between capsule biosynthesis and WTA biosynthesis (see also ref. 21). The presence of the capsule might thus also influence the characteristics of the periplasmic space. Considering that by far most pneumococcal strains are encapsulated, the authors should discuss this and why a capsule mutant was used in this study and how translatable their study using a capsule mutant is to S. pneumoniae in general.

A paragraph was added in the Introduction of V2 to present the complication and a sentence was added at the end of the discussion to mention that this should be studied in the future.

L102: Ref 29 should probably be cited here as well?

Since in Ref 29 (Flores-Kim et al. 2019) there is a detectable amount of LTA (presumably precursors TA) in the ∆tacL stain, we prefer to cite only Hess et al. 2017 regarding the absence of LTA in the absence of TacL. However, we added in V2 a reference to Flores-Kim et al. 2019 in the following paragraph regarding the role of the LTA/WTA ratio.

L106: dependent on the presence of the phosphotransferase LytR (21). --> dependent on the presence of the phosphotransferase LytR, whose expression is upregulated during competence (21).

Corrected in V2.

L119: I fail to see how the conclusions drawn by other groups (I assume the authors mean work from the Vollmer, Rudner, Bernhardt, Hammerschmidt, Havarstein, Veening groups?) are invalid if they compared WTA:LTA ratios between strains and conditions if they underestimated the LTA levels? Supposedly, the LTA levels were underestimated in all samples equally so the relative WTA/LTA ratio changes will qualitatively give the same outcome? I agree that these findings will allow for a reassessment of previous studies in which presumably too low LTA levels were reported, but I would not expect a difference in outcome when people compared WTA:LTA ratios between strains?

The sentence was rephrased in V2 to be neutral regarding previous work and rather emphasize future possibilities.

L131: Perhaps it would be good to highlight that such a conspicuous space has been noticed before by other EM methods (see e.g. Figs.4 and 5 or ref 19, or one of the most clear TEM S. pneumoniae images I have seen in Fig. 1F of Gallay et al, Nat. Micro 2021). However, always some sort of staining had previously been performed so it was never clear this was a real periplasmic space. CEMOVIS has this big advantage of being label free and imaging cells in their presumed native state.

Thanks for pointing out these beautiful data that we had overlooked. We have added a few sentences and references in the Discussion of V2.

L201: References are not numbered.

Corrected in V2.

L271/L892: Change section title. 'Evolution' can have multiple meanings. It would be more clear to write something like 'Increased TA chain length in stationary phase cells' or something like that.

Changed in V2.

L275: harvested

Corrected in V2.

L329: add, as suggested shown previously (I guess refs 24 and 29)

Reference to Hess et al. 2017 has been added in V2. A sentence and further references to Flores-Kim, 2019, 2022 and Wu et al. 2014 were added at the end of the discussion with respect to the LTA-like signal observed in these studies of ∆tacL strains.

L337: I think a concluding sentence is warranted here. These experiments demonstrate that membrane-bound TA precursors accumulate on the outside of the membrane, and are likely polymerized on the outside as well, in line with the model proposed in ref. 20.

From the point of view of formal logic, the accumulation of membrane-bound TA precursors on the outer face of the membrane does not prove that they were assembled there. They could still be polymerized inside and translocated immediately. However, since this is extremely unlikely for the reasons discussed by Gibson and Veening, we have added a mild conclusion sentence and the reference in V2.

L343: How accurate are these quantifications? Just by looking at the gel, it seems there is much less WTA in the lytR mutant than 50% of the wild type?

Yes, the 51% value was a calculation error. This was changed to 41%. Likewise, the decrease of the WTA amount relative to LTA was corrected to 5- to 7-fold.

Apart from the titration of TA in the WT strain, we haven’t yet carried out a careful quantification neither of TA nor of the LTA/WTA ratio in different strains and conditions, although we intend to do so in the near future using the method presented here.

However, to better substantiate our statement regarding the ∆lytR strain, we have quantified two experiments of growth in C-medium with azido-choline, and two experiments of pulse labeling in BHI medium. The results are presented in the additional supplementary Fig. S14.

L342: although WTA are less abundant and LTA appear to be longer (Fig. 6A). although WTA are less abundant and LTA appear to be longer (Fig. 6A), in line with a previous report showing that LytR the major enzyme mediating the final step in WTA formation (ref. 21). (or something like that). Perhaps better is to start this paragraph differently. For instance: Previous work showed that LytR is the major enzyme mediating the final step in WTA formation (ref. 21). As shown in Fig. 6A, the proportion of WTA significantly decreased in the lytR mutant. However, there was still significant WTA present indicating that perhaps another LCP protein can also produce WTA.

Changed in V2.

Of note, WTA levels would be a lot lower in encapsulated strains as used in Ref. 21 (assuming WTA and capsule compete for the same linkage on PG). So perhaps it would be hard to detect any residual WTA in a encapsulated lytR mutant?

Investigation of the relationship between TA and capsule incorporation or O-acetylation is definitely a future area of study using this method of TA monitoring.

L371: see my comments related to L131. Some TEM images clearly show the presence of a periplasmic space.

Comments and references have been added in V2.

L402: It would be really interesting to perform these experiments on a wild type encapsulated strain. Would these have much more LTA? (I understand you cannot do these experiments perhaps due to biosafety, but it might be interesting to discuss).

Yes. It would be interesting to compare the TA in D39 and D39 ∆cps strains. We have added this perspective at the end of the discussion in V2.

L418: ref lacks number

Corrected in V2.

L423: refs missing.

References added in V2.

L487: See my comments regarding L46. I do not see one valid point in the current paper why underestimating LTA levels would change any of the conclusions drawn in Ref. 21. I do not know the other papers cited well enough, but it seems highly unlikely that their conclusions would be wrong by systematically underestimating LTA levels. As far as I understand it, this current work basically confirms the major conclusions drawn by these 'doubtful' papers (that TacL makes LTA and LytR is the main WTA producer). As such, I find this sentence highly unfair without precisely specifying what the exact doubts are. Sure, this current paper now shows that probably people have discarded unknowingly LTA and therefore underestimated LTA levels, so any quantitative assessment of LTA levels are probably wrong. That is one thing. But to say this casts doubts on these studies is very serious and unfair (unless the authors provide good arguments to support these serious claims).

Yes indeed. The sentence was rephrased to be strictly factual in V2.

Table 2: I assume these strains are delta cps? Would be relevant to list this genotype.

The Table 2 was completed in V2.

The authors should comment on why the mutants have not been complemented, especially for lytR as it's the last gene in a complex operon. It would be great to see WTA levels being restored by ectopic expression of LytR.

Yes. We think this could be part of an in-depth study of the attachment of WTA, together with the investigation of the other LCP phosphotransferases.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

Pradhan et al investigated the potential gustatory mechanisms that allow flies to detect cholesterol. They found that flies are indifferent to low cholesterol and avoid high cholesterol. They further showed that the ionotropic receptors Ir7g, Ir51b, and Ir56d are important for the cholesterol sensitivity in bitter neurons. The figures are clear and the behavior result is interesting. However, I have several major comments, especially on the discrepancy of the expression of these Irs with other lab published results, and the confusing finding that the same receptors (Ir7g, Ir51b) have been implicated in the detection of various seemingly unrelated compounds.

Strengths:

The results are very well presented, the figures are clear and well-made, text is easy to follow.

Weaknesses:

(1) Regarding the expression of Ir56d. The reported Ir56d expression pattern contradicts multiple previous studies (Brown et al., 2021 eLife, Figure 6a-c; Sanchez-Alcaniz et al., 2017 Nature Communications, Figure 4e-h; Koh et al., 2014 Neuron, Figure 3b). These studies, using three different driver lines, consistently showed Ir56d expression in sweet-sensing neurons and taste peg neurons. Importantly, Sanchez-Alcaniz et al. demonstrated that Ir56d is not expressed in Gr66a-expressing (bitter) neurons. This discrepancy is critical since Ir56d is identified as the key subunit for cholesterol detection in bitter neurons, and misexpression of Ir7g and Ir51b together is insufficient to confer cholesterol sensitivity (Fig.4b,d). Which Ir56d-GAL4 (and Gr66a-I-GFP) line was used in this study? Is there additional evidence (scRNA sequencing, in-situ hybridization, or immunostaining) supporting Ir56d expression in bitter neurons?

We agree that the expression pattern of Ir56d diverges from two prior reports . The studies by Brown et al. and Koh et al. employed the same Ir56d-GAL4 driver line, which exhibited expression in sweet-sensing gustatory receptor neurons (GRNs) and taste peg neurons, but not bitter GRNs (the Sanchez-Alcaniz et al. paper did not use an Ir56d-Gal4).

In our study, we used a Ir56d-GAL4 driver line (KDRC:2307) and the Gr66a-I-GFP reporter line (Weiss et al., 2011 Neuron). This is a crucial distinction, as differences in the regulatory regions used to generate different driver lines are well known to underlie differences in expression patterns. Our double-labeling experiments revealed co-expression of Ir56d with Gr66a-positive bitter GRNs specifically within the S6 and S7 sensilla—types previously shown to exhibit strong electrophysiological responses to cholesterol (Figure 2—figure supplement 1F).

We believe this observation is biologically significant and consistent with our functional data. Specifically, targeted expression of Ir56d in bitter neurons using the Gr33a-GAL4 was sufficient to rescue cholesterol avoidance behavior in Ir56d¹ mutants (Figure 3G). These results demonstrate that Ir56d plays a functional role in bitter GRNs for cholesterol detection. The convergence of genetic, behavioral, and electrophysiological data presented in our study provides compelling support for this previously unappreciated expression pattern and function of Ir56d.

(2) Ir51b has previously been implicated in detecting nitrogenous waste (Dhakal 2021), lactic acid (Pradhan 2024), and amino acids (Aryal 2022), all by the same lab. Additionally, both Ir7g and Ir51b have been implicated in detecting cantharidin, an insect-secreted compound that flies may or may not encounter in the wild, by the same lab. Is Ir51b proposed to be a specific receptor for these chemically distinct compounds or a general multimodal receptor for aversive stimuli? Unlike other multimodal bitter receptors, the expression level of Ir51b is rather low and it's unclear which subset of GRNs express this receptor. The chemical diversity among nitrogenous waste, amino acids, lactic acid, cantharidin, and cholesterol raises questions about the specificity of these receptors and warrants further investigation and at a minimum discussion in this paper. Given the wide and seemingly unrelated sensitivity of Ir51b and Ir7g to these compounds I'm leaning towards the hypothesis that at least some of these is non-specific and ecologically irrelevant without further supporting evidence from the authors.

While it is true that IR51b and IR7g are responsive to a range of compounds, they share chemical features such as nitrogen-containing groups, hydrophobicity, or amphipathic structures suggesting that recognition of these chemicals may be mediated by the same or overlapping domains within the receptor complexes. These features could facilitate binding to a structurally diverse yet chemically related groups of aversive ligands.

In the case of cholesterol, while its sterol ring system is distinct from the other compounds, it shares hydrophobic and amphipathic properties that may enable interaction with these receptors via similar structural motifs. Importantly, our data demonstrates that Ir51b and Ir7g are necessary but not sufficient on their own to confer cholesterol sensitivity, indicating that additional co-factors or receptor subunits are required for full functionality (Figure 4B, D). Furthermore, our dose-response analysis (Figure 3F) shows that Ir7g is particularly important at higher cholesterol concentrations, supporting the idea of graded sensitivity rather than indiscriminate activation. This suggests that these receptors may have evolved to recognize cholesterol and its analogs (e.g., phytosterols such as stigmasterol, yet to be tested), which are naturally found in the fly’s diet (e.g., yeast and plant-derived matter), as ecologically relevant cues signaling microbial contamination, lipid imbalance, or dietary overconsumption.

We acknowledge the reviewer’s concern regarding the relatively low expression levels of Ir51b and Ir7g. However, we note that low transcript abundance does not necessarily equate to diminished physiological relevance. Finally, we agree that the chemical diversity of ligands associated with Ir51b and Ir7g warrants deeper investigation, particularly through structure-function studies aimed at identifying ligand-binding domains and receptor-ligand interactions at atomic resolution.

(3) The Benton lab Ir7g-GAL4 reporter shows no expression in adults. Additionally, two independent labellar RNA sequencing studies (Dweck, 2021 eLife; Bontonou et al., 2024 Nature Communications) failed to detect Ir7g expression in the labellum. This contradicts the authors' previous RT-PCR results (Pradhan 2024 Fig. S4, Journal of Hazardous Materials) showing Ir7g expression in the labellum. Additionally the Benton and Carlson lab Ir51b-GAL4 reporters show no expression in adults as well. Please address these inconsistencies.

With respect to Ir7g, we acknowledge that the Ir7g-GAL4 reporter line from the Benton lab does not exhibit detectable expression in adult labella. Furthermore, two independent transcriptomic studies—Dweck et al., 2021 (eLife) and Bontonou et al., 2024 (Nature Communications) also did not detect Ir7g transcripts in bulk RNA-seq datasets derived from adult labella. However, our previously published RT-PCR data (Pradhan et al., 2024, Journal of Hazardous Materials, Fig. S4) revealed Ir7g expression in labellar tissue, albeit at low levels. Our RT-PCR includes an internal control (tubulin) with the same reaction tube with control and the Ir7g mutant as a negative control. Therefore, we stand behind the findings that Ir7g is expressed in the labellum.

We would like to point out that RT-PCR is more sensitive and better-suited to detect low-abundance transcripts than bulk RNA-seq, which may fail to capture transcripts due to limitations in depth of coverage. Moreover, immunohistochemistry can have limitations in detecting very low expression levels. Costa et al. 2013 (Translational lung cancer research) states that “RNA-Seq technique will not likely replace current RT-PCR methods, but will be complementary depending on the needs and the resources as the results of the RNA-Seq will identify those genes that need to then be examined using RT-PCR methods”.

Similarly, regarding Ir51b, while the GAL4 reporter lines from the Benton and Carlson labs do not show robust adult expression, our RT-PCR and functional data strongly support a role for Ir51b in labellar bitter GRNs. Specifically, Ir51b¹ mutants display electrophysiological deficits in response to cholesterol (Figure 2A–B), and these defects are rescued by expressing Ir51b in Gr33a-positive bitter neurons (Figure 3G), providing functional validation of the RT-PCR expression.

(4) The premise that high cholesterol intake is harmful to flies, which makes sensory mechanisms for cholesterol avoidance necessary, is interesting but underdeveloped. Animal sensory systems typically evolve to detect ecologically relevant stimuli with dynamic ranges matching environmental conditions. Given that Drosophila primarily consume fruits and plant matter (which contain minimal cholesterol) rather than animal-derived foods (which contain higher cholesterol), the ecological relevance of cholesterol detection requires more thorough discussion. Furthermore, at high concentrations, chemicals often activate multiple receptors beyond those specifically evolved for their detection. If the cholesterol concentrations used in this study substantially exceed those encountered in the fly's natural diet, the observed responses may represent an epiphenomenon rather than an ecologically and ethologically relevant sensory mechanism. What is the cholesterol content in flies' diet and how does that compare to the concentrations used in this paper?

Drosophila melanogaster cannot synthesize sterols de novo, and must acquire them from its diet. In natural environments, flies acquire sterols from fermenting fruit, decaying plant matter, and yeast, which contain trace amounts of phytosterols (e.g., stigmasterol, β-sitosterol) and ergosterol. While the exact sterol concentrations in these sources remain uncharacterized, our behavioral assays used concentrations (0.001–0.01% by weight) that align with the low levels expected in such nutrient-limited ecological niches.

In our study, the cholesterol concentrations tested ranged from 0.001% to 0.1%, thereby spanning both the physiologically relevant and slightly elevated range. Importantly, avoidance behaviors and receptor activation were most prominent at 0.1% cholesterol. While it is true that high chemical concentrations may elicit off-target effects via broad receptor activation, our genetic and electrophysiological data indicate that the observed responses are mediated by specific ionotropic receptors (Ir51b, Ir7g, Ir56d) and not merely generalized chemical stress.

Ecologically, elevated sterol levels may also signal conditions unsuitable for egg-laying or larval development. For example, high levels of cholesterol or other sterols may occur in substrates colonized by pathogenic microbes, decaying animal tissue, or in cases of abnormal microbial fermentation, which could represent a nutritional or microbial hazard. The avoidance of cholesterol may help signal the flies to avoid consuming decaying animal tissue. In this context, sensory detection of excessive cholesterol might serve as a protective function.

Reviewer #2 (Public review):

Summary:

In Cholesterol Taste Avoidance in Drosophila melanogaster, Pradhan et al. used behavioral and electrophysiological assays to demonstrate that flies can: (1) detect cholesterol through a subset of bitter-sensing gustatory receptor neurons (GRNs) and (2) avoid consuming food with high cholesterol levels. Mechanistically, they identified five members of the IR family as necessary for cholesterol detection in GRNs and for the corresponding avoidance behavior. Ectopic expression experiments further suggested that Ir7g + Ir56d or Ir51b + Ir56d may function as tuning receptors for cholesterol detection, together with the Ir25a and Ir76b co-receptors.

Strengths:

The experimental design of this study was logical and straightforward. Leveraging their expertise in the Drosophila taste system, the research team identified the molecular and cellular basis of a previously unrecognized taste category, expanding our understanding of gustation. A key strength of the study was its combination of electrophysiological recordings with behavioral genetic experiments.

Weaknesses:

My primary concern with this study is the lack of a systematic survey of the IRs of interest in the labellum GRNs. Consequently, there is no direct evidence linking the expression of putative cholesterol IRs to the B GRNs in the S6 and S7 sensilla.

Specifically, the authors need to demonstrate that the IR expression pattern explains cholesterol sensitivity in the B GRNs of S6 and S7 sensilla, but not in other sensilla. Instead of providing direct IR expression data for all candidate IRs (as shown for Ir56d in Figure 2-figure supplement 1F), the authors rely on citations from several studies (Lee, Poudel et al. 2018; Dhakal, Sang et al. 2021; Pradhan, Shrestha et al. 2024) to support their claim that Ir7g, Ir25a, Ir51b, and Ir76b are expressed in B GRNs (Lines 192-194). However, none of these studies provide GAL4 expression or in situ hybridization data to substantiate this claim.

Without a comprehensive IR expression profile for GRNs across all taste sensilla, it is difficult to interpret the ectopic expression results observed in the B GRN of the I9 sensillum or the A GRN of the L-sensillum (Figure 4). It remains equally plausible that other tuning IRs-beyond the co-receptor Ir25a and Ir76b-could interact with the ectopically expressed IRs to confer cholesterol sensitivity, rather than the proposed Ir7g + Ir56d or Ir51b + Ir56d combinations.

We provide electrophysiological data demonstrating that the S6 and S7 sensilla respond to cholesterol (Figure 1D). This finding is consistent with the hypothesis that these sensilla harbor the complete receptor complexes necessary for cholesterol detection. In our electrophysiological recordings, only those bitter GRNs that co-express Ir56d along with either Ir7g or Ir51b generate action potentials in response to cholesterol. Other S-type sensilla lacking one or more of these subunits remain unresponsive, reinforcing the idea that these components are necessary for receptor function and sensory coding of cholesterol. Moreover, in the cholesterol-insensitive I9 sensillum (based on our mapping results using electrophysiology), co-expression of either Ir7g + Ir56d or Ir51b + Ir56d conferred de novo cholesterol sensitivity (Figure 4B). Importantly, no cholesterol response was observed when any of these IRs was expressed alone or when Ir7g + Ir51b were co-expressed without Ir56d. These findings strongly argue against the possibility that endogenous tuning IRs in I9 sensilla (e.g., Ir25a, Ir76b) are sufficient to generate cholesterol responsiveness.

Furthermore, based on the literature, Ir25a and Ir76b are endogenously expressed in I- and L-type sensilla. Thus, their presence alone is insufficient for cholesterol responsiveness. These data support the model that cholesterol sensitivity depends on a specific, multi-subunit receptor complex (e.g., Ir7g + Ir25a + Ir56d + Ir76b or Ir51b + Ir25a + Ir56d + Ir76b).

In conclusion, while we acknowledge that our data do not provide a full anatomical map of IR expression across all sensilla, our results strongly support the idea that cholesterol sensitivity in S6 and S7 sensilla arises from specific combinations of IRs expressed in the B GRNs.

Reviewer #3 (Public review):

Summary:

Whether and how animals can taste cholesterol is not well understood. The study provides evidence that 1) cholesterol activates a subset of bitter-sensing gustatory receptor neurons (GRNs) in the fly labellum, but not other types of GRNs, 2) flies show aversion to high concentrations of cholesterol, and this is mediated by bitter GRNs, and 3) cholesterol avoidance depends on a specific set of ionotropic receptor (IR) subunits acting in bitter GRNs. The claims of the study are supported by electrophysiological recordings, genetic manipulations, and behavioral readouts.

Strengths:

Cholesterol taste has not been well studied, and the paper provides new insight into this question. The authors took a comprehensive and rigorous approach in several different parts of the paper, including screening the responses of all 31 labellar sensilla, screening a large panel of receptor mutants, and performing misexpression experiments with nearly every combination of the 5 IRs identified. The effects of the genetic manipulations are very clear and the results of electrophysiological and behavioral studies match nicely, for the most part. The appropriate controls are performed for all genetic manipulations.

Weaknesses:

The weaknesses of the study, described below, are relatively minor and do not detract from the main conclusions of the paper.

(1) The paper does not state what concentrations of cholesterol are present in Drosophila's natural food sources. Are the authors testing concentrations that are ethologically Drosophila melanogaster primarily feeds on fermenting fruits and associated microbial communities, especially yeast, which serve as major sources of dietary sterols. These natural food sources are known to contain phytosterols such as stigmasterol and β-sitosterol. One study quantified phytosterols (e.g., stigmasterol, sitosterol) in fruits, reporting concentrations between 1.6–32.6 mg/100 g edible portion (~0.0016–0.0326% wet weight) (Han et al 2008). The range we tested falls within this range. Additionally, ergosterol, the principal sterol in yeast and a structural analog of cholesterol, is present at levels of about 0.005% to 0.02% in yeast-rich environments.

To ensure physiological relevance, we designed our behavioral assays to include a broad concentration range of cholesterol, from 10^-5% to 10^-1%. This spans both physiological levels (0.001–0.01%), which are comparable to those found in the natural diet, and supra-physiological levels (e.g., 0.1%), which exceed natural exposure but help define the threshold for aversive behavior.

Our results demonstrate that flies begin to avoid cholesterol at concentrations ≥10^-3% more (Figure 3A), which falls within the upper physiological range and may reflect the threshold beyond which cholesterol or related sterols become deleterious. At these higher concentrations, excess sterols may disrupt membrane fluidity, interfere with hormone signaling, or promote microbial overgrowth—all of which could compromise fly health.

(2) The paper does not state or show whether the expression of IR7g, IR51b, and IR56d is confined to bitter GRNs. Bitter-specific expression of at least some of these receptors would be necessary to explain why bitter GRNs but not sugar GRNs (or other GRN types) normally show cholesterol responses.

We show the Ir56d-Gal4 is co-expressed with Gr66a-GFP in S6/S7 sensilla, indicating that it is expressed in bitter GRNs (Figure 2—figure supplement 1F). In the case of Ir7g and Ir51b, there are no reporters or antibodies to address expression. However, previously they have been shown to be expressed in bitter GRNs using RT-PCR (Dhakal et al. 2021, Communications Biology; Pradhan et al. 2024, Journal of Hazardous Materials). In addition, we provide functional evidence that bitter GRNs are required for the cholesterol response since silencing bitter GRNs abolishes cholesterol-induced action potentials (Figure 1E–F). Moreover, we showed that we could rescue the Ir7g¹, Ir51b¹ and Ir56d¹ mutant phenotypes only when we expressed the cognate transgenes in bitter GRNs using the Gr33a-GAL4 (Figure 3G). Thus, while Ir7g/Ir51b are not exclusive to bitter GRNs, their functional role in cholesterol detection is bitter-GRN-specific.

(3) The authors only investigated the responses of GRNs in the labellum, but GRN responses in the leg may also contribute to the avoidance of cholesterol feeding. Alternatively, leg GRNs might contribute to cholesterol attraction that is unmasked when bitter GRNs are silenced. In support of this possibility, Ahn et al. (2017) showed that Ir56d functions in sugar GRNs of the leg to promote appetitive responses to fatty acids.

This is an interesting idea. Indeed, when bitter GRNs are hyperpolarized, the flies exhibit a strong attraction to cholesterol. Nevertheless, the cellular basis for cholesterol attraction and whether it is mediated by GRNs in the legs will require a future investigation.

(4) The authors might consider using proboscis extension as an additional readout of taste attraction or aversion, which would help them more directly link the labellar GRN responses to a behavioral readout. Using food ingestion as a readout can conflate the contribution of taste with post-ingestive effects, and the regulation of food ingestion also may involve contributions from GRNs on multiple organs, whereas organ-specific contributions can be dissociated using proboscis extension. For example, does presenting cholesterol on the proboscis lead to aversive responses in the proboscis extension assay (e.g., suppression of responses to sugar)? Does this aversion switch to attraction when bitter GRNs are silenced, as with the feeding assay?

We thank the reviewer for the suggestion regarding the use of the proboscis extension reflex (PER) assay to strengthen the link between labellar GRN activity and behavioral responses to cholesterol.

Author response image 1.

Our PER assay results shown above indicate that cholesterol presentation on the labellum or forelegs leads to an aversive response, as evidenced by a significant reduction in proboscis extension when compared to control stimuli (Author response image 1A. 2% sucrose or 2% sucrose with 10^-1% cholesterol was applied to labellum or forelegs and the percent PER was recorded. n=6. Data were compared using single-factor ANOVA coupled with Scheffe’s post-hoc test. Statistical significance was compared with the control. Means ± SEMs. **p<0.01). This finding supports the idea that cholesterol is detected by labellar and leg GRNs and elicits behavioral avoidance. In contrast, sucrose stimulation robustly induces proboscis extension, as expected for an appetitive stimulus. We confirmed the defects of due to each Ir mutant by presenting the stimuli to the labellum (Author response image 1B). Together, these PER results provide a more direct behavioral correlate of labellar and leg GRN activation and reinforce our conclusion that cholesterol is sensed as an aversive tastant through the labellar bitter GRNs.

(5) The authors claim that the cholesterol receptor is composed of IR25a, IR76b, IR56d, and either IR7g or IR51b. While the authors have shown that IR25a and IR76b are each required for cholesterol sensing, they did not show that both are required components of the same receptor complex. If the authors are relying on previous studies to make this assumption, they should state this more clearly. Otherwise, I think further misexpression experiments may be needed where only IR25a or IR76b, but not both, are expressed in GRNs.

In our study, we relied on prior work demonstrating that Ir25a and Ir76b function as broadly required co-receptors in most IR-dependent chemosensory pathways (Ganguly et al., 2017; Lee et al., 2018). These studies showed that Ir25a and Ir76b are co-expressed in many GRNs across multiple taste modalities. Functional IR complexes often fail to form or signal properly in the absence of these co-receptors. Thus, it is widely accepted in the field that Ir25a and Ir76b function together as a core heteromeric scaffold for diverse IR complexes, akin to co-receptors in other ionotropic glutamate receptor families. We state that while Ir25a and Ir76b are presumed co-receptors in the cholesterol receptor complex based on their conserved roles, their direct physical interaction with Ir7g, Ir51b, and Ir56d remains to be demonstrated.

In support of this model, we note that in our ectopic expression experiments using I9 sensilla, which endogenously express Ir25a and Ir76b, introduction of either Ir7g + Ir56d or Ir51b + Ir56d was sufficient to confer cholesterol sensitivity (Figure 4B). We obtained a similar result in L6 sensilla (Figure 4D), which also endogenously express Ir25a and Ir76b. These findings imply that both co-receptors are already present in these sensilla and are likely part of the functional complex. However, we agree that we have not directly tested the requirement for both co-receptors in a minimal reconstitution context, such as expressing only Ir25a or Ir76b alongside tuning IRs in an otherwise null background. Such an experiment would indeed provide more direct evidence of their joint requirement in the receptor complex. Future studies, including heterologous expression experiments, will be necessary to define the cholesterol-receptor complexes.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public Review):

The aim of this study is to test the overarching hypothesis that plasticity in BNST CRF neurons drives distinct behavioral responses to unpredictable threat in males and females. The manuscript provides evidence for a possible sex-specific role for CRF-expressing neurons in the BNST in unpredictable aversive conditioning and subsequent hypervigilance across sexes. As the authors note, this is an important question given the high prevalence of sex differences in stress-related disorders, like PTSD, and the role of hypervigilance and avoidance behaviors in these conditions. The study includes in vivo manipulation, bulk calcium imaging, and cellular resolution calcium imaging, which yield important insights into cell-type specific activity patterns. However, it is difficult to generate an overall conclusion from this manuscript, given that many of the results are inconsistent across sexes and across tests and there is an overall lack of converging evidence. For example, partial conditioning yields increased startle in males but not females, yet, CRF KO only increases startle response in males after full conditioning, not partial, and CRF neurons show similar activity patterns between partial and full conditioning across sexes. Further, while the study includes a KO of CRF, it does not directly address the stated aim of assessing whether plasticity in CRF neurons drives the subsequent behavioral effects unpredictable threat.

We appreciate the reviewer’s summary and agree that there is a large amount of complexity to the results, and that it was difficult to generate a simple model/conclusion to summarize our work. This is the unfortunate side effect of looking across both sexes at different conditioning paradigms, however, we believe that it is important to convey this information to the field even without a simple answer.  Our data reinforces the very important findings from the Maren and Holmes groups that partial fear is a different process than full fear, and that the BNST plays a differential role here. We have reworded the manuscript to better convey this complexity.

A major strength of this manuscript is the inclusion of both males and females and attention to possible behavioral and neurobiological differences between them throughout. However, to properly assess sex-differences, sex should be included as a factor in ANOVA (e.g. for freezing, startle, and feeding data in Figure 1) to assess whether there is a significant main effect or interaction with sex. If sex is not a statistically significant factor, both sexes should be combined for subsequent analyses. See, Garcia-Sifuentes and Maney, eLife 2021 https://elifesciences.org/articles/70817. There are additional cases where t-tests are used to compare groups when repeated measures ANOVAs would be more appropriate and rigorous.

We agree with the reviewer that this is the more appropriate analysis and have changed the analysis and figures throughout the revised manuscript to better assess sex differences as well as differences between fear conditions.

Additionally, it's unclear whether the two sexes are equally responsive to the shock during conditioning and if this is underlying some of the differences in behavioral and neuronal effects observed. There are some reports that suggest shock sensitivity differs across sexes in rodents, and thus, using a standard shock intensity for both males and females may be confounding effects in this study.

This is a great point. We have conducted appropriate analysis (Sex by Tone Repeated measures two-way ANOVAS for each of the groups: Ctrl, Full, Part) and there are no sex differences in freezing between males and females. The extent of conditioning is not different between the groups suggesting that if there was a difference in shock sensitivity, it is not driving any discernible differences in behavioral performance. However, it is possible that the experience of the shock differs for the animals even in the absence of any measurable behavior.

The data does not rule out that BNST CRF activity is not purely tracking the mobility state of the animal, given that the differences in activity also track with differences in freezing behavior. The data shows an inverse relationship between activity and freezing. This may explain a paradox in the data which is why males show a greater suppression of BNST activity after partial conditioning than full conditioning, if that activity is suspected to drive the increased anxiety-like response. Perhaps it reflects that activity is significantly suppressed at the end of the conditioning session because animals are likely to be continuously freezing after repeated shock presentations in that context. It would also explain why there is less of a suppression in activity over the course of the recall session, because there is less freezing as well during recall compared with conditioning.

While it is possible that the BNST may be tracking activity, we believe it is not purely tracking mobility state. For instance, while freezing increases across tone exposures in Part fear regardless of sex, males show an increase while females show a reduction in BNST response during tone 5 (Fig 2K). The data the reviewer refers to showing the inverse relationship with BNST activity and freezing would have suggested the opposite response if it were purely tracking the mobility state of the animal. This is also the case with BNST^CRF activity to first and last tone during recall. Despite the suppression of activity over the course of recall (Fig 5K), we see an increase in BNST^CRF tone response when comparing tone 1 and 6 in males and a decrease in females (Fig 6M), again suggesting the BNST is responding to more than just activity.

A mechanistic hypothesis linking BNST CRF neurons, the behavioral effects observed after fear conditioning, and manipulation of CRF itself are not clearly addressed here.

We disagree with this assertion. The data suggests a model in which males respond with increased arousal and Part fear males show persistent activation of the BNST and BNST^CRF neurons during fear conditioning and recall while female Part fear mice show the opposite response. This female response differs from what the field believes to be the role of the BNST in sustained fear. Additionally, we show that CRF knockdown is not involved in fear differentiation or fear expression in males, while it enhances fear learning and recall in females. We have reworded the manuscript to highlight these novel findings.

Reviewer #2 (Public Review):

This study examined the role of CRF neurons in the BNST in both phasic and sustained fear in males and females. The authors first established a differential fear paradigm whereby shocks were consistently paired with tones (Full) or only paired with tones 50% of the time (Part), or controls who were exposed to only tones with no shocks. Recall tests established that both Full and Part conditioned male and female mice froze to the tones, with no difference between the paradigms. Additional studies using the NSF and startle test, established that neither fear paradigm produced behavioral changes in the NSF test, suggesting that these fear paradigms do not result in an increase in anxiety-like behavior. Part fear conditioning, but not Full, did enhance startle responses in males but not females, suggesting that this fear paradigm did produce sustained increases in hypervigilance in males exclusively.

Thank you for this clear summary of the behavioral work.

Photometry studies found that while undifferentiated BNST neurons all responded to shock itself, only Full conditioning in males lead to a progressive enhancement of the magnitude of this response. BNST neurons in males, but not females, were also responsive to tone onset in both fear paradigms, but only in Full fear did the magnitude of this response increase across training. Knockdown of CRF from the BNST had no effect on fear learning in males or females, nor any effect in males on fear recall in either paradigm, but in females enhanced both baseline and tone-induced freezing only in Part fear group. When looking at anxiety following fear training, it was found in males that CRF knockdown modulated anxiety in Part fear trained animals and amplified startle in Fully trained males but had no effect in either test in females. Using 1P imaging, it was found that CRF neurons in the BNST generally decline in activity across both conditioning and recall trials, with some subtle sex differences emerging in the Part fear trained animals in that in females BNST CRF neurons were inhibited after both shock and omission trials but in males this only occurred after shock and not omission trials. In recall trials, CRF BNST neuron activity remained higher in Part conditioned mice relative to Full conditioned mice.

Overall, this is a very detailed and complex study that incorporates both differing fear training paradigms and males and females, as well as a suite of both state of the art imaging techniques and gene knockdown approaches to isolate the role and contributions of CRF neurons in the BNST to these behavioral phenomena. The strengths of this study come from the thorough approach that the authors have taken, which in turn helped to elucidate nuanced and sex specific roles of these neurons in the BNST to differing aspects of phasic and sustained fear. More so, the methods employed provide a strong degree of cellular resolution for CRF neurons in the BNST. In general, the conclusions appropriately follow the data, although the authors do tend to minimize some of the inconsistencies across studies (discussed in more depth below), which could be better addressed through discussion of these in greater depth. As such, the primary weakness of this manuscript comes largely from the discussion and interpretation of mixed findings without a level of detail and nuance that reflects the complexity, and somewhat inconsistency, across the studies. These points are detailed below:

- Given the focus on CRF neurons in the BNST, it is unclear why the photometry studies were performed in undifferentiated BNST neurons as opposed to CRF neurons specifically (although this is addressed, to some degree, subsequently with the 1P studies in CRF neurons directly). This does limit the continuity of the data from the photometry studies to the subsequent knockdown and 1P imaging studies. The authors should address the rationale for this approach so it is clear why they have moved from broader to more refined approaches.

The reviewer raises a good point.  We did some preliminary photometry studies with BNST CRF neurons and found that there was poor time locked signal. We reasoned that this was due to the heterogeneity of the cell activity, as we saw in our previous publication (Yu et al). Because of this, we moved to the 1p imaging work in place of continued BNST CRF photometry. We have also reworded the manuscript to better discuss the complexities and inconsistencies in findings across the studies.

- The CRF KD studies are interesting, but it remains speculative as to whether these effects are mediated locally in the BNST or due to CRF signaling at downstream targets. As the literature on local pharmacological manipulation of CRF signaling within the BNST seems to be largely performed in males, the addition of pharmacological studies here would benefit this to help to resolve if these changes are indeed mediated by local impairments in CRF release within the BNST or not. While it is not essential to add these experiments, the manuscript would benefit from a more clear description of what pharmacological studies could be performed to resolve this issue.

We agree with the reviewer that the addition of this experiment would be highly informative for differentiating the role of CRF in the BNST. This is something that will need to be considered moving forward and we have added this as a point of discussion.

- While I can appreciate the authors perspective, I think it is more appropriate to state that startle correlates with anxiety as opposed to outright stating that startle IS anxiety. Anxiety by definition is a behavioral cluster involving many outputs, of which avoidance behavior is key. Startle, like autonomic activation, correlates with anxiety but is not the same thing as a behavioral state of anxiety (particularly when the startle response dissociates from behavior in the NSF test, which more directly tests avoidance and apprehension). Throughout the manuscript the use of anxiety or vigilance to describe startle becomes interchangeable, but then the authors also dissociate these two, such as in the first paragraph of the discussion when stating that the Part fear paradigm produces hypervigilance in males without influencing fear or anxiety-like behaviors. The manuscript would benefit from harmonization of the language used to operationally define these behaviors and my recommendation would be to remain consistent with the description that startle represents hypervigilance and not anxiety, per se.

The reviewer raises an excellent point, we have clarified in the revised manuscript.

- The interpretation of the anxiety data following CRF KD is somewhat confusing. First, while the authors found no effect of fear training on behavior in the NSF test in the initial studies, now they do, however somewhat contradictory to what one would expect they found that Full fear trained males had reduced latency to feed (indicative of an anxiolytic response), which was unaltered by CRF KD, but in Part fear (which appeared to have no effect on its own in the NSF test), KD of CRF in these animals produced an anxiolytic effect. Given that the Part fear group was no different from control here it is difficult to interpret these data as now CRF KD does reduce latency to feed in this group, suggesting that removal of CRF now somehow conveys an anxiolytic response for Part fear animals. In the discussion the authors refer to this outcome as CRF KD "normalizing" the behavior in the NSF test of Part fear conditioned animals as now it parallels what is seen after Full fear, but given that the Part fear animals with GFP were no different then controls (and neither of these fear training paradigms produced any effect in the NSF test in the first arm of studies), it seems inappropriate to refer to this as "normalization" as it is unclear how this is now normalized. Given the complexity of these behavioral data, some greater depth in the discussion is required to put these data in context and describe the nuance of these outcomes, in particular a discussion of possible experimental factors between the initial behavioral studies and those in the CRF KD arm that could explain the discrepancy in the NSF test would be good (such as the inclusion of surgery, or other factors that may have differed between these experiments). These behavioral outcomes are even more complex given that the opposite effect was found in startle whereby CRF KD amplified startle in Full trained animals. As such, this portion of the discussion requires some reworking to more adequately address the complexity of these behavioral findings.

The reviewer raises a good point, and we agree that there are many inconsistencies in the behaviors. We believe it is still good to show these results but have expanded the manuscript on potential reasons for these behavioral inconsistencies.

Reviewer #3 (Public Review):

Hon et al. investigated the role of BNST CRF signaling in modulating phasic and sustained fear in male and female mice. They found that partial and full fear conditioning had similar effects in both sexes during conditioning and during recall. However, males in the partially reinforced fear conditioning group showed enhanced acoustic startle, compared to the fully reinforced fear conditioning group, an effect not seen in females. Using fiber photometry to record calcium activity in all BNST neurons, the authors show that the BNST was responsive to foot shock in both sexes and both conditioning groups. Shock response increased over the session in males in the fully conditioned fear group, an effect not observed in the partially conditioned fear group. This effect was not observed in females. Additionally, tone onset resulted in increased BNST activity in both male groups, with the tone response increasing over time in the fully conditioned fear group. This effect was less pronounced in females, with partially conditioned females exhibiting a larger BNST response. During recall in males, BNST activity was suppressed below baseline during tone presentations and was significantly greater in the partially conditioned fear group. Both female groups showed an enhanced BNST response to the tone that slowly decayed over time. Next, they knocked CRF in the BNST to examine its effect on fear conditioning, recall and anxiety-like behavior after fear. They found no effect of the knockdown in either sex or group during fear conditioning. During fear recall, BNST CRF knockdown lead to an increase in freezing in only the partially conditioned females. In the anxiety-like behavior tasks, BNST CRF knockdown lead to increased anxiolysis in the partially reinforced fear male, but not in females. Surprisingly, BNST CRF knockdown increased startle response in fully conditioned, but not partially conditioned males. An effect not observed in either female group. In a final set of experiments, the authors single photon calcium imaging to record BNST CRF cell activity during fear conditioning and recall. Approximately, 1/3 of BNST CRF cells were excited by shock in both sexes, with the rest inhibited and no differences were observed between sexes or group during fear conditioning. During recall, BNST CRF activity decreased in both sexes, an effect pronounced in male and female fully conditioned fear groups.

Overall, these data provide novel, intriguing evidence in how BNST CRF neurons may encode phasic and sustained fear differentially in males and females. The experiments were rigorous.

We thank you for this positive review of our manuscript.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

There are several graphs representing different analyses of (presumably) the same group of subjects, but which have different N/group. For example, in Figure 2:

(1) Fig 2P seems to have n=10 in Part Male group (Peak), but 2Q only has n=9 in Part Male group (AUC)

(2) Fig 2S seems to have n=10 in Part Female group (Peak), but 2T only has n=7 in Part Female group (AUC)

(3) Fig 2G (Tone Resp) has n=6 Full Males but 2F (Tone Resp), 2H (Shock Resp), and 2I (Shock Resp) have n=7 Full Males

(4) Fig 2K (Tone Resp) has n=7 Full Females but 2L (Tone Resp), 2M (Shock Resp), and 2N (Shock Resp) have n=8 Full Females

(5) Fig 2L (Tone Resp) has n=9 Part Females but 2K (Tone Resp), 2M (Shock Resp), and 2N (Shock Resp) have n=10 Part Females

It's possible that this is just due to overlapping individual data points which are made harder to see due to the low resolution of the figures. If so, this can be easily rectified. However, there may also be subjects missing from some analyses which must be clarified or corrected.

We thank you for catching these. We have gone through and fixed any issues with data points and have added statistics and exclusions in datasets to figure legends to further explain inconsistencies.

Regarding statistical tests:

(2) Data in Figs 2G and 2I should be analyzed using a two-way RM ANOVA.

We have now included sex as a factor in most of our analysis and are now using appropriate statistical tests.

(3) Data in Fig 3K should be analyzed using a two-way RM ANOVA.

We are now using appropriate statistical tests.

Calcium activity in response to the shock during conditioning and in response to the tone during recall should be included in Figure 5. Given partial and full animals also receive unequal presentations of the cue, it would be useful to see the effects trial by trial or normalized to the first 3 presentations only.

The reviewer raises a great point. We have changed this figure and have now added the response to shock and tones. Since we are most interested in the difference between sustained and phasic fear, we decided to compare tone 3 in Full fear and tone 4 in Part fear, which differ in the ambiguity of their cue and only have one tone difference.

Histology maps should be included for all experiments depicting viral spread and implant location for all animals, in addition to the included representative histology images. These can be placed in the supplement.

We agree this is helpful. While we have confirmed all of the experiments are hits, the tissue is no longer in condition for this analysis.

Referring to the quantification of peaks in fiber photometry and cellular resolution calcium imaging data as "spikes" is a bit misleading given the inexact relationship between GCAMP sensor dynamics/calcium binding and neuronal action potentials, perhaps calling it "event" frequency would be more clear.

We have changed the references of spikes to events as suggested.

The legend for Figure 2S is mislabeled as A.

Thank you for catching this mistake, it has been fixed.

The methods refer to CRFR1 fl/fl animals but it seems no experiments used these animals, only CRF fl/fl.

We have fixed this, thank you.

Reviewer #2 (Recommendations For The Authors):

As stated in the public review, while I think the addition of local pharmacological studies blocking CRF1 and 2 receptors in the BNST in both males and females, done under the same conditions as all of the other testing herein, would help to resolve some of the speculation of interpreting the CRF KD data, I dont think these studies are essential to do, but it would be good for the authors to more explicitly state what studies could be done and how they could facilitate interpretation of these data.

Thank you for this suggestion. We have added this discussion into the manuscript.

Asides from this, my other recommendations for the authors are to more clearly address the discrepancies in behavioral outcomes across studies and explicitly describe their rationale for the sequence of experiments performed and to harmonize their operationalization of how they define anxiety.

Again, we appreciate these great suggestions. We have added more discussion on the behavioral discrepancies as well as rationale for the experiments. We have also changed the wording to remain consistent that the NSF test relates to anxiety and the Startle test relates to vigilance.

- In Figure 2, Panel S is listed as Panel A in the caption and should be corrected.

Thank you for catching this mistake, we have fixed it.

Reviewer #3 (Recommendations For The Authors):

My biggest concerns I have regard the interpretations and some conclusions from this data set, which I have stated below.

(1) It was surprising to see minimal and somewhat conflicting behavioral effects due to BNST CRF knockdown. The authors provide a representative image and address this in the conclusion. They mention the role of local vs projection CRF circuits as well as the role of GABA. I don't think those experiments are necessary for this manuscript. However, it may be worthwhile to see through in situ hybridization or IHC, to see BNST CRF levels after both full and partial conditioned fear paradigms. Additionally, it would help to see a quantification of the knockdown of the animals.

Thank you for these great suggestions. We will consider these for future experiments. We piloted out some CRF sensor experiments to probe this, but it was unclear if the signal to noise for the sensor was sufficient. We hope to do more of this in the future if we ever manage to get funding for this work.

The authors can add a figure showing deltaF/F changes from control.

We did not have control mice in these in-vivo experiments Our main interests lie in understanding the differences in Full and Part Fear conditioning paradigms specifically.

(2) Related to the previous point, it was surprising to see an effect of the CRF deletion in the full fear group compared to the partial fear in the acoustic startle task. To strengthen the conclusion about differential recruitment of CRF during phasic and sustained fear, the experiment in my previous point could help elucidate that. Conversely, intra-BNST administration of a CRF antagonist into the BNST before the acoustic startle after both conditioning tasks could also help. Or patch from BNST CRF neurons after the conditioning tasks to measure intrinsic excitability. Not all these experiments are needed to support the conclusion, it's some examples.

We thank the reviewer for these suggestions and agree that these are important experiments. We will consider this in future experiments exploring the role of BNST CRF in fear conditioning.

(3) In Figure 5 F and K, the authors report data combined for both part and full fear conditioning. Were there any differences between the number of excited or inhibited neurons b/t the conditioning groups?

We are only looking at the first shock exposure in these figures. These were combined because the first tone and shock exposure is identical in Full and Part fear conditioning. Differences in these behavioral paradigms emerge after Tone 3 exposure, where Part fear does not receive a shock while Full fear does.

Also, can the authors separate male and female traces in Fig 5 E and P?

Traces in Fig E are from females only. We did not include male traces because males and females had identical responses to first shock, and we felt only one trace was needed as an example. Traces in Figure P are from males. We did not show female traces because females did not show differential effects from baseline to end.

(4) Also, regarding the calcium imaging data, what was the average length of a transient induced by shock? Were there any differences between the sexes?

We have many cells in each condition, and the length of traces after shock were all different and hard to quantify, as for example, sometimes cells were active before shock and thus trace length would be difficult to quantify. Therefore, to keep consistency and reduce ambiguity regarding trace lengths, we focused on keeping the time consistent across mice and focused on the 10 second window post shock to be consistent across conditions.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1 (Public Review):

Summary:

In this study, Osiurak and colleagues investigate the neurocognitive basis of technical reasoning. They use multiple tasks from two neuroimaging studies and overlap analysis to show that the area PF is central for reasoning, and plays an essential role in tool-use and non-tool-use physical problem-solving, as well as both conditions of mentalizing task. They also demonstrate the specificity of the technical reasoning and find that the area PF is not involved in the fluid-cognition task or the mentalizing network (INT+PHYS vs. PHYS-only). This work suggests an understanding of the neurocognitive basis of technical reasoning that supports advanced technologies.

Strengths:

-The topic this study focuses on is intriguing and can help us understand the neurocognitive processes involved in technical reasoning and advanced technologies.

-The researchers obtained fMRI data from multiple tasks. The data is rich and encompasses the mechanical problem-solving task, psychotechnical task, fluid-cognition task, and mentalizing task.

-The article is well written.

We sincerely thank Reviewer 1 for their positive and very helpful comments, which helped us improve the MS. Thank you.

Weaknesses:

- Limitations of the overlap analysis method: there are multiple reasons why two tasks might activate the same brain regions. For instance, the two tasks might share cognitive mechanisms, the activated regions of the two tasks might be adjacent but not overlapping at finer resolutions, or the tasks might recruit the same regions for different cognition functions.

Thus, although overlap analysis can provide valuable information, it also has limitations.

Further analyses that capture the common cognitive components of activation across different

tasks are warranted, such as correlating the activation across different tasks within subjects for a region of interest (i.e. the PF).

We thank Reviewer 1 for this comment. We added new analyses to address the two alternative interpretations stressed here by Reviewer 1, namely, the same-region-but-differentfonction interpretation and the adjacency interpretation. The new analyses ruled out both alternative interpretations, thereby reinforcing our interpretation.

“The conjunction analysis reported was subject to at least two key limitations that needed to be overcome to assure a correct interpretation of our findings. The first was that the tasks could recruit the same regions for different cognition functions (same-region-but-different-function interpretation). The second was that the activated regions of the different tasks could be adjacent but did not overlap at finer resolutions (adjacency interpretation). We tested the same-region-but-different-function interpretation by conducting additional ROI analyses, which consisted of correlating the specific activation of the left area PF (i.e., difference in terms of mean Blood-Oxygen Level Dependent [BOLD] parameter estimates between the experimental condition minus the control condition) in the psychotechnical task, the fluid-cognition task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task. This analysis did not include the mechanical problem-solving task because the sample of participants was not the same for this task. As shown in Fig. 5, we found significant correlations between all the tasks that were hypothesized as recruiting technical reasoning, i.e., the psychotechnical task and the PHYS-Only and INT+PHYS conditions of the mentalizing task (all p < .05). By contrast, no significant correlation was obtained between these three tasks and the fluid-cognition task (all p > .15). This finding invalidates the same-region-but-different-function interpretation by revealing a coherent pattern in the activation of the left area PF in situations in which participants were supposed to reason technically. We examined the adjacency interpretation by analysing the specific locations of individual peak activations within the left area PF ROI for the mechanical problemsolving task, the psychotechnical task, the fluid-cognition task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task. These peaks, which corresponded to the maximum value of activation obtained for each participant within the left area PF ROI, are reported in Fig. 6. As can be seen, the peaks of the fluid-cognition task were located more anteriorly, in the left area PFt (Parietal Ft) and the postcentral cortex, compared to the peaks of the other four tasks, which were more posterior, in the left area PF. Statistical analyses based on the y coordinates of the individual activation peaks confirmed this description (Fig. 6). Indeed, the y coordinates of the peaks of the mechanical problem-solving task, the psychotechnical task and the PHYS-Only and INT+PHYS conditions of the mentalizing task were posterior to the y coordinates of the peaks of the fluid-cognition task (all p < .05), whereas no significant differences were reported between the four tasks (all p > .05). These findings speak against the adjacency interpretation by revealing that participants recruited the same part of the left area PF to perform tasks involving technical reasoning.” (p. 11-13)

Control tasks may be inadequate: the tasks may involve other factors, such as motor/ actionrelated information. For the psychotechnical task, fluid-cognition task, and mentalizing task, the experiment tasks need not only care about technical-cognition information but also motor-related information, whereas the control tasks do not need to consider motor-related information (mainly visual shape information). Additionally, there may be no difference in motor-related information between the conditions of the fluid-cognition task. Therefore, the regions of interest may be sensitive to motor-related information, affecting the research conclusion.

We thank Reviewer 1 for this comment. We added a specific section in the discussion that addresses this limitation.

“The second limitation concerns the alternative interpretation that the left area PF is not central to technical reasoning but to the storage of sensorimotor programs about the prototypical manipulation of common tools. Here we show that the left area PF is recruited even in situations in which participants do not have to process common manipulable tools. For instance, some items of the psychotechnical task consisted of pictures of tractor, boat, pulley, or cannon. The fact that we found a common activation of the left area PF in such tasks as well as in the mechanical problem-solving task, in which participants could nevertheless simulate the motor actions of manipulating novel tools, indicates that this brain area is not central to tool manipulation but to physical understanding. That being said, some may suggest that viewing a boat or a cannon is enough to incite the simulation of motor actions, so our tasks were not equipped to distinguish between the manipulation-based approach and the reasoning-based approach. We have already shown that the left area PF is more involved in tasks that focus on the mechanical dimension of the tool-use action (e.g., the mechanical interaction between a tool and an object) than its motor dimension (i.e., the interaction between the tool and the effector [e.g., 24, 40]). Nevertheless, we recognize that future research is still needed to test the predictions derived from these two approaches.” (p. 18-19)

-Negative results require further validation: the cognitive results for the fluid-cognition task in the study may need more refinement. For instance, when performing ROI analysis, are there any differences between the conditions? Bayesian statistics might also be helpful to account for the negative results.

We agree that our negative results required further validation. We conducted the ROI analyses suggested by Reviewer 1, which confirmed the initial whole-brain analyses.

“Region of interest (ROI) results. We conducted additional analyses to test the robustness of our findings. One of our results was that we did not report any specific activation of the left area PF in the fluid-cognition task contrary to the mechanical problem-solving task, the psychotechnical task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task. However, this negative result needed exploration at the ROI level. Therefore, we created a spherical ROI of the left area PF with a radius of 12 mm in the MNI standard space (–59; –31; 40). This ROI was literature-defined to ensure the independence of its selection (40). ROI results are shown in Fig. 4. The analyses confirmed the results obtained with the whole-brain analyses by indicating a greater activation of the left area PF in the mechanical problem-solving task, the psychotechnical task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task (all p < .001), but not in the fluid-cognition task (p \= .35).” (p. 10-11)

Reviewer #1 (Recommendations For The Authors):

(1) I may not fully grasp some of the arguments. In the abstract, what does the term "intermediate-level" mean, and why is it an intermediate-level state? In the sentence "the existence of a specific cognitive module in the human brain dedicated to materiality", I cannot see a clear link between technical cognition and the word "materiality".

We used the term materiality to refer to a potential human trait that allows us to shape the physical world according to our ends, by using, making tools and transmiting them to others. This is a reference to Allen et al. (2020; PNAS): “We hope this empirical domain and modeling framework can provide the foundations for future research on this quintessentially human trait: using, making, and reasoning about tools and more generally shaping the physical world to our ends” (p. 29309). Scientists (including archaeologists, economists, psychologists, neuroscientists) interested in human materiality have tended to focus on how we manipulate things according to our thought (motor cognition) or how we conceptualize our behaviour to transmit it to others (language, social cognition). However, little has been said on the intermediate level, that is, technical cognition. We added the term “technical cognition” here, which should help to make the connection more quickly.

“Yet, little has been said about the intermediate-level cognitive processes that are directly involved in mastering this materiality, that is, technical cognition.” (p. 2)

(2) The introduction could provide more details on why the issue of "generalizability and specificity" is important to address, to clarify the significance of the research question.

We followed this comment and added a sentence to explain why it is important to address this research question. Again, we thank Reviewer 1 for their helpful comments.

“Here we focus on two key aspects of the technical-reasoning hypothesis that remain to be addressed: Generalizability and specificity. If technical reasoning is a specific form of reasoning oriented towards the physical world, then it should be implicated in all (the generalizability question) and only (the specificity question) the situations in which we need to think about the physical properties of our world.” (p. 5)

Reviewer #2 (Public Review):

Summary:

The goal of this project was to test the hypothesis that a common neuroanatomic substrate in the left inferior parietal lobule (area PF) underlies reasoning about the physical properties of actions and objects. Four functional MRI (fMRI) experiments were created to test this hypothesis. Group contrast maps were then obtained for each task, and overlap among the tasks was computed at the voxel level. The principal finding is that the left PF exhibited differentially greater BOLD response in tasks requiring participants to reason about the physical properties of actions and objects (referred to as technical reasoning). In contrast, there was no differential BOLD response in the left PF when participants engaged in fMRI variant of the Raven's progressive matrices to assess fluid cognition.

Strengths:

This is a well-written manuscript that builds from extensive prior work from this group mapping the brain areas and cognitive mechanisms underlying object manipulation, technical reasoning, and problem-solving. Major strengths of this manuscript include the use of control conditions to demonstrate there are differentially greater BOLD responses in area PF over and above the baseline condition of each task. Another strength is the demonstration that area PF is not responsive in tasks assessing fluid cognition - e.g., it may just be that PF responds to a greater extent in a harder condition relative to an easy condition of a task. The analysis of data from Task 3 rules out this alternative interpretation. The methods and analysis are sufficiently written for others to replicate the study, and the materials and code for data analysis are publicly available.

We sincerely thank Reviewer 2 for their precious comments, which helped us improve the MS. 

Weaknesses:

The first weakness is that the conclusions of the manuscript rely on there being overlap among group-level contrast maps presented in Figure 2. The problem with this conclusion is that different participants engaged in different tasks. Never is an analysis performed to demonstrate that the PF region identified in e.g., participant 1 in Task 2 is the same PF region identified in Participant 1 in Task 4.

We added new analyses that demonstrated that “the PF region identified in e.g., participant 1 in Task 2 is the same PF region identified in Participant 1 in Task 4”. We thank Reviewer 2 for this comment, because these new analyses reinforced our interpretation.

“The conjunction analysis reported was subject to at least two key limitations that needed to be overcome to assure a correct interpretation of our findings. The first was that the tasks could recruit the same regions for different cognition functions (same-region-but-different-function interpretation). The second was that the activated regions of the different tasks could be adjacent but did not overlap at finer resolutions (adjacency interpretation). We tested the same-region-but-different-function interpretation by conducting additional ROI analyses, which consisted of correlating the specific activation of the left area PF (i.e., difference in terms of mean Blood-Oxygen Level Dependent [BOLD] parameter estimates between the experimental condition minus the control condition) in the psychotechnical task, the fluid-cognition task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task. This analysis did not include the mechanical problem-solving task because the sample of participants was not the same for this task. As shown in Fig. 5, we found significant correlations between all the tasks that were hypothesized as recruiting technical reasoning, i.e., the psychotechnical task and the PHYS-Only and INT+PHYS conditions of the mentalizing task (all p < .05). By contrast, no significant correlation was obtained between these three tasks and the fluid-cognition task (all p > .15). This finding invalidates the same-region-but-different-function interpretation by revealing a coherent pattern in the activation of the left area PF in situations in which participants were supposed to reason technically. We examined the adjacency interpretation by analysing the specific locations of individual peak activations within the left area PF ROI for the mechanical problemsolving task, the psychotechnical task, the fluid-cognition task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task. These peaks, which corresponded to the maximum value of activation obtained for each participant within the left area PF ROI, are reported in Fig. 6. As can be seen, the peaks of the fluid-cognition task were located more anteriorly, in the left area PFt (Parietal Ft) and the postcentral cortex, compared to the peaks of the other four tasks, which were more posterior, in the left area PF. Statistical analyses based on the y coordinates of the individual activation peaks confirmed this description (Fig. 6). Indeed, the y coordinates of the peaks of the mechanical problem-solving task, the psychotechnical task and the PHYS-Only and INT+PHYS conditions of the mentalizing task were posterior to the y coordinates of the peaks of the fluid-cognition task (all p < .05), whereas no significant differences were reported between the four tasks (all p > .05). These findings speak against the adjacency interpretation by revealing that participants recruited the same part of the left area PF to perform tasks involving technical reasoning.” (p. 11-13)

A second weakness is that there is a variance in accuracy between tasks that are not addressed. It is clear from the plots in the supplemental materials that some participants score below chance (~ 50%). This means that half (or more) of the fMRI trials of some participants are incorrect. The methods section does not mention how inaccurate trials were handled. Moreover, if 50% is chance, it suggests that some participants did not understand task instructions and were systematically selecting the incorrect item.

It is true that the experimental conditions were more difficult than the control conditions, with some participants who performed at or below 50% in the experimental conditions. We added a section in the MS to stress this aspect. To examine whether this potential difficulty effect biased our interpretation, we conducted new ROI analyses by removing all the participants who performed at or below the chance level. These analyses revealed the same results as when no participant was excluded, suggesting that this did not bias our interpretation.

“As mentioned above, the experimental conditions of all the tasks were more difficult than their control conditions. As a result, the specific activation of the left area PF documented above could simply reflect that this area responds to a greater extent in a harder condition relative to an easy condition of a task. This interpretation is nevertheless ruled out by the results obtained with the fluid-cognition task. We did not report a specific activation of the left area PF in this task while its experimental condition was more difficult than its control condition. To test more directly this effect of difficulty, we conducted new ROI analyses by removing all the participants who performed at or below 50% (Fig. S2). These new analyses replicated the initial analyses by showing a greater activation of the left area PF in the mechanical problem-solving task, the psychotechnical task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task (all p < .001), but not in the fluid-cognition task (p \= .48). In sum, the ROI analyses corroborated the wholebrain analyses and ruled out the potential effect of difficulty.” (p. 11)

A third weakness is related to the fluid cognition task. In the fMRI task developed here, the participant must press a left or right button to select between 2 rows of 3 stimuli while only one of the 3 stimuli is the correct target. This means that within a 10-second window, the participant must identify the pattern in the 3x3 grid and then separately discriminate among 6 possible shapes to find the matching stimulus. This is a hard task that is qualitatively different from the other tasks in terms of the content being manipulated and the time constraints.

We acknowledge that the fluid-cognition task involved a design that differed from the other tasks. However, this was also true for the other tasks, as the design also differed between the mechanical problem-solving task, the psychotechnical task, and the mentalizing task. Nevertheless, despite these distinctions, we found a consistent activation of the left area PF in these tasks with different designs including in the psychotechnical task, which seemed as difficult as the fluid-cognition task.

“Region of interest (ROI) results. We conducted additional analyses to test the robustness of our findings. One of our results was that we did not report any specific activation of the left area PF in the fluid-cognition task contrary to the mechanical problem-solving task, the psychotechnical task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task. However, this negative result needed exploration at the ROI level. Therefore, we created a spherical ROI of the left area PF with a radius of 12 mm in the MNI standard space (–59; –31; 40). This ROI was literature-defined to ensure the independence of its selection (40). ROI results are shown in Fig. 4. The analyses confirmed the results obtained with the whole-brain analyses by indicating a greater activation of the left area PF in the mechanical problem-solving task, the psychotechnical task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task (all p < .001), but not in the fluid-cognition task (p \= .35).” (p. 10-11)

In sum, this is an interesting study that tests a neuro-cognitive model whereby the left PF forms a key node in a network of brain regions supporting technical reasoning for tool and non-tool-based tasks. Localizing area PF at the level of single participants and managing variance in accuracy is critically important before testing the proposed hypotheses.

We thank Reviewer 2 for this positive evaluation and their suggestions. As detailed in our response, our revision took into consideration both the localization of the left area PF at the level of single participants and the variance in accuracy. 

Reviewer #2 (Recommendations For The Authors):

Did the fMRI data undergo high-pass temporal filtering prior to modeling the effects of interest? Participants engaged in a long (17-24 minutes) run of fMRI data collection. Highpass filtering of the data is critically important when managing temporal autocorrelation in the fMRI response (e.g., see Shinn et al., 2023, Functional brain networks reflect spatial and temporal autocorrelation. Nature Neuroscience).

Yes. We added this information.

“Regressors of non-interest resulting from 3D head motion estimation (x, y, z translation and three axes of rotation) and a set of cosine regressors for high-pass filtering were added to the design matrix.” (p. 25-26)

Including scales in Figure 2 would help the reader interpret the magnitude of the BOLD effects.

We added this information in Figure 3 (Figure 2 in the initial version of the MS).

It was difficult to inspect the small thumbnail images of the task stimuli in Figure 1. Higher resolution versions of those stimuli would help facilitate understanding of the task design and trial structure.

We changed both Figure 1 and Figure S1.

Reviewer #3 (Public Review):

Summary:

This manuscript reports two neuroimaging experiments assessing commonalities and differences in activation loci across mechanical problem-solving, technical reasoning, fluid cognition, and "mentalizing" tasks. Each task includes a control task. Conjunction analyses are performed to identify regions in common across tasks. As Area PF (a part of the supramarginal gyrus of the inferior parietal lobe) is involved across 3 of the 4 tasks, the investigators claim that it is the hub of technical cognition.

Strengths:

The aim of finding commonalities and differences across related problem-solving tasks is a useful and interesting one.

The experimental tasks themselves appear relatively well-thought-out, aside from the concern that they are differentially difficult.

The imaging pipeline appears appropriate.

We thank Reviewer 3 for their constructive comments, which helped us improve the MS.

Weaknesses:

(1) Methodological

As indicated in the supplementary tables and figures, the experimental tasks employed differ markedly in 1) difficulty and 2) experimental trial time. Response latencies are not reported (but are of additional concern given the variance in difficulty). There is concern that at least some of the differences in activation patterns across tasks are the result of these fundamental differences in how hard various brain regions have to work to solve the tasks and/or how much of the trial epoch is actually consumed by "on-task" behavior. These difficulty issues should be controlled for by 1) separating correct and incorrect trials, and 2) for correct trials, entering response latency as a regressor in the Generalized Linear Models, 3) entering trial duration in the GLMs.

We thank Reviewer 3 for this comment. It is true that the experimental conditions were more difficult than the control conditions, with some participants who performed at or below 50% in the experimental conditions. We added a section in the MS to stress this aspect. We could not conduct new analyses by separating correct and incorrect trials because, for each task, participants had to respond only on the last item of the block. Therefore, we did not record a response for each event. Nevertheless, we could examine whether this potential difficulty effect biased our interpretation, by conducting new ROI analyses in which we removed all the participants who performed at or below the chance level. These analyses revealed the same results as when no participant was excluded, suggesting that this did not bias our interpretation. 

“As mentioned above, the experimental conditions of all the tasks were more difficult than their control conditions. As a result, the specific activation of the left area PF documented above could simply reflect that this area responds to a greater extent in a harder condition relative to an easy condition of a task. This interpretation is nevertheless ruled out by the results obtained with the fluid-cognition task. We did not report a specific activation of the left area PF in this task while its experimental condition was more difficult than its control condition. To test more directly this effect of difficulty, we conducted new ROI analyses by removing all the participants who performed at or below 50% (Fig. S2). These new analyses replicated the initial analyses by showing a greater activation of the left area PF in the mechanical problem-solving task, the psychotechnical task, and the PHYS-Only and INT+PHYS conditions of the mentalizing task (all p < .001), but not in the fluid-cognition task (p \= .48). In sum, the ROI analyses corroborated the wholebrain analyses and ruled out the potential effect of difficulty.” (p. 11)

A related concern is that the control tasks also differ markedly in the degree to which they were easier and faster than their corresponding experimental task. Thus, some of the control tasks seem to control much better for difficulty and time on task than others. For example, the control task for the psychotechnical task simply requires the indication of which array contains a simple square shape (i.e., it is much easier than the psychotechnical task), whereas the control task for mechanical problem-solving requires mentally fitting a shape into a design, much like solving a jigsaw puzzle (i.e., it is only slightly easier than the experimental task).

It is true that some control conditions could be easier than other ones. These differences reinforced the common activation found in the left area PF in the tasks hypothesized as involving technical reasoning, because this activation survived irrespective of the differences in terms of experimental design. For us, the rationale is the same as for a meta-analysis, in which we try to find what is common to a great variety of tasks. The only detrimental consequence we identified here is that this difference explained why we did not report a specific activation of the left area PF in the fluid-cognition task, as if the left area PF was more responsive when the task was difficult. This possibility assumes that the experimental condition of the fluid-cognition task is much more difficult than its control condition compared to what can be seen in the other tasks. As Reviewer 2 stressed in Point 1, this interpretation is unlikely, because the differences between the experimental and control conditions were similar to the fluid-cognition task in the mechanical problem-solving and psychotechnical tasks. In addition, again, the new ROI analyses in which we removed all the participants who performed at or below the chance level in expetimental conditions reproduced our initital results.

(2) Theoretical 

The investigators seem to overlook prior research that does not support their perspective and their writing seems to lack scientific objectivity in places. At times they over-reach in the claims that can be made based on the present data. Some claims need to be revised/softened.

As this comment is also mentioned below, please find our response to it below.

Reviewer #3 (Recommendations For The Authors):

(1) Because of the high level of detail, Figures 1 and S2 (particularly the mentalizing task and mechanical problem-solving task, and their controls) are very hard to parse, even when examined relatively closely. It is suggested that these figures be broken down into separate panels for Experiment 1 and Experiment 2 to facilitate understanding.

We changed both Figure 1 and Figure S1.

(2) The behavioral data (including response latencies) should be reported in the main results section of the paper and not in a supplement.

The behavioural data are now reported in the main results. We did not report response latencies because participants were not prompted to respond as quickly as possible.

“Behavioural results. All the behavioural results are given in Fig. 2. As shown, scores were higher in the experimental conditions than for the control conditions for all the tasks (all p < .05). In other words, the experimental conditions were more difficult than the control conditions. This difference in terms of difficulty can also be illustrated by the fact that some participants performed at or below the chance level in the experimental conditions whereas none did so in the control conditions.” (p. 8)

(3) The investigators seem to overlook prior research that does not support their perspective and their writing seems to lack scientific objectivity in places. At times they over-reach in the claims that can be made based on the present data. For example, claims that need to be revised/softened include:

Abstract: "Area PF... can work along with social-cognitive skills to resolve day-to-day interactions that combine social and physical constraints". This statement is overly speculative.

This statement is based on the fact that we reported a combined activation of the technical-reasoning network and the mentalizing network in the INT+PHYS condition of the mentalizing task. This suggests that both networks need to work together for solving a day-today problem in which both the physical constraints of the situation and the intention of the individual must be integrated. Our findings replicated previous ones with a similar task (e.g., Brunet et al. 2000; Völlm et al., 2006), in which the authors gave an interpretation similar to ours in considering that this task requires understanding physical and social causes. Perhaps that the reference to the results of the mentalizing task was not explicit enough. We added “dayto-day” before “problem” in the part of the discussion in which we discuss this possibility to make this aspect clearer.

“In broad terms, the results of the mentalizing task indicate that causal reasoning has distinct forms and that it recruits distinct networks of the human brain (Social domain: Mentalizing; Physical domain: Technical reasoning), which can nevertheless interact together to solve day-to-day problems in which several domains are involved, such as in the INT+PHYS condition of the mentalizing task.” (p. 16)

Introduction: "The manipulation-based approach... remains silent on the more general cognitive mechanisms...that must also encompass the use of unfamiliar or novel tools". This statement seems to be based on an overly selective literature review. There are a number of studies in which the relationship between a novel and familiar tool selection/use has been explored (e.g., Buchman & Randerath, 2017; Mizelle & Wheaton, 2010; Silveri & Ciccarelli, 2009; Stoll, Finkel et al., 2022; Foerster, 2023; Foerster, Borghi, & Goslin, 2020; Seidel, Rijntjes et al., 2023).

We thank Reviewer 3 for this comment. Even if we accept the idea that we possess specific sensorimotor programs about tool manipulation, it remains that these programs cannot explain how an individual decides to bend a wire to make a hook or to pour water in a recipient to retrieve a target. As a matter of fact, such behaviour has been reported in nonhuman animals, such as crows (Weir et al., 2002, Nature) or orangutans (Mendes et al., 2007, Biology Letters). In these studies, the question is whether these nonhuman animals understand the physical causes or not, but the question of sensorimotor programs is never addressed (to our knowledge). This is also true in developmental studies on tool use (e.g., Beck et al., 2011, Cognition; Cutting et al., 2011, Journal of Experimental Child Psychology). This is what we meant here, that is, the manipulation-based approach is not equipped to explain how people solve physical problems by using or making tools – or any object – or by building constructions or producing technical innovations. However, we agree that some papers have been interested in exploring the link between common and novel tool use and have suggested that both could recruit common sensorimotor programs. It is noteworthy that these studies do not test the predictions from the manipulation-based approach versus the reasoning-based approach, so both interpretations are generally viable as stressed by Seidel et al. (2023), one of the papers recommended by Reviewer 3.

“Apparently, the presentation of a graspable object that is recognizable as a tool is sufficient to provoke SMG activation, whether one tends to see the function of SMG to be either “technical reasoning” (Osiurak and Badets 2016; Reynaud et al. 2016; Lesourd et al. 2018; Reynaud et al. 2019) or “manipulation knowledge” (Sakreida et al. 2016; Buxbaum 2017; Garcea et al. 2019b).” (Seidel et al., 2023; p. 9)

Regardless, as suggested by Reviewer 3, these papers deserve to be cited and this part needed to be rewritten to insist on the “making, construction, and innovation” dimension more than on the “unfamiliar and novel tool use” dimension to avoid any ambiguity.

“This manipulation-based approach has provided interesting insights (12–16) and even elegant attempts to explain how these sensorimotor programs could support the use of both unfamiliar or novel tools (17–20), but remains silent on the more general cognitive mechanisms behind human technology that include the use of common and unfamiliar or novel tools but must also encompass tool making, construction behaviour, technical innovations, and transmission of technical content.” (p. 3)

Introduction: "Here we focus on two important questions... to promote the technicalreasoning hypothesis as a comprehensive cognitive framework..."(italics added). This and other similar statements should be rewritten as testable scientific hypotheses rather than implying that the point of the research is to promote the investigators' preferred view.

We agree that our phrasing could seem inappropriate here. What we meant here is that the technical-reasoning hypothesis could become an interesting framework for the study of the cognitive bases of human technology only if we are able to verify some of its key facets. As suggested, we rewrote this part. We also rewrote the abstract and the first paragraph of the discussion.

“Here we focus on two key aspects of the technical-reasoning hypothesis that remain to be addressed: Generalizability and specificity. If technical reasoning is a specific form of reasoning oriented towards the physical world, then it should be implicated in all (the generalizability question) and only (the specificity question) the situations in which we need to think about the physical properties of our world.” (p. 5)

Introduction: The Goldenberg and Hagmann paper cited actually shows that familiar tool use may be based either on retrieval from semantic memory or by inferring function from structure (mechanical problem solving); in other words, the investigators saw a role for both kinds of information, and the relationship between mechanical problem solving and familiar tool use was actually relatively weak. This requires correction.

We disagree with Reviewer 3 on this point. The whole sentence is as follows:

“This silence has been initially broken by a series of studies initiated by Goldenberg and Hagmann (9), which has documented a behavioural link in left brain-damaged patients between common tool use and the ability to solve mechanical problems by using and even sometimes making novel tools (e.g., extracting a target out from a box by bending a wire to create a hook) (9, 17).” (p. 3-4)

We did not mention the interpretations given by Goldenberg and Hagmann about the link with the pantomime task, but only focused on the link they reported between common tool use and novel tool use. This is factual. In addition, we also disagree that the link between common tool use and novel tool use was weak.

“The hypothesis put forward in the introduction predicts that knowledge about prototypical tool use assessed by pantomime of tool use and the ability to infer function from structure assessed by novel tool selection can both contribute to the use of familiar tools. Indeed results of both tests correlated signicantly with the use of familiar tools pantomime of tool use: r \= 0.77, novel tool selection: r \= 0.62; both P < 0.001), but there was also a signicant correlation between the two tests r \= 0.64, P < 0.001).” (Goldenberg & Hagmann, 1998; p. 585)

As can be seen in this quote, they reported a significant correlation between novel tool selection and the use of familiar tools. It is also noteworthy that the novel tool selection test and the pantomime test correlated together. Georg Goldenberg told one of the authors (F. Osiurak; personal communication) that this result incited him to revise its idea that pantomime could assess “semantic knowledge”, which explains why he did not use it again as a measure of semantic knowledge. Instead, he preferred to use a classical semantic matching task in his 2009 Brain paper with Josef Spatt, in which they found a clearer dissociation between semantic knowledge and common/novel tool use not only at the behavioral level but also at the cerebral level.

Introduction: Please expand and clarify this sentence "However, this involvement seems to be task-dependent, contrary to the systematic involvement of left are PF. The IFG and LOTC activations observed in prior studies are of interest as well. Were they indeed all taskdependent in these studies?

We agree that this sentence is confusing. We meant that, in the studies reported just above in the paragraph, these regions were not systematically reported contrary to the left area PF. As we think that this information was not crucial for the logic of the paper, we preferred to remove it. 

Introduction: If implicit mechanical knowledge is acquired through interactions with objects, how is that implicit knowledge conveyed to pass on the material culture to others?

We thank Reviewer 3 for this comment. Although mechanical knowledge is implicit, it can be indirectly transmitted to other individuals, as shown in two papers we published in Nature Human Behaviour (Osiurak et al., 2021) and Science Advances (Osiurak et al., 2022). Actually, verbal teaching is not the only way to transmit information. There are many other ways of transmitting information such as gestural teaching (e.g., pointing the important aspects of a task to make them salient to the learner), observation without teaching (i.e., when we observe someone unbeknown to them) or reverse engineering (i.e., scrutinizing an artifact made by someone else). We have shown that even in reverse-engineering conditions, participants can benefit from what previous participants have done to increase their understanding of a physical system. In other words, all these forms of transmission allow the learners to understand new physical relationships without waiting that these relationships randomly occur in the environment. There is a wide literature on social learning, which describes very well how knowledge can be transmitted without using explicit communication. In fact, it is very likely that such forms of transmission were already present in our ancestors, allowing them to start accumulating knowledge without using symbolic language. We did not add this information in the MS because we think that this was a little bit beyond the scope of the MS. Nevetheless, we cited relevant literature on the topic to help the reader find it if interested in the topic.

“Yet, recent accounts have proposed that non-social cognitive skills such as causal understanding or technical reasoning might have played a crucial role in cumulative technological culture (6, 29, 66). Support for these accounts comes from micro-society experiments, which have demonstrated that the improvement of technology over generations is accompanied by an increase in its understanding (67, 68), or that learners’ technical-reasoning skills are a good predictor of cumulative performance in such micro-societies (33, 69).” (p. 19)

What distinguishes this implicit mechanical knowledge from stored knowledge about object manipulation? Are these two conceptualizations really demonstrably (testably) different?

We agree that it is complex to distinguish between these two hypotheses as suggested by Seidel et al. (2023) cited above (see Reviewer 3 Point 8). We have conducted several studies to test the opposite predictions derived from each hypothesis. The main distinction concerns the understanding of physical materials and forces, which is central to the technical-reasoning hypothesis but not to the manipulation-based approach. Indeed, sensorimotor programs about tool manipulation are not assumed to contain information about physical materials and forces. In the present study, the understanding of physical materials and forces was needed in the four tasks hypothesized as requiring technical reasoning, i.e., the mechanical problem-solving task, the psychotechnical task and the PHYS-Only and INT+PHYS conditions of the mentalizing task. We can illustrate this aspect with items of each of these tasks. Figure 1A is of the mechanical problem-solving task. 

As explained in the MS, participants had memorized the five possible tools before the scanner session. Thus, for 4 seconds, they had to imagine which of these tools could be used to extract the target out from the box. We did so to incit them to reason about mechanical solutions based on the physical properties of the problem. Then, they had 3 seconds to select the tool with the appropriate shape, here the right one. In this case, the motor action remains the same (i.e., pulling). Another illustration can be given, with the psychotechnical task (Figure 1B).

In this task, the participant had to reason as to whether the boat-tractor connection was better in the left picture or in the right picture. This needs to reason about physical forces, but there is no need to recruit sensorimotor programs about tool manipulation. Finally, a last example can be given with the PHYS-Only condition of the mentalizing task (but the logic is the same for the INT+PHYS condition except that the character’s intentions must also be taken into consideration) Figure 1D).

Here the participant must reason about which picture shows what is physically possible. In this task, there is no need to recruit sensorimotor programs about tool manipulation. In sum, what is common between these three tasks is the requirement to reason about physical materials and forces. We do not ignore that motor actions could be simulated in the mechanical problemsolving task, but no motor action needed to be simulated in the other three tasks. Therefore, what was common between all these tasks was the potential involvement of technical reasoning but not of sensorimotor programs about tool manipulation. Of course, an alternative is to consider that motor actions are always needed in all the situations, including situations where no “manipulable tool” is presented, such as a tractor and a boat, a pulley, or a cannon. We cannot rule out this alternative, which is nevertheless, for us, prejudicial because it implies that it becomes difficult to test the manipulation-based approach as motor actions would be everywhere. We voluntarily decided not to introduce a debate between the reasoning-based approach and the manipulation-based approach and preferred a more positive writing by stressing the insights from the present study. Note that we stressed the merits of the manipulation-based approach in the introduction because we sincerely think that this approach has provided interesting insights. However, we voluntarily did not discuss the debate between the two approaches. Given Reviewer 3’s comment (see also Reviewer 1 Point 2), we understand and agree that some words must be nevertheless said to discuss how the manipulation-based approach could interpret our results, thus stressing the potential limitations of our interpretations. Therefore, we added a specific section in the discussion in which we discussed this aspect in more details.

“The second limitation concerns the alternative interpretation that the left area PF is not central to technical reasoning but to the storage of sensorimotor programs about the prototypical manipulation of common tools. Here we show that the left area PF is recruited even in situations in which participants do not have to process common manipulable tools. For instance, some items of the psychotechnical task consisted of pictures of tractor, boat, pulley, or cannon. The fact that we found a common activation of the left area PF in such tasks as well as in the mechanical problem-solving task, in which participants could nevertheless simulate the motor actions of manipulating novel tools, indicates that this brain area is not central to tool manipulation but to physical understanding. That being said, some may suggest that viewing a boat or a cannon is enough to incite the simulation of motor actions, so our tasks were not equipped to distinguish between the manipulation-based approach and the reasoning-based approach. We have already shown that the left area PF is more involved in tasks that focus on the mechanical dimension of the tool-use action (e.g., the mechanical interaction between a tool and an object) than its motor dimension (i.e., the interaction between the tool and the effector [e.g., 24, 40]). Nevertheless, we recognize that future research is still needed to test the predictions derived from these two approaches.” (p. 18-19)

Introduction and throughout: The framing of left Area PF as a special area for technical reasoning is overly reductionistic from a functional neuroanatomic perspective in that it ignores a large relevant literature showing that the region is involved with many other tasks that seem not to require anything like technical cognition. Indeed, entering the coordinates - 56, -29, 36 (reported as the peak coordinates in common across the studied tasks) in Neurosynth reveals that 59 imaging studies report activations within 3 mm of those coordinates; few are action-related (a brief review indicated studies of verbal creativity, texture processing, reading, somatosensory processing, stress reactions, attentional selection etc). Please acknowledge the difficulty of claiming that a large brain region should be labeled the brain's technical reasoning area when it seems to also participate in so much else. The left IPL (including area PF) is densely connected to the ventral premotor cortex, and this network is activated in language and calculation tasks as well as tool use tasks (e.g., Matsumoto, Nair, et al., 2012). What other constructs might be able to unite this disparate literature, and are any of these alternative constructs ruled out by the present data? Lacking this objective discussion, the manuscript does read as a promotion of the investigators' preferred viewpoint.

We thank Reviewer 3 for this comment. As stressed in the initial version of the MS, we did not write that the left area PF is sufficient but central to the network that allows us to reason about the physical world. Regardless, we agree that an objective discussion was needed on this aspect to help the reader not misunderstand our purpose. We added a section in this aspect as suggested. 

“Before concluding, we would like to point out two potential limitations of the present study. The first limitation concerns the fact that the literature has documented the recruitment of the left area PF in many neuroimaging experiments in which there was no need to reason about physical events (e.g., language tasks). This can be easily illustrated by entering the left area PF coordinates in the Neurosynth database.

This finding could be enough to refute the idea that this brain area is specific to technical reasoning. Although this limitation deserves to be recognized, it is also true for many other findings. For instance, sensory or motor brain regions such as the precentral or the postcentral cortex have been found activated in many non-motor tasks, the visual word form area in non-language tasks, or the Heschl’s gyrus in nonmusical tasks. This remains a major challenge for scientists, the question being how to solve these inconsistencies that can result from statistical errors or stress that considerable effort is needed to understand the very functional nature of these brain areas. Thus, understanding that the left area PF is central to physical understanding can be viewed as a first essential step before discovering its fundamental function, as suggested by the functional polyhedral approach (56).” (p. 18)

Discussion: The discussion of a small cluster in the IFG (pars opercularis) that nearly survived statistical correction is noteworthy in light of the above point. This further underscores the importance of discussing networks and not just single brain regions (such as area PF) when examining complex processes. The investigators note, "a plausible hypothesis is that the left IFG integrates the multiple constraints posed by the physical situation to set the ground for a correct reasoning process, such as it could be involved in syntactic language processing". In fact, the hypothesis that the IFG and SMG are together related to resolving competition has been previously proposed, as has the more specific hypothesis that the SMG buffers actions and that the context-appropriate action is then selected by the IFG (e.g., Buxbaum & Randerath, 2018). The parallels with the way the SMG is engaged with competing lexical or phonological alternatives (e.g., Peramunage, Blumstein et al., 2011) have also been previously noted.

We added the Buxbaum and Randerath (2018)’s reference in this section.

“The functional role of the left IFG in the context of tool use has been previously discussed (24) and a plausible hypothesis is that the left IFG integrates the multiple constraints posed by the physical situation to set the ground for a correct reasoning process, such as it could be involved in syntactic language processing (for a somewhat similar view, see [51]).” (p. 16-17)

Introduction and Discussion: Please clarify how the technical reasoning network overlaps with or is distinct from the tool-use network reported by many previous investigators.

We added a couple of sentences in the discussion to clarify this point.

“It should be clear here that we do not advocate the localizationist position simply stating that activation in the left area PF is the necessary and sufficient condition for technical reasoning. We rather defend the view according to which it requires a network of interacting brain areas, one of them – and of major importance – being the left area PF. This allows the engagement of different configurations of cerebral areas in different technical-reasoning tasks, but with a central process acting as a stable component: The left area PF. Thus, when people intend to use physical tools, it can work in concert with brain regions specific to object manipulation and motor control, thereby forming another network, the tool-use network. It can also interact with brain regions specific to intentional gestures to form a “social-learning” network that allows people to enhance their understanding about the physical aspects of a technical task (e.g., the making of a tool) through communicative gestures such as pointing gestures (42). The major challenge for future research is to specify the nature of the cognitive process supported by the left area PF and that might be involved in the broad understanding of the physical world.” (p. 14)

Discussion: All of the experimental tasks require a response from a difficult choice in an array, and all of the tasks except for the fluid cognition task are likely to require prediction or simulation of a motion trajectory-whether an embodied or disembodied trajectory is unclear. The Discussion does mention the related (but distinct) idea of an "intuitive physics engine", a "kind of simulator", Please clarify how this study can rule out these alternative interpretations of the data. If the study cannot rule out these alternatives, the claims of the study (and the paper title which labels PF as a technical cognition area) should be scaled back considerably. 

We thank Reviewer 3 for this comment. The authors of the papers on intuitive physics engine or associative learning do not suggest that these processes are embodied. As discussed above, we clarified our perspective on the role of the left area PF and hope that these modifications help the reader better understand it. We warmly thank Reviewer 3 for their comments, which considerably helped us improve the MS.

Author response:

The following is the authors’ response to the original reviews

Reviewer #1 (Public review):

Hüppe and colleagues had already developed an apparatus and an analytical approach to capture swimming activity rhythms in krill. In a previous manuscript they explained the system, and here they employ it to show a circadian clock, supplemented by exogenous light, produces an activity pattern consistent with "twilight" diel vertical migration (DVM; a peak at sunset, a midnight sink, and a peak in the latter half of the night).

They used light:dark (LD) followed by dark:dark (DD) photoperiods at two times of the year to confirm the circadian clock, coupled with DD experiments at four times of year to show rhythmicity occurs throughout the year along with DVM in the wild population. The individual activity data show variability in the rhythmic response, which is expected. However, their results showed rhythmicity was sustained in DD throughout the year, although the amplitude decayed quickly. The interpretation of a weak clock is reasonable, and they provide a convincing justification for the adaptive nature of such a clock in a species that has a wide distributional range and experiences various photic environments. These data also show that exogenous light increases the activity response and can explain the morning activity bouts, with the circadian clock explaining the evening and late-night bouts. This acknowledgement that vertical migration can be driven by multiple proximate mechanisms is important.

The work is rigorously done, and the interpretations are sound. I see no major weaknesses in the manuscript. Because a considerable amount of processing is required to extract and interpret the rhythmic signals (see Methods and previous AMAZE paper), it is informative to have the individual activity plots of krill as a gut check on the group data.

The manuscript will be useful to the field as it provides an elegant example of looking for biological rhythms in a marine planktonic organism and disentangling the exogenous response from the endogenous one. Furthermore, as high latitude environments change, understanding how important organisms like krill have the potential to respond will become increasingly important. This work provides a solid behavioral dataset to complement the earlier molecular data suggestive of a circadian clock in this species.

We appreciate the positive evaluation of our work by Reviewer 1, acknowledging our approach to record locomotor activity in krill and the importance of the findings in assessing krill’s potential to respond to environmental change in their habitat.

Reviewer #2 (Public review):

Summary:

This manuscript provides experimental evidence on circadian behavioural cycles in Antarctic krill. The krill were obtained directly from krill fishing vessels and the experiments were carried out on board using an advanced incubation device capable of recording activity levels over a number of days. A number of different experiments were carried out where krill were first exposed to simulated light:dark (L:D) regimes for some days followed by continuous darkness (DD). These were carried out on krill collected during late autumn and late summer. A further set of experiments was performed on krill across three different seasons (summer, autumn, winter), where incubations were all DD conditions. Activity was measured as the frequency by which an infrared beam close to the top of the incubation tube was broken over unit time. Results showed that patterns of increased and decreased activity that appeared synchronised to the LD cycle persisted during the DD period. This was interpreted as evidence of the operation of an internal (endogenous) clock. The amplitude of the behavioural cycles decreased with time in DD, which further suggests that this clock is relatively weak. The authors argued that the existence of a weak endogenous clock is an adaptation to life at high latitudes since allowing the clock to be modulated by external (exogenous) factors is an advantage when there is a high degree of seasonality. This hypothesis is further supported by seasonal DD experiments which showed that the periodicity of high and low activity levels differed between seasons.

Strengths

Although there has been a lot of field observations of various circadian type behaviour in Antarctic krill, relatively few experimental studies have been published considering this behaviour in terms of circadian patterns of activity. Krill are not a model organism and obtaining them and incubating them in suitable conditions are both difficult undertakings. Furthermore, there is a need to consider what their natural circadian rhythms are without the overinfluence of laboratory-induced artefacts. For this reason alone, the setup of the present study is ideal to consider this aspect of krill biology. Furthermore, the equipment developed for measuring levels of activity is well-designed and likely to minimise artefacts.

We would like to thank Reviewer 2 for their positive assessment of our approach to study the influence of the circadian clock on krill behavior. We are delighted, that Reviewer 2 found our mechanistic approach in understanding daily behavioral patterns of Antarctic krill using the AMAZE set-up convincing, and that the challenging circumstances of working with a polar, non-model species are acknowledged.

Weaknesses

I have little criticism of the rationale for carrying out this work, nor of the experimental design. Nevertheless, the manuscript would benefit from a clearer explanation of the experimental design, particularly aimed at readers not familiar with research into circadian rhythms. Furthermore, I have a more fundamental question about the relationship between levels of activity and DVM on which I will expand below. Finally, it was unclear how the observational results made here related to the molecular aspects considered in the Discussion.

(1) Explanation of experimental design - I acknowledge that the format of this particular journal insists that the Results are the first section that follows the Introduction. This nevertheless presents a problem for the reader since many of the concepts and terms that would generally be in the Methods are yet to be explained to the reader. Hence, right from the start of the Results section, the reader is thrown into the detail of what happened during the LD-DD experiments without being fully aware of why this type of experiment was carried out in the first place. Even after reading the Methods, further explanation would have been helpful. Circadian cycle type research of this sort often entrains organisms to certain light cycles and then takes the light away to see if the cycle continues in complete darkness, but this critical piece of knowledge does not come until much later (e.g. lines 369-372) leaving the reader guessing until this point why the authors took the approach they did. I would suggest the following (1) that more effort is made in the Introduction to explain the exact LD/DD protocols adopted (2) that a schematic figure is placed early on in the manuscript where the protocol is explained including some logical flow charts of e.g. if behavioural cycle continues in DD then internal clock exists versus if cycle does not continue in DD, the exogenous cues dominate - followed by - major decrease in cyclic amplitude = weak clock versus minor decrease = strong clock and so on

We want to thank Reviewer 2 for pointing out that the experimental design and its rationale are not becoming clear early in the manuscript, especially for people outside the field of chronobiology. We added a new figure (now Fig. 1), illustrating the basic principle of chronobiological study design and how we adopted it. We also extended the description at the beginning of the Results section to clarify the rationale behind the experimental design.

(2) Activity vs kinesis - in this study, we are shown data that (i) krill have a circadian cycle - incubation experiments; (ii) that krill swarms display DVM in this region - echosounder data (although see my later point). My question here is regarding the relationship between what is being measured by the incubation experiments and the in situ swarm behaviour observations. The incubation experiments are essentially measuring the propensity of krill to swim upwards since it logs the number of times an individual (or group) break a beam towards the top of the incubation tube. I argue that krill may be still highly active in the rest of the tube but just do not swim close to the surface, so this approach may not be a good measure of "activity". Otherwise, I suggest a more correct term of what is being measured is the level of "upward kinesis". As the authors themselves note, krill are negatively buoyant and must always be active to remain pelagic. What changes over the day-night cycle is whether they decide to expend that activity on swimming upwards, downwards or remaining at the same depth. Explaining the pattern as upward kinesis then also explains by swarms move upwards during the night. Just being more active at night may not necessarily result in them swimming upwards.

We believe there is a slight misunderstanding in how what we call “activity” is measured. The experimental columns are equipped with five detector modules, evenly distributed over the height of the column. In our analysis we count all beam breaks caused by upward movement, i.e. every time a detector module is triggered after a detector module at a lower position has been triggered, and not only when the top detector module is triggered. In this way, we record upward swimming movements throughout the column, and not only when the krill swims all the way to the top of the column. This still means that what we are measuring is swimming activity, caused by upward swimming. We use this measure, to deliberately separate increased swimming activity, from baseline activity (i.e. swimming, which solely compensates for negative buoyancy) and inactivity (i.e. passive sinking).

Higher activity is thus at first interpreted as an increase in swimming activity, which in the field may result in upwards-directed swimming but also could mean a horizontal increase in activity, for example, representing increased foraging and feeding activity. This would explain the daily activity pattern observed under LD cycles (now Fig. 3), which shows a general increase in activity during the dark phase. This nighttime increase could be used for both upward directed migration during sunset and horizontal directed swimming for feeding and foraging throughout the night.

We added the following sentence to the description of the activity metric in the Methods section to clarify this point (lines 465-469):

“To accomplish this, we organized the raw beam break data from all five detector modules in each experimental column in chronological order. We selected only those beam break detections that occurred after a detection in the detector module positioned lower on the column. Like this, we consider upward swimming movements throughout the full height of the column.”

(3) Molecular relevance - Although I am interested in molecular clock aspects behind these circadian rhythms, it was not made clear how the results of the present study allow any further insight into this. In lines 282 to 284, the findings of the study by Biscontin et al (2017) are discussed with regard to how TIM protein is degraded by light via the clock photreceptor CRYTOCHROME 1. This element of the Discussion would be a lot more relevant if the results of the present study were considered in terms of whether they supported or refuted this or any other molecular clock model. As it stands, this paragraph is purely background knowledge and a candidate for deletion in the interest of shortening the Discussion.

We agree that this part is not directly related to the data presented in the manuscript. We, therefore, omitted this part in the revised version of the manuscript to keep the discussion concise and focused on the results.

Other aspects

(i) 'Bimodal swimming' was used in the Abstract and later in the text without the term being fully explained. I could interpret it to mean a number of things so some explanation is required before the term is introduced.

We thank the Reviewer for pointing this out. We provided an explanation for the term “bimodal” in the Results section, where the two clock driven activity bouts are described first, by extending the sentence in lines 161-164, which now reads:

“This suggests that the circadian clock drives a distinct bimodal activity pattern with two activity peaks in one day, i.e. the evening and late-night activity bouts, while. In contrast, the morning activity bout is triggered by the onset of illumination in the experimental set-up.”.

(ii) Midnight sinking - I was struck by Figure 2b with regards to the dip in activity after the initial ascent, as well as the rise in activity predawn. Cushing (1951) Biol Rev 26: 158-192 describes the different phases of a DVM common to a number of marine organisms observed in situ where there is a period of midnight sinking following the initial dusk ascent and a dawn rise prior to dawn descent. Tarling et al (2002) observe midnight sinking pattern in Calanus finmarchicus and consider whether it is a response to feeding satiation or predation avoidance (i.e. exogenous factors). Evidence from the present study indicates that midnight sinking (and potential dawn rise) behaviour could alternatively be under endogenous control to a greater or lesser degree. This is something that should certainly be mentioned in the Discussion, possibly in place of the molecular discussion element mentioned above - possibly adding to the paragraph Lines 303-319.

We would like to thank the Reviewer for pointing this out and agree that adding the idea of an endogenous control of midnight sinking would be interesting to the discussion. We added the following section to the Discussion (lines 335-343):

“Interestingly, the decrease in clock-controlled swimming activity during the early night, right after the evening activity bout, may further facilitate a phenomenon called “midnight sinking”, which describes the sinking of animals to intermediate depths after the evening ascent, followed by a second rise to the surface before the morning descend. This behavior has been observed in a number of zooplankton species, including calanoid copepods (see 69, 70 and references therein) and krill (71). While previous studies suggested several exogenous factors, such as satiation or predator presence, as drivers of the midnight sink (69, 70), our study suggests that this pattern may be partly under endogenous control.”

(iii) Lines 200-207 - I struggled to follow this argument regarding Piccolin et al identifying a 12 h rhythm whereas the present study indicates a ~24 h rhythm. Is one contradicting the other - please make this clear.

In our study, we found that the circadian clock drives a bimodal pattern of swimming activity in krill, meaning it controls two bouts of activity in a 24-hour cycle. Piccolin et al. (2020) identified a swimming activity pattern of ~12 h (i.e. two peaks in 24 h) at the group level, which aligns with our findings at the individual level. We revised the Section in the discussion for more clarity, which now reads:

“Data from Piccolin et al. (20) showed a strong damping of the amplitude and indication of a remarkably short (~12 h) free running period (FRP) of vertical swimming behavior of a group of krill under constant darkness (20). The short period found in Piccolin et al. (20) complements is in line with our findings of a bimodal activity pattern the pattern of swimming activity under DD conditions on the individual level found in the present study, suggesting that the ~12 h rhythm in group swimming behavior in Piccolin et al. (20) could have resulted from a bimodal activity pattern at the individual level, as found in our study.” (lines 212-219).  

(iv) Although I agree that the hydroacoustic data should be included and is generally supportive of the results, I think that two further aspects should be made clear for context (a) whether there was any groundtruthing that the acoustic marks were indeed krill and not potentially some other group know to perform DVM such as myctophids (b) how representative were these patterns - I have a sense that they were heavily selected to show only ones with prominent DVM as opposed to other parts of the dataset where such a pattern was less clear - I am aware of a lot of krill research where DVM is not such a clear pattern and it is disingenuous to provide these patterns as the definitive way in which krill behaves. I ask this be made clear to the reader (note also that there is a suggestion of midnight sinking in Fig 5b on 28/2).

To clarify the mentioned points concerning the hydroacoustic data:

a) As mentioned in the Methods section, only hydroacoustic data during active fishing was included in the analysis. E. superba occurs in large monospecific aggregations, and the fishery actively targets E. superba and monitors their catch and the proportion of non-target species continuously with cameras. Krill fishery bycatch rates are very low (0.1–0.3%, Krafft et al. 2022), and fishing operations would stop if non-target species were caught in significant proportions at any time. Therefore, and supported by our own observations when we conducted the experiments, we argue that it is a valid assumption that E. superba predominantly causes the backscattering signal shown in Figure 5 (now Fig. 6).

b) We are aware of the fact that DVM patterns of Antarctic krill are highly variable and that normal DVM patterns do not need to be the rule (e.g. see our cited study on the plasticity of krill DVM by Bahlburg et al. 2023). The visualized data were not selected for their DVM pattern but represent the period directly preceding the sampling for behavioral experiments in four seasons (experiment 2), including the day of sampling. These periods were chosen to assess the DVM behavior of krill swarms in the field in the days before and during the sampling for behavioral experiments.

To improve understanding, we modified the description in the Results, Discussion, and Methods sections, as well as the caption of Figure 5 (now Fig. 6), which now read:

“To investigate whether krill swarms exhibited daily behavioral patterns in swimming behavior in the field before they were sampled for seasonal experiments, hydroacoustic data were recorded from the fishing vessel, continuously over a three-day period prior to sampling for the seasonal experiments described above…” (lines 191-194).

“Furthermore, hydroacoustic recordings demonstrate that most krill swarms sampled exhibited synchronized DVM in the field in the days directly before sampling for behavioral experiments, indicating that in this region, krill remain behaviorally synchronized across a wide range of photoperiods.” (lines 397-400).

“Hydroacoustic data were collected using a hull-mounted SIMRAD ES80 echosounder (Kongsberg Maritime AS) aboard the Antarctic Endurance, covering three days before the sampling for each of the seasonal behavioral experiments of experiment 2” (lines 512-515).

“We only included data during active fishing periods and the vessel is specifically targeting E. superba, which occurs in large monospecific aggregations. Further, krill fishery bycatch rates are very low (0.1-0.3%, 84), which makes it highly probable that the recorded signal represents krill swarms.” (lines 523-526).

“Hydroacoustic recordings showing the vertical distribution of krill swarms in the upper water column (<220 m) below the vessel, visualized by the mean volume backscattering signal (200 kHz), on the three days prior to krill sampling for experiments…” (lines 802-804).

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

As noted in the public review, this is a logical and well-written manuscript. I have very few comments to consider addressing.

The Results lead with a paragraph outlining the experimental approach. This is good, but you use the term "experiments" to refer to both the two sets, and the two or four subsets of experiments. Perhaps consider the subset experiments as "treatments"? I understood what you meant, but it took a few read-throughs to be sure I got it.

We thank the reviewer for pointing this out and changed the nomenclature of the experiments throughout the manuscript. We now refer to the two sets of experiments as experiment 1 and 2, to the subsets of experiment 1 as “short day treatment” and “long day treatment”, and to the subsets of experiment 2 as summer treatment, late summer treatment, autumn treatment, and winter treatment. We also believe that the new Figure 1 is now helping to follow the experimental design more efficiently.

Ln 140: "...off and decrease at lights-on."

We adjusted the sentence accordingly.

Ln 244: Can you define "extreme photic conditions"? I get what you mean, but to be clear to the reader this would help.

We adjusted the sentence, which now reads:

“This could confer a significant adaptive advantage to species inhabiting environments characterized by extreme photic conditions (53, 54, 60), such as phases of polar night or midnight sun as well as rapid changes in daylength, or species that rely on precise photoperiodic time measurement for accurate seasonal adaptation.” (lines 258-261).

Figures: Consider adding an LSP for groups in Fig 1. Also, it would be useful to have LSP period estimates for each individual tested. This could be a separate table, or it could be added to the individual activity plots. Should S3 and S4 be reversed?

We thank the reviewer for their suggestion and added an LSP as figure 1d (now Fig. 2d) to statistically support the group activity shown in Figure 1c (now Fig. 2c) as suggested. We added the individual animals' LSP period estimates to supplementary figures S2, S7, S8, S9, and S10. We also reversed Figures S3 and S4 to match the appearance in the main text. 

Fig 5: are the light regime bars for b and c correct? They look similar, but there are only 15 days apart, so perhaps they are correct as is.

We double checked the light regime bars in Fig. 5b and c (now 6b and c) and they are correct as is.

Author response:

The following is the authors’ response to the original reviews

Public reviews:

Reviewer #1 (Public review):

Summary:

This very interesting manuscript proposes a general mechanism for how activating signaling proteins respond to species-specific signals arising from a variety of stresses. In brief, the authors propose that the activating signal alters the structure by a universal allosteric mechanism.

Strengths:

The unitary mechanism proposed is appealing and testable. They propose that the allosteric module consists of crossed alpha-helical linkers with similar architecture and that their attached regulatory domains connect to phosphatases or other molecules through coiled-coli domains, such that the signal is transduced via rigidifying the alpha helices, permitting downstream enzymatic activity. The authors present genetic and structural prediction data in favor of the model for the system they are studying, and stronger structural data in other systems.

Weaknesses:

The evidence is indirect - targeted mutations, structural predictions, and biochemical data. Therefore, these important generalizable conclusions are not buttressed by impeccable data, which would require doing actual structures in B. subtilis, confirming experiments in other organisms, and possibly co-evolutionary coupling. In the absence of such data, it is not possible to rule out variant models.

We thank the reviewer for their feedback. A challenge of studying flexible proteins is that it is often not possible to directly obtain high resolution structural data. For the case of B. subtilis RsbU, the independent experimental approaches we applied (including two unbiased genetic screens, targeted mutagenesis, SAXS, enzymology, and structure prediction, which includes evolutionary coupling) converged upon a model for activation, which we feel is well supported. Frustratingly, our attempts at determining high resolution experimental structures have been unsuccessful, which we think is due to the flexibility of the proteins revealed by our SAXS experiments. For example, we collected X-ray diffraction data from crystals of a fragment of B. subtilis RsbU containing the N-terminal domain and linker in which the linker was almost entirely disordered in the maps. We agree that doing experiments in other organisms would be valuable next steps to test the hypothesis that this coiled-coil based transduction mechanism is conserved across species, and will modify the text to differentiate this more speculative section of the manuscript.

We have modified the abstract to read:

“This coiled-coil linker transduction mechanism additionally suggests a resolution to the mystery of how shared sensory domains control serine/threonine phosphatases, diguanylate cyclases and histidine kinases.”

We have modified the results to read:

"These predictions suggest a testable hypothesis that RsbP is controlled through an activation mechanism similar to that of RsbU (Fig. 5A)”

“From this analysis, we speculate that linker-mediated phosphatase domain dimerization is an evolutionarily conserved, adaptable mechanism to control PPM phosphatase activity.”

Based on this critique (and the critiques of the other reviewers), we plan to do energetic analysis of the predicted coiled coils from the enzymes we analyzed from other species and to incorporate this into the manuscript.

We have modified the results to read:

Consistent with a model in which the stability of the linker plays a conserved regulatory role, the AlphaFold2 models for many of the predicted structures have unfavorable polar residues buried in the coiled-coil interface (positions a and d, for which non-polar residues are most favorable) (Figure 5 – figure supplement 2).”

Finally, in the manuscript, we have highlighted that this mechanism is not the only mechanism for activation of other proteins with effector domains connected to linkers, but rather one of many mechanisms (Fig 5G). The reviewer additionally made helpful suggestions about the text in detailed comments that we will incorporate as appropriate.

Reviewer #2 (Public review):

Summary:

While bacteria have the ability to induce genes in response to specific stresses, they also use the General Stress Response (GSR) to deal with growth conditions that presumably include a larger range of stresses (for instance, stationary phase growth). The activation of GSR-specific sigma factors is frequently at the heart of the induction of a GSR. Given the range of stresses that can lead to GSR induction, the regulatory inputs are frequently complex. In B. subtilis, the stressosome, a multi-protein complex, contains a set of proteins that, upon appropriate stresses, initiate partner switching cascades that free the sigma B sigma factor from an anti-sigma. The focus here is on the mode of activation of RsbU, a serine/threonine phosphatase of the PPM family, leading to sigB activation. RbsT, a component of the degradosome interacts with RsbU upon stress, activating the phosphatase activity. Once active, RsbU dephosphorylates its target (RsbV, an anti-antisigma), which in turn binds the anti-sigma. The conclusion is that flexible linker domains upstream of the phosphatase domain are the target for activation, via binding of proteins to the N-terminal domain, resulting in a crossed-linker dimeric structure. The authors then use the information on RsbU to suggest that parallel approaches are used to activate PPM phosphatases for the GSR response in other bacteria. (Biology vs. Mechanism, evolution?)

Strengths and Weaknesses:

Many of these have to do with clarifying what was done and why. This includes the presentation and content of the figures.

One issue relates to the background and context. A bit more information on the stresses that release RsbT would be useful here. The authors might also consider a figure showing the major conclusions and parallels for SpoIIE activation and possibly other partner switches that are discussed, introducing the switch change more clearly to set the stage for the work here (and the generalization). There are a lot of players to keep track of.

We plan to carefully review the manuscript to improve the clarity of presentation and background. In particular, we thank the reviewer for pointing out the missing information about the release of RsbT from the stressosome. We will incorporate this information into the introduction and provide an additional figure.

We have added the following text to the introduction:

“RsbT is sequestered in a megadalton stress sensing complex called the stressosome, and is released to bind RsbU in response to specific stress signals including ethanol, heat, acid, salt, and blue light”

We have added a new figure panel (2C) that shows the model for how Q94L, M166V, and RsbT binding induce conformational change of the PPM domain to recruit metal cofactor and activate RsbU (analogous, but slightly different from the mechanism for SpoIIE).

The reviewer additionally provided detailed helpful comments that we will incorporate in the text and figures.

Reviewer #3 (Public review):

Summary:

The authors present a study building on their previous work on activation of the general stress response phosphatase, RsbU, from Bacillus subtilis. Using computed structural models of the RsbU dimer the authors map previously identified activating mutations onto the structure and suggest further protein variants to test the role of the predicted linker helix and the interaction with RsbT on the activation of the phosphatase activity.

Using in vivo and in vitro activity assays, the authors demonstrate that linker variants can constitutively activate RsbU and increase the affinity of the protein for RsbT, thus showing a link between the structure of the linker region and RsbT binding.

Small angle X-ray scattering experiments on RsbU variants alone, and in complex with RsbT show structural changes consistent with a decreased flexibility of the RsbU protein, which is hypothesised to indicate a disorder-order transition in the linker when RsbT binds. This interpretation of the data is consistent with the biochemical data presented by the authors.

Further computed structure models are presented for other protein phosphates from different bacterial species and the authors propose a model for phosphatase activation by partner binding. They compare this to the activation mechanisms proposed for histidine kinase two-component systems and GGDEF proteins and suggest the individual domains could be swapped to give a toolkit of modular parts for bacterial signalling.

Strengths:

The key mutagenesis data is presented with two lines of evidence to demonstrate RsbU activation - in vivo sigma-b activation assays utilising a beta-galactosidase reporter and in vitro activity assays against the RsbV protein, which is the downstream target of RsbU. These data support the hypothesis for RsbT binding to the RsbU linker region as well as the dimerisation domain to activate the RsbU activity.

Weaknesses:

Small angle scattering curves are difficult to unambiguously interpret, but the authors present reasonable interpretations that fit with the biochemical data presented. These interpretations should be considered as good models for future testing with other methods - hydrogen/deuterium exchange mass spectrometry, would be a good additional method to use, as exchange rates in the linker region would be affected significantly by the disorder/order transition on RsbT binding.

We agree with the reviewer that the SAXS data has inherent ambiguity due to the nature of the measurement. However, SAXS is one of the best techniques to directly assess conformational flexibility. Our scattering data for RsbU have multiple signatures of flexibility supporting a high confidence conclusion. While the scattering data support a reduction in flexibility for the RsbT/RsbU complex, we agree that a high resolution structure would be valuable. However the combination of the scattering data with our biochemical and genetic data supports the validity of the AlphaFold predicted model. We thank the reviewer for the suggestion of future hydrogen/deuterium exchange experiments that would be complementary, but which we feel are beyond the scope of this work.

The interpretation of the computed structure models should be toned down with the addition of a few caveats related to the bias in the models returned by AlphaFold2. For the full-length models of RsbU and other phosphatase proteins, the relationship of the domains to each other is likely to be the least reliable part of the models - this is apparent from the PAE plots shown in Supplementary Figure 8. Furthermore, the authors should show models coloured by pLDDT scores in an additional supplementary figure to help the reader interpret the confidence level of the predicted structures.

We thank the reviewer for suggestions on how to clarify the discussion of AlphaFold models. We will decrease the emphasis on the computed models in the text and will add figures with the models colored by the pLDDT scores to aid in the interpretation.

We have modified the text of the Abstract: “This coiled-coil linker transduction mechanism additionally suggests a resolution to the mystery of how shared sensory domains control serine/threonine phosphatases, diguanylate cyclases and histidine kinases.”

We have modified the text of the Results: “These predictions suggest a testable hypothesis that RsbP is controlled through an activation mechanism similar to that of RsbU (Fig. 5A).”

“From this analysis, we speculate that linker-mediated phosphatase domain dimerization is an evolutionarily conserved, adaptable mechanism to control PPM phosphatase activity”

We have also added Figure 1 – figure supplement 2 with the AlphaFold2 models colored by the pLDDT scores.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Baral and colleagues investigate the regulatory mechanisms of the General Stress Response (GSR) in Bacillus subtilis, focusing on the phosphatase RsbU and its regulation by the protein RsbT. The GSR is a critical adaptive mechanism that allows bacteria to survive under various stress conditions by reshaping their physiology through a broad transcriptional response. RsbU, a key player in the GSR, facilitates the activation of the transcription factor SigB by dephosphorylating RsbV. This activation is mediated through a partner-switching mechanism involving RsbT. Baral and colleagues use a combination of genetic screening, structural predictions via AlphaFold2, and biophysical techniques such as SAXS and MALS to present a model for how RsbT regulates RsbU. Key findings include the identification of specific amino acid substitutions that enhance RsbU activity, the role of the α-helical linker in RsbU dimerization and activation, and the potential broader conservation of these mechanisms across bacterial species. However, as described below, additional work is required to solidify the results.

Major Points

(1) The manuscript is misnamed--it dissects a single step of the signal-transduction pathway regulating the general stress response. Instead, it is rather seeking a generalizable mechanism for kinase -phosphatase interactions across stresses.

We have edited the title to “A General Mechanism for Initiating the General Stress Response in Bacteria” to reflect that that this study addresses the initiating event of the general stress response.

(2) The genetic screen likely has limitations in detecting all possible variants that could affect RsbU activity. The readout is specific to σ^B activation, and the focus on specific amino acid substitutions may overlook other significant regions or mechanisms involved in the regulation of RsbU, particularly those involving RsbV and RsbT.

Our screens were specifically designed to identify features of RsbU that contribute to regulation. Importantly, RsbU does not have any known targets other than RsbV and the downstream σ^B response but agree that substitutions in either RsbV or RsbT could influence RsbU activation. In principle our suppressor screen with RsbU^Y28I could have identified RsbT variants (rsbT was mutagenized in this screen), but we did not identify any such variants in the screen. We conducted a separate screen (published elsewhere) that specifically addressed how RsbU recognizes RsbV.

(3) The authors largely focus on the biochemical and structural aspects of RsbU regulation. There is limited discussion on the broader functional implications of these findings in the context of bacterial physiology and stress response. Incorporating more in vivo studies to show how these mechanisms impact bacterial survival and adaptation would provide a more comprehensive understanding.

We appreciate this comment, but did not conduct additional studies of survival and adaptation because the phenotypes of σ^B deletion in B. subtilis under laboratory conditions are relatively mild and therefore difficult to assay. Future studies to address this in other systems could be highly informative.

(4) The results primarily support the model of linker-mediated dimerization and rigidity. However, other potential regulatory mechanisms or interacting partners might also play significant roles in RsbU activation. A more thorough exploration of these possibilities would strengthen the study's conclusions.

One of the major advantages of RsbU as a model for initiation of the general stress response is that the system is discreet with all evidence pointing to there being a single primary input (RsbT) and output (dephosphorylation of RsbV). While there are other possible variations on the system (for example RsbU may be directly activated by manganese stress), we focused on this system precisely because of its simplicity.

(5) While the study presents evidence for the conservation of the described mechanism across different species, this assumption is based on structural predictions and limited experimental data. Broader experimental validation across diverse bacterial species would be necessary to substantiate this claim. Coevolution coupling along with conservation/evolutionary studies could be considered.

We have altered the language in the paper to emphasize where we are making inferences from predictions that are therefore more speculative. We agree that a more detailed analysis of the evolutionary coupling would likely be fruitful. We note that these couplings are the major driving force of AlphaFold predictions, suggesting that these couplings contributed to the models that we analyzed.

(6) The reliance on AlphaFold2 for structural predictions introduces potential biases and uncertainties inherent in computational models. Experimental validation of these models through additional techniques such as cryo-EM or X-ray crystallography would strengthen the conclusions.

We agree with this point, which is why we performed extensive analysis and validation of the models for RsbU using SAXS, genetics, and biochemistry. The proposed techniques are made more challenging by flexibility and heterogeneity, which we detected in our experiments. Our attempts thus far at experimental structure determination are consistent with this being a major technical hurdle.

(7) SAXS data provide low-resolution structural information, and the interpretation of flexibility versus rigidification might be overemphasized in its interpretation. This part of the study was difficult to interpret. Improving readability by breaking down the text into sections with clear headings for each figure panel and clarifying descriptions of the panels and methods would help. Complementary high-resolution techniques could provide a more definitive view of the linker's conformational changes.

We have modified the presentation of the figures to clarify the SAXS analysis. The fact that the SAXS analysis suggests flexibility rather than a discrete inactive conformation means that high-resolution techniques may not be appropriate for this system.

(8) The study primarily focuses on the model where RsbT binding rigidifies the RsbU linker. Alternative hypotheses, such as subtle conformational adjustments without complete rigidification, are not extensively explored or ruled out.

Our analysis of the SAXS data strongly suggests that a subtle conformational change could not account for the scattering data that we obtained. We have modified the text to clarify this point.

“Indicative of significant deviation between the RsbU structure in solution to the AlphaFold2 model, the scattering intensity profile (I(q) vs. q) was a poor fit (χ² 12.53) to a profile calculated from the AlphaFold2 model of an RsbU dimer using FoXS (Schneidman-Duhovny et al. 2016; Schneidman-Duhovny et al. 2013) (Fig. 4A). We therefore assessed the SAXS data for the RsbU dimer for features that report on flexibility (Kikhney & Svergun 2015). First, the scattering intensity data lacked distinct features caused by the multi-domain structure of RsbU from the AlphaFold2 model (Fig.4A).”

(9) Future studies should aim to validate the AlphaFold2 predictions with high-resolution structural techniques. This would provide definitive evidence for the proposed conformational states of RsbU with and without RsbT.

The fact that the SAXS analysis suggests flexibility rather than a discrete inactive conformation means that high-resolution techniques may not be appropriate for this system.

(10) Investigating the RsbU-RsbT interaction in vivo using techniques like FRET, co-immunoprecipitation, or live-cell imaging would provide a more comprehensive understanding of their functional dynamics in a cellular context.

We appreciate the reviewer’s suggestions for future experiments.

(11) Exploring and testing alternative models of RsbU activation, such as partial rigidification or different modes of conformational change, would strengthen the conclusions.

While our data strongly support that a flexible-to-rigid transition controls RsbU activation, we agree that it is possible that other mechanisms of linker modification could control other phosphatases and we discuss this at some length in the discussion.

(12) The figure legends are quite dense and could benefit from some streamlining.

We have edited the figure legends for clarity and length.

Reviewer #2 (Recommendations for the authors):

(1) Activation assays (Figures 1, 3, S2) are presented here as blue or white spots (reflecting a reporter activity). While off and on these are fairly clear, it is more difficult to compare the degree of activity (for instance that rsbU^Q94L is more active than M166V). It would also be good to clearly present in the text the logic of asking if the mutant is RsbT independent or not (and the interpretation of that). Quantitative assays of these would be very useful.

We chose not to perform quantitative-LacZ assays here because of several complications to interpreting these results that we encountered in our previously published study (Ho and Bradshaw, 2021). However, the level of blue pigmentation shown in Figure 1B for RsbU Q94L and RsbU M166V is qualitatively different, making the comparison possible. Most importantly, we observed cell density dependent changes in LacZ activity in the absence of rsbT for rsbU^M166V expressing cells, meaning that comparisons between strains would be difficult. Additionally, we found that it was important to make a chromosomal replacement of rsbU to see the full effect of the M166V substitution. However, we were not able to construct a similar rsbU^Q94L strain, likely because the high level σ^B activity is lethal (we were able to construct this strain when σ^B was deleted but only obtained strains with additional loss-of-function mutations in RsbU when σ^B was present.

We have modified the text to explain the logic of identifying RsbT independent variants: “We previously conducted a genetic screen (Ho & Bradshaw 2021) to identify features of RsbU that are important for phosphatase regulation by isolating gain-of-function variants that are active in the absence of RsbT.”

(2) Explain Figure S8 graphs: as much as Alphafold is now in use, the authors should provide some further explanation of what is shown here. Blue (low error) is good, presumably. What are the A, B, C, and D sections showing? Different parts of a given letter region (and between them)? What is the x-axis? Is the top-ranked model used in every case in the text? How different are these models? The Methods section could be used for some of this (but doesn't in its current form). This also becomes important for the models generated later in the paper (Figure S7), which look rather different here.

We have modified figure S8 to include additional labels and have added structures with the pLDDT scores shown. We have additionally modified the figure legends and methods to provide the requested information.

(3) Figure 1C, D, Figure S2: amino acid ends of linker domains could be shown (text discusses 83-97 the linker as a two-turn coiled coil; Q94 is pretty close to the end of this coiled-coil? Figure S2 is even less clear - addresses of other amino acids would help, and or an added sequence showing the full linker and coiled-coil region). Some explanation for positions for readers to focus on for full coiled-coil would be useful in the legend of Figure S2. How strong a coiled-coil prediction is there for this region?

We have added the sequence of the coiled-coil regions to the figures with numbering. For these analyses we used the Socket2 program, which analyzes a PDB file to identify coiled-coil regions and thus does not provide a confidence score. However, inspection of the sequence and the confidence scores of the AlphaFold2 models indicates that the coiled-coil regions are not ideal, consistent with this being a regulatory feature.

Is it clear that the fully inactive proteins are still properly folded and soluble?

In the case of RsbU, our biophysical analysis indicates that the inactive form of the protein is soluble. While phosphatase activity is substantially reduced, our unpublished comparison of single- and multiple-turnover reactions in the absence of RsbT indicates that nearly all of the enzyme is active.

Finally, are there other positions that would also be expected, from this model, to stabilize the coiled-coil and thus bypass the requirement for RsbT? If so, it would be good to test these. Is it the burial of amino acid at position 94 that is important, or the ability to form crossed helices?

Because of how short the predicted coiled-coil region is, we did not identify any obvious positions that would likely have the same effect as Q94 substitution. We considered making helix-breaking mutations, which would be predicted to block RsbU activation, but favored analysis of the wildtype protein because of limitations in interpreting the effects of loss-of-function mutations.

(4) Figure 2A, RsbT binding to RsbU: It was not entirely clear to this reviewer why one would expect the RsbT binding, not needed for activation, to be increased by the mutation that stabilizes the crossed alpha helices. The change is impressive but doesn't the lack of a need for RsbT suggest that this mutation bypasses the normal mechanism? (Is dimerization enuf? Or other protein cross helices?).

We have modified the text to clarify this point: “One prediction of our hypothesis that RsbT stabilizes the crossed alpha helices of the RsbU dimer, is that RsbT should bind more tightly to rsbU^Q94L than to RsbU because the coiled-coil conformation that RsbT binds would be more energetically favorable.” Another way of putting this is that if the Q94L substitution activates RsbU through an on-pathway mechanism, RsbT must bind more tightly.

(5) Figure 3A, Figure S3: Please label the yellow (interface) residues in RsbU and RsbT in Fig. S3 and the green (suppressor) spheres in Figure 3A.

We have added labels to the figures as suggested.

If RbsT interacts with the N-terminal dimerization domain and linker, why were residues 174 and 178 (from PPM domain) shown to be implicated in binding?

The fact that residues in the switch region suppress a mutation that decreases RsbT binding suggests that this region is part of an allosteric network that links RsbT binding, the linker, and dimerization of the phosphatase domains. For example, any substitution that promotes a conformation of the phosphatase domain that is more favorable for dimerization would also promote RsbT binding. However, the precise details of how each mutation fits into this network is not clear and we have therefore chosen to not specify a particular model to avoid over interpreting our data.

Are these marked in Figure S3?

We have added labels to make this clear.

Are these part of a dimerization interface in the C-terminal domain? Are any/all of these RsbU mutants suppressed by Q94L, as one might predict (apparently Y28I is since Q94L was again identified)?

We chose to focus on Y28I because it was the best studied previously, but we would predict that Q94L would suppress other RsbT binding mutations.

(6) Line 191-192: Is it surprising that no suppressors were isolated in RsbT?

We didn’t have a preconception of whether or not it would be possible to identify similar suppressors in RsbT. Explanations for why we did not identify such suppressors could include that RsbT may be destabilized more easily by substitution, that RsbT is more constrained because it has other interaction partners, or that the particular substitutions that would suppress Y28I are less common by the PCR mutagenesis strategy we used.

(7) Figure 3: Would the same mutants arise if the screen had been done in the absence of RsbT? Was RsbT-dependent tested for the rsbU alleles?

Our prediction is that we would not have identified any of these mutations except for Q94L in the absence of rsbT. We tested a few of the alleles and found them all to be rsbT dependent, but did not systematically test all of the alleles and therefore did not include this analysis in the manuscript.

Given the findings earlier in the paper for Q94L, suggesting that this stabilizes the coiled-coil and shows some activity in the absence of RsbT, it seems that the interpretation of other mutants in this region (and Q94L itself) as evidence that RsbT contacts the linker directly and that contact is necessary for activation may be an overinterpretation. If these are in fact RsbT independent, they support the importance of the linker (do they further stabilize coiled-coil formation?), rather than the role of RsbT here. Are G92 and T89 on the outside of the coiled-coil? If Q94 is buried, is it qualitatively different from these others?

G92 and T89 are predicted to be exposed. The fact that these mutations are near Q94 is part of the reason that we focused on R91 and the predicted contact with D92 of RsbT as another approach to validate the predicted interface.

(8) Figure 3C addresses the issue of direct interaction of RsbT with the RsbU linker to some extent, given that RsbU R91E doesn't appear to have a lot of activity without RsbT. It would be helped by telling the reader what the R91 contact is initially.

We have modified the text to clarify this point: “To test the model that RsbT activates RsbU by directly interacting with the linker to dimerize the RsbU phosphatase domains, we introduced a charge swap at position R91 that would abolish a predicted salt-bridge with RsbT D92 (Fig. 3C).”

(9) Figure 4 and the discussion of it in the text is not likely to be easily understandable for many readers. Aside from providing a bit more explanation of what these analyses are showing, it would be useful to start the whole section (or maybe even much earlier in the paper) with the information found on lines 261-264, that other studies show that the N-terminus dimerizes stably on its own (and is it known that the C-terminus does not?). Then the discussion of the alternative models early in this section would be clearer.

We have updated the introduction to emphasize this point “RsbU has an N-terminal four-helix bundle domain that dimerizes RsbU and is also the binding site for RsbT, which activates RsbU as a phosphatase (Fig. 1C,D) (Delumeau et al. 2004).”

We have also added clarification to the model presented at the beginning of this section: “A second possibility is that inactive RsbU is dimerized by the N-terminal domains but that the linkers of inactive RsbU are flexible and that the phosphatase domains only interact with each other when RsbT orders the linkers into a crossing conformation.”

Is the dimerization of the N-terminal domains previously determined similar/the same as what is seen in the AlphaFold models used here (or the AlphaFold dimerization derived primarily from that data?).

Yes, the dimerization in the AlphaFold models matches closely to the published structure.

(10) Discussion and Figure 5: The final part of this work predicts AlphaFold models for a set of other phosphatases involved in initiating GSR across bacterial species, and suggests that linked-mediated phosphatase dimerization is the critical factor to activate the phosphatase. Clearly, this is the most speculative but interesting aspect of the paper. A number of possible questions are suggested by some of this:

a. Do any of the activating mutants In RsbU and RsbP in the PPM domain (that apparently improve dimerization and thus activation) do a similar job in the other modeled proteins?

This is an interesting question, but unfortunately most of these proteins have not been biochemically characterized. We highlight examples of RsbP and E. coli RssB for which similar activating mutations have been characterized.

b. The legend (Figure 5G) suggests that all of the linker combinations will be coiled-coils, but that they will undergo different types of activating (and dimerizing?) transitions. Is that in fact what is being proposed here?

Yes, this is our working hypothesis.

c. If there is no dimerization (as noted, only weak dimerization has been reported for E. coli RssB), does that generalize the model to there are linkers and their structures are important? At the least, would the folding up of the E. coli RssB linker with antiadaptor binding be considered another mode of signal transduction or rather some sort of storage form?

Interestingly, the P. aeruginosa RssB constitutively dimerizes, suggesting the E. coli is the outlier.

d. Would the "toolkit" model, in which different changes occur in the linker regions, suggest that the interacting proteins are going to be critical for the type of linker changes that will be important? Or something about the nature of the linkers themselves?

This is an interesting question that we cannot yet answer. We have chosen to focus on the possible flexibility of this mechanism and anticipate that a variety of mechanisms will be used.

e. Given the extensive comparison to E. coli RssB, the authors might consider a figure to clarify the relative domain architecture, sequences that are akin to switch regions, and others important to the discussion here.

We tried to highlight this in Figure 5C including coloring the regions similar to the switch regions.

Reviewer #3 (Recommendations for the authors):

Given the caveats noted above related to the reliability of computed structure models, I would recommend the authors make the following additions/modifications to their manuscript:

(1) The authors should show alpha fold models coloured by pLDDT scores in an additional supplementary figure to help the reader interpret the confidence level of the predicted structures.

We have added these models to figure 1 – figure supplement 2.

(2) Because of the points mentioned above the authors should tone down the generalisation relating to the activation mechanism of this family of phosphatases presented in the discussion.

We have modified the paper throughout to emphasize where we are speculating.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1:

Summary:

Kimura et al performed a saturation mutagenesis study of CDKN2A to assess functionality of all possible missense variants and compare them to previously identified pathogenic variants. They also compared their assay result with those from in silico predictors.

Strengths:

CDKN2A is an important gene that modulate cell cycle and apoptosis; therefore it is critical to accurately assess functionality of missense variants. Overall, the paper reads well and touches upon major discoveries in a logical manner.

Weaknesses:

The paper lacks proper details for experiments and basic data, leaving the results less convincing. Analyses are superficial and does not provide variant-level resolution. Many of which were addressed during the revision process.

Comments on revisions:

The manuscript was improved during the revision process.

We thank the reviewer for their comments. We are grateful for the opportunity to provide additional information and data to clarify our approach and study results.

Reviewer #2:

Summary:

This study describes a deep mutational scan across CDKN2A using suppression of cell proliferation in pancreatic adenocarcinoma cells as a readout for CDKN2A function. The results are also compared to in silico variant predictors currently utilized by the current diagnostic frameworks to gauge these predictors' performance. The authors also functionally classify CDKN2A somatic mutations in cancers across different tissues.

Review:

The goal of this paper was to perform functional classification of missense mutations in CDKN2A in order to generate a resource to aid in clinical interpretation of CDKN2A genetic variants identified in clinical sequencing. In our initial review, we concluded that this paper was difficult to review because there was a lack of primary data and experimental detail. The authors have significantly improved the clarity, methodological detail and data exposition in this revision, facilitating a fuller scientific review. Based on the data provided we do not think the functional characterization of CDKN2A variants is robust or complete enough to meet the stated goal of aiding clinical variant interpretation. We think the underlying assay could be used for this purpose but different experimental design choices and more replication would be required for these data to be useful. Alternatively, the authors could also focus on novel CDKN2A variants as there seems to be potential gain of function mutations that are simply lumped into "neutral" that may have important biological implications.

Major concerns:

Low experimental concordance. The p-value scatter plot (Figure 2 Figure Supplement 3A) across 560 variants shows low collinearity indicating poor replicability. These data should be shown in log2fold changes, but even after model fitting with the gamma GLM still show low concordance which casts strong doubt on the function scores.

Concordance among non-significant p-values is generally low because most of the signal comes from random variability across repeats. If the observed log2 fold change between the repeats is entirely due to noise, one would expect two repeated p-values to behave like independent random uniforms. True concordance is typically more evident in significant p-values because they reflect consistent effects above random noise. Functionally deleterious variants are called when their associated p-value is significant. To confirm this statement, a scatter plot with the log2 normalized fold change was added in Figure 2 Supplement 3C. We see low concordance between repeats in the log2 normalized fold changes centered around 0, corresponding to log log2 normalized changes mainly due to noise. The concordance increases as the variants become significant. One can notice that the correlation coefficient between duplicate assay results was almost identical between the model-based p-values and log2normalized fold change (Figure 2-figure supplement 3A and 3C, Appendix 1-table 4, and Appendix 1-table 6). Also, importantly, no variant was functionally deleterious in one replicate and functionally neutral in another, implying a perfect concordance in calls if we exclude variants that were called indeterminate in one of the two repeats. Finally, of variants with discordant classifications, only 6/560 repeats (1.1%) were functionally deleterious (significant p-value) in one replicate and of indeterminate function in another. We have updated the text as follows:

“Of variants with discordant classifications, 6 (1.1%) were functionally deleterious in one replicate and of indeterminate function in another. While 102 variants (18.2%) were functionally neutral in one replicate and of indeterminate function in another. Importantly, no variant that was functionally deleterious in one replicate and functionally neutral in another (Appendix 1 -table 4). Furthermore, the correlation coefficient between duplicate assay results was similar using the gamma GLM and log2 normalized fold change (Figure 2-figure supplement 3A and 3C).”

The more detailed methods provided indicate that the growth suppression experiment is done in 156 pools with each pool consisting of the 20 variants corresponding to one of the 156 aa positions in CKDN2A. There are several serious problems with this design.

Batch effects in each of the pools preventing comparison across different residues. We think this is a serious design flaw and not standard for how these deep mutational scans are done. The standard would be to combine all 156 pools in a single experiment. Given the sequencing strategy of dividing up CDKN2A into 3 segments, the 156 pools could easily have been collapsed into 3 (1 to 53, 54 to 110, 111 to 156). This would significantly minimize variation in handling between variants at each residue and would be more manageable for performance of further replicates of the screen for reproducibility purposes. The huge variation in confluency time 16-40 days for each pool suggest that this batch effect is a strong source of variation in the experiment.

While there is variation in time to confluency between different amino acid residues, we do not anticipate this batch effect to significantly affect variant classifications in our study. For example, our results were generally consistent with previous classifications. All synonymous variants (one per residue) and benchmark benign variants assayed were classified as functionally neutral. Furthermore, of benchmark pathogenic variants assayed, none were classified as functionally neutral. 84% were classified as functionally deleterious and 16 percent were classified as indeterminate function.

Lack of experimental/biological replication: The functional assay was only performed once on all 156 CDKN2A residues and was repeated for only 28 out of 156 residues, with only ~80% concordance in functional classification between the first and second screens. This is not sufficiently robust for variant interpretation. Why was the experiment not performed more than once for most aa sites?

In our study we determined functional classifications for all CDKN2A missense variants while assessing variability with replicates across 28 residues. Of these variants, only 6 (1.1%) were functionally deleterious in one replicate and of indeterminate function in another. Furthermore, no variant was functionally deleterious in one replicate and functionally neutral in another (Appendix 1 -table 4).  As noted above, we provided additional context in the manuscript.

For the screen, the methods section states that PANC-1 cells were infected at MOI=1 while the standard is an MOI of 0.3-0.5 to minimize multiple variants integrating into a single cell. At an MOI =1 under a Poisson process which captures viral integration, ~25% of cells would have more than 1 lentiviral integrant. So in 25% of the cells the effect of a variant would be confounded by one or more other variants adding noise to the assay.

As noted previously, we are not able to differentiate effects due to multiple viral integrations per cells. However, we do not anticipate multiple viral integrations to significantly affect variant classifications in our study as our results are consistent with previous classifications, as described above.

While the authors provide more explanation of the gamma GLM, we strongly advise that the heatmap and replicate correlations be shown with the log2 fold changes rather than the fit output of the p-values.

Thank you for the suggestion. As noted, we provide additional explanation in the manuscript about why we classified variants using a gamma GLM. Using a gamma GLM, classification thresholds were determined using the change in representation of 20 non-functional barcodes in a pool of PANC-1 cells stably expressing CDKN2A after a period of in vitro proliferation. Our variant classifications were therefore not based on assay outputs for previously reported – benchmark – pathogenic or begin variants to determine thresholds. We strongly prefer using p-values and classifications using the gamma GLM in the manuscript. However, comparison of assay outputs using a gamma GLM and log2 fold change are included in the manuscript. Read counts, log2 fold change, and classifications based on log2 fold change are presented in the manuscript, for all variants. Readers who wish to use these data may do so and we refer them to the manuscript text, Appendix 1 -table 4, Appendix 1 -table 6, and Figure 2 -figure supplement 2.

In this study, the authors only classify variants into the categories "neutral", "indeterminate", or "deleterious" but they do not address CDKN2A gain-of-function variants that may lead to decreased proliferation. For example, there is no discussion on variants at residue 104, whose proliferation values mostly consist of higher magnitude negative log2fold change values. These variants are defined as neutral but from the one replicate of the experiment performed, they appear to be potential gain-of-function variants.

We have added a comment to the discussion to highlight that we did not identify potential gain-of-function variants. Specifically:

“We classified CDKN2A missense variants using a gamma GLM, as either functionally deleterious, indeterminate functional or functionally neutral. However, we did not classify variants that may have gain-of-function effects, resulting in decreased representation in the cell pool. Future studies are necessary to determine the prevalence and significance of CDKN2A gain-of-function variants.”

Minor concerns:

The differentiation between variants of "neutral" and "indeterminate" function seems unnecessary and it seems like there are too many variants that fall into the "indeterminate" category. The authors seem to have set numerical thresholds for CDKN2A function using benchmark variants of known function. While the benchmark variants are important as a frame of reference for the "dynamic range" of the assay, their function scores should not necessarily be used to define hard cutoffs of whether a variant's function score can be interpreted.

We did not utilize benchmark variants to define thresholds for functional classifications using a gamma GLM. This is one of the strengths of using a gamma GLM model for classification. As explained in our manuscript, classification thresholds were determined using the change in representation of 20 non-functional barcodes in a pool of PANC-1 cells stably expressing CDKN2A after a period of in vitro proliferation. Our variant classifications were therefore not based on assay outputs for previously reported – benchmark – pathogenic or begin variants. While not required when using a gamma GLM, we included indeterminate classifications, which are not uncommon.

Figure 2 supplement 2 - on the x-axis, should "intermediate" be "indeterminate"?

This, and a similar typographical error in Figure 2 -figure supplement 3, has been corrected.

Reviewer #3 (Public review):

This study investigates the characteristics of the autofluorescence signal excited by 740 nm 2-photon excitation, in the range of 420-500 nm, across the Drosophila brain. The fluorescence lifetime (FL) appears bi-exponential, with a short 0.4 ns time constant followed by a longer decay. The lifetime decay and the resulting parameter fits vary across the brain. The resulting maps reveal anatomical landmarks, which simultaneous imaging of genetically encoded fluorescent proteins helps to identify. Past work has shown that the autofluorescence decay time course reflects the balance of the redox enzyme NAD(P)H vs. its protein-bound form. The ratio of free-to-bound NADPH is thought to indicate relative glycolysis vs. oxidative phosphorylation, and thus shifts in the free-to-bound ratio may indicate shifts in metabolic pathways. The basics of this measure have been demonstrated in other organisms, and this study is the first to use the FLIM module of the STELLARIS 8 FALCON microscope from Leica to measure autofluorescence lifetime in the brain of the fly. Methods include registering the brains of different flies to a common template and masking out anatomical regions of interest using fluorescence proteins.

The analysis relies on fitting an FL decay model with two free parameters, f_free and t_bound. F_free is the fraction of the normalized curve contributed by a decaying exponential with a time constant of 0.4 ns, thought to represent the FL of free NADPH or NADH, which apparently cannot be distinguished. T_bound is the time constant of the second exponential, with scalar amplitude = (1-f_free). The T_bound fit is thought to represent the decay time constant of protein-bound NADPH but can differ depending on the protein. The study shows that across the brain, T_bound can range from 0 to >5 ns, whereas f_free can range from 0.5 to 0.9 (Figure 1a). These methods appear to be solid, the full range of fits are reported, including maximum likelihood quality parameters, and can be benchmarks for future studies.

The authors measure the properties of NADPH-related autofluorescence of Kenyon Cells (KCs) of the fly mushroom body. The results from the three main figures are:

(1) Somata and calyx of mushroom bodies have a longer average tau_bound than other regions (Figure 1e);

(2) The f_free fit is higher for the calyx (input synapses) region than for KC somata (Figure 2b);

(3) The average across flies of average f_free fits in alpha/beta KC somata decreases from 0.734 to 0.718. Based on the first two findings, an accurate title would be "Autofluorecense lifetime imaging reveals regional differences in NADPH state in Drosophila mushroom bodies."

The third finding is the basis for the title of the paper and the support for this claim is unconvincing. First, the difference in alpha/beta f_free (p-value of 4.98E-2) is small compared to the measured difference in f_free between somas and calyces. It's smaller even than the difference in average soma f_free across datasets (Figure 2b vs c). The metric is also quite derived; first, the model is fit to each (binned) voxel, then the distribution across voxels is averaged and then averaged across flies. If the voxel distributions of f_free are similar to those shown in Supplementary Figure 2, then the actual f_free fits could range between 0.6-0.8. A more convincing statistical test might be to compare the distributions across voxels between alpha/beta vs alpha'/beta' vs. gamma KCs, perhaps with bootstrapping and including appropriate controls for multiple comparisons.

I recommend the authors address two concerns. First, what degree of fluctuation in autofluorescence decay can we expect over time, e.g. over circadian cycles? That would be helpful in evaluating the magnitude of changes following conditioning. And second, if the authors think that metabolism shifts to OXPHOS over glycolosis, are there further genetic manipulations they could make? They test LDH knockdown in gamma KCs, why not knock it down in alpha/beta neurons? The prediction might be that if it prevents the shift to OXPHOS, the shift in f_free distribution in alpha/beta KCs would be attenuated. The extensive library of genetic reagents is an advantage of working with flies, but it comes with a higher standard for corroborating claims.

FLIM as a method is not yet widely prevalent in fly neuroscience, but recent demonstrations of its potential are likely to increase its use. Future efforts will benefit from the description of the properties of the autofluorescence signal to evaluate how autofluorescence may impact measures of FL of genetically engineered indicators.

In my opinion, design is framed as something everyone does, not just professionals, which makes it feel more universal and accessible. I think simple actions like rearranging a room or making a sign are forms of design, even if they lack the formal methods and expertise of professional work. However, while this perspective is valuable, it overlooks how structured processes and iteration differentiate professional design from everyday problem-solving.