This shows how separation from families is common because many don't see each other again which is horrible. He was talking about his personal experience about being sold into different places and suffering emotionally. It's sad and I wish to read and find out about the well-being of his sister and makes me question that did he ever got to see her again like he did last time.

To me, this is part is kind of dark. He was on the slave ship. He saw the men who are being tortured and chained up on the ship. The men were so cruel to them. Thats sad that he had to witness about that. It's heartbreaking that at first, he got kidnapped, his sister got kidnapped, he got departed from his sister, sold into slavery, sold into many different locations, then have to witness the men who are being tortured on the ship. This is very dark and traumatizing. And also, he still feeling hopeless of the desire of returning back home and also reunite with his sister again. It makes me question that would he ever reunite with his sister again.

From my readings in EDSCI so far, this is the first time I have seen someone address the heaviness that may come with being a transformative learner. Many of our biases and student's biases as well as oppressive ideologies might be the only way they have learned. The word trauma exemplifies the impact of unlearning things that perhaps have been the building blocks of your identity. I think about my 7th graders, primarily Latinx, primarily Christian, primarily male, and primarily from low income households surrounding our school;they might experience this trauma when presented with ideas that deviate from what they hold to be truth. However, their truth has been their reality, and rather than negate them, they have to be part of the conversation.

Social natives may be more interactive in social media and more inclined to socialize on their devices instead of in person. This is my hypothesis because I see a lot of younger people Facetiming and utilizing social media more than I do, and even though I could be as active as them, I am just not as inclined to participate on a more personal account. I would be more inclined to participate promoting a company I work for on social media than interacting with it as much as the younger generation for myself. I think that digital natives may be more skilled at media literacy because they have some background knowledge on what existed before the floods of mass misinformation from social media on platforms like Twitter and Facebook- while seeing how misinformation was perpetuated in the media for the generation older than them.

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

Learn more at Review Commons

---

Reply to the reviewers

We thank the reviewers for their thoughtful comments. Here we provide a point-by-point response to their reviews. All additional experiments that are present in the revised manuscript, or that we plan to include in the final manuscript, are numbered.

__Reviewer #1 (Evidence, reproducibility and clarity (Required)):

The concept introduced by this paper is exciting and novel. However, the current paucity of presented data can lead to incorrect interpretations of the findings and speculations that might not hold true after a more rigorous assessment of the observed phenomenon.

The premise of this study builds upon an interaction between the PAXT complex and nuclear YTH domain containing proteins. However, figures 1B and C should be improved. The interacting band for the ZFC3H1 presented in panel B does not seem to match the size of the construct used in panel C. Is the Flag version of ZFC3H1 expressing a smaller isoform for this protein? __

The reviewer is correct in that endogenous ZFC3H1 (which migrates at 250kD with a minor band at 150kD, see Figure 1B in the initial manuscript) appears to differ from the FLAG-tagged construct as expressed from a plasmid transfected in HEK293 cells (which migrates as two bands at 180kD and 200kD, see Figure 1C in the initial manuscript). For the endogenous protein, the predicted molecular weight is 226kD and the 250kD band disappears when cells are transduced with lentivirus containing shRNAs against ZFC3H1 (see Figure 4A in the initial manuscript), indicating that it is the correct protein. Both the 250kD endogenous protein (*) and the 200kD overexpressed protein (**) in transfected HEK293 and U2OS cells are detected in immunoblots using anti-ZFC3H1 antibodies (see Figure 1 in this document) indicating that the over-expressed protein is indeed ZFC3H1.

[ Figure 1]

_Figure 1. Molecular Weight Size Comparison of Endogenous ZFC3H1 and FLAG-ZFC3H1 (1-1233). _Lysates from HEK293 and U2OS that were either untransfected or transfected with FLAG-ZFC3H1 (1-1233) plasmid. We labelled the bands corresponding to the endogenous ZFC3H1 “*” and FLAG-ZFC3H1 (1-1233) “**”.

We have sequenced the plasmid, and discovered that it contains an additional sequence inserted within the middle of the ZFC3H1 cDNA with a premature stop codon. As such, the version of the protein that is expressed from the plasmid only contains amino acids 1-1233 of the endogenous protein and is missing amino acids 1234-1989. The deleted region only contains TPR repeats, and is not known to interact with any of the well characterized interactors of ZFC3H1 (Wang, Nuc Acid Res 2021, Figure 3). We have renamed this construct FLAG-ZFC3H1 (1-1233). Given these new considerations, our results are consistent with the idea that the N-terminal portion of ZFC3H1 interacts with U1-70K, YTHDC1 and YTHDC2. We will change the text to reflect this.

We are currently in the process of deleting the small insertion to obtain a plasmid that encodes a full length version of human ZFC3H1. For the final manuscript:

Experiment #1) We will repeat the co-immunoprecipitation experiment with the full length FLAG-ZFC3H1 to determine whether it interacts with YTHDC1 and YTHDC2. This will take a few weeks.

__Also, the YTHDC1-2 interaction in panel C is not as convincing considering the negative controls lane show some degree of binding. __

Although the reviewer is correct that there is substantial background binding in the YTHDC1 immunoblot, we disagree with their characterization of the results with the YTHDC2 immunoblot (see Figure 1B-C in the initial manuscript). In the new manuscript we have included:

Experiment #2) A new co-immunoprecipitate of the FLAG-tagged ZFC3H1 (1-1233) from HEK293 cells under more stringent conditions where the background level of YTHDC1 binding to beads is negligible. We have already completed this experiment (see Figure 1D in the revised manuscript).

__Additionally, can the authors test if their RNaseA treatment worked? __

In the new manuscript we have included:

Experiment #3) A new co-immunoprecipitate of FLAG-YTHDC1 immunoblotted for eIF4AIII from HEK293 cell lysates. We find that without RNAse, there is some eIF4AIII in the precipitates but that the levels diminish substantially after RNAse A/T1 treatment. We have already completed this experiment (see Figure 1B in the preliminary revised manuscript).

__Why do you need 18 hours to observe the nuclear export of your modifiable construct when inhibiting METTL3 in figure 3? Is it possible that your observation is secondary to phenotypes these cells develop as a result of blocking METTL3? __

We treated cells for this period of time so that during the expression of the reporter, all of the newly synthesized mRNA is expressed in the absence of m6A methyl transferase activity. For shorter treatment times, it is unclear whether the bulk of the reporter mRNA, which would be synthesized before the treatment, would lose any pre-existing m6A marks, making a negative result hard to interpret. Previously we found that although 50% of intronic polyadenylated (IPA) transcripts from our reporters are rapidly degraded, about 50% are stable and are nuclear retained over extended periods of time (see Lee at al., PLOS ONE 2015; https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0122743 Figure 3B-G). We believe that the bulk of the reporter mRNA that we are visualizing is stable and accumulates in the nucleus. Given that METTL3-depletion inhibits nuclear retention and that versions of the IPA reporter that lack m6A modification motifs are exported, we think that the most likely interpretation of the 18 hour STM2457 treatment experiments is that the lack of methyltransferase activity had a direct effect, rather than an indirect effect, on nuclear retention. We would be open to performing more experiments if the editors insist, however we ordered STM2457 four weeks ago and it has yet to arrive from Sigma-Millipore. Performing this experiment may substantially delay our ability to resubmit the manuscript in a timely manner.

__Is ALKBH5 nuclear and/or cytoplasmic in the cell system used? __

According to The Human Protein Atlas, ALKBH5 is predominantly nuclear in U2OS cells, with some present in the cytoplasm (https://www.proteinatlas.org/ENSG00000091542-ALKBH5/subcellular#human).

In the revised manuscript we have included:

Experiment #4) Data from subcellular fractionation demonstrating that ALKBH5 is present in both the nucleus and cytoplasm that we have already performed (see Figure 4J in the preliminary revised manuscript).

__Reviewer #1 (Significance (Required)):

The study is highly significant __

------

__Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Summary: In the manuscript by Lee et al. entitled "N-6-methyladenosine (m6A) Promotes the Nuclear Retention of mRNAs with Intact 5'Splice Site Motifs", the authors provide evidence that m6A modifications within specific regions of transcripts can confer nuclear retention. These results are important because they add to our understanding of how m6A modifications can contribute to post-transcriptional regulation. Although the authors do not quite come out and say this, data seem to be accumulating to suggest that the location of the m6A modifications within a given transcript can dictate the functional consequences of those modifications.__

We thank the reviewer for pointing this out. We have included a few sentences in the new preliminary revised manuscript pointing out that the location of the m6A modification in IPA transcripts, with respect to intact 5’SS and poly(A) signals, may play a role in promoting nuclear retention.

__The current work builds on previous findings from these authors identifying factors critical for retention of intronic polyadenylated (IPA) transcripts. The present study identified m6A modification as one of the signals for the retention of such transcripts. The authors use reporters for their analysis and also examine validated endogenous IPA transcripts. The data presented supports the conclusions albeit they show a surprising finding for one of the m6A erasers, ALKBH5. However, there is some controversy over the mechanism by which ALKBH5 functions and whether the m6A mark is truly reversible, so these results may continue to add to this point of view.

Major Comments: One experiment that might add to the argument would be overexpression of Mettl3 as compared to catalytically inactive Mettl3. The prediction would be that the reporter transcript with intact DRACH sequences would be even more retained in the nucleus in a manner that depends on Mettl3 catalytic activity. For some of the data presented, the reporter is already wholly nuclear so no difference could be detected, but in the U2OS cells shown in Figure 2B, it appears that an increase in nuclear localization might be evident. Such an experiment would add an orthogonal approach to demonstrate that the methylation by Mettl3 is required for retention. If such an experiment would work with the endogenous IPA transcripts shown in Figure 4, but these transcripts may already be too nuclear to detect any increase in nuclear retention.

__

We have performed two experiments that try to address this but they gave negative results:

Experiment #5) We have over-expressed wildtype and a methyl transferase mutant FLAG-METTL3 and assessed the nuclear export/retention of ftz-Δi-5’SS mRNA. There was no effect (see Figure 2 in this document).

[Figure 2]

__Figure 2. Over-expression of METTL3 does not increase the nuclear retention of ftz-Δi-5’SS. __U2OS cells were co-transfected with ftz-Δi-5’SS reporter and either FLAG-METTL3 or FLAG-METTL3-D395A, which lacks methyl-transferase activity (Wang, Mol Cell 2016). Cells were fixed, stained for ftz mRNA by fluorescent in situ hybridization and METTL3 using anti-FLAG antibodies. The nuclear and cytoplasmic distribution of ftz mRNA was quantified as described in the manuscript. Note that this is the average of one independent experiment (each bar consisting of the average of at least 50 cells). We plan to repeat this two more times, but we anticipate that these will show the same result.

We could include this negative data as a supplemental figure. We believe that there are two possible reasons for this experimental result. First, as the reviewer points out, the reporter transcripts are already too nuclear to detect any significant change. Second, METTL3 is part of a larger complex that includes several proteins including METTL14, WTAP and potentially other proteins (for example see Covelo-Molares, Nuc Acid Res 2021). We may need to co-express all of these proteins to see an effect.

Experiment #6) We have also expressed versions of ftz-Δi and ftz-Δi-5’SS mRNA with optimized m6A modification (i.e. DRACH) motifs (AGACT) to enhance methylation (“e-m6A-ftz”). We only observed a slight increase in nuclear retention but it is not significant (see Figure 2A,C in the revised manuscript).

Again, this result could be explained by the fact that the reporter is too nuclear to detect any significant increase in retention. We had originally performed this in parallel with the no-m6A-ftz-Δi-5’SS reporters but did not report this negative data in the original manuscript.

__Some rather minor changes to the presentation of the data could enhance the impact of this study.

Specific Comments:

The primary question in this manuscript is comparing reporters with m6A site (intact DRACH sequences) to those without. For this reason, organizing the data to the +/- DRACH sites are adjacent to one another might make the most sense. This point is evident in Figure 1C where perhaps simply changing the order of the bars presented to put the ones directly compared adjacent would be preferable. Then the p-value would compare sets of data directly adjacent to one another. __

We thank the reviewer for this suggestion and we have made these changes to the figures in the preliminary revised manuscript.

__While the authors show representative fields/cells for most assays, they do an excellent job of providing quantitation as well. One exception is Figure 3D, which shows a single cell image for the most key panel (the 5'SS-containing reporter upon Mettl3 depletion). If there is not a field with more cells, the authors could create a montage. __

In the revised manuscript, we have replaced this image with one containing multiple cells expressing the reporter.

__Minor Comments:

Figure presentation:

The text in a number of the figures is VERY small (Figures 1B,1C, and 4A) for example. __

We have fixed this in the new manuscript.

__Figure 3A includes the label "shRNA:" at the top, but these cells are treated with Mettl3 inhibitor and there does not appear to be any shRNA employed, so this seems like a labeling error. __

We have fixed this in the new manuscript.

__In Figure 3C, the immunoblot of Mettl3, there are three bands that all disappear completely upon knockdown of Mettl3- are these all Mettl3? This should at least be mentioned in the legend and perhaps indicated in the figure. The authors do mention in the text employing shRNAs to target multiple Mettl3 isoforms, so likely this is the case. __

We have clarified these issues in the new manuscript.

__Minor points (some really minor to just polish the presentation for clarity):

The word "since" should only be used if there is a time element- otherwise the word "as" is preferable.

For example on p. 4, the sentence: "Since inhibition of mRNA export typically enhances the nuclear retention of RNAs with intact 5'SS motifs (Lee et al. 2020),.." would more precisely read "As inhibition of mRNA export typically enhances the nuclear retention of RNAs with intact 5'SS motifs (Lee et al. 2020),..". __

We thank the reviewer for pointing this out. We have fixed this issue in the revised manuscript.

__Reviewer #2 (Significance (Required)):

Summary: In the manuscript by Lee et al. entitled "N-6-methyladenosine (m6A) Promotes the Nuclear Retention of mRNAs with Intact 5'Splice Site Motifs", the authors provide evidence that m6A modifications within specific regions of transcripts can confer nuclear retention. These results are important because they add to our understanding of how m6A modifications can contribute to post-transcriptional regulation. Although the authors do not quite come out and say this, data seem to be accumulating to suggest that the location of the m6A modifications within a given transcript can dictate the functional consequences of those modifications.

This study would be of significant interest to those that study gene expression in any context as well as cell biologists as the data add to our understanding of export of mRNA from the nucleus. This work also adds to our understanding of the biological consequences of m6A modification, which is an area of significant interest. In my opinion, the authors could make a broader conclusion that we do, which is that the location of the modification significantly dictates function- an extension of previous findings mostly focused on processed mRNA transcripts. __

-------

__Reviewer #3 (Evidence, reproducibility and clarity (Required)):

Quality control of mRNA is vital for all types of cells. In eukaryotic cells, nuclear export of misprocessed mRNAs containing the 5' splice site is prevented. In this manuscript, Lee and colleagues demonstrate that the nuclear retention of intronic polyadenylated transcripts is dependent on m6A modification. Based on the results shown in yeast, they perform immunoprecipitation experiments and demonstrate the interaction between ZFC3H1, a component of the PAXT complex, and YTHDC1 and YTHDC 2, nuclear YTH RNA-binding proteins that recognize m6A-modified transcripts. The study also shows the interaction of U1-70K with YTHDC1 and with ZFC3H1. Depletion of YTHDC1/2 prevents the nuclear retention of IPA transcripts. Additionally, CLIP-seq analysis is performed, demonstrating that m6A modification is enriched around the 5' splice site motif and the 3' polyadenylation site in IPAs. From these observations, they conclude that m6A modification contributes to the quality control of mRNA by promoting nuclear retention of misprocessed transcripts.

Major Points 1. The interaction between ZFC3H1 and YTHDC1 is clearly shown by immunoprecipitation of FLAG-tagged YTHDC1 in Figure 1B. However, the co-purification of YTHDC1 with FLAG-tagged ZFC3H1 in Figure 1C is rather ambiguous. Additionally, the immunoprecipitated samples do not appear to show signals corresponding to FLAG-tagged ZFC3H1, making it unclear if the immunoprecipitation is working. It is essential to provide a better quality result to clarify these observations. __

Please see our responses to reviewer #1. We have repeated the co-immunoprecipitation of FLAG-ZFC3H1 (1-1233) with YTHDC1 under more stringent conditions and have reduced the background binding (see Figure 1B and D in the new manuscript). We have also determined why the FLAG-ZFC3H1 is smaller than expected as the construct contains a premature stop codon. As explained above, we are in the midst of generating a full-length FLAG-ZFC3H1 and we plan to repeat the co-immunoprecipitation with this new construct.

2. While the authors demonstrate that the m6A modification is dispensable for the targeting of IPA reporter transcripts to the nuclear speckles, it would be valuable to investigate whether m6A is required for their exit from the nuclear speckles. Do reporter transcripts with m6A motifs remain in the nuclear speckles at later time points?

We have now analyzed the colocalization of nuclear speckles (SC35) with ftz-Δi-5’SS, which contains both a 5’SS and DRACH motifs, and no-m6A-ftz-Δi-5’SS, which contains a 5’SS but lacks DRACH motifs, at steady state – i.e. after 18-24 hours of transfection (as opposed to at early time points as shown in Figure 2D-E of the initial manuscript). Unexpectedly, we see that both mRNAs continue to colocalize with nuclear speckles, although the no-m6A-ftz-Δi-5’SS mRNA is well exported from the nucleus and its signal in nuclear speckles is faint (see Figure 2F-H in the new manuscript).

Previously, we observed that ftz-Δi-5’SS required the 5’SS motif to remain in nuclear speckles at these later time points (Lee PLOS ONE 2015 and Lee RNA 2022). Upon closer inspection, ftz-Δi-5’SS mRNA also accumulates in additional nuclear foci that are not SC35-positive. Our new results may indicate that m6A marks promote the transfer of mRNAs from nuclear speckles to other foci, but more data is required to make a firm statement. Given this, we plan to conduct further experiments which may take a month to complete:

Experiment #7) We are now assessing whether these additional ftz-Δi-5’SS foci correspond to either YTHDC-positive foci which were previously shown to partially overlap nuclear speckles and sequester m6A-rich mRNAs (Cheng Cancer Cell 2022), or “pA+ RNA foci” which accumulate MTR4/ZFC3H1-targetted RNAs when the nuclear exosome is inhibited (Silla Cell Reports 2018). These foci are enriched in ZFC3H1. We plan on co-staining ftz-Δi, ftz-Δi-5’SS, no-m6A-ftz-Δi and no-m6A-ftz-Δi-5’SS with SC35, YTHDC1 and ZFC3H1 to determine whether m6A may help to transfer mRNAs from nuclear speckles to YTHDC1 or ZFC3H1-enriched foci.

__3. Figures 5B and 5C suggest that ZFC3H1 is required for the degradation of IPA transcripts. However, the range of the vertical axis is inappropriate and it is difficult to assess the extent of the increase in expression levels. Please adjust the vertical axis range for improved clarity. __

We thank the reviewer for the feedback we have added additional graphs with an expanded vertical axis to demonstrate that ZFC3H1 is required for the degradation IPA transcripts.

Minor Points 1. page 4, line 2 "RNAse" should be corrected to "RNase".

We thank the reviewer for catching this error. We have fixed this.

__ 2. page 7, line 5: Is the statement "prevents the nuclear export and decay of non-functional and misprocessed RNAs" correct? m6A modification promotes the decay of such RNAs. __

We thank the reviewer for pointing this out. We have altered the text to clarify that m6A promotes decay.

__3. Figure 2E: ftz-∆i should be ftz-∆i-5'SS. __

We thank the reviewer for catching this error. We have fixed this.

__4. Figure 5A: It would be helpful to indicate the number of IPA transcripts analyzed. __

We have included this information.

__Reviewer #3 (Significance (Required)):

Overall, the work is sound and generally well-controlled. This study advances our understanding of the quality control of misprocessed transcripts in higher eukaryotes. This reviewer suggests a few points for clarification or improvement. __

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

Learn more at Review Commons

---

Reply to the reviewers

1. General Statements

We would like to thank the editorial staff and the reviewers for their handling of our manuscript. We were very pleased with the timely communications from Review Commons, and we are grateful to have been assigned this insightful and constructive group of reviewers.

The reviewers were well-suited to evaluate our work based on their stated areas of expertise (cancer biology, image analysis, machine learning, cell-based screening, etc.). As such, we received thoughtful and constructive feedback, which we have already incorporated into our attached revision. We are confident that these reviews have improved our manuscript.

Our goal with this manuscript is to present a proof-of-concept study where high-content imaging and morphological profiling are used to characterize drug resistance in clonal cell lines. The main criticism from reviewers was that our original manuscript may have overstated our method’s ability to discriminate the signal of bortezomib resistance and that any extension beyond cultured cells (to patient samples for example) would require significant follow-up studies. The reviewers suggested that such work would be beyond the scope of our study, and recommended toning down our language to better reflect the limitations of this proof-of-concept work. We have embraced this suggestion, extensively revising our text, and we now believe our language and tone more accurately reflects our results. The reviewers also suggested follow-up computational analyses to more robustly characterize the bortezomib resistance signature. We have performed these analyses and added their description to our revised manuscript. We feel that these analyses have improved understanding of the signature, and will help a reader to gain a deeper understanding of our results and methodology.

The reviewers also suggested several minor changes; many of which we embraced fully, but others that we chose not to incorporate. We felt that a lack of clarity in our text contributed to these reviewer suggestions. In these cases, we improved clarity in the text and responded to each comment point-by-point in the “prefer not to carry out” section. Further, we address all reviewer comments in the following document point-by-point, grouped by common themes across reviewers (e.g., tone, clarity, analyses, etc.).

Lastly, a common theme among reviewer comments was their appreciation for our strong methodology and data transparency (examples pasted below). We are extremely gratified by this observation as we feel this is a particular strength of our manuscript. In addition, we were pleased to see reviewers engaged by our work, acknowledging the interest this manuscript is likely to generate among a broad range of scientific disciplines.

Examples of reviewer appreciation of our strong methodology and data transparency:

Reviewer 1: “However, this does not imply that the same approach can not achieve the goal, perhaps by using other cell painting markers for bortezomib-sensitivity, or with the same markers to assess sensitivity of different drugs. The cell painting + analysis approaches are not new and the clinical impact is questionable, but the technical aspects (data, analysis) are exceptional and the concept may hold as I described above.”

Reviewer 2: “The paper is well written, and the text is clear, as is the presentation of data and transparency of methods being utilized. The methods were applied appropriately and followed established standards in the field. The paper's premise is timely and interesting, addressing a pressing issue in cancer therapy: making informed treatment decisions fast, based on markers found in tumors early in tumor development, and using image-based screening for characterizing drug resistance before treatment could be an option. A fascinating bit of the manuscript is the description of the feature selection from the screen is done systematically, considering the technical and biological variability and technical artifacts and modeling covariates using linear models seems a very appropriate way of doing so and could serve as another proof of concept that this is indeed the most robust way of modeling and removing signal of technical covariates from the data.”

Reviewer 3: “The strengths of this study are the machine learning best practice and detailed methodology. The experiments could be reproduced and statistical analysis is more than adequate. The analysis takes into account batch effects, well position, differences in cell numbers, and other sources of technical variation that complicate high-content image analysis. It is a good exemplar of how unsupervised morphological profiling can be applied to imaging data. The major limitation is the generalizability of this particular method for patient samples. This could be addressed in the Discussion.”

1. Description of the planned revisions

We have incorporated all planned revisions.

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

Text revisions already carried out

1. [Text revision] We have materially toned down our claims in the manuscript in two distinct areas: A) model performance and B) potential clinical application. A) Model performance. We specifically balanced our discussion of the discriminative signal of the Bortezomib Signature. While the signature adequately separated never-before-seen wildtype and resistant clones with metrics well above randomly permuted baselines (accuracy near 80%, average precision about 70%, area under the ROC curve (AUROC) about 84%), there were many limitations that we should have more explicitly highlighted. For example, many individual profiles were incorrectly classified, some clones were predicted entirely incorrectly, and many profiles did not receive Bortezomib Signature scores above the randomly permuted baseline. We have more clearly discussed these limitations and used more balanced language (see key examples of text-based changes below). Additionally, we modified a figure (now Figure 3) to include boxplots of clones that explicitly show the Bortezomib Signature scores of each well profile and permit examination of the strength of the signature for each clone (previously found in Figure 2-Supplement 9). Lastly, we add a new supplementary figure (now Figure 5-Supplement 1) that describes a feature space analysis of misclassified samples. Please note that this figure rearrangement and new analysis helped to balance our claims, but were also performed in response to other tangential reviewer comments. B) Clinical application. In the abstract, introduction, and discussion, we further emphasized that this work is a proof of concept, and that more advances must be made prior to clinical application.

We made these changes in direct response to the following reviewer comments:

Reviewer 1 - Major Comment 1 (relevant excerpts)

While I am convinced that the signature captures morphological phenotypes associated with drug resistance, at the cumulative scale, the discriminative signal of a single cell type seems weak… With Fig. 4, the data fully supports the argument that the bortezomib-signature encodes bortezomib-resistance, but the signal is weak. Thus statements such as "We found the Bortezomib Signature could predict whether a cell line was bortezomib-resistant or bortezomib-sensitive" (line #172) and the specificity statements in the abstract" (line #28) are not supported by the data in my opinion. I would recommend the authors to tune down these and other related statements throughout the manuscript.

Reviewer cross-commenting - Reviewer 1

My main critic is regarding "over selling" a weak discriminative signal. Specifically, I am not convinced that the major claims regarding predicting sensitivity and specificity at the single cell types scales are supported by the data. Since reviewer #2 and #3 did not raise this concern I think it is worth discussion here.

Once these statements are tuned down - I think no significant additional work is needed to make the point that they can measure a discriminative signal. If they want to make these claims, perhaps they'd like to collect more data to gain statistical power (but I am not optimistic this will work at the single cell level).

Personally, I was happy with the authors' choice of cell lines not included in the training dataset. I am not convinced that additional cell lines + validations are necessary for making the point of a proof of principle.

Reviewer cross-commenting - Reviewer 2

I agree that, perhaps, my major criticism of the paper was the manuscript's 'overselling' of claims that were only weakly supported by the data. Yes, if the authors tune down their claims and clearly state that this is an interesting starting point and proof of concept study, it might be ok to publish with only minor revisions. If the claims should be more generalized, then this study needs more data supporting the conclusions and the method's predictive power.

Reviewer 2 - Major Comment 8

Lastly, I find some misfits between the question, the model used, and the conclusions drawn. The authors start by exploring the problem of bortezomib resistance in cancer treatment, which they say is a devastating issue for patients with, e.g., multiple myeloma. Yet, the authors use HCT116 as their model cell line, a microsatellite instable, colorectal cell line with several intrinsic mutations that make it a difficult model to address physiologically relevant medical problems after all. The authors then go on to suppose that their method might be suitable to diagnose resistance in patient samples, but I am not convinced this conclusion can be speculated based on data from HCT cells. I suggest the authors test their approach on at least two other cell lines (maybe from different tissues) and benchmark their results against a dataset of digital pathology where such predictions are made from stained and analyzed tissue slices. This way, after a thorough benchmark against related third-party data sets, the method would significantly gain relevance, the paper would appeal to a broader audience, and the advance gains more merit.

Reviewer 3 - Major Comment 5

It is not clear from the Discussion whether this type of analysis is more broadly applicable to cell lines derived from patients, rather than selected from a parental cell line, or if this approach would be more efficient than genotyping or next-gen sequencing. How many replicates and ground truth cell lines would be necessary for predictive confidence?

We edited the last two sentences of the abstract to tone down specificity claims (“provide evidence”) and clarify that we are establishing a “proof-of-concept framework”.

• This signature predicted bortezomib resistance better than resistance to other drugs targeting the ubiquitin-proteasome system. Our results establish a proof-of-concept framework for the unbiased analysis of drug resistance using high-content microscopy of cancer cells, in the absence of drug treatment.

We revised the last paragraph of the introduction to contrast bortezomib predictions with ixazomib/CB-5083 predictions, and to remove claims about “using microscopy to guide therapy”.

• This morphological signature correctly predicted the bortezomib resistance of seven out of ten clones not included in the signature training dataset. Overall, our results establish a proof-of-concept framework for identifying unbiased signatures of drug resistance using high-content microscopy. The ability to identify drug-resistant cells based on morphological features provides a valuable orthogonal method for characterizing resistance in the absence of drug treatment.

To tone down claims in the figures, we added boxplots to Figure 3 (previous Figure 2) showing specific distribution of signature scores per well profile and updated Figure 4 legend (previous Figure 3).

• Figure 4. Bortezomib Signature has limited ability to characterize clones resistant to other ubiquitin-proteasome system inhibitors.

We modify the following text in the discussion to tone down claims of specificity and clinical utility:

• This Bortezomib Signature correctly predicted the bortezomib resistance of seven out of ten clones not included in the training dataset and was more specific to bortezomib-resistance given its limited ability to identify clones that were resistant to other UPS-targeting drugs.

Though it is unclear whether this method can be extended to patient samples, where identifying intrinsic drug resistance in cells prior to treatment has the potential to improve targeted cancer therapy, our results are an encouraging proof of concept. We expect that further refinement may develop Cell Painting as a tool for identifying drug-resistant cells, perhaps even guiding strategies to overcome intrinsic resistance.

1. [Text revision] We defined LD50 in text (originally line #97), changed description of resistant clone selection to remove main text references to LD90 (originally line #87), and stated drug concentrations used for selection in Methods. We also defined LD90 in the Methods and described its role in determining the drug concentrations to use for clone selection. This change was in response to the following comments:

Reviewer 1 - Minor Comment 2

What is LD90 (line #87)? LD50 (line #97)?

Reviewer 2 - Minor Comment 5

What was the LD 90 per drug on HCT cells? Rather than LD90 foldchanges, absolute concentrations should be used in the results and discussion to allow the reader to vet the conclusions.

• To determine the appropriate drug concentrations to use in order to isolate drug-resistant clones, we performed proliferation assays on HCT116 parental cells with our drugs of interest: bortezomib (proteasome inhibitor), ixazomib (proteasome inhibitor), or CB-5083 (p97 inhibitor) (Fig. 1-Supplement 1 A-D).
• We characterized the bortezomib-resistant clones and found that the median lethal doses (LD50s) were ~2.8- to ~9-fold that of HCT116 parental cells (Fig. 1-Supplement 2 B).
• Briefly, HCT116 cells were plated in 150 mm dishes and grown in the presence of the desired drug at a concentration that resulted in the death of the majority of cells (selection concentrations: bortezomib, 12 nM; ixazomib, 150 nM; CB-5083, 600 and 700 nM).

• Using the data from our proliferation assays, we calculated the median lethal dose (LD50) for each of our drugs of interest by fitting data of normalized growth vs. log[drug concentration] to a sigmoidal dose-response curve using GraphPad Prism (v.9.2.0) (Fig. 1-Supplement 1 D).

• [Text revision] We thank the reviewer for allowing us an opportunity to improve clarity on the clones we used. We now describe the total number of clones generated and removed unnecessary references to specific clones for ease of reading (originally lines #96-98) (We maintain all references to specific clones in the figures, legends, supplement, and methods)

Reviewer 1 - Minor Comment 3

It was not clear to me in the text which and how many cell lines were evaluated and the reader is forced to go to the SI. For example, "(BZ01-10 and BZ clones A and E)" (line #96-97) and "wild-type clones (WT01-05, 10, and 12-15)" (line #98) appeared when presenting the results without a clear explanation and made it harder for me to follow. Summary of the data (for example, based on Figure 2-Supplement 8) can be briefly mentioned in the text to make it more clear for the reader.

We added the following to the second paragraph of the results:

• Together these methods provided a total of twelve bortezomib-resistant, five ixazomib-resistant, five CB-5083-resistant, and twelve bortezomib-sensitive clones as well as HCT116 parental cells for our experiments.

[Text revision] We removed duplicate text (originally lines #115-125).

Reviewer 1 - Minor Comment 5

1. Lines #104-111 were duplicated in lines #114-122.

Reviewer 3 - Minor Comment 4

Ten lines of text are duplicated on page 5.

Reviewer 2 - Minor Comment 4

on page 5, paragraph 4, there is a sizeable copy-and-paste error of text being identically replicated.

1. [Text revision] We provided more intuition of the Bortezomib Signature in the results section (originally lines #150-151).

Reviewer 1 - Minor Comment 6

The "Bortezomib Signature" is a critical measurement but is only briefly mentioned in lines 150-151 ("..based on the direction-sensitive ranking method for phenotype analysis, singscore (Foroutan et al., 2018)"). Please provide more information/intuition.

• We used these 45 features to compute a rank-based resistance score or “Bortezomib Signature” for each well profile based on the direction-sensitive method called singscore (Foroutan et al. 2018). Singscore ranks these 45 resistance-related features on a per sample basis and calculates a normalized score between -1 and 1, with higher values expected for bortezomib-resistant clones and lower values expected for bortezomib-sensitive clones.

• [Text revision] We clarified that DNA sequencing had been performed solely on clones A and E in a previous study (originally lines #88-90). Furthermore, one of the strengths of our approach is that it can identify resistant clones in an unbiased fashion prior to molecular characterization. It is beyond scope to perform these sequencing studies in the present paper.

Reviewer 2 - Minor Comment 3

The authors talk about validating the mutation - PSMB5 by RNA-seq. However, the data for the genotyping/sequencing/characterization of these newly generated BZ-resistant lines are missing.<br />

In the results, we clarify DNA sequencing that was previously performed on clones A and E

• We also isolated bortezomib-sensitive (wild-type; WT) clones by dilution of the HCT116 parental cell line and acquired two bortezomib-resistant clones (BZ clones A and E) both with mutations in PSMB5 identified by RNA sequencing performed in previous work (Fig. 1-Supplement 1 E) (Wacker et al. 2012).

In the last paragraph of the discussion, we highlight the strength of our unbiased approach

• Together, our work has demonstrated the potential for morphological profiling with Cell Painting to be used as an unbiased method to characterize resistance in the absence of drug treatment. Our results indicate that different mechanisms of bortezomib resistance may generate distinct morphological profiles; with larger and broader training datasets, it may be possible to identify signatures for distinct mechanisms of bortezomib resistance as well as signatures of resistance to other drugs. Though it is unclear whether this method can be extended to patient samples, where identifying intrinsic drug resistance in cells prior to treatment has the potential to improve targeted cancer therapy, our results are an encouraging proof of concept. We expect that further refinement may develop Cell Painting as a tool for identifying drug-resistant cells, perhaps even guiding strategies to overcome intrinsic resistance.

• [Text revision] We thank the reviewers for their suggestions. We agree that the description of the experimental design was somewhat unclear and have provided greater detail and clarity, particularly regarding the generation of clones. We used the HCT116 parental cell line to generate drug-resistant clones by identifying single surviving cells after drug treatment and allowing these cells to expand prior to isolating colonies for experimentation. We did not perform experiments to confirm whether these “clones” were isogenic and can not exclude cell migration during expansion or genetic drift as convoluting factors. However, we have provided greater detail in the descriptions of our method for clone isolation in order to address this concern.

Reviewer 1 - Minor Comment 1

More information in Fig. 1's legend would be helpful to follow the experimental design. I found it hard to follow in its current form and had to go back to carefully reading the main text to fully understand.

Reviewer 2 - Minor Comment 6

The description of the resistant clonal populations is confusing. As I understand, no single-cell clones were isolated during the selection procedure. Thus, the training lines are not yet isogenic clones but oligoclonal sub-populations of the parental cell line. The authors could provide more details here and discuss the different characteristics of their sub-populations, e.g., their growth kinetics or molecular alterations.

We bolstered the description in the results.

• We first isolated and characterized drug-resistant cells (Fig. 1 A). To isolate drug-resistant clones, we used an approach we have described previously (Wacker et al. 2012; Kasap, Elemento, and Kapoor 2014) and the HCT116 cell line. These cancer cells express multidrug resistance pumps at low levels and are mismatch repair deficient, providing a genetically heterogeneous polyclonal population of cells (Umar et al. 1994; Papadopoulos et al. 1994; Teraishi et al. 2005) allowing for isolation of drug-resistant clones in 2-3 weeks. We hypothesize that a rapid selection of resistance could favor the isolation of clones with intrinsic resistance. To determine the appropriate drug concentrations to use in order to isolate drug-resistant clones, we performed proliferation assays on HCT116 parental cells with our drugs of interest: bortezomib, ixazomib, or CB-5083 (Fig. 1-Supplement 1 A-D). We also isolated bortezomib-sensitive (wild-type; WT) clones by dilution of the HCT116 parental cell line and acquired two published bortezomib-resistant clones (BZ clones A and E) both with mutations in PSMB5 identified by RNA sequencing performed in previous work (Fig. 1-Supplement 1 E) (Wacker et al. 2012). We characterized the bortezomib-resistant clones and found that the median lethal doses (LD50s) for bortezomib were ~2.8- to ~9-fold that of HCT116 parental cells (Fig. 1-Supplement 2 B). In contrast, bortezomib-sensitive clones had LD50s for bortezomib that ranged from ~0.7- to ~1.2-fold that of HCT116 parental cells (Fig. 1-Supplement 2 A). Together these methods provided a total of twelve bortezomib-resistant, five ixazomib-resistant, five CB-5083-resistant, and twelve bortezomib-sensitive clones as well as HCT116 parental cells for our experiments.

We also updated the legend for Figure 1A.

• Figure 1. Experimental design for using Cell Painting to examine morphological profiles of drug-resistant cells. (A) Graphic of the experimental workflow: we isolated drug-resistant clones by treating parental HCT116 cells with a high dose of the desired drug and then expanded them for experiments. We isolated drug-sensitive clones by diluting HCT116 cells and then expanded them for experiments. We then performed proliferation assays on select clones to screen for multidrug resistance. Next, we performed Cell Painting on both drug-resistant and -sensitive clones, using multiplexed high-throughput fluorescence microscopy of fixed cells followed by feature extraction and morphological profiling to search for features that contribute to a signature of drug resistance.

• [Text revision] We clarified that the Bortezomib Signature did not correspond to well position (originally lines #155-157).

Reviewer 1 - Minor Comment 9

Line #155-156: "We found that the pattern of Bortezomib Signatures corresponded to the cell identity plate layout", the word "not" is missing before "corresponded".

We found that the pattern of Bortezomib Signatures did not correspond to well position relative to the plate (Fig. 2-Supplement 7 B), indicating that the well position for each clone was not strongly contributing to its Bortezomib Signature.

1. [Text revision] We explicitly described the result that some misclassified clones (WT10, WT15, and BZ06) did not have unexpected bortezomib sensitivity as determined by proliferation assays. We also moved the supplementary figure to an updated Figure 3 to better highlight this result (described below in “Figure revisions already carried out”). Lastly, we add a new figure (Figure 5-Supplement 1) to more explicitly analyze the misclassified lines (described below in “New analyses already carried out”).

Reviewer 3 - Minor Comment 3

The bortezomib sensitivity of the WT lines used in the last experiments was determined and did not seem to be greater than parental. This could be mentioned in the text; the figure raises the question and the answer is provided, but it's in the supplemental material.

While the Bortezomib Signature correctly characterized the bortezomib sensitivity of most clones, it consistently misclassified others (WT10, WT15, and BZ06) (Fig 5-Supplement 1 A). Proliferation assays conducted in earlier experiments showed that WT10 and WT15 were sensitive to bortezomib while BZ06 was resistant (Fig. 1-Supplement 2 A and B). By comparing these incorrect predictions with high-confidence correct predictions, we observed differences that varied by clone type, suggesting unique morphology may be driving each of these misclassifications (Fig. 5-Supplement 1 B and C). These results are consistent with the Bortezomib Signature being generalizable to clones not included in the training dataset and suggest that morphological profiling has the potential to identify bortezomib-resistant clones based on the morphological features of cells in the absence of drug treatment.

1. [Text revision] We clarified that the metrics (accuracy and average precision) were based on median Bortezomib Signature scores of all replicate well-level profiles per clone. We can compare samples based on rank, and difference from 95% confidence interval of permuted data. There is no current way for our method to assign a likelihood. Also note that we have updated the discussion to discuss alternative metrics (see Reviewer 1 - Minor Comment 7) These are very important distinctions, and we are grateful to the reviewer for bringing them up.

Reviewer 3 - Major Comment 3

The study classifies cells as binary sensitive or resistant, but would results be improved by scoring based on likelihood of being resistant/sensitive?

Reviewer 3 - Minor Comment 2

It is not clear whether the accuracy was based on a percentage of replicates per cell line that were classified correctly or whether that was referring to classification of the cell line overall as sensitive/resistant.

• We next examined whether the Bortezomib Signature was able to predict the bortezomib resistance of a clone based on morphological profiling data (Fig. 3 A-E and Fig. 3-Supplement 2 A and B). We called the clone bortezomib-resistant if the median Bortezomib Signature of all replicate well profiles was greater than zero and bortezomib-sensitive if the median Bortezomib Signature less than zero. In the training dataset, the Bortezomib Signature correctly predicted the bortezomib resistance of all ten clones, with median Bortezomib Signatures for eight out of ten clones beyond the 95% confidence interval for the randomly permuted data (Fig. 3 A). The accuracy of the Bortezomib Signature was 88% while the average precision was 81% for the training dataset (Fig. 3-Supplement 2 A and B) (see Methods). The signature performed similarly well in the validation dataset (Fig. 3 B), with an accuracy of 92% and an average precision of 89% (Fig. 3-Supplement 2 A and B). In the test dataset the Bortezomib Signature correctly predicted the bortezomib resistance of all clones, though only HCT116 parental cells had a median Bortezomib Signature outside the 95% confidence interval for the randomly permuted data (Fig. 3 C). The test dataset had an accuracy of 80% and an average precision of 68% (Fig. 3-Supplement 2 A and B). Similarly, in the holdout dataset the Bortezomib Signature had an accuracy of 78% and an average precision of 69% (Fig.3 -Supplement 2 A and B), and correctly predicted the bortezomib resistance of twelve out of thirteen clones, with WT01 misclassified as bortezomib-resistant (Fig. 3 D). In the holdout dataset, four of the twelve correctly characterized clones had median Bortezomib Signatures outside the 95% confidence interval for the randomly permuted data.

We also mirrored language when discussing the ixazomib and CB-5083 results.

• However, only two of the four correctly identified ixazomib-resistant clones and one of the three CB-5083-resistant clones had median Bortezomib Signatures outside the 95% confidence interval of the randomly permuted data. The area under the ROC (AUROC) curve for ixazomib-resistant and CB-5083-resistant clones (0.63 and 0.60, respectively) was lower than those calculated for the training, validation, test, and holdout datasets. In addition, many of the Bortezomib Signatures for well profiles of ixazomib- and CB-5083-resistant clones, particularly those for CB-5083-resistant clones, landed within the 95% confidence interval of the randomly permuted data. These results suggest that the Bortezomib Signature is not a general signature of UPS-targeting drug resistance and instead has some specificity for bortezomib.

• [Text revision] We added an explicit note that our image analysis pipelines are also publicly available. Our reporting of our data processing pipelines are documented fully and well above standards in our field. Linking the publicly-available resources with these methods maximizes reproducibility.

Reviewer 1 - Minor Comment 10

Additional details on the processing steps in the analysis pipeline in the Methods will be highly appreciated.

We include all image analysis pipelines at https://github.com/broadinstitute/profiling-resistance-mechanisms (G. Way et al. 2023).

1. [Text revision] We have compared our approach to the on-disease/off-disease scores as introduced in (Heiser et al. 2020). We agree with the reviewer that a discussion of these two methods would help clarify our phenotypic signature concept. The on/off score is about the degree to which a perturbation pushes disease towards a healthy state. In this case we have 3 sets of data: healthy samples (used for training), disease samples (used for training), and the sample we want to score, which should be of the form "disease + perturbation". With our approach, based on singscore, we also have 3 sets of data: sensitive samples (used for training), resistance samples (used for training), and the sample we want to score. Here, our sample we want to score could be anything, not necessarily of the form "resistance + perturbation". Furthermore, singscore does not have the concept of orthogonality to resistance/sensitivity. This would become relevant if we were exploring perturbations or conditions that would induce a resistant cell line to become sensitive, but we are not doing that here. There are other statistical differences (projection vs. rank based etc.) but the key difference is the applicability of the method to the specific problem at hand.

Reviewer 1 - Minor Comment 7

How is the Bortezomib Signature related to the "on-disease"/"off-disease" scores described in https://www.biorxiv.org/content/10.1101/2020.04.21.054387v1.full? Are there other alternatives used for similar binary phenotypic signatures? What is the justification for using these measurements? I would love to see this generalized concept explicitly discussed in the Discussion.

We added the following to the discussion.

• The Bortezomib Signature is conceptually similar to the on-disease/off-disease score (Heiser et al. 2020). Both require three phenotypic measurements: a target phenotype representing ideal, a disease phenotype, and a new phenotype to classify. However, our approach is technically different (non-parametric compared to linear projection) and our goals are different (phenotypic classification compared to perturbation alignment). Other methods also enable phenotype labeling, but they focus on single-sample annotation without regard to a target phenotype (Wawer et al. 2014; Rohban et al. 2017; Simm et al. 2018; Nyffeler et al. 2020).

Figure revisions already carried out

1. [Figure revision] We moved all boxplots from the original Fig. 2-Supplement 9 to the main text (also splitting Fig. 2 into Fig. 2 and 3). From the original Figure 2, we moved the accuracy and average precision bar graphs to the supplement. We also note that this change increases transparency of the discriminative signal of our signature.

Reviewer 1 - Minor Comment 8

I would highly recommend showing the Bortezomib Signatures from Figure 2-Supplement 9. in Fig. 2. This was the main measurement used throughout the manuscript and in my opinion, it is very important to consistently visualize the data along the manuscript, for clarity and easier reader interpretation.

1. [Figure revision] We adjusted the position of the legend in the accuracy and average precision bar graphs (originally Fig. 2 C and D, now Fig. 3-Supplement 2) for clarity. We also note that keeping the bar chart here is standard best practice (compared to a dot plot).

Reviewer 1 - Minor Comment 4

I found the visualization in Fig. 2C-D not intuitive (it is properly explained in the legend). I suggest replacing the accuracy colorbar with a color marker to make it more distinct from the random permutation (|--*--|) The location of the text "mean +- SD of 100 random permutation" made me first think that it is linked to the holdout.

1. [Figure revision] We changed the point distribution in the boxplots (from expanded to standard) to minimize overlap with the boxplot lines. We also updated the legend text to indicate that individual points in boxplots represent the Bortezomib Signature for well profiles. Note, we paste a representative example of this change above (new Figure 3).

Reviewer 3 - Minor Comment 1

I found the box plots somewhat difficult to interpret (especially where the WT lines had a lot of overlap with the red shaded area). Do the points in these charts correspond to replicate wells?

We also update the figure legend.

• Plots show values for individual well profiles (points), range (error bars), 25th and 75th percentiles (box boundaries), and median.

• [Figure revision] [Response to Reviewer 2 - Major Comment 7] We thank the reviewer for allowing us an opportunity to clarify the mechanism. We feel that it is beyond scope of this manuscript to disentangle the molecular alterations that cause bortezomib resistance based on our Cell Painting insights. This wet lab experimental process is arduous and cost prohibitive, and we argue that one of the benefits of taking a morphology approach to resistance status is that we can detect resistant cells (and therefore cells that won’t die when presented with a treatment) without knowing the molecular mechanism.

Nevertheless, the reviewer has encouraged us to enhance the ability for a reader to view and interpret the signature to perhaps more easily facilitate future work. Previously, we presented our signature in text form in Figure 2-Supplement 4 and in heatmap form in Figure 2-Supplement 5. Here, we add a new figure (Figure 2-Supplement 6; pasted below) which will improve interpretability.

Reviewer 2 - Major Comment 7:

Next to feature importance, the authors do not discuss (or I missed) what biology the features represent. Such the reader is left wondering what the actual mechanism of bortezomib resistance could be and if cell painting could shed light on the molecular alterations that cause the treatment resistance. While reviewing, I thus wondered which audience the authors targeted with their manuscript. A more focused analysis of their data that highlights aspects of the study either for the machine learning community, the cell biology community, or the precision oncology community would greatly benefit the manuscript's impact. In its current form, the study's findings seem diluted and spread across a wide range of research questions.<br />

• Figure 2-Supplement 6. Bortezomib Signature visualized by CellProfiler features. Visualization of CellProfiler features contributing to the Bortezomib Signature. Features with high values (mean signature estimates) in resistant cells are purple while features with low values in resistant cells are green. The mean signature estimates were based on Tukey's Honestly Significant Difference test score and the number in each box represents the number of features used to calculate the mean signature estimate.

Additionally, we add the following to the results section:

• We then examined the grouping of features across compartments and channels and found radial distribution features were higher in resistant cells (Fig 2-Supplement 6).

The code change to generate the signature visualization summary is available at: https://github.com/broadinstitute/profiling-resistance-mechanisms/pull/131

New analyses already carried out

1. [New analysis] [Response to Reviewer 2 - Major Comment 5] We agree that a systematic analysis of feature selection methods will provide additional insights not already in the manuscript. Therefore, we have performed two new computational experiments to compare our linear modeling feature selection approach against other standard approaches. We demonstrate that our linear modeling approach is effective at isolating the core differences between resistant and sensitive classes.

Specifically, we performed two analyses: A) UMAP and B) k-means cluster analysis. We analyzed profiles defined by four different feature selection approaches: 1) Using all traditional CellProfiler features; 2) Using the traditional CellProfiler feature selection approach (removing low variance features, high correlating features, etc.); 3) Using 45 random features (same size as Bortezomib Signature); and 4) Using only the bortezomib signature features. We performed Fisher’s exact tests to derive odds ratios of cluster membership by resistance status and calculated Silhouette widths to quantify relative proximity of clusters.

This analysis generates a new supplementary figure (see below), and demonstrates that the linear-modeling-based feature selection isolated the features driving the differences between the clone types (resistance vs. wildtype) while the standard approaches do not as effectively separate.

Reviewer 2 - Major Comment 5:

A fascinating bit of the manuscript is the description of the feature selection from the screen is done systematically, considering the technical and biological variability and technical artifacts and modeling covariates using linear models seems a very appropriate way of doing so and could serve as another proof of concept that this is indeed the most robust way of modeling and removing signal of technical covariates from the data. Yet, I wondered why the authors do not discuss other means of feature selection or dimensionality reduction; further, they need to show how the features cluster the cell lines or why impact (information content) different features deliver. For an audience interested in the technical aspects of cell painting analysis and machine learning based on the data, that would, IMHO, be the most exciting questions.

• Figure 3-Supplement 3. Benchmarking linear-modeling feature selection to separate clones by bortezomib resistance. Uniform Manifold Approximation and Projection (UMAP) analysis of the qualitative separability of (A) resistance status and (B) Bortezomib Signature scores across four different feature spaces. (C) k-means clustering from k=2 to k=14 of average odds ratio, maximum odds ratio (Fisher’s exact test), and Silhouette width using Bortezomib Signature features.

Additionally, we add the following to the results section:

• We then compared our linear-modeling approach to feature selection against other feature spaces and found that the Bortezomib Signature clusters same-type clones (bortezomib-resistant vs. bortezomib-sensitive) with higher enrichment compared to the full feature space, standard feature selection (see Methods), or a random selection of 45 features (Fig 3-Supplement 3).

And methods section, describing this analysis:

• We were also interested in comparing the ability of different feature spaces to cluster clones of the same type (resistant vs. sensitive). This analysis would determine if the Bortezomib Signature features, which we derived using linear modeling to isolate biological from technical variables, had a greater ability to cluster. We compared the Bortezomib Signature against three other feature spaces: 1) the full feature space, 2) standard feature selection (see Image data processing methods), and 3) 45 randomly selected features. We performed two analyses using these four feature spaces including Uniform Manifold Approximation and Projection (UMAP) (McInnes et al. 2018) and k-means clustering. For UMAP, we used default umap-learn parameters to identify two UMAP coordinates per feature space. We then visualized the clusters by their resistance status and Bortezomib Signature score. The UMAP analysis represents a qualitative analysis. Next, we applied k-means clustering with 25 initializations across a range of 2-14 clusters (k). Prior to clustering and for each feature space, we applied principal component analysis (PCA) and transformed each feature space into 30 principal components. This step was necessary to compare k-means clustering metrics, which are sensitive to the feature space dimensionality. We applied a Fisher’s exact test to each cluster using a two-by-two contingency matrix that specified cluster membership for each clone classification (resistant vs. sensitive). We visualized the mean odds ratio and max cluster odds ratio for each feature space across k. A high odds ratio tells us that the feature space effectively clusters clones of the same resistance status. Lastly, we calculated Silhouette width (the average proximity between samples in one cluster to the second nearest cluster) for each feature space across k.

The code change to derive the UMAP coordinates, perform clustering, and generate the figure is available at https://github.com/broadinstitute/profiling-resistance-mechanisms/pull/132

1. [New analysis] [Response to Reviewer 3 - Major Comment 1] We thank the reviewer for this suggestion, which allowed us to explore the misclassified samples in more depth. We added a new supplementary figure in which we summarized all bortezomib clones (wildtype and resistant) in their accuracy based on the bortezomib signature (panel A). We did not include training set samples in this analysis. Using samples that were consistently incorrectly classified with high confidence (three samples: WT15, BZ06, WT10) we performed two separate two-sample Kolmogorov–Smirnov (KS) tests. Specifically, we compared high incorrect wildtype to high correct wildtype and high incorrect resistant to high correct resistant. Our results indicate that most bortezomib signatures were significantly different between correct and incorrect assignments (panel B), and that the signature features varied between resistant and wildtype misclassification tests (panel C).

Reviewer 3 - Major Comment 1:

While the claims are largely substantiated, there are a few points where further consideration would improve the manuscript. Several cell lines were mis-classified with what appears to be a high degree of certainty. Can the authors tell what was driving those predictions? Was there something in the morphological signature that weighed more heavily in those cases?

• Figure 5-Supplement 1. Examining the accuracy of clone classification and misclassification of clones. (A) Proportion of high-confidence correct, low-confidence correct, low-confidence incorrect, and high-confidence incorrect predictions of well profiles across clones in the test, holdout, and validation sets. High-confidence predictions (high) had a Bortezomib Signatures greater (resistant clones) or less than (sensitive) the 95% confidence interval of randomly permuted data while low-confidence predictions (low) had Bortezomib Signatures within the 95% confidence interval of randomly permuted data. (B) Visualization of Kolmogorov-Smirnov (KS) test statistic means of feature groups across channels and cellular compartments. (C) Plot of the KS test statistic means for feature groups in bortezomib-resistant vs. -sensitive cells. Each feature group is color coded by the imaging channel.

Additionally, we add the following to the results section:

• While the Bortezomib Signature correctly characterized the bortezomib sensitivity of most clones, it consistently misclassified others (WT10, WT15, and BZ06) (Fig 5-Supplement 1 A). Proliferation assays conducted in earlier experiments showed that WT10 and WT15 were sensitive to bortezomib while BZ06 was resistant (Fig. 1-Supplement 2 A and B). By comparing these incorrect predictions with high-confidence correct predictions, we observed differences that varied by clone type, suggesting unique morphology may be driving each of these misclassifications (Fig. 5-Supplement 1 B and C). These results are consistent with the Bortezomib Signature being generalizable to clones not included in the training dataset and suggest that morphological profiling has the potential to identify bortezomib-resistant clones based on the morphological features of cells in the absence of drug treatment.

And methods section, describing this analysis:

Some profiles were consistently predicted incorrectly with high confidence but in the opposite direction (see Figure 5-Supplement 1). For a well-level profile to be categorized as high-confidence (in either the correct or incorrect directions), it needed to score beyond the 95% confidence interval of the randomly permuted data range. For example, a high-confidence incorrect resistant profile would have a Bortezomib Signature below 95% confidence interval of the randomly permuted data. To evaluate the features driving the differences in these samples, we applied two-sample Kolmogorov–Smirnov (KS) tests per Bortezomib Signature feature. We applied these tests to two separate groups: 1) misclassified bortezomib-sensitive vs. high-confidence accurate bortezomib-sensitive and 2) misclassified bortezomib-resistant vs. high-confidence accurate bortezomib-resistant.

The code change to generate the UMAP coordinates and figure is available at https://github.com/broadinstitute/profiling-resistance-mechanisms/pull/130

Description of analyses that authors prefer not to carry out

1. [Response to Reviewer 2 - Minor Comments 1 and 2]: These are interesting suggestions! Still, we prefer not to speculate on the biological mechanism of the Bortezomib signature. Connecting morphological features identified as contributing to the Bortezomib Signature by Cell Painting to specific biological pathways would demand considerable cell-based assays to validate. In addition, our analyses suggest that the features contributing to the Bortezomib Signature are spread across a range of cellular compartments and channels, making it difficult to pin down specific mechanisms or pathways as likely contributors to bortezomib resistance. However, we are adding a figure to increase interpretability of the signature, which will aid in developing future hypotheses. Note that the signature was not possible to detect by eye (Fig. 2 A).

Reviewer 2 - Minor Comment 1:

There could be some speculation on the mechanism of Bortezomib resistance concerning the literature with the existing image data. For example, Bortezomib resistance is connected to serine synthesis and how a particular feature could contribute to the known mechanism.<br />

Reviewer 2 - Minor Comment 2:

Along the same lines, the authors could show that larger cells lead to resistance with microscopic images.

2. [Response to Reviewer 2 - Major Comment 8]: We appreciate the reviewer’s concern that our work using HCT116 clonal cells lines may not directly reflect results from patient samples. Our choice was based on previously published work demonstrating the efficiency with which HCT116 cells generate resistant clones due to diminished DNA mismatch repair and decreased expression of drug efflux pumps. Since our work is a proof of concept rather than a comprehensive demonstration of translating morphological profiling into clinical practice, we believe that experiments using multiple patient cell lines from different tissues as well as digital pathology records to be beyond the scope of this work. We instead chose to tone down the language of our manuscript to more clearly acknowledge the limitations of our work and clarify this as a proof of concept.

Reviewer 2 - Major Comment 8 (relevant excerpt):

I suggest the authors test their approach on at least two other cell lines (maybe from different tissues) and benchmark their results against a dataset of digital pathology where such predictions are made from stained and analyzed tissue slices. This way, after a thorough benchmark against related third-party data sets, the method would significantly gain relevance, the paper would appeal to a broader audience, and the advance gains more merit.<br />

3. [Response to Reviewer 3 - Major Comment 2]: The bortezomib sensitivity of ixazomib- and CB-5083-resistant clones was not determined, and hence can not be ruled out as a possible explanation for their high Bortezomib Signature scores. However, we prefer not to conduct additional proliferation assays for the misclassified clones (IX02, WT06, CB14, CB16) in the presence of bortezomib to determine whether coincidental bortezomib resistance might explain the signature performance. Our rationale is that three other misclassified clones (WT10, WT15, and BZ06) had the expected bortezomib sensitivity in proliferation assays (Fig. 1-Supplement 2), meaning that additional proliferation assays may not reveal any insights regarding the signature performance.

Reviewer 3 - Major Comment 2:

Was the bortezomib sensitivity of the IX (or CB) resistant cell lines determined? If there were differences, this could explain some of the variation in the morphological signatures. This could be easily done in one or two growth experiments.

4. [Response to Reviewer 2 - Major Comment 7]: Thank you for pointing this out. Our goal is to keep the study multi-disciplinary. We are adding a figure to increase interpretability of the signature, and adding text-based clarifications.

Reviewer 2 - Major Comment 7 (relevant excerpt):

While reviewing, I thus wondered which audience the authors targeted with their manuscript. A more focused analysis of their data that highlights aspects of the study either for the machine learning community, the cell biology community, or the precision oncology community would greatly benefit the manuscript's impact. In its current form, the study's findings seem diluted and spread across a wide range of research questions.<br />

5. [Response to Reviewer 2 and 3 - Major Comments 6 and 4]: We prefer not to expand the scope of the model to predict other drug signatures. This would require a substantial amount of work to generate the appropriate drug-resistant clones, collect the imaging data, and analyze it, and we think it important to convey the purpose of our paper is proof of concept. We do not feel that the time invested in performing this analysis would result in adequate returns beyond what we already demonstrate.

Reviewer 2 - Major Comment 6.

Interestingly, the Bortezomib signature is specific to the drug and not a broad range of proteasomal inhibitors. However, seeing the common features between all the proteasomal inhibitors would be interesting.

Reviewer 3 - Major Comment 4

There was some predictive ability of the Bortezomib Signature for ixazomib resistance. Were there some features that were correlated with IX-resistance, i.e. UPS pathway, versus specific to bortezomib? Do the features suggest anything about resistance mechanisms or is the feature set too abstruse to interpret?

References

Foroutan, Momeneh, Dharmesh D. Bhuva, Ruqian Lyu, Kristy Horan, Joseph Cursons, and Melissa J. Davis. 2018. “Single Sample Scoring of Molecular Phenotypes.” BMC Bioinformatics 19 (1): 404.

Heiser, Katie, Peter F. McLean, Chadwick T. Davis, Ben Fogelson, Hannah B. Gordon, Pamela Jacobson, Brett Hurst, et al. 2020. “Identification of Potential Treatments for COVID-19 through Artificial Intelligence-Enabled Phenomic Analysis of Human Cells Infected with SARS-CoV-2.” bioRxiv. https://doi.org/10.1101/2020.04.21.054387.

McInnes, Leland, John Healy, Nathaniel Saul, and Lukas Großberger. 2018. “UMAP: Uniform Manifold Approximation and Projection.” Journal of Open Source Software 3 (29): 861.

Nyffeler, Johanna, Clinton Willis, Ryan Lougee, Ann Richard, Katie Paul-Friedman, and Joshua A. Harrill. 2020. “Bioactivity Screening of Environmental Chemicals Using Imaging-Based High-Throughput Phenotypic Profiling.” Toxicology and Applied Pharmacology 389 (January): 114876.

Rohban, Mohammad Hossein, Shantanu Singh, Xiaoyun Wu, Julia B. Berthet, Mark-Anthony Bray, Yashaswi Shrestha, Xaralabos Varelas, Jesse S. Boehm, and Anne E. Carpenter. 2017. “Systematic Morphological Profiling of Human Gene and Allele Function via Cell Painting.” eLife 6 (March). https://doi.org/10.7554/eLife.24060.

Simm, Jaak, Günter Klambauer, Adam Arany, Marvin Steijaert, Jörg Kurt Wegner, Emmanuel Gustin, Vladimir Chupakhin, et al. 2018. “Repurposing High-Throughput Image Assays Enables Biological Activity Prediction for Drug Discovery.” Cell Chemical Biology 25 (5): 611–18.e3.

Wacker, Sarah A., Benjamin R. Houghtaling, Olivier Elemento, and Tarun M. Kapoor. 2012. “Using Transcriptome Sequencing to Identify Mechanisms of Drug Action and Resistance.” Nature Chemical Biology 8 (3): 235–37.

Wawer, Mathias J., Kejie Li, Sigrun M. Gustafsdottir, Vebjorn Ljosa, Nicole E. Bodycombe, Melissa A. Marton, Katherine L. Sokolnicki, et al. 2014. “Toward Performance-Diverse Small-Molecule Libraries for Cell-Based Phenotypic Screening Using Multiplexed High-Dimensional Profiling.” Proceedings of the National Academy of Sciences of the United States of America 111 (30): 10911–16.

Way, Gregory, Yu Han, David Stirling, and Shantanu Singh. 2023. Broadinstitute/profiling-Resistance-Mechanisms: Analysis for Preprint. Zenodo. https://doi.org/10.5281/ZENODO.7803787.

Way, Gregory P., Maria Kost-Alimova, Tsukasa Shibue, William F. Harrington, Stanley Gill, Federica Piccioni, Tim Becker, et al. 2021. “Predicting Cell Health Phenotypes Using Image-Based Morphology Profiling.” Molecular Biology of the Cell 32 (9): 995–1005.

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 #1

Evidence, reproducibility and clarity

The authors use Cell Painting, a high-content image-based phenotypic assay, to distinguish between clonal cancer cell lines that are resistant versus sensitive to a proteasome inhibitor anti-myeloma drug called bortezomib. The authors characterized a high-dimensional cell morphology signature for bortezomib-resistance, evaluated it on an independent subset of cell lines, and evaluated specificity in respect to other drugs targeting the ubiquitin-proteasome system. The authors thus propose image-based morphology characterization as an alternative method for characterizing drug resistance.

Strengths: solid methodology - cell lines validation of drug resistance, extensive data collection, thorough validation of the analysis pipeline, avoiding potential confounders, biases and proper data partitioning to test and hold-out (what the authors refer to as "machine learning best practices").

Weakness: weak discriminative signal. Some aspects of the writing could be improved to make the manuscript easier to follow (see Minor comments).

Major comments:

While I am convinced that the signature captures morphological phenotypes associated with drug resistance, at the cumulative scale, the discriminative signal of a single cell type seems weak. Specifically, it is not clear whether the signature can effectively capture the drug resistance of a single cell line. In Figure 2-Supplement 9, considering the test (C) and the holdout (D), only 1/9 BZ clones' median signatures were beyond the 95% confidence interval, with 4/6 and 2/6 WT cell types with median signatures beyond the positive and negative 95% confidence interval correspondingly. When defining bortezomib-sensitivity according to the median signatures' sign (>0 or <0) of a cell line, Figure 2-Supplement 9 shows that in the test+holdout there are 9/9 correct bortezomib-resistance (BZ) and 6/7 correct bortezomib-sensitive (WT) predictions. However, similar discrimination levels also appeared in the other drugs (ixazomib, CB-5083), making the statements about specificity less grounded. When the authors evaluate the AUROC they report ~0.6 (line #194) for the non-specific (ixazomib, CB-5083) drugs versus ~0.75 for bortezomib-resistance (line #202). With Fig. 4, the data fully supports the argument that the bortezomib-signature encodes bortezomib-resistance, but the signal is weak. Thus statements such as "We found the Bortezomib Signature could predict whether a cell line was bortezomib-resistant or bortezomib-sensitive" (line #172) and the specificity statements in the abstract" (line #28) are not supported by the data in my opinion. I would recommend the authors to tune down these and other related statements throughout the manuscript. An alternative would be to increase the number of wells and see whether this weak signal can indeed be statistically amplified with many replicates to make a robust and specific characterization of a cell line's bortezomib-sensitivity (but I assume this is a lot of work and probably out of scope of this manuscript). I think it is also important to discuss in more detail the interpretation of these results (including Figure 2-Supplement 9), in this context, in the Discussion.

Minor comments:

Suggested clarifications (some might be less relevant if the manuscript is designed for experts in the more clinical domain who are familiar with these terms / style):

1. More information in Fig. 1's legend would be helpful to follow the experimental design. I found it hard to follow in its current form and had to go back to carefully reading the main text to fully understand.
2. What is LD90 (line #87)? LD50 (line #97)?
3. It was not clear to me in the text which and how many cell lines were evaluated and the reader is forced to go to the SI. For example, "(BZ01-10 and BZ clones A and E)" (line #96-97) and "wild-type clones (WT01-05, 10, and 12-15)" (line #98) appeared when presenting the results without a clear explanation and made it harder for me to follow. Summary of the data (for example, based on Figure 2-Supplement 8) can be briefly mentioned in the text to make it more clear for the reader.
4. I found the visualization in Fig. 2C-D not intuitive (it is properly explained in the legend). I suggest replacing the accuracy colorbar with a color marker to make it more distinct from the random permutation (|--*--|) The location of the text "mean +- SD of 100 random permutation" made me first think that it is linked to the holdout.
5. Lines #104-111 were duplicated in lines #114-122.
6. The "Bortezomib Signature" is a critical measurement but is only briefly mentioned in lines 150-151 ("..based on the direction-sensitive ranking method for phenotype analysis, singscore (Foroutan et al., 2018)"). Please provide more information/intuition.
7. How is the Bortezomib Signature related to the "on-disease"/"off-disease" scores described in https://www.biorxiv.org/content/10.1101/2020.04.21.054387v1.full? Are there other alternatives used for similar binary phenotypic signatures? What is the justification for using these measurements? I would love to see this generalized concept explicitly discussed in the Discussion.
8. I would highly recommend showing the Bortezomib Signatures from Figure 2-Supplement 9. in Fig. 2. This was the main measurement used throughout the manuscript and in my opinion, it is very important to consistently visualize the data along the manuscript, for clarity and easier reader interpretation.
9. Line #155-156: "We found that the pattern of Bortezomib Signatures corresponded to the cell identity plate layout", the word "not" is missing before "corresponded".
10. Additional details on the processing steps in the analysis pipeline in the Methods will be highly appreciated.

Referees cross-commenting

My main critic is regarding "over selling" a weak discriminative signal. Specifically, I am not convinced that the major claims regarding predicting sensitivity and specificity at the single cell types scales are supported by the data. Since reviewer #2 and #3 did not raise this concern I think it is worth discussion here.

Once these statements are tuned down - I think no significant additional work is needed to make the point that they can measure a discriminative signal. If they want to make these claims, perhaps they'd like to collect more data to gain statistical power (but I am not optimistic this will work at the single cell level).

Personally, I was happy with the authors' choice of cell lines not included in the training dataset. I am not convinced that additional cell lines + validations are necessary for making the point of a proof of principle.

Significance

Cell Painting was applied to many applications, but as far as I am aware this is the first attempt for an image-based phenotypic characterization of drug resistance. While the authors established that this approach can measure, to some extent, bortezomib-sensitivity, at the current state of the results, I am not convinced that cell painting can be practically used to assess bortezomib-sensitivity of a single cell line. However, this does not imply that the same approach can not achieve the goal, perhaps by using other cell painting markers for bortezomib-sensitivity, or with the same markers to assess sensitivity of different drugs. The cell painting + analysis approaches are not new and the clinical impact is questionable, but the technical aspects (data, analysis) are exceptional and the concept may hold as I described above.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

Summary: Forer and Otsuka provide first-rate evidence for tethers fixed in place between separating anaphase chromosomes using electron tomography. The authors traced the anaphase movement of a number of living cells before fixation for examination using electron tomography. The manuscript is clearly written and provides an excellent introduction and discussion of the known literature. The reader will have an excellent background to see the importance of this work.

Major comments:<br /> - Are the claims and the conclusions supported by the data or do they require additional experiments or analyses to support them?

No further experiments are needed. The data are very supportive, and extremely clear.<br /> - Are the data and the methods presented in such a way that they can be reproduced? Yes.<br /> - Are the experiments adequately replicated and statistical analysis adequate? Yes.

Minor comments:<br /> - Are prior studies referenced appropriately? Yes.<br /> - Are the text and figures clear and accurate? Absulotely.<br /> - Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

The authors are to congratulated on their major contribution to this study on tethers between separated daughter chromosomes. It is a tpur deforce to go from the living cells to fixing and identifying the same separated chromosomes using electron tomography to see the ultrastructure of the fibers seen fir.

Referees cross-commenting<br /> Thank you reviewer #2. The manuscript should be published. It is an excellent contribution.

We thank the reviewer for the appreciation of the clarity and quality of our work.

Reviewer #1 (Significance):

Provide contextual information to readers (editors and researchers) about the novelty of the study, its value for the field and the communities that might be interested.

This manuscript is the first to use electron tomography to identify the tethers between separated anaphase chromosomes. Forer and the laetMichael Berns and their co-authors have published a number of papers using phase microscopy and lasers to report on the physical nature and elastic properties of these fibres in the past. Forer and Otsuka have presented first-rate evidence for the reality of these structures using electron tomography. This manuscript should highlighted in the published journal.<br /> The chemical identity of these fibers as the authors state is unclear.

The following aspects are important:

• Audience: describe the type of audience ("specialized", "broad", "basic research", "translational/clinical", etc...) that will be interested or influenced by this research; how will this research be used by others; will it be of interest beyond the specific field?

This exciting contribution will be read by anyone interested in mitosis. It will be of interest to all Cell Biologists because of the careful manner in which the living cells were studied before they were fixed for examination using electron tomography. The readers will be dreaming how they can use this process on their Cell Biology problems._

• Please define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate.

I am a cell Biologist who has made contributions, both in light microscopy and in transmission microscopy on diving cells, both in tissue culture and in situ in aviav and zebrafish embryos.

We thank the reviewer for appreciating the significance of our work.

Reviewer #2 (Evidence, reproducibility and clarity):

In this paper, the authors use light microscopy and electron tomography to study anaphase chromosomes in crane fly spermatocytes. They find that there are two "tether" structures that connect telomeres of sister chromatids. One tether is thicker (denser) and extends between sister chromatids during early but not late anaphase, whereas a second, less-dense tether maintains contact with both sister chromatids in all examined stages of anaphase. The paper makes arguments as to what the tethers could or could not be. Specifically, they are too numerous to be ultrafine DNA bridges seen in various normal or abnormal segregation events and they also do not affect anaphase chromosome motion the same way ultrafine DNA bridges do.

Major comments:<br /> The major claim that there are tethers that connect sister chromatids in anaphase is supported by the data. Moreover, the data resolves two types of tethers on the basis of their density. While it is unclear what the composition of the tethers are, the paper makes a convincing case that they cannot be the DNA ultrafine bridges seen in other studies. The discussion has sufficient caveats that most readers will see that more work is needed to identify the composition of the two tethers. In my opinion, no further experiments are needed to support the modest claims of this paper. Therefore, I only have minor comments that may hopefully improve the paper's clarity.

We thank the reviewer for the positive evaluation of our work.

Minor comments:<br /> It was argued that the tethers reported here were also seen in other species and cellular contexts, where the imaging work was done with projection EM imaging. Presumably, what is new here is the usage of electron tomography. It would help readers if the authors explained why the electron tomography done here was essential to arrive at key conclusions.

Thank you for the useful comment. We have added the explanation of why electron tomography was critical to visualise small tether structures to the last paragraph of the Discussion on page 7.

p.3 mitochondria appeared to be fixed properly ... (e.g., Figs. 1C, 2B) - I don't see any mitochondria in any figures. Perhaps this observation should be noted as "not shown"?

We thank the reviewer for pointing this out. We have added an electron micrograph of mitochondria to the Supplementary Figure 1.

p.3 The images shown in Figs. 1, 2, 4 - The figures should be called out in the order; in this case, Fig 3 has not been called out yet.

We have corrected the order of the figures.

p.4 we did not find any other connecting structures - Because the sample was processed by traditional EM methods, it's safer to add a caveat that other connecting structures could be missed if they were disrupted by sample prep or if they did not pick up stain as well as the two structures presented in this paper.

We have clarified that our sample was chemically fixed in the first paragraph of the Discussion on page 4. Because the details of how our samples were prepared are described in the Method section, we did not add further details to this paragraph.

p.7 we expect such structures to be commonly seen in other cell types as well if they are examined carefully - Instead of saying that examinations should be done "carefully", it would be more helpful to specify how other cell types should be examined. This work shows that the bridges can be found if the cells are either sectioned parallel to the spindle axis or if a sufficiently large volume is sampled.

We have now clarified that 3D electron microscopy techniques such as electron tomography are critical to visualise small tether structures in the last paragraph of the Discussion on page 7.

Please use consistent spelling/hyphenation of ultrafine/ultra-fine and word choice (strands vs. bridges).

Referees cross-commenting<br /> I agree with my co-reviewers's comments and have no further suggestions._

Reviewer #2 (Significance):

This may be the first use of electron tomography to study the structural details of tethers that connect chromosomes in anaphase cells. The data is of sufficient quality to reveal differences in density. Namely, one class of tether appears to be an extension of the chromosome while the other class is composed of thin filaments. This study is novel in that it characterizes a mitosis-associated complex that is poorly studied compared to the microtubule-based spindle apparatus and the kinetochore. Hopefully, the tethers will draw more attention and further characterization by methods like super-resolution microscopy and cryo-electron microscopy. My expertise is in chromatin, mitotic machines, and cryo-electron tomography.

We thank the reviewer for appreciating the novelty and the impact of our work.

Reviewer #3 (Evidence, reproducibility and clarity):

Summary:

Tethers between telomeres of chromosomes in anaphase were inferred from earlier studies of laser microbeam cutting experiments. The current paper presents images from electron tomography of crane fly spermatocytes that substantiates the earlier inference. The authors deduce that the darker filaments and the lighter filaments that they visualize may be the structural tethers at telomeres.

Major comments:

The experiments are carefully done, and the conclusions are appropriately worded to qualify any caveats. This short communication is well-presented, and I have only a few comments._

We thank the reviewer for appreciating the clarity and quality of our work.

The authors should expand their list of references on bridges to include those listed by Warecki et al (Curr Biol 33:1-17, 2023; refs 15-26, etc).

We do not think it is necessary to expand the list of references for ultra-fine DNA bridges. In the article we submitted, we discussed the Warecki at al article in the penultimate paragraph of the Discussion; we concluded that the bridges that Warecki at al described are different from ours in having so few per cell that they couldn’t be tethers, and further that there was no evidence that those bridges were elastic. For those reasons, we do not find discussion of those proteins relevant to tethers, any more than would listing all the proteins associated with ultra-fine DNA bridges be relevant to the elastic tethers.

In the Discussion, we discussed data suggesting that a known elastic protein titin was present; that is as far as we wanted to go on speculation of what the elastic component of tethers might be.

The authors present arguments that the tethers are not the DNA bridges observed by others. However, they should try to address this experimentally by treatment of their preparations with DNase to see if the thick and/or thin filaments disappear.

While we agree that it would be important to identify the components of the tethers, we are concerned that those experiments are beyond the scope of this manuscript. Nevertheless, we appreciate the constructive suggestion for the future research direction.

Moreover, they should discuss in more detail the possible functions of (DNA) bridges, including the recent model from Bill Sullivan's lab (Warecki et al, Curr Biol, 2023) that they help to retain fragments of broken chromosomes. In addition, the authors should summarize the various proteins that may be associated with the bridges (as enumerated in the Warecki et al 2023 paper).

As we describe above, we concluded that the bridges Warecki at al described are different from the tethers that we report in our manuscript. Therefore, we do not think it is necessary to expand the discussion on the proteins and functions associated with ultra-fine DNA.

The authors could add a sentence to the Results or Discussion of whether the thicker tethers might become stretched as anaphase progresses to become the thinner tethers (Fig. 4G).

We thank the reviewer for this suggestion. We actually mentioned this possibility in the third paragraph of our Discussion on page 7.

The authors may want to add a few sentences to the Discussion about the "chromosomal bouquet" stage of leptotene of meiosis prophase I where the telomeres of chromosomes seem pulled together and associate with the nuclear envelope --- they could speculate if this might also be due to the tethers that they describe in spermatocytes.

This is a very interesting possibility. While we would refrain from adding this speculation to our manuscript as it is beyond the scope of the main points, it is certainly an interesting avenue of future research.

Minor comments:

A few additional comments are as follows:

p. 2 last sentence of first paragraph -modify the wording about "no structural evidence that identifies physical connections between separating telomeres", since there is some information from genetic and cell biology light microscopy experiments. Perhaps simply change "structural" to "ultrastructural".

We have changed the wording as the reviewer recommended

p. 6, 5th line of second paragraph - change "ribosome DNA" to "ribosomal DNA"

We have corrected it.

Figure 1D - add the chromosome to the right of the schematic model (as suggested by Fig. 1B).

We are sorry for the confusion. In Figure 1D, the left half of the tethers are 3D modelled and shown. We have clarified this point by modifying the legend of Figure 1D

p. 17 (Methods), line 10 of first paragraph - state if this is light or heavy Halocarbon oil (give details).

It is a mixture of heavy and light Halocarbon oil. We have clarified it on page 17.

p. 17 (Methods), line 12 of first paragraph- state the concentration for fibrinogen and for thrombin.

As we wrote in the original manuscript, the procedures are described in detail in our previous publication (Forer A. & Pickett-Heaps J. (2005) Fibrin clots keep non-adhering living cells in place on glass for perfusion or fixation. Cell Biology International 29: 721–730). Nonetheless, to clarify this point, we have modified the text on page 17.

p. 17 (Methods), line 4 of second paragraph - is there any data to show that the filaments (tethers) occur if there is no cold shock?

Yes, we do see similar filamentous structures in the sample without cold shock. For your information, we show one of the electron micrographs below. In our manuscript, we show the data from the samples prepared with cold shock, because it better visualizes the filamentous structures. We now show these electron micrographs in the Supplementary Figure 2.

Referees cross-commenting<br /> I concur with Reviewers #1 and #2 that this is a fine paper that should be published. My detailed comments submitted with my review are simply meant as revisions to further strengthen this paper.

We thank the reviewer for supporting the publication of our manuscript.

Reviewer #3 (Significance):

Strengths: This is an important conceptual advance and the carefully done ultrastructural imaging provides the foundation for future studies that could delve into the molecular composition and functional significance of the tethers at telomeres of anaphase chromosomes seen here by 3D electron microscopy.

Limitations: the molecular composition and functional roles are not yet known for the tethers seen here by 3D electron microscopy, but to do so would involve an entire new program of experimentation.

Advances: there have only been two earlier ultrastructural papers on tethers at telomeres, and the tethers were peripheral to the main focus of those papers. Thus, the current paper extends our ultrastructural information about tethers.

Audience: this work is of importance for scientists who study the mechanics of chromosome movement on spindles, including regulation to combat aneuploidy. This work will also be important for a broader audience to inform them about transmission of the hereditary information to daughter cells._

We thank the reviewer for appreciating the significance and the impact of our work.

With 126 published definitions of "communication," touching on other forms of communication other than merely speaking in a speech class is vital. With humans having some of the widest range of speech (i.e. various languages) that often times are not seamless, other universal abstract symbolism in conjunction with spoken communication is necessary to bridge the gap. Even our written language and assigned meaning to certain methodic squiggles displayed on paper varies widely, as well as other less obvious ways of communicating like gestures and body languages that could seem inconsequential to one may be monumentally offensive to others, the intricate woven methods to communicate within the complexities we as a human species have created is a fascinating study beyond merely standing in front of a group of peers and talking at them for 3-5 minutes about a chosen topic.

Everyone on this planet has intrapersonal communication. I talk to myself every day, and I have conversations with myself on what I'm going to do or what I need to do. Some people talk to themselves to calm down, or they journal to ease their minds. When something surprising happens people usually react somehow in their head, basically when anything happens people react to themselves. Just as the text states, "We also use intrapersonal communication or “self-talk” to let off steam, process emotions, think through something, or rehearse what we plan to say or do in the future." Intrapersonal communication happens almost every second throughout one person's day.

Making things messy from the start as advice for getting started.

I've seen this before in other settings, particularly in starting new notebooks. Some have suggested scrawling on the first page to get over the idea of perfection in a virgin notebook. I also think I've seen Ton Ziijlstra mention that his dad would ding every new car to get over the new feeling and fear of damaging it. Get the damage out of the way so you can just move on.

The fact that a notebook is damaged, messy, or used for the smallest things may be one of the benefits of a wastebook. It averts the internal need some may find for perfection in their nice notebooks or work materials.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

The manuscript by Rigger and Brenner details the role of vimentin network, in advancing OA pathogenesis by exacerbating premature senescence. The data is well presented and the study of interest, in that there is little known about vimentin in cartilage biology.<br /> The authors used OA derived cartilage explants and chondrocytes cultures, were graded for severity and compared accordingly. Figure 1 shows that markers of senescence are increased with structural damage, which is well established and consistant with the literature. Using a DOX model the authors induce premature senescence and exhibit a disrupted vimentin network. However, upon KD of CDKN2A, a marker of senescence, but did not observe complete reversal of CSV presentation.<br /> Next the authors show in figure 4 and 5, that the reduction or dismemberment of vimentin structures are linked to senescence and may act as contributing factors.<br /> Figures 6 and 7 then go on to show that upon advanced passage chondrocytes lose their vimentin network, and tend to senesce and mineralize.

Reviewer #1 (Significance):

Strength:<br /> This is a very novel study showing a link between vimentin and senescence in chondrocytes. The data are in line with other data. The work is clearly written structured and well displayed.

Author´s response:<br /> We thank reviewer #1 for their interest in our work and their overall positive report.

Suggestions for improvement:

While the study is very thorough ought in describing the markers of senescence and vimentin network, it lacks insight regarding mechanism which isn't completely deciphered. Are there links to key transcription factors?

Author´s response:<br /> The transcriptional regulation of vimentin in human cells is very complex. The VIM promoter region comprises multiple elements, such as a NF-kB- binding site, a PEA3-binding site and two AP1-binding sites (Zhang et al., 2003). Moreover, it was recently demonstrated that redox signaling is involved in vimentin expression at the wound margin after tissue injury in zebra fish (LeBert et al., 2018). However, it has also been reported that IL-1ß stimulation results in reduced gene expression of vimentin via p38-signalling in cartilage degeneration and OA progression (see manuscript REF. 36,37).

In our study, we observed that enhanced CSV levels are associated with a decreased vimentin gene expression, indicating a lower stability of the mRNA or decreased transcription of VIM in senescent chondrocytes (maybe due to enhanced p38-signalling as mentioned above). Since the transcriptome in senescent cells is radically changed, this question cannot be answered easily.

In future studies, we will rather try to clarify the underlying mechanism of vimentin externalization. There are still many questions to be answered: is the CSV anchored in the cell membrane (which anchor protein?) and is there still a connection to the intracellular vimentin network? Which proteins are involved in the externalization process: maybe comparable to phosphatidylserine exposure, mediated by flippases, scramblases, and lipid transfer proteins or rather by vesicles?

Literature mentioned above (not included in manuscript):

LeBert et al., 2018: Damage-induced reactive oxygen species regulate vimentin and dynamic collagen-based projections to mediate wound repair. DOI: 10.7554/eLife.30703

Zhang et al., 2003: ZBP-89 represses vimentin gene transcription by interacting with the transcriptional activator, Sp1. DOI: 10.1093/nar/gkg380

It is also unclear if disruption of the network is more detrimental than KD in promoting senescence.

Author´s response:<br /> KD of Vimentin led to a gradually decrease of intracellular Vimentin content and consequent stress. The cells were analyzed 7 days after induction of the KD and exhibited a stable senescent phenotype, comparable to Doxorubicin-treated chondrocytes (treated with very low concentrations over several days to produce only mild but ongoing stress). These models might reflect the pathophysiologic situation: We think that cellular stress due to mechanical impact and subsequent oxidative stress/ low-grade inflammation might lead to a gradual disruption or re-organization of the vimentin network, which is accompanied by decreased vimentin gene expression.

In case of the disruption of the vimentin network by Simvastatin, the stress response was very intense and rapid (24 h), and was only conducted as a proof-of-principle experiment. Despite the upregulation of some senescence-associated markers, we don`t think that permanent Simvastatin treatment would be suitable to obtain a stable senescent phenotype, but rather expect the cells to die due to excessive stress.

It would have been good to include models OA murine models to understand these processes better, and make a stronger physiological connection with OA of the joint.

Author´s response:<br /> The CSV antibody is only suitable for human cells and cannot be used for immunohistochemistry. Therefore, all previous reports of CSV are based on human (isolated) cells. At the current time point, it would not be possible to stain CSV in joints of mice after induction of PTOA due to the methodological limitations. We actually tested the CSV-antibody in isolated lapine chondrocytes and found a high percentage of CSV-positive cells, even at low passages. Although stress increased the amount of CSV-positive lapine cells, we did not consider the results as reliable due to the high percentage in un-stressed cells, which might result from unspecific antibody binding.

Overall, we think that the usage of clinical OA samples is convincing and reflect the pathophysiologic situation in the human OA joint.

Reviewer #2 (Evidence, reproducibility and clarity):

The manuscript provides solid evidence for an association between cell surface vimentin (CSV) and chondrocyte senescence. Human cartilage and cultured chondrocytes are used with a wide range of approaches to provoke senescence: natural osteoarthritis, traumatic loading ex vivo, doxorubicin to cells in monolayer, vimentin siRNA, and simvastatin. In contrast, relatively little was done to try and interrupt or reverse the role of CSV in senescence, with CDKN2A siRNA representing one attempted intervention. The manuscript is well written and the data are presented in a logical and clear manner, with a high likelihood of being reproduced in subsequent studies.

Author´s response:<br /> We thank reviewer #2 for their interest in our work and their mainly positive report.<br /> Regarding their comment on our attempts to reverse CSV on senescent chondrocytes, we would like to add the following: Reversal of cellular senescence is a very ambitious challenge. But in fact, we are currently preparing a manuscript in which we characterize an appropriate senolytic strategy to “rejuvenate” human chondrocytes and plan to use this approach to reduce the amount of senescent and thus CSV-positive cells in future experiments.

_Major comments:

In the doxorubicin experiments, the senescent cells show a spread morphology as expected. Given the importance of vimentin in cell spreading (as the authors own data show), the possibility that spread morphology itself (and not senescence) leads to CSV should probably be examined. This could perhaps be achieved by plating with different concentrations of fibronectin or other matrix proteins that produce a spread morphology to a degree that matches the doxo. If the cells remain spread for ~10 days but don't become senescent and don't have CSV, this would provide further support for a direct relationship.

Author´s response:<br /> We agree that cell spreading is associated with various cellular processes (for example by the YAP signaling pathway). Moreover, we would like to thank the reviewer for the proposed experiment.

Seeding of cartilage cells on fibronectin coated plates is a commonly used procedure to isolate chondrogenic stem progenitor cells, due to their higher affinity to fibronectin. The cells are usually cultured for several days on the coated plates and do not exhibit a flattened, senescent-like phenotype (as we observe for Doxorubicin-treated cells), but an elongated, fibroblast-/ stem cell-like shape. Our results (Figure 6E) demonstrate that CSPC have no increased CSV levels, despite their elongated (not flat) morphology.

There are some findings supporting the assumption that CSV leads to enhanced cell adhesion, but not that adhesion or cell spreading promotes CSV: we included experiments with HeLa (low CSV levels) and SaOS-2 (high CSV levels), which demonstrated that high CSV levels are associated with increased plastic adhesion (Figure S5). In line with this, we demonstrated that higher CSV levels on chondrocytes were associated with enhanced fibronectin and vitronectin binding, which might explain increased plastic adhesion. Moreover, Simvastatin stimulation and subsequent cellular stress by Vimentin disruption resulted in enhanced CSV but did not lead to cell spreading (Actin not affected, cells rather elongated, not flattened).

Minor comments:

The CSV antibody and staining method appeared to have generated some signal from debris, which makes it challenging to assess the localization of true staining. Presumably the true staining would be present only on the cell surface. While the widefiled view is appreciated, perhaps insets with a higher magnification would clarify.

Author´s response:<br /> In Figure 2h and Figure 2i, we provide insets of the IF-staining and an exemplary image made by scanning electron microscopy (SEM). CSV is not localized on debris – Figure 2h, actually represents the cell surface. The magnified, Doxo-treated cell is highly senescent and thus flattened. The uneven (rather spotted) staining pattern of CSV and the unusual shape of the cell might suggest that this is debris, not the cell membrane.

For figure 1k, it is a bit surprising that CDKN2A would peak so early after injury and then drop off. Most studies in other systems show a gradual increase in CDKN2A levels with persistent stress as opposed to a rapid increase in response to acute stress. Could the drop-off be due to preferential death of these cells? The CSV % in 1m was taken from 7d after trauma (plus 7 days in monolayer it appears). Further discussion on the timing of traditional senescence markers as compared to the emergence of CSV would be useful.

Author´s response:

We would like to thank the reviewer for this comment. That CDKN1A was induced by mechanical trauma without significant decrease at the later time points was in line with the P53 expression, which we detected via immunohistochemistry (IHC; positive staining of chondrocyte nuclei in cartilage). P53 and P21 are regarded as interconnected senescence markers. Interestingly, P53 is not regulated on gene expression level upon cartilage trauma or Doxorubicine stimulation – but there is a significant increase in P53 nuclear translocation.

Although such a discrepancy between gene expression and protein activity has not been reported in case of P16 or P21, we plan to investigate the dynamics of these cell cycle regulators and its connection to CSV after cartilage trauma in more detail in future studies.

We included the following statement in the discussion part:

“In the current study, we observed that CSV on chondrocytes was reduced by siRNA-mediated silencing of CDKN2A and increased after Doxo treatment or cartilage trauma. While we confirmed that mRNA levels of both CDKN1A and CDKN2A were significantly enhanced upon injury but exhibited different expression levels over time, we determined CSV-positive cells only at one time point after ex vivo cartilage trauma. Future studies might also consider earlier and later time points after cartilage injury to identify a potential time-dependent peak or decline in CSV-positive chondrocytes. In this way a potential association between CSV and the expression levels of CDKN1A and CDKN2A, which are thought to play differential roles in initiating and maintenance of senescence, respectively [50], might be clarified.”

[50] Stein G, Drullinger L, Soulard A, and Dulić V. Differential Roles for Cyclin-Dependent Kinase Inhibitors p21 and p16 in the Mechanisms of Senescence and Differentiation in Human Fibroblasts. Mol Cell Biol. 1999;19(3): 2109–2117. https://doi.org/10.1128/mcb.19.3.2109.

There is no CSV staining shown for figures 4 and 5. While the quantification of CSV was done by flow cytometry, it would nice confirmation to see the increase in CSV on the surface of cells with either siRNA for vimentin or the simvastatin.

Author´s response:

CSV-IF of simvastatin-treated chondrocytes is provided in Figure 5 (b). We did not perform exemplary staining of CSV after VIM-KD, because the quantification was performed via flow cytometry.

Reviewer #2 (Significance):

The strengths of the study include a rigorous design and the establishment of a potential new cell surface marker of chondrocyte senescence. The main limitation is that the conclusions are largely descriptive in nature.

If CSV is confirmed as a robust marker of senescence, this would be of value to the field. While this marker has been explored previously in other systems, there is value in this manuscript given the wide range of contexts investigated for a cell type in which senescence likely has an important role.

Reviewer #3 (Evidence, reproducibility and clarity):

This study presents a sound piece of science in the puzzle about extracellular vimentin in the differentiation/dedifferentiation of human chondrocytes and senescence and osteoarthritis. Eventhough, no mechanism is elucidated, the results clearly point towards a correlation of the amount of extra cellular vimentin and the level of chondrocyte senescence, and therefore signs of osteoarthritic changes in the cultivated chondrocytes. The methods applied are state-of-the art and provide the means to generate meaningful results in this experimental setting. The paper is concise and clearly written, there are only minor remarks.

Author´s response:

We thank reviewer #3 for their interest in our work and their overall positive report.

Minor comments:

1. The main clue of the paper is extra cellular vinemtin around chondrites in culture, please provide better pictures (1g) to support this. Why is the extra cellular staining seen so broad and not concentrated on the cells surface? The picture chosen imply a huge amount of vimentin to be externilized in disease states. It also indicates that in diseased chondrocytes no intact or semi-intact vimentin network is found intracellular. Please comment.

Author´s response:

In Figure 1g, CSV is located on the cell membrane. The pattern of the staining was surprising to us, as well. CSV was not equally distributed on the membrane, but rather represented an inconsistent pattern. Sometimes the staining was located at the filopodia of the cells, sometimes the whole cell was covered by spots. We also observed this on cancer cells, which was in line with other studies using this antibody. It remains unclear whether the distribution of the CSV has any effect. But we assume that the high abundance in filopodia might be connected with cell adhesion and mobility, which was positively associated with CSV.

Yes, chondrocytes isolated from highly degenerated tissue exhibited higher CSV levels as compared to cells derived from macroscopically intact regions. Although we did not investigate the vimentin network of these cells, our observations in Doxo-treated cells imply, indeed, that intracellular vimentin might be altered in diseased chondrocytes. According to this, Blain et al (Ref. 13) reported that there is a disassembly of the intracellular vimentin network in OA chondrocytes, which can disturb the chondrocyte phenotype and contributes to the development of OA (see discussion).

1. In the doxo experiment no extracellular vimentin is found? Please explain.

Author´s response:

Doxo-treated cells are highly positive for CSV (= extracellular vimentin on membrane). However, the intracellular vimentin is strongly decreased and some cells seem to be negative. We have not clarified the underlying mechanism by now, but it seems that senescence/ disease progression negatively affects the transcription of vimentin and, at the same time, promotes the externalization of the existing intracellular vimentin. Altogether, this might result in a decline in intracellular vimentin.

1. The SEM picture is showing what. IGH? The red dots are colloidal gold particles? In any case the quantity of stain gathered EM level would not correlate to the huge amount seen in LM staining. Please comment.

Author´s response:

For the SEM analysis, a gold particle-coated secondary antibody was used. The positive signal usually appears in white and was subsequently colored via a software. In IF and ICC staining, we had a signal amplification due to the biotin-streptavidin system and the magnification makes, of course, a huge difference.

1. Why the ICC in Fig. 3c? The siRNA is not detected in the KD? A reduction of Vimentin could be shown via WB.

Author´s response:

In Figure 3c, the KD of P16 was confirmed on protein level. In addition to the gene expression analysis, we chose the ICC (IF) to confirm that there is a decline in active (nuclear) CDKN2A. In case of P53, we made the experience that gene expression and the amount of cytoplasmic/ nuclear protein might not be consistent.

In Figure 4, we confirmed the successful KD of vimentin on mRNA and protein level (flow cytometry plus IF). Of course, WB would also be possible, but we decided to use the methods in which the antibody was well established and we wanted to visualize the disturbance of the intracellular vimentin network upon KD.

1. Fig. 4c, why are there no remnants of the vimentin networks seen in the chondrocytes? A Knock-down, not a KO is shown.

Author´s response:

In fact, most of the intracellular vimentin seems to be gone. However, there are some remnants (condensed fibers/ bundles) of the former vimentin network. We applied the VIM-KD over seven days. Usually, a KD experiment is only conducted for 2-3 days. But since we were not sure how stable the vimentin protein would be, we chose seven days. This long-lasting KD might have resulted in a strong decline of the protein. Moreover, the CSV levels on these cells were very high, indicating that existing vimentin was externalized and additionally decreased the amount of intracellular vimentin.

1. Please comment of the concentration of simvastatin, why not nmolar?

Author´s response:

The concentration of Simvastatin was chosen in accordance with Trogden et al. (Ref. 26), who first described the effects of simvastatin on the vimentin network. A lower concentration might have had the advantage, that the effects were less severe, allowing a longer observation time than 24h. However, as a proof-of-principle model to demonstrate the connection between vimentin network collapse ant CSV expression, the concentration worked quite well.

1. CSV+ is misleading in Fig. 6g, it's not an over expression.

Author´s response:

We would like to thank the reviewer for this comment and removed the “+” to make it less misleading.

1. The concept of EMT is debatable, at least in kidney fibrosis, and chondrocytes are not epithelial cells. Please add a more critical discussion point.

Author´s response: The authors agree with the reviewer’s argument that chondrocytes are no epithelial cells ant that the term EMT doesn’t seem to be appropriate. However, this is one leading hypothesis proposed by the working group of Prof. Mayán, who described CX43 and other EMT-markers on/ in senescent chondrocytes (see reference 31; more recently: Cell Death Dis. 2022;13(8):681. doi: 10.1038/s41419-022-05089-w).

We added the following passage in the discussion part to indicate that this hypothesis is a controversial concept:

“Nevertheless, the hypothesis that chondrocytes might undergo an EMT-like process remains controversially discussed, because chondrocytes are mesenchymal and not epithelial cells. In a recent review, Gems and Kern propose to consider senescent chondrocytes as activated and hyperfunctional remodeling cells occurring during OA progression [49]. Accordingly, chondrosenescence might represent an unsuccessful attempt of tissue repair. They further suppose that the senescent or activated chondrocytes are associated with a hypertrophic, bone-forming phenotype, following the process of bone development rather than hyaline cartilage formation. In line with this, we observed that CSV was associated with enhanced osteogenic capacities and a decline in chondrogenic properties.”

[49] Gems and Kern, 2022): Geroscience. 2022;44(5):2461-2469. doi: 10.1007/s11357-022-00652-x.

Reviewer #3 (Significance):

The manuscript provides novel insight in the role of intermediary filaments, i.e. vimentin, on chondrocyte senescence and osteoarthritic changes in vitro. It's strength is a thorough elucidation of the connection with a wealth of experimental data, a weakness is the missing elucidation, or first experiments in the direction, of the cell biological mechanism.<br /> It is well suited for a broad audience, because it deals with fundamental cell biological phenomena, definitely it's important for the OA /chondrocyte biology community.

Author Response

The following is the authors’ response to the original reviews.

Reviewer #1 (Recommendations For The Authors):

• Introduction: "In plants, IP7 and IP8 decrease upon Pi starvation and mutants lacking either of the enzymes necessary for their synthesis induce the phosphate starvation response, leaving it open whether IP7 , IP8 , or both, control the Pi starvation program (10, 13, 14)" In plants, ITPK enzymes catalyze the formation of 5-InsP7 from InP6 https://pubmed.ncbi.nlm.nih.gov/34274522/ and VIH enzymes the formation of 1,5-InsP8 from 5-InsP7 https://pubmed.ncbi.nlm.nih.gov/25901085/ . Loss-of-function of both ITPK1 https://pubmed.ncbi.nlm.nih.gov/34274522/ and of VIH1/2 https://pubmed.ncbi.nlm.nih.gov/31436531/ https://pubmed.ncbi.nlm.nih.gov/31419530/ results in constitutive Pi starvation responses. vih1/vih2 mutants lack 1,5-InsP8 and hyperaccumulate 5-InsP7 https://pubmed.ncbi.nlm.nih.gov/31436531/ and act directly upstream of the transcription factors PHR1/PHL1 https://pubmed.ncbi.nlm.nih.gov/31436531/ https://pubmed.ncbi.nlm.nih.gov/33452263/ https://pubmed.ncbi.nlm.nih.gov/34857773/ , the case for 1,5-InsP8 is settled in plants. I would suggest revising this statement and citing all relevant literature.

We don't see the case for 1,5-IP8 as settled in plants, and none of the papers mentioned above draws this strong conclusion. This may be due to several limitations in the available data. The mentioned studies do not allow to differentiate the effects of 1-IP7 and 1,5-IP8 and, where binding or competition experiments have been performed, e.g. on the transcription factors, the differences in the Kd values for IP7 and IP8 were minor. Furthermore,1,5-IP8 levels and Pi starvation response do not always correlate. IPTK1 mutants, for example, show Pi overaccumulation, and low 5-IP7, but normal 1,5-IP8 (Riemer et al., 2021). Finally, plants are complex organisms with multiple tissue types that serve for accumulating, exporting, transporting or finally consuming Pi. Therefore, correlating inositol pyrophosphate levels from whole-plant extracts with a Pi starvation response is problematic, except if these data could both be obtained from the same cell types or at least tissues.

The comment of the reviewer made us recognize that the complex situation in plants deserves a more detailed coverage and we have therefore adjusted the introduction accordingly.

Results: "We determined the corresponding lysines in Pho81 (Fig. S3), created a point mutation in the genomic PHO81 locus that substitutes one of them, K154, by alanine, and investigated the impact on the PHO pathway."

In my opinion, it would be important to test here in a quantitative in vitro binding assay if (i) the SPX domain of Pho81 can bind PP-InsPs including 1,5-InsP8, (ii) if the dissociation constant is in agreement with the cellular levels of 1,5-InsP8 in yeast (compare Fig. 2) and (iii) if the K154A mutation blocks or reduces the binding of 1,5-InsP8. Without such experimentation, I find the statement "this result underlines the efficiency of the K154A substitution in preventing PP-IP binding to the Pho81 SPX domain." to be overly speculative, as no binding experiment has been conducted.

We agree with the comment of the reviewer concerning the overstatement in the phrase. It has been deleted.

As mentioned already in our previous work (Wild et al., 2016), Pho81SPX counts among the SPX domains that we could not express recombinantly. Likewise, full-length Pho81, which would be the relevant object for correlating in vitro binding studies with the cellular concentrations, has not been accessible. Expression in yeast did not provide sufficient material for ITC or other quantitative techniques. Therefore, we refrained from pursuing binding studies. Nevertheless, given the high conservation of the positively charged patch on SPX domains and the fact that, in every case where it has been tested so far, SPX domains showed inositol polyphosphate binding activity, we find it a conservative assumption that the Pho81SPX binds them as well. This is supported by the effects of the binding site mutant, which mimics the effect of ablating IP8 synthesis.

Results: "Inositol pyrophosphate binding to the SPX domain labilizes the Pho81-Pho80 interaction." Again, in the absence of any protein - protein interaction assay I find this statement not to be supported by the experiments outlined in the manuscript. The best way to address this point would be to perform either co-IP or in vitro pull-down experiments between Pho81-SPX and Pho81-85, in the pre- and absence of 1,5-InsP8 and/or using the Pho81 point-mutants described in the text.

Since Pho81 could not be produced recombinantly, neither by us nor by others who worked on this protein previously, quantitative in vitro binding assays are not accessible for now. A simple IP suffers from the problem that Pho81 interacts with Pho85-Pho80 not only through the SPX domain but also through the minimum domain. The latter interaction may be constitutive. Since the main point of the manuscript is not to dissect the exact mechanisms of Pho85-Pho80 regulations, but only to address the point why the postulated inactivation of this kinase by an 1-IP7/minimum domain complex makes no sense, we prefer not to show a profound (and more complex) analysis of how the different Pho81 domains contribute to binding.

To test the potential of the SPX domain for binding Pho85/Pho80 in vivo, we have created a GFP-fusion of the SPX domain of Pho81. This fusion protein localizes mainly to the cytosol when cells are on high-Pi. Upon Pi starvation, it concentrates in the nucleus. This concentration is not observed in pho80 mutant background (New Fig. S7).

In line with this, I would suggest to move the molecular modelling/docking studies from the discussion into the results section and to use these models to design some interface mutations that could be tested in coIP and/or pull-down assays. Alternatively, the authors may choose to omit the discussion section starting with: "Even though the minimum domain is unlikely to function as a receptor for PP-IPs this does not ... and ending with . In sum, multiple lines of evidence support the view that the SPX domain exerts dominant, 1,5-IP8 mediated control over Pho81 activity in response to Pi availability."

We have now moved the modelling data to the Results section. The structure prediction of the interface is experimentally validated. Data on the effect of interface substitutions are already published, although these substitutions had not been recognized as affecting a common interface at the time. Substituting the interface residues either on the side of Pho80 or of Pho81 constitutively activates Pho85-Pho80 kinase and destabilizes its interaction with Pho81. This was shown by Co-IP experiments from cell extracts by Huang et al. We mention the respective substitutions in the manuscript and cite the paper in which their effect on PHO pathway activation had been described.

Reviewer #2 (Recommendations For The Authors):

Some points need additional attention by the authors:

• In general, it would be helpful to introduce abbreviations more thoroughly (certain enzyme names, PA, MD, ...)

We paid more attention to this.

• Also in general, the authors may want to think about the nomenclature of inositol pyrophosphates. Given the expansion of PP-IPs that are being detected in different organisms these days it may be a good time to convert to a more precise nomenclature, i.e. 5PP-IP5 instead of 5-IP7; and 1,5(PP)2-IP4, instead of 1,5-IP8. The latter could just be stated once, and then be abbreviated as IP8.

To our understanding the field has not yet come up with a unified nomenclature. Therefore, we prefer to stick with the more practical nomenclature that we have chosen, which also corresponds to what is commonly used in presentations and discussions among colleagues. We have now introduced a sentence making the link to the nomenclature that the reviewer has proposed.

• p. 1, Abstract: "negative bioenergetic impacts" - the phrasing seems really vague

Agreed, but we find it difficult to be more explicit and precise in the abstract while remaining concise and not distracting from the main message. This aspect is better explained in the introduction.

• p. 3, Significance statement: "... unified model across all eukaryotic kingdoms" While the intended meaning of this wording is better explained in the text later, the phrasing here suggests a more all-encompassing study at hand, instead of a conclusion that fits more closely with established reports from other organisms. Please rephrase.

We have adapted the phrase to avoid this impression.

• p. 4: "IPTKs" - are the ITPKs meant here?

Yes, that was a typo.

• p. 7, the introduction ends abruptly and could use a concluding sentence.

Done

• p.7, "enzymes diphosphorylation either the..."; I understand what the authors are trying to say with diphosphorylating, but the enzymes are phosphorylating a phosphorylated substrate.

Yes. We changed the phrase to "....adding phosphate groups at the 1- or 5-positions....".

• p. 7, subtitle "...concentrations and kinetics of..."; kinetics of what? Synthesis/turnover?

We corrected this subtitle

• p. 8, with regards to the recovery experiment: Was this recovery determined elsewhere (please cite)? Otherwise it would be beneficial to include an extra figure to illustrate these recoveries in the supplementary information. And do the authors suspect some hydrolysis of IP8 given the lower recovery?

We have now added the experiment testing recovery of IPPs as the new Fig. S1.

• p. 9: It is appreciated that the authors point out the concentration of IP6 in S. cerevisiae. I found that concentration rather low, and the authors could highlight this a bit more, given their ability to carry our absolute quantification.

This was a leftover from a previous version of the paper. Since the paper does not treat IP6 or lower inositol polyphosphates, we have deleted this phrase.

• p. 9, Fig 2: The exponential decay of 5-IP7 is very nicely shown in Figure 2c. But one of the most important discussion points is IP8 being the key controller of the PHO pathway - it would therefore be beneficial for the argument to also show the same kind of graph for IP8 and if possible, fit a function to the data points to better quantify and compare the decay processes (e.g. via "half-life time" of PP-IPs during starvation, in addition to the suggested "critical concentration" which was only discussed for 5-IP7 thus far).

Kinetic resolution is an issue here. The approach shown in Figs. 2 and 5 is not apt to determine a critical concentration of IP8 because the decline upon transfer to starvation conditions is too fast and difficult to relate to the equally rapid induction of the PHO pathway. We shall address this point in a more appropriate setup in a future study.

• p.9, Fig 2a: Where does the 5-IP7 come from in the kcs1Δ strain? In the text the authors state that 5-IP7 in kcs1Δ was not detected, but the figure suggests otherwise. Please explain.

Currently, we do not know where these residual signals stem from. One possibility is that they represent other isomers that exist in minor concentrations and that are not resolved from 5-IP7 in CE. We added a sentence to the figure legend to indicate this.

• p. 10: "IP8 was undetectable in kcs1Δ and decreased by 75% in vip1Δ. kcs1Δ mutants also showed a 2 to 3-fold decrease in 1-IP7, suggesting that the synthesisof 1-IP7 depends on 5-IP7. This might be explained by assuming that a significant source of 1-IP7 is synthesis of 1,5-IP8 through successive action of Kcs1 and Vip1, followed by dephosphorylation to 1-IP7." - Please specify this statement. Do the authors mean that 1,5-IP8 is only produced transiently below the detection capabilities of the method but that there still is a (reduced) flux from 5-IP7 to 1,5-IP8 to 1-IP7? Otherwise it would seem paradoxical to have a dependency on a non-existing metabolite in that cell line.

This was not clearly expressed. The revised version now says: " ... a 2 to 3-fold decrease in 1-IP7, suggesting that the synthesis of 1-IP7 depends on 5-IP7. This might be explained by assuming that, in the wildtype, most 1-IP7 stems from the conversion of 5-IP7 to 1,5-IP8, followed by dephosphorylation of 1,5-IP8 to 1-IP7.". We hope that this clarifies the matter.

• p. 10: "pulse-labeling approaches are not available for PP-IPs." While this statement is correct, a recent paper co-authored by Qui and Jessen showed nice pulse-labeling data for the lower Ips and could be cited here (PMID: 36589890)

Yes, indeed, we should have been more precise here. What we wanted to express was that rapid pulse-labeling methods for following phosphate group turnover were lacking, with a temporal resolution of minutes rather than hours. Existing pulse labeling approaches, including the study mentioned by the reviewer, do not provide that. We have changed the phrase accordingly.

• p. 10: continuation of caption of Fig 2: "were extracted [and] analyzed"

Corrected. Thank you.

• p. 12: How is 1-IP7 made in the vip1 kcs1 double mutant?

As explained above, we suspect that these may be side products of IPMKs, which accumulate in the absence of vip1 phosphatase.

• p. 13, caption to Figure 3: "XXX cells were analyzed" please replace the place holder XXX.

Done. Thank you.

• p. 13, Fig 3B, C, D and p. 50, Fig. S4: On screen the contrast between the different shades of grey of the bars are just visible enough, but not on paper, I suggest using a higher contrast/ different colouring scheme.

We enhanced the contrast.

• p. 24, 25, Fig 7.: I could not really appreciate the AlphaFold part, and found it unnecessary. No docking or molecular dynamics simulations were carried out here, and it was not clear to me what information should be gleaned from this part.

Following this comment, we have modified the respective part of the text. This part refers to a publication from the O'Shea lab (Nat. Chem Biol. 4,25) proposing the model that 1-IP7 and the Pho81 minimum domain bind competitively to the active site of Pho85 to inhibit its kinase activity. Modeling of complexes between Pho81, Pho80 and Pho85, which we present in the manuscript, rather suggests binding of the minimum domain to a groove in Pho80. This is important because it provides a viable alternative model for the action of the minimum domain. It suggests the minimum domain as a constitutive linker that attaches Pho80 to Pho85. Importantly, this model accounts perfectly for the results of previous random mutagenesis studies on Pho80 and on the minimum domain, which had independently identified both the Pho80 groove and the minimum domain residues that bind it in the prediction as critical residues for inhibition of Pho85, and for integrity of the Pho85/Pho80/Pho81 complex. We find this alternative explanation for Pho85-Pho80 regulation by Pho81, which we can derive by combining the predictions with already published experimental data, an important element to re-evaluate the relevance of 1-IP7 in PHO pathway regulation and resolve one of the existing discrepancies.

• p. 28: No experiments were carried out with plants or mammals. The relevance for plants or mammalian systems therefore seems to be overstated at this point in time.

We are not quite sure how to interpret this remark. We do not claim that our data support a role for IP8 in mammals and plants. But we refer to and cite studies providing the strongest evidence in favor of it in these systems. The relevance of our current study relies in refuting seemingly strong evidence from yeast, which had been diametrically opposed to the data obtained in plants and mammals. The revision of the situation in yeast now paves the way to drawing a coherent concept for fungi, plants and mammals. We feel that this is important and should be underlined.

• p. 31: "300 mL of 3% ammonium" - 300 µL?

Yes. Thank you.

• p. 45, CE-ESI-MS parameters: "1IP8"

Corrected.

• p. 47: Figure S1: Please include more experimental details in the caption and/or methods section. Was a similar analysis software used as e.g. Figure S2 (NIS Elements Software)? Please also include all the analysis software in the Methods section under "fluorescence microscopy". Unless these additional experimental details already clarify the following point: Can the authors briefly comment on why the morphological determination in S1 requires trypan blue staining while in later experiments the yeast cells are readily recognized by the software in "simple" brightfield images?

Trypan blue staining is not strictly required for this. It is just a simple method to fluorescently stain the cell wall. There are many other ways of delineating the cells. It could also have been done in a brightfield image.

We updated the figure legend to better describe how these measurements were done and deposited the script and training file on figshare.

• p. 48: "can be downloaded from **" please insert the link once the script is available online.

It has been deposited at Figshare under DOI 10.6084/m9.figshare.c.6700281

Reviewer #3 (Recommendations For The Authors):

1) Italicize the scientific names of the organisms; this was inconsistent throughout the manuscript. Also, gene names should be italicized; this was also inconsistent (e.g., p.12 "... did not induce the PHO84 and PHO5 [sic] promoters...).

Done

2) Summary of the Figure 2A data in the text (p.9) probably has swapped the determined concentrations for 1-IP7 and IP8 (0.3 µM or 0.5 µM) as compared with the data figure.

Yes, indeed. We have corrected this.

3) Figure 2A: which of the mutant PP-IP levels are significantly different from the WT control?

We have now added asterisks to indicate the significance for every mutant.

4) In the discussion on the data (Fig. 2A), I was tripped up by the verb tense in this phrase "5-IP7 has not been detected in the kcs1Δ mutant and 1-IP7 has been strongly reduced..."; I think you want to use the past tense "was" in both cases [as is used in the next sentence]. It made me wonder if there was a difference in the detection of 5-IP7 and IP8 in the kcs1Δ mutant, you could detect 5-IP7 but not IP8; if so, where did the 5-IP7 come from?

We have corrected the tense. Thank you for highlighting this. For the residual inositol pyrophosphate signal in kcs1Δ. We do not know its origin. One possibility, which we now mention in the text, is that it stems from IPMK side activity. It should be underlined that all signals disappear upon PI starvation.

Figure 2C, include the data points that the lines are built from (suggestion).

We refrained from that for the line graphs. For reasons of consistency, we should do this for every line graph. If we did that, Fig. 4B would become quite hard to read.

6) Figure 3B-D, please check that the stipples or hatches are in the figure - the printed copy lacked them although I could see them in the electronic version; this was also true for Figures 5 and 6 (I do not know if it is a printer issue, but other hatches were visible: e.g., not seen in S4 but seen in S5).

They are visible in our copies, also after printing. They may have been lost during file conversion at the journal.

7) The text description of the Pho4-yEGFP, Pho5-yEGFP and Pho84-yEGFP says that the kcs1Δ mutant "showed Pho4-yEGFP constitutively in the nucleus already ... and PHO5 and PHO84 were activated". However, the data is more complex than that: whereas the localization of Pho4-yEGFP is constitutively nuclear, there is a higher basal (repressed) expression of both Pho5 and Pho84 as well as increased expression of both proteins under -Pi conditions. What accounts for the increased expression when Pho4 is already nuclear? This is also seen in the vip1Δ kcs1Δ mutant.

We agree with the reviewer, but we cannot explain this effect with certainty. One possibility could be a wider dysregulation of Pi metabolism in kcs1 mutants. To name a few possibilities: Wildtype cells have polyphosphate reserves that are gradually mobilized during the first hours of P-starvation. kcs1 mutants don't have those and might fall into a "deeper" state of starvation faster. It should be kept in mind that the starvation response is also regulated at the level of chromatin structure, and by antisense transcripts. The influence of kcs1 on these processes is unclear.

8) Figure 9 legend: please add a definition of the MP region (in red) and include it more explicitly in the described model.

We now mention the relevant region also in the legend and have labeled the relevant regions in the images (Huang et al., 2001).

9) Figure S2 legend: information is missing (downloading link).

It has been deposited at Figshare under DOI 10.6084/m9.figshare.c.6700281

10) Figure S4 and S5, missing statistics.

They have been added to the new Fig. S6, which interprets differences between strains and conditions. Fig. S4 (now S3) shows timecourses of IPPs down to zero. Adding statistics for all pairwise differences between the timepoints would be almost an overkill.

Author Response

The following is the authors’ response to the original reviews.

eLife assessment

It is very important to find practical and efficient means in order to increase agricultural productivity. Drawing on data from variable field environments, this study provides a useful theoretical framework to identify new factors that could increase agricultural production. There is solid evidence to support the authors' claims, though following the fate of candidate species after introduction into rice fields would have strengthened the study. Plant biologists and ecologists working in nature and fields will find the work interesting.

Thank you so much for your careful evaluation of our manuscript. We are very pleased to hear that you found our framework useful. We have revised our manuscript according to the "Recommendations for the Authors" to improve our manuscript.

Public Review

Reviewer #1 (Public Review):

This manuscript describes the identification of influential organisms on rice growth and an attempt of validation. The analysis of eDNA on rice pot and mimic field provides rice growth promoting organisms. This approach is novel for plant ecology field. However current results did not fully support whether eDNA analysis-based detection of influencing organism.

Thank you so much for evaluating our manuscript. We have carefully read and responded to your comments. We hope our responses resolve your concerns on our study.

The strength of this manuscript is to attempt application of eDNA analysis-based plant growth differentiation. The weakness is too preliminary data and experimental set-up to make any conclusion. The trials of authors experiments are ideal. However, the process of data analysis did not meet certain levels. For example, eDNA analysis of different time points on rice growth stages resulted in two influential organisms for rice growth. Then they cultivate two species and applied rice seedlings. Without understanding of fitness and robustness, how we can know the effect of the two species on rice growth.

We agree with your comments that we did not have the fitness data of the two species and/or rice seedlings. Thus, it is still difficult to obtain deep understanding of the mechanisms of our findings that the species introduced in the system would influence rice growth. Nonetheless, our study demonstrated the effectiveness of our research framework as we found evidence that the species that were discovered by the eDNA monitoring and time series analysis indeed cause changes in the system. We believe that the first step is to show that the framework is workable and that detailed understanding of the mechanisms or genetic pathway was not a focus of our study. To avoid misunderstanding, we have added several explanations regarding this point in L426–431 and L447. For example, in L426, we have added the following statement: "... the detailed dynamics of the two introduced species was unclear (i.e., the fate of the introduced species). This is particularly important for understanding how the introduced organisms affected rice performance...".

The authors did not check the fate of two species after introducing into rice. If this is true, it is difficult to link between the rice gene expression after treatments and the effectiveness of two species. I think the validation experiment in 2019 needs to be re-conducted.

We did not check the fate of the two species (except measuring the eDNA concentrations of the species), and it is true that we cannot show evidence of "how" these two species influence the rice gene expression. Understanding molecular mechanisms of the phenomenon that we found is important (especially from the viewpoint of molecular biology), but our primary objective was to demonstrate that our "eDNA x time series analysis" framework is feasible for detecting previously overlooked but influential organisms. To this end, we believe that we achieved our objective and repeating the validation experiment should be for a different purpose (i.e., for understanding molecular mechanisms). We have clarified these points in L426–431 and L447 as explained above.

Reviewer #2 (Public Review):

The manuscript "Detecting and validating influential organisms for rice growth: An ecological network approach" explores the influence of biotic and abiotic entities that are often neglected on rice growth. The study has a straightforward experimental design, and well thought hypothesis for explorations. Monitoring data is collected to infer relationships between species and the environment empirically. It is analyzed with an up-to-date statistical method. This allowed the manuscript to hypothesize and test the effects most influential entities in a controlled experiment.

Thank you so much for your careful evaluations. We are pleased to see that you evaluated our manuscript positively. We have further revised our manuscript according to your comments and hope the revision has resolved your concerns.

The manuscript is interesting and sets up a nice framework for future studies. In general, the manuscript can be improved significantly, when this workflow is smoothly connected and communicated how they follow each other more than the sequence and dates provided. It is valuable philosophical thinking, and the research community can benefit from this framework.

Thank you for your suggestions. In order to improve the logic flow and readability of our manuscript, we have revised the descriptions of workflow and clarified how the experimental and statistical steps were connected to each other. To do so, we have added brief explanations about what/how we did at the first sentence of Results subsections (some of these explanations were only in Materials and Methods in the original manuscript). Also, we have moved all of the Supplementary Materials and Methods to the main text. We have thoroughly revised the manuscript, and we hope that all the parts of our manuscript have been connected more smoothly than in the original manuscript.

I understand the length and format of the manuscript make it difficult to add more details, but I am sure it can refer to/clear some concepts/methods that might be new for the audience. How/why variables are selected as important parts of the system, a tiny bit of information about the nonlinear time series analysis in the early manuscript, and the biological reasoning behind these statistically driven decisions are some examples.

We have explained how/why variables are selected (in L125), added more information about the nonlinear time series analysis (in L129 and L175) , and added the biological reasoning behind the statistical decisions (L195).

Reviewer #3 (Public Review):

Most farming is done by subtracting or adding what people want based in nature. However, in nature, crops interact with various objects, and mostly we are unaware of their effects. In order to increase agricultural productivity, finding useful objects is very important. However, in an uncontrolled environment, it coexists with so many biological objects that it is very inefficient to verify them all experimentally. It is therefore necessary to develop an effective screening method to identify external environmental factors that can increase crop productivity. This study identified factors presumed to be important to crop growth based on metabarcoding analysis, field sampling, and non-linear analysis/information theory, and conducted a mesocosm experiment to verify them experimentally. In conclusion, the object proposed by the author did not increase rice yield, but rather rice growth rate.

Thank you so much for your evaluation of our manuscript. We have revised our manuscript based on your comments, and hope it has been improved compared with the original version.

Strength

In actual field data, since many variables are involved in a specific phenomenon, it is necessary to effectively eliminate false positives. Based on the metabarcoding technique, various variables that may affect rice growth were quantitatively measured, although not perfectly, and the causal relationship between these variables and rice growth was analyzed by using information transfer analysis. Using this method, two new players capable of manipulating rice growth were verified, despite their unknown functions until now. I found this process to be very logical, and I think it will be valuable in subsequent ecological studies.

We are very pleased to see that you found our framework is very logical and potentially beneficial for future ecological studies.

Weaknesses

CK treatment's effectiveness remains questionable. Rice's growth was clearly altered by CK treatment. The validation of the CK treatment itself is not clear compared to the GN treatment, and the transcriptome data analysis results do not show that DEG is not present. The possibility of a side effect caused by a variable that the author cannot control remains a possibility in this case. Even though this part is mentioned in Discussion, it is necessary to discuss various possibilities in more detail.

We agree that the effectiveness of the CK treatment was questionable. We have added some more discussion about this point in L376: "The unclear effects of the CK treatment relative to those of the GN treatment could be due to the relatively unstable removal method (i.e., C. kiiensis larvae were manually removed by a hand net) or incomplete removal of the larvae (some larvae might have remained after the removal treatment)."

Reviewer #1 (Recommendations For The Authors):

Comment #1-1 This manuscript describes identification of influential organisms on rice growth and an attempt of validation. The analysis of eDNA on rice pot and mimic field provides rice growth promoting organisms. This approach is novel for plant ecology field. However current results did not fully support whether eDNA analysis-based detection of influencing organism.

Thank you for your careful evaluations of our manuscript. We are pleased to see you found that our approach is novel. We have revised our manuscript in accordance with your comments, and we hope that the revision and responses resolved your concerns.

Comment #1-2 1. Experimental setting: Authors made up small scale pot system in 2017 and then expanded manipulative experiment. I do not understand how two influencing organism sequences were identified from the single treatment depending on different time points. How they can be convince the two organisms affect the rice growth rather than other biological and environmental factors.

In 2017, we performed an intensive monitoring of the experimental rice plots and obtained large time series data (122-day consecutive monitoring x 5 plots = 610 data points). The time series data were analyzed using the information-theoretic causal analysis. The analysis is critically different from correlational analyses and designed to identify causal relationships among variables. Although we understand that field manipulation experiments are a common and straightforward approach to identify causal relationships among organisms, we chose the "fieldmonitoring + time-series-based causal analysis" approach. This is because, as explained in the main text, there are numerous factors that could influence rice performance, and it is practically impossible to perform manipulative experiments for all the potential factors that could influence rice growth. On the other hand, our "field-monitoring + timeseries-based causal analysis" approach has a potential to identify multiple factors under field conditions, even by the single experimental treatment.

Nonetheless, we must admit that our time-series-based approach still has a chance to misidentify causal factors. Our framework relies on statistics, so the chance of false-positive detection of causality cannot be zero. This was exactly the reason why we performed the "validation" experiment in 2019. To complement the statistical results of the 2017 experiments, we performed another experiment in 2019.

Comment #1-3 2. eDNA technology: The eDNA analysis based on four universal primers 16s rRNA, 18s rRNA, ITS, and COI regions must not be enough to identify specific species. The resolution of species classification may not meet to confirm exact species. Thus, the accuracy of two species that they selected for further experiment is difficult to be confirmed. Authors also referred to "putative Globisporangium".

Your point is correct. The DNA barcoding regions we selected are short and it is often difficult to identify species. However, this limitation could not have been overcome even if we had chosen a different genetic marker. The long-read sequencing technology could partially solve the issue, but the number of sequence reads generated by the long-read technique is less than that by the short-read sequencing technology, and comprehensive detection of all species in an ecological community was still challenging. Our approach struck a balance among the identification resolution, comprehensiveness of the analysis, and sequencing costs. In addition, even though we could not identify most ASVs at the species level, some ASVs could be identified at the species level (52 ASVs among the 718 ASVs which had causal influences on rice growth), and we selected the two species (G. nunn and C. kiiensis) from the 52 species.

Further, the taxa assign algorithm we used here (i.e., Claident; Tanabe & Toju 2012 PLoS ONE 10.1371/journal.pone.0076910) adopted conservative criteria for species identification and has a low falsepositive probability.

More importantly, this is also the reason why we performed the "validation" experiment in 2019. The species identified in the 2017 experiment are still "potential" organisms that influence rice growth (i.e., the hypothesis-generating phase), and we tested the hypothesis in 2019.

Nonetheless, we must admit that clear description of potential limitations is important. Thus, we have discussed this in L418: "As for the second issue, short-read sequencing has dominated current eDNA studies, but it is often not sufficient for lower-level taxonomic identification. Using long-read sequencing techniques (e.g., Oxford Nanopore MinION) for eDNA studies is a promising approach to overcome the second issue".

Comment #1-4 3. Biological relevance 1: Authors identify two organisms as influencing organism for rice growth. As conducting the first experiment in 2017, the 2019 experiment was different from natural condition. The two experiments in 2017 and 2019 were conducted under different conditions. How do they compare the experiments? At least, the eDNA analyses in 2017 and 2019 should be very similar. I cannot find such data.

The experimental conditions were different between 2017 and 2019 because they were conducted in different years. Theoretically, it is ideal if the experimental conditions in 2019 are covered by the range of experimental conditions in 2017 (e.g,. rice variety, air temperature, rainfall, and solar radiation). If this condition were satisfied, the attractor (i.e., rice growth trajectory delineated in the state space) in 2019 would be within that in 2017, and our model prediction in 2017 would be used to predict dynamics in 2019 accurately. To fulfill the conditions, we made as much effort as possible: we used the same rice variety and soils in 2019 as those used in 2017, and started our experiment at the same timing in 2019 as that in 2017.

Although natural ecological dynamics cannot be precisely controlled, our monitoring revealed that the ecological dynamics in 2019 was qualitatively similar to that in 2017. To demonstrate that the experimental conditions and eDNA community data were similar between the two experiments, we have presented the climate and eDNA data in an inset figure in Figure 3a, Figure 1–figure supplement 2, Figure 3–figure supplement 2. We must admit that these dynamics are not identical, but we hope that this resolves your concern.

Comment #1-5 4. Lack of detail description: In the Materials and Methods, there are many parts which lack on detail description. For instance, authors must described the two species cultivation, application concentrations, and application methods.

We have moved Supplementary Materials and Methods to the main text and added more detailed descriptions in Materials and Methods. Also, to improve the logical flow and readability of our manuscript, we have added brief explanations about what/how we did at the first sentence of Results subsections (some of these explanations were only in Materials and Methods in the original manuscript). We have added the reference for how to cultivate G. nunn in L608 (Kobayashi et al., 2010; Tojo et al., 1993) (C. kiiensis was not cultivated but removed from the system as in Materials and Methods), and application concentrations. Application methods were described in Materials and Methods, the section Field manipulation experiments in 2019 in L596.

Comment #1-6 5. Validation: Application of one species clearly resulted to promote rice growth. They must include appropriate control treatment. If they pick same genus but different species that identified no specific effect on rice growth through eDNA analysis, no effect on growth can be provided. Generally application of large population of certain non-harmful organism confer plant growth promotion. It is not surprising result. Authors need to prove effectiveness of eDNA analysis. In addition, the field experiments required at least two years of consistent data for publication because environmental factors are so dynamic.

Thank you for pointing this out. We agree with your comment that species that were predicted to have no effect should not promote rice growth in a validation experiment. It was also one of our inititial experimental plans to include such species in our manipulation experiment in 2019, but we could not include them because of the limitation of time, labor, and money. More extensive validation of the statistical results of the 2017 data, including multi-year experiments, would further validate the effectiveness of our approach, which should be done as future studies. To clarify this point, we have added statements in the paragraph starting at L396.

Comment #1-7 In conclusion, I suggest that authors need more large data analysis and validate with more accurate and meaningful protocol.

As we explained in the revised manuscript and the Response to Comments #1-2 to #1-7, our study demonstrated a novel research framework to detect previously overlooked influential organisms under field conditions. We agree that larger data analysis would be ideal to further validate our approach, but whether and how to collect larger data is constrained by time, money, and labor. We believe that our study was designed carefully and could provide meaningful avenues for developing an ecological-network based, novel, and environment-friendly agriculture solutions.

Reviewer #2 (Recommendations For The Authors):

Comment #2-1 Lines 97-110: This is so cool. Modeling with empirical data is very powerful. But a rice field is an open system consisting of metacommunity dynamics. Maybe a tiny bit of biological and biogeochemical background here would be good.

Thank you for your comments. We have added a few examples of how and in which systems these methods were used to evaluate community dynamics and detect biological interactions in L109-L118.

Comment #2-2 Lines 111-126: I like the summary of the study here. I think the influential species concept can be a little more elevated. Paine's famous keystone species work has been cited but a couple more pieces of literature can help to enhance the ecological importance of this work.

We have explained the work by Paine (1966) a bit more and added one more paper that showed the effect of multiple predator species on the system dynamics at L88. We have also added a relevant sentence at L137 to emphasize the ecological/agricultural significance of our work.

Comment #2-3 Experimental design/Figure 1:

Is there any rationale behind choosing red individuals to measure the growth?

Is there any competition between the individuals in the pots?

Figure 1e: It is nice to show the ASVs in time. I wonder how the plot would look like when normalized by biomass/DNA content/coverage/rarefaction because of the seasonality.

As for the first question, we chose the four individuals to minimize the edge effects (i.e., effects of microclimates and neighboring rice would be different between the four rice individuals and those planted in the edge regions). We have mentioned this in the legend of Figure 1.

As for the second question, there might be competition among the individuals in the pot. However, we did not measure the effect of competition (e.g., by comparing the growth with/without other rice individuals).

As for the third question, we published detailed dynamics of ecological community in the Supplementary Figures in Ushio (2022) Proceedings B https://doi.org/10.6084/m9.figshare.c.5842766.v1. In addition, we have uploaded a video showing the temporal dynamics of some top (= most abundant) ASVs in https://doi.org/10.6084/m9.figshare.23514150.v2.

We have mentioned the supporting information in L153.

Comment #2-4 Line 146-147: Is this damage influence the inferences? Maybe it is better to justify.

While we occasionally observed physical damages, it is unlikely that they affected our causal inference because the changes in the rice heights due to the damages were smaller and less frequent than those due to growth. We have noted this at L151.

Comment #2-5 Line 161-162: Maybe refer readers to the methods section where you explain UIC analysis. It'd be easier to interpret the figures.

Mentioned.

Comment #2-6 Line 175-176: I believe very brief information in the intro about the organisms might help explain the hypothesis and interpret the results better.

We have included brief information of the two species at L197.

Comment #2-7 Figure 2: Species interaction strength: Are these proxies to the Jacobians? Is there a threshold for the influence we can consider strong/weak? For example, influential species compared to diagonal elements of the Jacobians (intraspecies interactions) could be shown as a mean vertical line in Figure 2b.

"Influences to rice growth" in Figure 2b is transfer entropy (TE) from a target ASV to rice growth. They are not proxies of the Jacobians, but they might positively correlate with the absolute value of the Jacobians. We have clarified this point in the legend (L953). More direct estimations of the Jacobian can be done using the MDR S-map method (Chang et al. 2021 DOI:10.1111/ele.13897), but we did not perform the MDR S-map in the present manuscript (see Ushio et al. 2023 https://doi.org/10.7554/eLife.85795 for the application of the MDR S-map). As for TE, there is no clear threshold to distinguish strong/weak interactions.

Comment #2-8 Figure 2: Looking at panels c and d, it looks like there is a negative frequency selection between two influential species. Is it a reasonable observation?

This is an interesting point. In this manuscript, we have not carefully examined the interspecific relationship between these two particular species. However, the interspecific interactions were examined in detail and reported in Ushio (2022) Proceedings of the Royal Society B DOI:10.1098/rspb.2021.2690). We re-checked the result in Ushio (2022); although there is a negative correlation between them, we did not find any (statistical) causal relationship between them.

Comment #2-9 Line 209: What is t-SNE analysis? Because of the manuscript's format, maybe methods should be shortly referred to in the relevant section or explained in brackets.

We have spelled out t-SNE.

Comment #2-10 Line 212-214: Maybe briefly explain what the hypotheses are for the alternative analysis, and what is the contribution of the results to the study.

We have added a brief explanation at L241: "Alternative statistical modeling that included the treatments (the control versus GN or CK treatments) and manipulation timing (i.e., before or after the manipulation), which simultaneously took the temporal changes of all the treatments into account, also showed qualitatively similar results (Supplementary file 4), further supporting the results."

Comment #2-11 Figure 3b/c: Maybe species names as panel titles could be helpful. d: Treatment names with initials in the legend could be also helpful to read the plots.

We have added species name as panel titles of Figure 3b,c. Treatment names were included in the legend of Figure 3.

Comment #2-12 Line 233: Maybe mention why the manuscript uses the word "clear".

We have mentioned this in L185.

Comment #2-13 Line 234-236: I think that these alternative tests should be explained somewhere.

We have revised the sentence so that it includes some explanations (L241). Also, we have referred to Materials and Methods.

Comment #2-14 Figure 4: The title says ecological community compositions, and panels show the growth rates and cumulative growth.

Thank you for pointing this out. This was a typo and we have corrected it.

Comment #2-15 Lines 246-269: Can these expression patterns be transient and relevant to the time point that the sample is taken?

Yes, these expression patterns were transient. We collected rice leaf samples for RNA-seq 1 day before the first manipulation and 1, 14, and 38 days after the third manipulation (see Supplementary file 3 for the sampling design). When we merged the pot locations, we observed no difference in the gene expression for samples 1 day before the first manipulation and 14 and 38 days after the third manipulation (except for two genes in samples 38 days after the manipulation), and thus, we consider the DEGs that appeared only in the short period after the manipulation. We have mentioned this in L278 and L383: "We found almost no DEGs for leaf samples taken one day before and 14 and 38 days after the third manipulation (the leaf sampling event 1, 3, and 4), suggesting that the influences of the treatments on the gene expression patterns were transient." (L278) and "These changes were observed relatively quickly and transient." (L383)

Comment #2-16 I wonder if a conceptual framework figure would help to generalize the workflow that can be used for other studies.

Thank you for your suggestion. Although we agree with your comment that such a figure would be helpful to generalize the workflow, we believe that our framework is clear and decided not to include it in the present manuscript. We might consider including such a figure (like Figure 1a in Ushio 2022) if we have an opportunity to write a review paper regarding this topic.

Comment #2-17 Lines 329-335: I feel this information is unclear in the early manuscript. Maybe it's necessary to clearly communicate in the beginning.

We have explained that we could not find any relevant information at least at the time we detected the ASVs in L189.

Comment #2-18 Lines 336-337: Can these species be identified in the previous data set from the ASV sequences?

Yes, these species were identified in the DNA data set obtained in 2017.

Comment #2-19 Lines 387-397: Are there any measurements such as total biomass, and statistical methods to help with the eDNA bias and data compositionality?

We have confirmed that our quantitative eDNA metabarcoding generates comparable results with the fluorescence-based method and quantitative PCR (e.g., see Supplementary Figures in Ushio 2022) (mentioned in L310 in the revised manuscript). However, at least in this study, we could not perform a direct comparison of the eDNA data with species abundance and/or biomass. This is partly because the number of our target species was too large (> 1,000 species). The accurate estimation of species abundance and/or biomass is one of our next goals.

Comment #2-20 Line 472: Maybe mention transfer entropy somewhere in the early manuscript.

We have mentioned this in L175.

Comment #2-21 Lines 494-503: Maybe a summary of this reasoning should be mentioned somewhere in the early manuscript too.

We have described a brief summary of the reasoning in L195.

Comment #2-22 Lines 29-33 If this sentence is simplified it might be easier to follow.

The sentence has been divided into two sentences in L28. Also, each sentence has been simplified.

Comment #2-23 Line 38 Maybe "macrobes" can be explicitly mentioned. Fungi, protozoa, etc.

Mentioned.

Comment #2-24 Line 139: I am not sure if the date should be in the title.

Similar monitoring was done in 2017 and 2019. Thus, we think the date is necessary in the section title.

Comment #2-25 Figure 1: There are 4 red individuals in the design but 5 measurements in the plots.

Heights and SPAD of the four individuals were measured for each plot and the averaged values were used as representative values for each plot. Therefore, 20 measurements (= 4 rice individuals 5 plots) were done every day, but each plot has one rice height for each day. We have clarified this in the legend of Figure 1: "the average values of the four individuals were regarded as representative values for each plot."

Comment #2-26 Figure 1b: Maybe use the same axis length for the temperature as the other plots?

Corrected.

Comment #2-27 Lines 259-261: Are there the names of the genes in databases?

Yes, these are gene names used in the rice databases (e.g., The Rice Annotation Project Database; https://rapdb.dna.affrc.go.jp/inde x.html).

Reviewer #3 (Recommendations For The Authors):

Comment #3-1 Additionally, RGR is not statistically significant, but statistical significance is observed only in cumulative growth because data presentation does not reflect plant characteristics. RGR changes according to the developmental stage of the plant. Therefore, if RGR data are shown separately according to the rice growing season, the cumulative growth pattern and the pattern will appear similar.

RGRs were calculated daily (i.e., cm/day) and they changed depending on the developmental stage of the rice (Figure 1 and Figure 4–figure supplement 1). Therefore, we might find similar RGR patterns if we focus on a specific period of the growing season. However, unfortunately, we performed the intensive (i.e., daily) monitoring in 2019 only during the field manipulation period (middle June to middle July 2019), and we cannot investigate the changes in cumulative growth throughout the growing season (this depends on how many days we add up RGR to calculate the cumulative growth, though). We agree that, if we had investigated the detailed pattern of RGR throughout the growing season in 2019, we could have found similar pattens between RGR and cumulative growth rate at a certain period in the growing season. In Figure 4, the cumulative growths were calculated based on the RGRs before the third manipulation or during 10 days after the third manipulation. We clarified this in the legend of Figure 4.

Author Response

The following is the authors’ response to the original reviews.

Reviewer #1 (Public Review):

This paper describes the development and initial validation of an approach-avoidance task and its relationship to anxiety. The task is a two-armed bandit where one choice is 'safer' - has no probability of punishment, delivered as an aversive sound, but also lower probability of reward - and the other choice involves a reward-punishment conflict. The authors fit a computational model of reinforcement learning to this task and found that self-reported state anxiety during the task was related to a greater likelihood of choosing the safe stimulus when the other (conflict) stimulus had a higher likelihood of punishment. Computationally, this was represented by a smaller value for the ratio of reward to punishment sensitivity in people with higher task-induced anxiety. They replicated this finding, but not another finding that this behavior was related to a measure of psychopathology (experiential avoidance), in a second sample. They also tested test-retest reliability in a sub-sample tested twice, one week apart and found that some aspects of task behavior had acceptable levels of reliability. The introduction makes a strong appeal to back-translation and computational validity, but many aspects of the rationale for this task need to be strengthened or better explained. The task design is clever and most methods are solid - it is encouraging to see attempts to validate tasks as they are developed. There are a few methodological questions and interpretation issues, but they do not affect the overall findings. The lack of replicated effects with psychopathology may mean that this task is better suited to assess state anxiety, or to serve as a foundation for additional task development.

We thank the reviewer for their kind comments and constructive feedback. We agree that the approach taken in this paper appears better suited to state anxiety, and further work is needed to assess/improve its clinical relevance.

Reviewer #1 (Recommendations For The Authors):

1) For the introduction, the authors communicate well the appeal of tasks with translational potential, and setting up this translation through computational validity is a strong approach. However, I had some concerns about how the task was motivated in the introduction:

a) The authors state that current approach-avoidance tasks used in humans do not resemble those used in the non-human literature, but do not provide details on what exactly is missing from these tasks that makes translation difficult.

Our intention for the section that the reviewer refers to was to briefly convey that historically, approach-avoidance conflict would have been measured either using questionnaires or joystick-based tasks which have no direct non-human counterpart. However, we note that the phrasing was perhaps unfair to recent tasks that were explicitly designed to be translatable across species. Therefore, we have amended the text to the following:

In humans, on the other hand, approach-avoidance conflict has historically been measured using questionnaires such as the Behavioural Inhibition/Activation Scale (Carver & White, 1994), or cognitive tasks that rely on motor biases, for example by using joysticks to approach/move towards positive stimuli and avoid/move away from negative stimuli, which have no direct non-human counterparts (Guitart-Masip et al., 2012; Kirlic et al., 2017; Mkrtchian et al., 2017; Phaf et al., 2014).

b) Although back-translation to 'match' human paradigms to non-animal paradigms is useful for research, this isn't the end goal of task development. What really matters is how well these tasks, whether in humans or not, capture psychopathology-relevant behavior. Many animal paradigms were developed and brought into extensive use because they showed sensitivity to pharmacological compounds (e.g., benzodiazepines). The introduction accepts the validity of these paradigms at face value, and doesn't address whether developing human tests of psychopathology based on sensitivity to existing medication classes is the best way to generate new insights about psychopathology.

We agree that whilst paradigms with translational and computational validity have merits of their own for neuroscientific theory, clinical validity (i.e. how well the paradigm reflects a phenomenon relevant to psychopathology) is key in the context of clinical applications. While our findings of associations between task performance and self-reported (state) anxiety suggest that our approach is a step in the right direction, the lack of associations with clinical measures was disappointing. Although future work is needed to more directly test the sensitivity of the current approach to psychopathology, this may mean that it, and its non-human counterparts, do not measure behaviours relevant to pathological anxiety. Since our primary focus in this paper was on translational and computational validity, we have opted to discuss the author’s suggestion in the ‘Discussion’ section, as follows:

Further, it is worth noting that many animal paradigms were developed and widely adopted due to their sensitivity to anxiolytic medication (Cryan & Holmes, 2005). Given the lack of associations with clinical measures in our results, it is possible that current translational models of anxiety may not fully capture behaviours that are directly relevant to pathological anxiety. To develop translational paradigms of clinical utility, future research should place a stronger emphasis on assessing their clinical validity in humans.

c) The authors may want to bring in the literature on the description-experience gap (e.g., PMID: 19836292) when discussing existing decision tasks and their computational dissimilarity to non-human operant conditioning tasks.

We thank the reviewer for this useful addition to the introduction. We have now added the following to the 'Introduction’ section:

Moreover, evidence from economic decision-making suggests that explicit offers of probabilistic outcomes can impact decision-making differently compared to when probabilistic contingencies need to be learned from experience (referred to as the ‘description-experience gap’; Hertwig & Erev, 2009); this finding raises potential concerns regarding the use of offer-based tasks in humans as approximations of non-human tasks that do not involve explicit offers.

d) How does one evaluate how computationally similar human vs. non-human tasks are? What are the criteria for making this judgement? Specific to the current tasks, many animal learning tasks are not learning tasks in the same sense that human learning tasks are, in terms of the number of trials used and if the animals are choosing from a learned set of contingencies versus learning the contingencies during the testing.

The computational similarity of human and non-human strategies in a given translational task can be tested empirically. This can be done by fitting models to the data and assessing whether similar models explain choices, even if parameter distributions might vary across species due to, for example, physiological differences. Indeed, non-human animals require much more training to perform even uni-dimensional reinforcement learning, but once they are trained, it should be possible to model their responses. In fact, it should even be possible to take training data into account in some cases. For example, the training phase of the Vogel/Geller-Seifter preclinical tests require an animal to learn to emit a certain action (e.g. lever press) simply to obtain some reward. In the next phase, an aversive outcome is introduced as an additional outcome, but one could model both the training and test phase together – the winning model in our studies would be a suitable candidate to model behaviour here. As we also discuss predictive validity in the ‘Discussion’ section, we opted to add the following text there too:

… computational validity would also need to be assessed directly in non-human animals by fitting models to their behavioural data. This should be possible even in the face of different procedures across species such as number of trials or outcomes used (shock or aversive sound). We are encouraged by our finding that the winning computational model in our study relies on a relatively simple classical reinforcement learning strategy. There exist many studies showing that non-human animals rely on similar strategies during reward and punishment learning (Mobbs et al., 2020; Schultz, 2013); albeit to our knowledge this has never been modelled in non-human animals where rewards and punishment can occur simultaneously.

2) What do the authors make of the non-linear relationship between probability of punishment and probability of choosing the conflict stimulus (Fig 2d), especially in the high task-induced anxiety participants? Did this effect show up in the replication sample as well?

Figures 2c-e were created by binning the continuous predictors of outcome probabilities into discrete bins of equal interval. Since punishment probability varied according to Gaussian random walks, it was also distributed with more of its mass in the central region (~ 0.4), and so values at the extreme bins were estimated on fewer data and with greater variance. The non-linear relationships are likely thus an artefact of our task design and plotting procedure. The pattern was also evident in the replication sample, see Author response image 1:

Author response image 1.

However, since these effects were estimated as linear effects in the logistic regression models, and to avoid overfitting/interpretations of noise arising from our task design, we now plot logistic curves fitted to the raw data instead.

3) How correlated were learning rate and sensitivity parameters? The EM algorithm used here can sometimes result in high correlations among these sets of parameters.

As the reviewer suspects the parameters were strongly correlated, especially across the punishment-specific parameters. The Pearson’s r estimates for the untransformed parameter values were as follows:

Reward parameters: discovery sample r = -0.39; replication sample r = -0.78

Punishment parameters: discovery sample r = -0.91; replication sample r = -0.85

We have included the correlation matrices of the estimated parameters as Supplementary Figure 2 in the ‘Computational modelling’ section of the Supplement.

We have now also re-fitted the winning model using variational Bayesian inference (VBI) via Stan, and found that the cross-parameter correlations were much lower than when the data were fitted using EM. We also ran a sensitivity analysis assessing whether using VBI changed the main findings of our studies. This showed that the correlation between task-induced anxiety and the reward-punishment sensitivity index was robust to fitting method, as was the mediating effect of reward-punishment sensitivity index on anxiety’s effect on choice. This indicates that overall our key findings are robust to different methods of parameter-fitting.

We now direct readers to these analyses from the new ‘Sensitivity analyses’ section in the manuscript, as follows:

As our procedure for estimating model parameters (the expectation-maximisation algorithm, see ‘Methods’) produced high inter-parameter correlations in our data (Supplementary Figure 2), we also re-estimated the parameters using Stan’s variational Bayesian inference algorithm (Stan Development Team, 2023) – this resulted in lower inter-parameter correlations, but our primary computational finding, that the effect of anxiety on choice is mediated by relative sensitivity to reward/punishment was consistent across algorithms (see Supplement section 9.8 for details).

We have included the relevant analyses comparing EM and VBI in the Supplement, as follows:

[9.8 Sensitivity analysis: estimating parameters via expectation maximisation and variational Bayesian inference algorithms]

Given that the expectation maximisation (EM) algorithm produced high inter-parameter correlations, we ran a sensitivity analysis by assessing the robustness of our computational findings to an alternative method of parameter estimation – (mean-field) variational Bayesian inference (VBI) via Stan (Stan Development Team, 2023). Since, unlike EM, the results of VBI are very sensitive to initial values, we fitted the data 10 times with different initial values.

Inter-parameter correlations

The VBI produced lower inter-parameter correlations than the EM algorithm (Supplementary Figure 8).

Sensitivity analysis

Since multicollinearity in the VBI-estimated parameters was lower than for EM, indicating less trade-off in the estimation, we re-tested our computational findings from the manuscript as part of a sensitivity analysis. We first assessed whether we observed the same correlations between task-induced anxiety and punishment learning, and reward-punishment sensitivity index (Supplementary Figure 9a). Punishment learning rate was not significantly associated with task-induced anxiety in any of the 10 VBI iterations in the discovery sample, although it was in 9/10 in the replication sample. On the other hand, the reward-punishment sensitivity index was significantly associated with task-induced anxiety in 9/10 VBI iterations in the discovery sample and all iterations in the replication sample. This suggests that the correlation of anxiety and sensitivity index is robust to these two fitting approaches.

We also re-estimated the mediation models, where in the EM-estimated parameters, we found that the reward-punishment sensitivity index mediated the relationship between task-induced anxiety and task choice proportions (Supplementary Figure 9b). Again, we found that the reward-punishment sensitivity index was a significant mediator in 9/10 VBI iterations in the discovery sample and all iterations in the replication sample. Punishment learning rate was also a significant mediator in 9/10 iterations in the replication sample, although it was not in the discovery sample for all iterations, and this was not observed for the EM-estimated parameters.

Overall, we found that our key results, that anxiety is associated with greater sensitivity to punishment over reward, and this mediates the relationship between anxiety and approach-avoidance behaviour, were robust across both fitting methods.

As an aside, we were unable to run the model fitting using Markov chain Monte Carlo sampling approaches due to the computational power and time required for a sample of this size (Pike & Robinson, 2022, JAMA Psychiatry).

4) What is the split-half reliability of the task parameters?

We thank the reviewer for this query. We have now included a brief section on the (good-to-excellent) split-half reliability of the task in the manuscript:

We assessed the split-half reliability of the task by correlating the overall proportion of conflict option choices and model parameters from the winning model across the first and second half of trials. For overall choice proportion, reliability was simply calculated via Pearson’s correlations. For the model parameters, we calculated model-derived estimates of Pearson’s r values from the parameter covariance matrix when first- and second-half parameters were estimated within a single model, following a previous approach recently shown to accurately estimate parameter reliability (Waltmann et al., 2022). We interpreted indices of reliability based on conventional values of < 0.40 as poor, 0.4 - 0.6 as fair, 0.6 - 0.75 as good, and > 0.75 as excellent reliability (Fleiss, 1986). Overall choice proportion showed good reliability (discovery sample r = 0.63; replication sample r = 0.63; Supplementary Figure 5). The model parameters showed good-to-excellent reliability (model-derived r values ranging from 0.61 to 0.85 [0.76 to 0.92 after Spearman-Brown correction]; Supplementary Figure 5).

5) The authors do a good job of avoiding causal language when setting up the cross-sectional mediation analysis, but depart from this in the discussion (line 335). Without longitudinal data, they cannot claim that "mediation analyses revealed a mechanism of how anxiety induces avoidance".

Thank you for spotting this, we have now amended the text to:

… mediation analyses suggested a potential mechanism of how anxiety may induce avoidance.

Reviewer #2 (Public Review):

Summary:

The authors develop a computational approach-avoidance-conflict (AAC) task, designed to overcome limitations of existing offer based AAC tasks. The task incorporated likelihoods of receiving rewards/ punishments that would be learned by the participants to ensure computational validity and estimated model parameters related to reward/punishment and task induced anxiety. Two independent samples of online participants were tested. In both samples participants who experienced greater task induced anxiety avoided choices associated with greater probability of punishment. Computational modelling revealed that this effect was explained by greater individual sensitivities to punishment relative to rewards.

Strengths:

Large internet-based samples, with discovery sample (n = 369), pre-registered replication sample (n = 629) and test-retest sub group (n = 57). Extensive compliance measures (e.g. audio checks) seek to improve adherence.

There is a great need for RL tasks that model threatening outcomes rather than simply loss of reward. The main model parameters show strong effects and the additional indices with task based anxiety are a useful extension. Associations were broadly replicated across samples. Fair to excellent reliability of model parameters is encouraging and badly needed for behavioral tasks of threat sensitivity.

We thank the reviewer for their comments and constructive feedback.

The task seems to have lower approach bias than some other AAC tasks in the literature. Although this was inferred by looking at Fig 2 (it doesn't seem to drop below 46%) and Fig 3d seems to show quite a strong approach bias when using a reward/punishment sensitivity index. It would be good to confirm some overall stats on % of trials approached/avoided overall.

The range of choice proportions is indeed an interesting statistic that we have now included in the manuscript:

Across individuals, there was considerable variability in overall choice proportions (discovery sample: mean = 0.52, SD = 0.14, min/max = [0.03, 0.96]; replication sample: mean = 0.52, SD = 0.14, min/max = [0.01, 0.99]).

Weaknesses:

The negative reliability of punishment learning rate is concerning as this is an important outcome.

We agree that this is a concerning finding. As reviewer 3 notes, this may have been due to participants having control over the volume used to play the aversive sounds in the task (see below for our response to this point). Future work with better controlled experimental settings will be needed to determine the reliability of this parameter more accurately.

This may also have been due to the asymmetric nature of the task, as only one option could produce the punishment. This means that there were fewer trials on which to estimate learning about the occurrence of a punishment. Future work using continuous outcomes, as the reviewer suggests below, whilst keeping the asymmetric relationship between the options, could help in this regard.

We have included the following comment on this issue in the manuscript:

Alternatively, as participants self-determined the loudness of the punishments, differences in volume settings across sessions may have impacted the reliability of this parameter (and indeed punishment sensitivity). Further, the asymmetric nature of the task may have impacted our ability to estimate the punishment learning rate, as there were fewer occurrences of the punishment compared to the reward.

The Kendall's tau values underlying task induced anxiety and safety reference/ various indices are very weak (all < 0.1), as are the mediation effects (all beta < 0.01). This should be highlighted as a limitation, although the interaction with P(punishment|conflict) does explain some of this.

We now include references to the effect sizes to emphasise this limitation. We also note, as the reviewer suggests, that this may be due to crudeness of overall choice proportion as a measure of approach/avoidance, as it is contaminated with variables such as P(punishment|conflict).

One potentially important limitation of our findings is the small effect size observed in the correlation between task-induced anxiety and avoidance (Kendall's tau values < 0.1, mediation betas < 0.01). This may be attributed to the simplicity of using overall choice proportion as a measure of approach/avoidance, as the effect of anxiety on choice was also influenced by punishment probability.

The inclusion of only one level of reward (and punishment) limits the ecological validity of the sensitivity indices.

We agree that using multi-level outcomes will be an important question for future work and now explicitly note this in the manuscript, as below:

Using multi-level or continuous outcomes would also improve the ecological validity of the present approach and interpretation of the sensitivity parameters.

Appraisal and impact:

Overall this is a very strong paper, describing a novel task that could help move the field of RL forward to take account of threat processing more fully. The large sample size with discovery, replication and test-retest gives confidence in the findings. The task has good ecological validity and associations with task-based anxiety and clinical self-report demonstrate clinical relevance. The authors could give further context but test-retest of the punishment learning parameter is the only real concern. Overall this task provides an exciting new probe of reward/threat that could be used in mechanistic disease models.

We thank the reviewer again for helping us to improve our analyses and manuscript.

Reviewer #2 (Recommendations For The Authors):

Additional context:

In the introduction "cognitive tasks that bear little semblance to those used in the non-human literature" seems a little unfair. One study that is already cited (Ironside et al, 2020) used a task that was adapted from non-human primates for use in humans. It has almost identical visual stimuli (different levels of simultaneous reward and aversive outcome/punishment) and response selection processes (joystick) between species and some overlapping brain regions were activated across species for conflict and aversiveness. The later point that non-human animals must be trained on the association between action and outcome is well taken from the point of view of computational validity but perhaps not sufficient to justify the previous statement.

Our intention for this section was to briefly convey that historically, approach-avoidance conflict would have been measured either using questionnaires or joystick-based tasks which have no direct non-human counterpart. However, we agree that this phrasing is unfair to recent studies such as those by Ironside and colleagues. Therefore, we have amended the text to the following:

In humans, on the other hand, approach-avoidance conflict has historically been measured using questionnaires such as the Behavioural Inhibition/Activation Scale (Carver & White, 1994), or cognitive tasks that rely on motor biases to approach/move towards positive stimuli and avoid/move away from negative stimuli which have no direct non-human counterparts (Guitart-Masip et al., 2012; Kirlic et al., 2017; Mkrtchian et al., 2017; Phaf et al., 2014).

It would be good to speculate on why task induced anxiety made participants slower to update their estimates of punishment probability.

Although a meta-analysis of reinforcement learning studies using reward and punishment outcomes suggests a positive association between punishment learning rate and anxiety symptoms (and depressed mood), we paradoxically found the opposite effect. However, previous work has suggested that distinct forms of anxiety associate differently with anxiety (Wise & Dolan, 2020, Nat. Commun.), where somatic anxiety was negatively correlated with punishment learning rate whereas cognitive anxiety showed the opposite effect. We have now added the following to the manuscript, and noted that future work is needed to understand the potentially complex relationship between anxiety and learning from punishments:

Notably, although a recent computational meta-analysis of reinforcement learning studies showed that symptoms of anxiety and depression are associated with elevated punishment learning rates (Pike & Robinson, 2022), we did not observe this pattern in our data. Indeed, we even found the contrary effect in relation to task-induced anxiety, specifically that anxiety was associated with lower rates of learning from punishment. However, other work has suggested that the direction of this effect can depend on the form of anxiety, where cognitive anxiety may be associated with elevated learning rates, but somatic anxiety may show the opposite pattern (Wise & Dolan, 2020) and this may explain the discrepancy in findings. Additionally, parameter values are highly dependent on task design (Eckstein et al., 2022), and study designs to date may be more optimised in detecting differences in learning rate (Pike & Robinson, 2022) – future work is needed to better understand the potentially complex association between anxiety and punishment learning rate. Lastly, as punishment learning rate was severely unreliable in the test-retest analyses, and the associations between punishment learning rate and state anxiety were not robust to an alternative method of parameter estimation (variational Bayesian inference), the negative correlation observed in our study should be treated with caution.

Were those with more task-based anxiety more inflexible in general?

The lack of associations across reward learning rate and task-induced anxiety suggest that this was not a general inflexibility effect. To test the reviewer’s hypothesis more directly, we conducted a sensitivity analysis by examining the model with a general learning rate – this did not support a general inflexibility effect. Please see the new section in the Supplement below:

[9.10 Sensitivity analysis: anxiety and inflexibility]

As anxious participants were slower to update their estimates of punishment probability, we determined whether this was due to greater general inflexibility by examining the model including two sensitivity parameters, but one general learning rate (i.e. not split by outcome). The correlation between this general learning rate and task-induced anxiety was not significant in either samples (discovery: tau = -0.02, p = 0.504; replication: tau = -0.01, p = 0.625), suggesting that the effect is specific to punishment.

Was the 16% versus 20% of the two samples with clinically relevant anxiety symptoms significantly different? What about other demographics in the two samples?

The difference in proportions were not significantly different (χ2 = 2.33, p = 0.127). The discovery sample included more females and was older on average compared to the replication sample – information which we now report in the manuscript:

The discovery sample consisted of a significantly greater proportion of female participants than the replication sample (59% vs 52%, χ2 = 4.64, p = 0.031). The average age was significantly different across samples (discovery sample mean = 37.7, SD = 10.3, replication sample mean = 34.3, SD = 10.4; t785.5 = 5.06, p < 0.001). The differences in self-reported psychiatric symptoms across samples did not reach significance (p > 0.086).

It would be interesting to know how many participants failed the audio attention checks.

We have now included information about what proportion of participants fail each of the task exclusion criteria in the manuscript:

Firstly, we excluded participants who missed a response to more than one auditory attention check (see above; 8% in both discovery and replication samples) – as these occurred infrequently and the stimuli used for the checks were played at relatively low volume, we allowed for incorrect responses so long as a response was made. Secondly, we excluded those who responded with the same response key on 20 or more consecutive trials (> 10% of all trials; 4/6% in discovery and replication samples, respectively). Lastly, we excluded those who did not respond on 20 or more trials (1/2% in discovery and replication samples, respectively). Overall, we excluded 51 out of 423 (12%) in the discovery sample, and 98 out of 725 (14%) in the replication sample.

There doesn't appear to be a model with only learning from punishment (i.e. no reward learning) included in the model comparison. It would be interesting to see how it compared.

We have fitted the suggested model and found that it is the least parsimonious of the models. Since participants were monetarily incentivised based on the rewards only, this was to be expected. We have now added this ‘punishment learning only’ model and its variant including a lapse term into the model comparison. The two lowest bars on the y-axis in Author response image 2 represent these models.

Author response image 2.

Were sex effects examined as these have been commonly found in AAC tasks. How about other covariates such as age?

We have now tested the effects of sex and age on behaviour and on parameter values. There were indeed some significant effects, albeit with some inconsistencies across the two samples, which for completeness we have included in the manuscript, as follows:

While sex was significantly associated with choice in the discovery sample (β = 0.16 ± 0.07, p = 0.028) with males being more likely to choose the conflict option, this pattern was not evident in the replication sample (β = 0.08 ± 0.06, p = 0.173), and age was not associated with choice in either sample (p > 0.2).

Comparing parameters across sexes via Welch’s t-tests revealed significant differences in reward sensitivity (t289 = -2.87, p = 0.004, d = 0.34; lower in females) and consequently reward-punishment sensitivity index (t336 = -2.03, p = 0.043, d = 0.22; lower in females i.e. more avoidance-driven). In the replication sample, we observed the same effect on reward-punishment sensitivity index (t626 = -2.79, p = 0.005, d = 0.22; lower in females). However, the sex difference in reward sensitivity did not replicate (p = 0.441), although we did observe a significant sex difference in punishment sensitivity in the replication sample (t626 = 2.26, p = 0.024, d = 0.18).

Minor: Still a few placeholders (Supplementary Table X/ Table X) in the methods

We thank the reviewer for spotting these errors. We have now corrected these references.

Reviewer #3 (Public Review):

This study investigated cognitive mechanisms underlying approach-avoidance behavior using a novel reinforcement learning task and computational modelling. Participants could select a risky "conflict" option (latent, fluctuating probabilities of monetary reward and/or unpleasant sound [punishment]) or a safe option (separate, generally lower probability of reward). Overall, participant choices were skewed towards more rewarded options, but were also repelled by increasing probability of punishment. Individual patterns of behavior were well-captured by a reinforcement learning model that included parameters for reward and punishment sensitivity, and learning rates for reward and punishment. This is a nice replication of existing findings suggesting reward and punishment have opposing effects on behavior through dissociated sensitivity to reward versus punishment.

Interestingly, avoidance of the conflict option was predicted by self-reported task-induced anxiety. This effect of anxiety was mediated by the difference in modelled sensitivity to reward versus punishment (relative sensitivity). Importantly, when a subset of participants were retested over 1 week later, most behavioral tendencies and model parameters were recapitulated, suggesting the task may capture stable traits relevant to approach-avoidance decision-making.

We thank the reviewer for their useful analysis of our study. Indeed, it was reassuring to see that performance indices were reliable across time.

However, interpretation of these findings are severely undermined by the fact that the aversiveness of the auditory punisher was largely determined by participants, with the far-reaching impacts of this not being accounted for in any of the analyses. The manipulation check to confirm participants did not mute their sound is highly commendable, but the thresholding of punisher volume to "loud but comfortable" at the outset of the task leaves substantial scope for variability in the punisher delivered to participants. Indeed, participants' ratings of the unpleasantness of the punishment was moderate and highly variable (M = 31.7 out of 50, SD = 12.8 [distribution unreported]). Despite having this rating, it is not incorporated into analyses. It is possible that the key finding of relationships between task-induced anxiety, reward-punishment sensitivity and avoidance are driven by differences in the punisher experienced; a louder punisher is more unpleasant, driving greater task-induced anxiety, model-derived punishment sensitivity, and avoidance (and vice versa). This issue can also explain the counterintuitive findings from re-tested participants; lower/negatively correlated task-induced anxiety and punishment-related cognitive parameters may have been due to participants adjusting their sound settings to make the task less aversive (retest punisher rating not reported). It can therefore be argued that the task may not actually capture meaningful cognitive/motivational traits and their effects on decision-making, but instead spurious differences in punisher intensity.

We thank the reviewer for raising this important potential limitation of our study. We agree that how participants self-adjusted their sound volume may important consequences for our interpretations of the data. Unfortunately, despite the scalability of online data collection, this highlights one of its major weaknesses in the lack of controllability over experimental parameters. The previous paper from which we obtained our aversive sounds (Seow & Hauser, 2021, Behav Res, doi.org/10.3758/s13428-021-01643-0) contains useful analyses with regards to this discussion. When comparing the unpleasantness of the sounds played at 50% vs 100% volume, the authors indeed found that the lower volumes lead to lower unpleasantness ratings. However, the magnitude of this effect did not appear to be substantial (Fig. 4 from the paper), and even at 50% volume, the scream sounds we used were rated in the top quartile for unpleasantness, on average. This implies that the sounds have sufficient inherent unpleasantness, even when played at half intensity. We find this reassuring, in the sense that any self-imposed volume effects may not be large. Of note, our instructions to participants to adjust the volume to a ‘loud but comfortable’ level was based on the same phrasing used in this study.

To the reviewers point on how this might affect the reliability of the task, we have included the following in the ‘Discussion’ section:

Alternatively, as participants self-determined the loudness of the punishments, differences in volume settings across sessions may have impacted the reliability of this parameter (and indeed other measures).

Please see below for analyses accounting for punishment unpleasantness ratings.

This undercuts the proposed significance of this task as a translational tool for understanding anxiety and avoidance. More information about ratings of punisher unpleasantness and its relationship to task behavior, anxiety and cognitive parameters would be valuable for interpreting findings. It would also be of interest whether the same results were observed if the aversiveness of the punisher was titrated prior to the task.

As suggested, we have now included sensitivity analyses using the unpleasantness ratings that show their effect is minimal on our primary inference. We report relevant results below in the ‘Recommendations For The Authors’ section. At the same time, we think it is important to acknowledge that unpleasantness is a combination of both the inherent unpleasantness of the sound and the volume it is presented at, where only the latter is controlled by the participant. Therefore, these analyses are not a perfect indicator of the effect of participant control. For convenience, we reproduce the key findings from this sensitivity analysis here:

Approach-avoidance hierarchical logistic regression model

We assessed whether approach and avoidance responses, and their relationships with state anxiety, were impacted by punishment unpleasantness, by including unpleasantness ratings as a covariate into the hierarchical logistic regression model. Whilst unpleasantness was a significant predictor of choice (positively predicting safe option choices), all significant predictors and interaction effects from the model without unpleasantness survived (Supplementary Figure 11). Critically, this suggests that punishment unpleasantness does not account for all of the variance in the relationship between anxiety and avoidance.

Mediation model

When unpleasantness ratings were included in the mediation models, the mediating effect of the reward-punishment sensitivity index did not survive (discovery sample: standardised β = 0.003 ± 0.003, p = 0.416; replication sample: standardised β = 0.004 ± 0.003, p = 0.100; Supplementary Figure 12). Pooling the samples resulted in an effect that narrowly missed the significance threshold (standardised β = 0.004 ± 0.002, p = 0.068).

More generally, whether or not to titrate the punishments (and indeed the rewards) is an interesting experimental decision, which we think should be guided by the research question. In our case, we were interested in individual differences in reward/punishment learning and sensitivity and their relation to anxiety, so variation in how aversive the sounds affected approach-avoidance decisions was an important aspect of our design. In studies where the aim is to understand more general processes of how humans act under approach-avoidance conflict, it may be better to tightly control the salience of reinforcers.

Ultimately, the best test of the causal role of anxiety on avoidance, and against the hypothesis that our results were driven by spurious volume control effects, would be to run within-subjects anxiety interventions, where these volume effects are naturally accounted for. This will be an important direction for future studies using similar measures. We have added a paragraph in the ‘Discussion’ section on this point:

Relatedly, participants had some control over the intensity at which the punishments were presented, which may have driven our findings relating to anxiety and putative mechanisms of anxiety-related avoidance. Sensitivity analyses showed that our finding that anxiety is positively associated with avoidance in the task was robust to individual differences in self-reported punishment unpleasantness, whilst the mediation effects were not. Future work imposing better control over the stimuli presented, and/or using within-subjects designs will be needed to validate the role of reward/punishment sensitivities in anxiety-related avoidance.

Although the procedure and findings reported here remain valuable to the field, claims of novelty including its translational potential are perhaps overstated. This study complements and sits within a much broader literature that investigates roles for aversion and cognitive traits in approach-avoidance decisions. This includes numerous studies that apply reinforcement learning models to behavior in two-choice tasks with latent probabilities of reward and punishment (e.g., see doi: 10.1001/jamapsychiatry.2022.0051), as well as other translationally-relevant paradigms (e.g., doi: 10.3389/fpsyg.2014.00203, 10.7554/eLife.69594, etc).

We agree with the reviewer that our approach builds on previous work in reinforcement learning, approach-avoidance conflict and translational measures of anxiety. Whilst there are by now many studies using two-choice learning tasks with latent reward and punishment probabilities, our main, and which we refer to as ‘novel’, aim was to bring these fields together in such a way so as to model anxiety-related behaviour.

We note that we do not make strong statements about whether these effects speak to traits per se, and as Reviewer 1 notes, the evidence from our study suggests that the present measure may be better suited to assessing state anxiety. While computational model parameters can and are certainly often interpreted as constituting stable individual traits, a more simple interpretation of our findings may be that state anxiety is associated with a momentary preference for punishment avoidance over reward pursuit. This can still be informative for the study of anxiety, especially given the notion of a continuous relationship between adaptive/state anxiety and maladaptive/persistent anxiety.

Having said that, we agree with the underlying premise of the reviewer’s point that how the measure relates to trait-level avoidance/inhibition measures will be an interesting question for future work. We appreciate the importance of using tasks such as ours and those highlighted by the reviewer as trait-level measures, especially in computational psychiatry. We have now included a discussion on the potential roles of cognitive/motivational traits, in line with the reviewer’s recommendation – briefly, we have included the suggested references by the reviewer, discussed the measure’s potential relevance to cognitive/motivational traits, and direct interested readers to the broader literature. Please see below for details.

Reviewer #3 (Recommendations For The Authors):

As stated in the public review, punisher unpleasantness and its relationship to key findings (including for retest) should be reported and discussed.

We signpost readers to our new analyses, incorporating unpleasantness ratings into the statistical models, from the main manuscript as follows:

Since participants self-determined the volume of the punishments in the task, and therefore (at least in part) their aversiveness, we conducted sensitivity analyses by accounting for self-reported unpleasantness ratings of the punishment (see the Supplement). Our finding that anxiety impacts approach-avoidance behaviour was robust to this sensitivity analysis (p < 0.001), however the mediating effect of the reward-sensitivity sensitivity index was not (p > 0.1; see Supplement section 9.9 for details).

We reproduce the relevant section from the Supplement below. Overall, we found that the effect of anxiety on choices (via its interaction with punishment probability) remained significant after accounting for unpleasantness, however the mediating effect of reward-punishment sensitivity was no longer significant when unpleasantness ratings were included in the model. As noted above, unpleasantness ratings are not a perfect measure of self-imposed sound volume, and indeed punishment sensitivity is essentially a computationally-derived measure of unpleasantness, which makes it difficult to interpret the mediation model which contains both of these measures. However, since we found that anxiety affected choice over and above and effects of self-imposed sound volume (using unpleasantness ratings as a proxy measure), we argue that the task still holds value as a model of anxiety-related avoidance.

[Supplement Section 9.9: Sensitivity analyses of punishment unpleasantness]

Distribution of unpleasantness

The punishments were rated as unpleasant by the participants, on average (discovery sample: mean rating = 31.1 [scored between 0 and 50], SD = 13.1; replication sample: mean rating = 32.1, SD = 12.7; Supplementary Figure 10).

Approach-avoidance hierarchical logistic regression model

We assessed whether approach and avoidance responses, and their relationships with state anxiety, were impacted by punishment unpleasantness, by including unpleasantness ratings as a covariate into the hierarchical logistic regression model. Whilst unpleasantness was a significant predictor of choice (positively predicting safe option choices), all significant predictors and interaction effects from the model without unpleasantness ratings survived (Supplementary Figure 11). Critically, this suggests that punishment unpleasantness does not account for all of the variance in the relationship between anxiety and avoidance.

Mediation model

When unpleasantness ratings were included in the mediation models, the mediating effect of the reward-punishment sensitivity index did not survive (discovery sample: standardised β = 0.003 ± 0.003, p = 0.416; replication sample: standardised β = 0.004 ± 0.003, p = 0.100; Supplementary Figure 12). Pooling the samples resulted in an effect that narrowly missed the significance threshold (standardised β = 0.004 ± 0.002, p = 0.068).

Test-retest reliability of unpleasantness

The test-retest reliability of unpleasantness ratings was excellent (ICC(3,1) = 0.75), although participants gave significantly lower ratings in the second session (t56 = 2.7, p = 0.008, d = 0.37; mean difference of 3.12, SD = 8.63).

Reliability of other measures with/out unpleasantness

To assess the effect of accounting for unpleasantness ratings on reliability estimates of task performance, we extracted variance components from linear mixed models, following a standard approach (Nakagawa et al., 2017) – note that this was not the method used to estimate reliability values in the main analyses, but we used this specific approach to compare the reliability values with and without the covariate of unpleasantness ratings. The results indicated that unpleasantness ratings did not have a material effect on reliability (Supplementary Figure 14).

We discuss the findings of these sensitivity analyses in the ‘Discussion’ section, as follows:

Relatedly, participants had some control over the intensity at which the punishments were presented, which may have driven our findings relating to anxiety and putative mechanisms of anxiety-related avoidance. Sensitivity analyses showed that our finding that anxiety is positively associated with avoidance in the task was robust to individual differences in self-reported punishment unpleasantness, whilst the mediation effects were not. Future work imposing better control over the stimuli presented, and/or using within-subjects designs will be needed to validate the role of reward/punishment sensitivities in anxiety-related avoidance.

Introduction and discussion should spend more time relating the task and current findings to existing procedures and findings examining individual differences in avoidance and cognitive/motivational correlates.

We thank the reviewer for the opportunity to expand on the literature. Whilst there are numerous behavioural paradigms in both the human and non-human literature that involve learning about rewards and punishments, our starting point for the introduction was the state-of-the-art in translational models of approach-avoidance conflict models of anxiety. Therefore, for the sake of brevity and logical flow of our introduction, we have opted to bring in the discussion on other procedures primarily in the ‘Discussion’ section of the manuscript.

We have now included the reviewer’s suggested citations from their ‘Public Review’ as follows:

Since we developed our task with the primary focus on translational validity, its design diverges from other reinforcement learning tasks that involve reward and punishment outcomes (Pike & Robinson, 2022). One important difference is that we used distinct reinforcers as our reward and punishment outcomes, compared to many studies which use monetary outcomes for both (e.g. earning and losing £1 constitute the reward and punishment, respectively; Aylward et al., 2019; Jean-Richard-Dit-Bressel et al., 2021; Pizzagalli et al., 2005; Sharp et al., 2022). Other tasks have been used that induce a conflict between value and motor biases, relying on prepotent biases to approach/move towards rewards and withdraw from punishments, which makes it difficult to approach punishments and withdraw from rewards (Guitart-Masip et al., 2012; Mkrtchian et al., 2017). However, since translational operant conflict tasks typically induce a conflict between different types of outcome (e.g. food and shocks/sugar and quinine pellets; Oberrauch et al., 2019; van den Bos et al., 2014), we felt it was important to implement this feature. One study used monetary rewards and shock-based punishments, but also included four options for participants to choose from on each trial, with rewards and punishments associated with all four options (Seymour et al., 2012). This effectively requires participants to maintain eight probability estimates (i.e. reward and punishment at each of the four options) to solve the task, which may be too difficult for non-human animals to learn efficiently.

We have also included a discussion on the measure’s potential relevance to cognitive/motivational traits as follows:

Finally, whilst there is a broad literature on the roles of behavioural inhibition and avoidance tendency traits on decision-making and behaviour (Carver & White, 1994; Corr, 2004; Gray, 1982), we did not replicate the correlation of experiential avoidance and avoidance responses or the reward-punishment sensitivity index. Since there were also no significant correlations across task performance indices and clinical symptom measures, our findings suggest that the measure may be more sensitive to behaviours relating to state anxiety, rather more stable traits. Nevertheless, how performance in the present task relates to other traits such as behavioural approach/inhibition tendencies (Carver & White, 1994), as has been found in previous studies on reward/punishment learning (Sharp et al., 2022; Wise & Dolan, 2020) and approach-avoidance conflict (Aupperle et al., 2011), will be an important question for future work.

We also now direct readers to a recent, comprehensive review on applying computational methods to approach-avoidance behaviours in the ‘Introduction’ section:

A fundamental premise of this approach is that the brain acts as an information-processing organ that performs computations responsible for observable behaviours, including approach and avoidance (for a recent review on the application of computational methods to approach-avoidance conflict, see Letkiewicz et al., 2023).

I am curious why participants were excluded if they made the same response on 20+ consecutive trials. How does this represent a cut-off between valid versus invalid behavioral profiles?

We apologise for the lack of clarity on this point in our original submission – this exclusion criterion was specifically if participants used the same response key (e.g. the left arrow button) on 20 or more consecutive trials, indicating inattention. Since the left-right positions of the stimuli were randomised across trials, this did not exclude participants who repeatedly chose the same option frequently. However, as we show in the Supplement, this, along with the other exclusion criteria, did not affect our main findings.

We have now clarified this as follows:

… we excluded those who responded with the same response key on 20 or more consecutive trials (> 10% of all trials; 4%/6% in discovery and replication samples, respectively) – note that as the options randomly switched sides on the screen across trials, this did not exclude participants who frequently and consecutively chose a certain option.

12:3 Those who are wi se[a] will shine like the brightness of the heavens, and those who lead many to righteousness, like the stars for ever and ever.

• https://www.americamagazine.org/politics-society/2020/05/08/its-time-rethink-electoral-college

• https://www.npr.org/sections/itsallpolitics/2011/12/20/144016912/we-the-people-npr-readers-would-ratify-four-new-amendments

• https://www.americamagazine.org/politics-society/2020/05/08/its-time-rethink-electoral-college

• https://www.npr.org/sections/itsallpolitics/2011/12/20/144016912/we-the-people-npr-readers-would-ratify-four-new-amendments

• https://constitutioncenter.org/blog/vote-now-an-amendment-to-end-the-electoral-college

• https://www.nytimes.com/2020/02/09/opinion/letters/electoral-college.html

• https://www.latimes.com/opinion/readersreact/la-ol-le-electoral-college-20180904-story.html

you are offline

• https://slate.com/news-and-politics/2014/05/amending-the-constitution-is-much-too-hard-blame-the-founders.html

we the people rise again

• https://slate.com/news-and-politics/2012/06/fix-the-constitution-amending-by-national-referendum.html

safe souls, safe fu

• https://slate.com/news-and-politics/2012/06/fixing-the-constitution-protecting-informational-privacy.html

---

• https://slate.com/news-and-politics/2020/05/new-reconstruction-constitution-democracy.html

We the People of Slate ...

The U.S. Constitution, as you [mighta been, shoulda "come" on ... its someday] rewrϕte it.

"Politicians talk about the Constitution as if it were as sacrosanct as the Ten Commandments [interjection: spec. it is actually almost exactly related!]. But the document itself invites change and revision. What if the president served only one six-year term instead two four-year terms? What if your state's population determined how many senators represent it? What if the Constitution included a right to health care? We asked legal scholars and Slate readers to cross out what they didn't like in the Constitution and pencil in their hearts' desires. Here's what the document would look like with their best ideas."

• Slate: u_s_constitution as_rewritten by_slate_legal_experts_and_readers

多也了了夕 "with a ~~wand~~ of scheffilara, 并#亦太 he begins ... "I am now on the Staff of Menelaus, the Spears of Longinus and Lancelot; and the name "Mosche ex Nashon."

Logically the recent mentions of Gilgamesh and the simultaneous 同時 overlaping 場道 of the eventual link between the famous ruling of Solomon on the separation of babies and mothers and waters and land ... to a story of many "two cities" that culminates in a cultural or societal or "evolutionary" link to Sodom and Gomorrah and the city-state of Babylon (and it's Hanging Gardens) and also of course to Paris and Troy and "Masstodon" and city-states [ciudadestado] and perhaps planet-cities; from Cambridge to Cambridge across the "Cable" to see state to "London" ... recently I called it "the city of realms" ... I started out logically intending to link "game theory" and John Nash to the mathematical story of Sputnik and a revival of American physics; but in my usual way of rambling into the woods [I mean neighborhood] of stream of consciousness ... turned into a premonitory discourse of "two cities" and how sometimes even things as obvious as the number of letters in the word "two" don't do a good enough job of conveying ... how and/or why one is simply never enough, and two isn't much better--but in the end a circle ... is drawn; the perfect circle in our imaginary mathematical perfection ... I see a parted "line" in the letter pronounced "tea" (and beginning that word); and two "vee" (pron. of "v") symbols joined together in a word we pronounce as "double-you" ... and symbolically because I know "V" is the Roman Numeral for 5 (five) and I know not how to multiply in Roman numerals--

It's important to pause; here. I am going to write a more detailed piece on "the two cities" as I work through this maze like crossroads between "them" and "demo..." ... here demorigstrably I am trying to fuse together an evolutionary change in ... lit. biological evolution as well as an echelon leap forward in "self-government" ... in a place where these two things are unfathomable and unspokenly* connected.

• https://www.google.com/search?q=prometheuslocke+%2Bsite%3Agodlikeproductions.com

• "Silence is betrayal" -MLK

To a question on the idiom; is Bablyon about "the law" or "of the land of Nod?"

"What is democracy" ... the song, Metallica's "ONE" echoes and repeats; as we apparently scrive together the word "THEM" ... I question myself ... if Babylon were the capital city of some mythical Nation of Time ... if it were the central "turning point" of Sheol; ... >|<

Can you not see that in this place; in a world that should see and does there is a gigantic message proving that we are not in reality and trying to show us how and why that's the best news since ... ever---that it's as simple as conjoining "the law of the land" with a basic set of rules that automatically turn Hell into something so much closer to Heaven I just do not understand---why we cant stand up together and say "bullets will not kill innocent children" and "snowflakes will not start avalanches ...." that cover or bury or hide the road from Earth to Verital)e .... or from the mythical Valis to Tanis---or from Rigel to Beth-El ... "guess?"

## as "an easy" answer; I'm looking for a fusion of "law and land" that somehow remembers a "jok'er a scene" about "lawn" seats; and "where the girls are green;"

It's as simple as night and day; Heaven and Hell ... the difference between survival and--what we are presented with here; it's "doing this right"--that ends the Hell of representative democracy and electoral college--the blindness and darkness of not seeing "EXTINCTION LEVEL EVENT" encoded in these words and in our governments foundation ... *by the framers [not just of the USA; but English .. and every language] *

... is literally just as simple as "not caring" or thinking we are at the beginning of some long process--or thinking it will never be done--that special "IT" that's the emancipation of you and I.

Here words like "gnosis" and "gaudeamus" pair with my/ur "new ntersanding*" of the difference between Asgard and Medgard and really understanding our purpose here is to end "evil" ... things like "simulating disease and pain" (here, simulating meaning ... intentionally causing, rather than "gamifying away") and successfully linking the "Pillars of Hercules" to Plato's vision of Atlantis and the letter sequences "an" and "as" ... unlock a fusion of religion and mythology and "cryptographic truth" that connects "messianic" and "Christian" to "Roman" ... "Chinese" and "American" ... literally the key to the difference between the phrases "we are" and "we were" ....

in "sight" of "silicon" in simulation and Israel, Genesis, and "silence" ... trying to the raising of Asgardian enlightenment ... and seeing "simple cypher" connecting to "Norse" ...

and the "I AM THAT" surer than shit ... the intention and design of all religion and creation is to end "simulated reality" and also not seeing "SR" ... in Israel and Norse ... "for instance."

• https://www.google.com/search?q=%22I+AM%22+%22WE+ARE%22+%2Bsite%3Afromtaws
• "SOIS" a key--in two languages conjugated literally as both "I AM" and "WE ARE" simultaneously;
• Search: I know that if I am than so are you ... and it is because we have overcome .... something I truly cannot figure out, fathom, or believe ... was truly here before us--a spiralling series of failures ... speaking: to the heavens; but in secret and in action; "doing everything possible to succeed."

It's a simple linguistic concept; the "singularity" and the "plurality" of a simple word--"to be"--but it goes to the heart of everything that we are and everything that is around us. This is a message about understanding and preserving individuality as well as liberty; and literally seeing "ARXIV" and understanding "often" and failing to connect God and prescience to "IV" and the Fourth Amendment ... it's about blindness and ... "curing the blind instantly" ... and fathoming how and why this message has been etched into our entire history and and all religions and myths and music--to help us "to be THAT we" that actually "are responsible" for the end of Hell.

• I neglected to mention "Har-Wer" and "Tower of Babel" which are both related lingusitically, religiously and topically: "to who ..." and while we're on "four score and [seven years from now]" seeing the fourth "living thing" in Eden and it's (the name, Abel) connection to Babel and Abraham Lincoln; slavery and ... understanding we live in a place where the history of the United States also, like Monoceros and "Neil Armstrong's first step" are a time shifted ... overlayed map to achieving freedom ... it's about becoming a father-race ... and actually "doing" the technological steps required to "emancipate the e's of 'me&e'" and survive in exo-planetary space---

it might be as simple as adding "because we did this" here and now; and having it be something we are truly proud of .... forevermore™ ... for certain in the heart of this story about cyclicality and repetition of error--its not because we did "this" or something over and over again; it's about changing "the problem" and then helping others to also overcome ... "things like time travel ... erasing speech" --- however that happenecl.

• I also failed to mention that "I am in Hell" ... as in this world is hellacious to me; in an overlay with the Hellenic period and this message that we are in the Trojan Horse ... a small gem .... "planet" truly is the Ark of the Covenant---and it's the simple understanding that "reality is hell" is to "living without air conditioning and plumbing is hell" just as soon as you achieve ... "rediscovering" those things---

• I can't figure out why I am the only person screaming "this is Hell." That's also, Hell.

... but recently suggested an old joke about "there being 10 kinds of people in the world (obv an anti-tautology and a tautology simultaneously)" only after that brief bit of singularity and duality mentioning the rest of the joke: "those that understand binary and those that don't know how to base convert between counting with two hands and counting with only an 'on and off.'" It's not obvious if you aren't trying to figure it out, I suppose; but 10 is decimal notation for "kiss" and the "often" without "of" ... and binary notation for the decimal equivalent of "2." A long long time ago in a state that simply non-randomly ties to the heart of the name of our galaxy ... I was again thinking of the "perfect imperfections" of things like saying "three equals one equals one" (which, of course was related to the Holy Trinity and it's "prescient/anachronistic Adamic presence encoded in the name Ab|ra|ha|m" which means "father of a great multitude") ... I brought that one back in the last few months; connecting the letter K and in this "logos-rythmic" tie to the "base of a number system" embellish the truth just a bit and suggest a more accurate rendition of the original [there is no such thing as equality, "is" of separate objects--as in no two snowflakes are the same unless they are literally the same one; true of ancient weights and with the advent of (thinking about) time no two "planets" are the same even if they're the exact same one--unless it's at a fixed moment in time.

• This name may be viewed either as meaning "father of many" in Hebrew or else as a contraction of ABRAM (1) and הָמוֹן (hamon) meaning "many, multitude". The biblical patriarch Abraham was originally named Abram but God changed his name (see Genesis 17:5).

• https://en.wikipedia.org/wiki/Yeshua#Yeshua,_Yehoshua,_and_Yeshu_in_the_Talmud

K=3:11 ... to a handle on the music, the DHD of the gate and the *ring of David's "sling" ...

---and that's a relationship of "3 is to 11" as [the SAT style "analogy)]y" as a series of alpha, two mathematic, and two numeric symbols ... may only tie in my mind alone to the books of Genesis and Matthew and the phrase "chapter and verse" and to the stories of Lot and Job ... again in Genesis and the eponymous "Book of Job." So ... "tying up loose ends one 10b [III] iv. " as it appears I've taken it upon myself to call a Job and suggest is my "Lot in life [x]i* [3]"

• I worry sometimes that important things are missing, or will disappear---for instance Mirriam Webster, which is a "canonical/standard dictionary) should probably have an entry for "lot in life" non-idiomatically as "granny apples to sour apples" as

2 MANY ALSO ICI; 1twoⅱ ... following in Mitnick's bold introductory word steps; the curve and the complement ... the missiles and the canoes; the line and the blank space ... "supposedly two examples of two kinds, which could be three not nothings ... Today I write about something monumental; as if as important as the singularity depicted in Arthur C. Clarke's 2001 "A Space Odyssey" ... and remember a day when I thought it very novel and interesting to see the words "stillborn and yet still born" connected in a single piece of writing to "Stillwater and yet still water" ... today adding in another phrase noting the change wrought only by one magical single "space" (also a single capital letter; and a third phrase): "block chains with a great blockchain."

• http://www.goodmath.org/blog/2015/07/21/arabic-numerals-have-nothing-to-do-with-angle-counting/

• https://gizmodo.com/no-this-viral-image-does-not-explain-the-history-of-ar-1719306568

• https://en.wikipedia.org/wiki/Chinese_word_for_%22crisis%22

• https://dictionary.hantrainerpro.com/chinese-english/translation-ji_howmany.htm
• https://dictionary.hantrainerpro.com/chinese-english/translation-duo_many.htm

• https://en.wikipedia.org/wiki/Euripides, Iphigenia in Aulis or Iphigenia at Aulis[1] (Ancient Greek: Ἰφιγένεια ἐν Αὐλίδι, Iphigeneia en Aulidi; variously translated, including the Latin Iphigenia in Aulide) is the last of the extant works by the playwright Euripides. Written between 408, after Orestes, and 406 BC, the year of Euripides' death, the play was first produced the following year[2] in a trilogy with The Bacchae and Alcmaeon in Corinth by his son or nephew, Euripides the Younger,[3] and won first place at the City Dionysia in Athens.

• The play revolves around Agamemnon, the leader of the Greek coalition before and during the Trojan War, and his decision to sacrifice his daughter, Iphigenia, to appease the goddess Artemis and allow his troops to set sail to preserve their honour in battle against Troy. The conflict between Agamemnon and Achilles over the fate of the young woman presages a similar conflict between the two at the beginning of the Iliad. In his depiction of the experiences of the main characters, Euripides frequently uses tragic irony for dramatic effect.

J.K. Rowling spurred just this past week a series of explanations about just exactly what is a blockchain coin worth ... and why is it so; her final words on the subject (artistic liberty taken, obviously not the last she'll say of this magic moment) "I don't think I trust this."

Taken directly from an off the cuff email to ARXM titled: "Slow the S is ... our Hypothes.is"

I imagine I'll be adding some wiki/ipfs stuff to it--and try to keep it compatible; the design and layout is almost exactly what I was dreaming about seeing--as a "first rough draft product." Lo, and behold. It's been added to the many places I host my tome; the small compilation of nearly every important email that has gone out ... all the way back to the days of the strange looking Margarita glass ... that now very much resembles the "Cantonese character 'le'" which I've come to associate with a "handle" on multiple corners of a room--something like an automatic coat rack conveyor belt connecting different versions of "what's in the box." I'm planning on using that symbol 了 to denote something like multiple forks of the same page. Obviously I'm thinking forward to things like "the Transhumaist Chain Party" (BDSM, right?)'s version of some particular piece of legislation, let's say everything starts with the sprawling "bulbing" of "Amendment M" ideas and specific verbiage ... and then we'll of course need some kind of new git/subversion/cvs style version control mechanism to merge intelligently into something that might actually .... really should ... make it into that place in history--the first constitutional amendment ratified by a "Continental Congress of All People" ... but you could also see it as an ongoing sort of forking of something like the "wikipedia page" on what some specific term, say "technocracy" means, and how two parties might propagandize and change the meaning of such thing; to suit the more intelligent and wise times we now live in. For instance, we might once have had a "democracy" and a "democractic" party that had some Anarchist Cook Book version of the history of it ending in something like Snipes and Stallone's "DEMOLITION MAN."

Just kidding, we all know "democracy" has everything to do with "d is cl ... and not th" ... to be the them that is the heart of the start of the first true democracy. At least the first one I've ever seen, in my old "to a republic" ... style. As it is you can play around with commenting and highlighting and annotating all the stuff I've written and begged and begged for comments on--while I work on layering the backend to to perma-store our ideas and comments on both a blockchain (probably a new one; now that i've worked a little with ethereum) with maybe some key-merkle-tree-walk-search stuff etched into the original Rinkeby ... and then of course distributed data in the "public owned and operated" IPFS. To be clear, I plan on rewriting the backend storage so that we will have a permanent record of all comments; all versions of whatever is being commented on; and changes/revisions to those documents--sort of turning the web into a massive instant "place of collaboration, discussion, and co-authoring" ... if you use the wonderful LEGO pieces that have been handed to us in ideas from places like me, lemma--dissenter, and of course hypothes.is who has brought you and i such a polished and nice to look at "first draft" of something like the living Constitution come repository of all human knowledge. I do sort of secretly wich they would have called this project something like "annotating and reflecting (or real or ...) knowledge" just so the movement could have been called ARK. ... or something .... but whatever join the "calling you a reporter" group or ... "supposedly a scientist?"

NOIR INgR .. I CITE SITE OF ENUDRICAM; a rekindling of the dream of a city appearing high above in the sky, now with a boldly emblazened smiling rainbow and upsidown river ... specifically the antithesis of "angel falls," there's a lagoon too--actually a chain of several ponds underneith the floating rock ... and in some versions of this waking dream there are rings around the thing; you might imagine an artificial set of centripetal orbitals something like a fusion of the ring Eslyeum and the "Six-Axis ride" of the JKF Center's "Spacecamp." I write as I dream, and though I cannot for certain explain exactly how; it's become a strong part of my mythology that this spectacular rendition of "what ends the silence" has something to do with the magical delivery of "a book" ... something not of this Earth but an unnatural thing; one I've dreamt of creating many times. This book is something like the DSM-IV and something like a Merck diagnostic manual; but rather than the old antiquated cures of "the Norse Medgard" this spectacle nearly "itsimportant" autoprints itself and lands on something like every doorpost; what it is is a list of reasons why "simply curing all disease" with no explanation and no conversation would be a travesty of morality--how it would render us half-blind to the myriad of new solutions that can come from truly understanding why "ITIS" to me has become a kind of magical marker: an "it is special" as in, it's cure could possibly solve a number of other problems.

Through that missing "o," English on the ball, we see a connection between a number of words that shine bright light including Exodus itself which means "let there be light," the word for Holy Fire and the Burning Bush.. .reversed to hSE'Ah, and a story about the Second Coming parting our holy waters.

This answer connects the magical Rod's of Aaron in Exodus and the Iron Rod of Jesus Christ to the Sang Rael itself... in a fusion that explains how the Periodic Table element for Iron links not just to Total Recall and Mars, but also to this key

my dream of what the first day of the Second Coming might be like; were the Rod of Christ... in the right hands. In a story that also spans the Bible, you might understand better how stone to bread and your input make all the difference in the world between Heaven and Adam's Hand. Once more, what do you think He ....

Since the very earliest days of this story, I have asked for better for you, even than see

Nearly all of the original parts of the original "post-origination dream" remain intact; there's a walkway that magically creates new paths and "attractions" based on where you walk, something like an inversion of the artificial intelligence term "a random walk down a binary tree" ... for instance going left might bring you to the Internet Cafetornaseum of the Earl of Sandwich; and going to the right might bring you to the ICIMAX/Auditorium of Science and Discovery--there's a walkway to "Magical GLAS D'elevators" that open a special "instantiation" of the Japan Room of the Potter and the Toolmaker ... complete with a special [second level and hidden staircase] Pool of Bethesdaibo verily delivering something like youth of mind and body ... or at least as close to such a thing as a sip of Holy Water or Ambrosia or a dip in the pool of Coccoon and Ponce De'Leon could instantly bring ... to those that have seen Jupiter Ascending ... the questions of "nature versus nurture" and what it means to be "old and wise" and "young at heart" truly mean---

• https://www.youtube.com/watch?v=M8CyN1awWls
• https://link.springer.com/chapter/10.1057/9780230366688_16
• https://www.youtube.com/watch?v=YDo5zvYNn3A

Somewhere between the outdoor rafting ride and the level with the special "ballroom of the ancient gallery" ... perhaps now being named or renamed or recalled as something about "Face [of] the Music" lies a magical "mini-maize" ... a look at a mock-up (or #isitit) of Merlink and Harthor's "round table" that displays a series of ... (at least to me) magical appearing holographic displays and controls that my dreams have stolen from Phillip K. Dick's Minority Report and something of what I hope Microsoft's Dynamics/Hololens/Surface will become---a series of short "focus groups" .... to guage and discuss the information in the "CITIES-D5AM-MERCK" ... how to end world hunger and nearly all disease with the press of a magical buzzer--castling churches to something like "political-party-town-hall-meeting centers" and replacing jails and prisons and hospitals with something like the "Hospitalier's PRIDE and DOJOY's I practiced "Kung-fun-dance" ... a fusion of something like a hotel and a school that probably looks very much like a university with classrooms and dorms and dining hall's all fit into a single building. I imagine a series of 2 or 3 "room changes" as in you walk from the one where you get the book and talk about it ... to the one where you talk about "what everyone else said about it" and maybe another one that actually connects you to other people with something like Facebook's Portal; the point of the whole thing to really quickly "rubber stamp" the need for an end to "bars in the sky" nonalcoholic connotation--as in "overcoming the phrase the sky is the limit" and showing us the need for a beacon of glowing hope fulfilled--probably actually the vision of a holographic marker turning into actual rings around the single moon of Earth, the focus of the song annoucing the dawn of the age of Aquarius---

It might lead us also to Ceres; and another set of artificial rings, or to Monoceros and a rehystorical understanding of the birthplace and birthing of the "river roads" that bridge the "space gaps" in the galaxy from our "one giant leap for mankind" linking the Apollo moon landing to the mythological connection to the sun; and connecting how the astrological charts of the ancients might detail a special kind of overlapping--the link between Earth's SOL and something like Proxima or Alpha Centauri; and how that "monostar bridge" might overlap to Orion and from there through Sagitarius and the center of the Milky Way ... all the way to Andromeda and more dreams of being in a place where there's a map to a tri-galactic system in the constellation Cancer and a similar one in Leo ... and just incase you haven't noticed it--a special marker here, I thought to myself it might be cool to "make an acronymic tie to Monoceros" and without even thinking auto-wrote Orion (which was the obvious constellation next to Monoceros, in the charts) and then to Sagitarrius; which is the obvious ... heart of our astrological center and link to "other galaxies."

----I've dreamt or scriven or reguessed numerous times how the Milky Way's map to an "Atlas marked through time by the ages and the ancients" might tie this place and this actual map to the creation of the railways between stars to the beginning and the end of time and of course to this message that links it all to time travel. There's a few "guesses" I've contemplated; that perhaps the Milky Way chart is a metal-cosmic or microcosmic map to the dawn of time in the galactic vision of ... just after the big bang; or it might tie to a map of something like the unthinkable--a civilization that became so powerful it was able to reverse the entropy of "cosmic expansion" and reverse the thing Asimov wrote of in "The Last Question" as the end of life and the ability to survive basically due to "heat loss."

"The Last Question." (And if you read two, why not "The Last Answer"?). Find these readings added to our collection, 1,000 Free Audio Books: Download Great Books for Free.

• https://archive.org/details/texts
• http://zlibraryexau2g3p.onion.pet/

Looking for free, professionally-read audio books from Audible.com, including ones written by Isaac Asimov?

* all "asterisks" in the abovə document denote a sort of Adamic unspoken relationship between notations and meanings; here adding the "Latin word for three" and source of the phrase "t.i.d." (which is doctor/pharmacy latin for "three times a day") where the "t" there is an abbreviation of "ter" ... and suppose the link between K and 11 and 3 noting it's alphanumeric position in the English alphabet as the 11th letter and only linking cognitively to three via the conversion between hex, and binarryy ... aberrative here is the overlapping "hakkasan" style (or ZHIV) lack of mention of the answer in "state of Kansas" and the "citystate of Slovakia" as described in the ICANN document linked [in] the related subsection or slice of the word "binarry" for the state of India. Tetris could be spelled with the addition of only a single letter [in] "tea"---the three letters "ris" are the hearts of the words "Christ" and "wrist" [and arguably of Osiris where you also see the round table character of the solar-system/sun glyph and the chemical element for The Fifth Element (as def. by i) via "Sinbad" and "Superman." The ERIS Free Network should also be mentioned here in connection with the IRC network I associate in the place between skipping stones and sacred hearts defined by "AOL" and "Kdice" in my life. In the lexicon of modern HTML, curly braces are generally relative to "classes" and "major object definitions (javascript/css)" while square brackets generally only take on computer-interpreted meaning in "Markdown" which is clearly (by definition, by this character set "[]") a superset (or at least definately not a subset) of HTML.

Dr. Will Caster (Johnny Depp) is a scientist who researches the nature of sapience, including artificial intelligence. He and his team work to create a sentient computer; he predicts that such a computer will create a technological singularity, or in his words "Transcendence". His wife, Evelyn (played by Rebecca Hall), is also a scientist and helps him with his work.

Following one of Will's presentations, an anti-technology terrorist group called "Revolutionary Independence From Technology" (R.I.F.T.) shoots Will with a polonium-laced bullet and carries out a series of synchronized attacks on A.I. laboratories across the country. Will is given no more than a month to live. In desperation, Evelyn comes up with a plan to upload Will's consciousness into the quantum computer that the project has developed. His best friend and fellow researcher, Max Waters (Paul Bettany), questions the wisdom of this choice, reasoning that the "uploaded"

Just from my general understanding and memory "st" is not ... to me (specifically) an abbreviation of "state" but "ste" is a U.S. Postal code (also "as I understand it") for the name of a special room or set of rooms called a "suite" and in Adamic "connotation" I sometimes read it as "sweet" ... which has several meanings that range from "cool" to "a kind of taste sensation" to "easy to sway or fool."

If you asked me though, for instance if "it" was an abbreviation or shorthand notation or acronym for either "a United state" or "saint" ... you'd be sure.

While it's clear from studying linguistic cryptography ... (If I studied it a little here and some there, its also from the "universal translator of Star Trek") and the personal understanding that language is a kind of intelligent code, and "any code is crackable" ... that I caution here that "meaning" and "face value" often differ widely and wildly ... even in the same place or among the same group of people ... either varying over time or heritage.

Menelaus, in Greek mythology, king of Sparta and younger son of Atreus, king of Mycenae; the abduction of his wife, Helen, led to the Trojan War. During the war Menelaus served under his elder brother Agamemnon, the commander in chief of the Greek forces. When Phrontis, one of his crewmen, was killed, Menelaus delayed his voyage until the man had been buried, thus giving evidence of his strength of character. After the fall of Troy, Menelaus recovered Helen and brought her home. Menelaus was a prominent figure in the Iliad and the Odyssey, where he was promised a place in Elysium after his death because he was married to a daughter of Zeus. The poet Stesichorus (flourished 6th century BCE) introduced a refinement to the story that was used by Euripides in his play Helen: it was a phantom that was taken to Troy, while the real Helen went to Egypt, from where she was rescued by Menelaus after he had been wrecked on his way home from Troy and the phantom Helen had disappeared.

• https://www.britannica.com/topic/Menelaus-Greek-mythology

This article is about the ancient Greek city. For the town of ancient Crete, see Mycenae (Crete). For the hamlet in New York, see Mycenae, New York.

Μυκῆναι, Μυκήνη

The Lion Gate at Mycenae, the only known monumental sculpture of Bronze Age Greece

37°43′49"N 22°45′27"ECoordinates: 37°43′49"N 22°45′27"E

This article contains special characters. Without proper rendering support, you may see question marks, boxes, or other symbols.

Mycenae (Ancient Greek: Μυκῆναι or Μυκήνη, Mykēnē) is an archaeological site near Mykines in Argolis, north-eastern Peloponnese, Greece. It is located about 120 kilometres (75 miles) south-west of Athens; 11 kilometres (7 miles) north of Argos; and 48 kilometres (30 miles) south of Corinth. The site is 19 kilometres (12 miles) inland from the Saronic Gulf and built upon a hill rising 900 feet (274 metres) above sea level.[2]

In the second millennium BC, Mycenae was one of the major centres of Greek civilization, a military stronghold which dominated much of southern Greece, Crete, the Cyclades and parts of southwest Anatolia. The period of Greek history from about 1600 BC to about 1100 BC is called Mycenaean in reference to Mycenae. At its peak in 1350 BC, the citadel and lower town had a population of 30,000 and an area of 32 hectares.[3]

3. Chew 2000, p. 220; Chapman 2005, p. 94: "...Thebes at 50 hectares, Mycenae at 32 hectares..."

• https://en.wikipedia.org/wiki/Clymene_(mythology)

Melpomene (/mɛlˈpɒmɪniː/; Ancient Greek: Μελπομένη, romanized: Melpoménē, lit. 'to sing' or 'the one that is melodious'), initially the Muse of Chorus, she then became the Muse of Tragedy, for which she is best known now.[1] Her name was derived from the Greek verb melpô or melpomai meaning "to celebrate with dance and song." She is often represented with a tragic mask and wearing the cothurnus, boots traditionally worn by tragic actors. Often, she also holds a knife or club in one hand and the tragic mask in the other.

Melpomene is the daughter of Zeus and Mnemosyne. Her sisters include Calliope (muse of epic poetry), Clio (muse of history), Euterpe (muse of lyrical poetry), Terpsichore (muse of dancing), Erato (muse of erotic poetry), Thalia (muse of comedy), Polyhymnia (muse of hymns), and Urania (muse of astronomy). She is also the mother of several of the Sirens, the divine handmaidens of Kore (Persephone/Proserpina) who were cursed by her mother, Demeter/Ceres, when they were unable to prevent the kidnapping of Kore (Persephone/Proserpina) by Hades/Pluto.

In Greek and Latin poetry since Horace (d. 8 BCE), it was commonly auspicious to invoke Melpomene.[2]

See also [AREXMACHINA]

• Muses in popular culture
• The Nine Muses

Flagstaff (/ˈflæɡ.stæf/ FLAG-staf;[6] Navajo: Kinłání Dookʼoʼoosłííd Biyaagi, Navajo pronunciation: [kʰɪ̀nɬɑ́nɪ́ tòːkʼòʔòːsɬít pɪ̀jɑ̀ːkɪ̀]) is a city in, and the county seat of, Coconino County in northern Arizona, in the southwestern United States. In 2018, the city's estimated population was 73,964. Flagstaff's combined metropolitan area has an estimated population of 139,097.

Flagstaff lies near the southwestern edge of the Colorado Plateau and within the San Francisco volcanic field, along the western side of the largest contiguous ponderosa pine forest in the continental United States. The city sits at around 7,000 feet (2,100 m) and is next to Mount Elden, just south of the San Francisco Peaks, the highest mountain range in the state of Arizona. Humphreys Peak, the highest point in Arizona at 12,633 feet (3,851 m), is about 10 miles (16 km) north of Flagstaff in Kachina Peaks Wilderness. The geology of the Flagstaff area includes exposed rock from the Mesozoic and Paleozoic eras, with Moenkopi Formation red sandstone having once been quarried in the city; many of the historic downtown buildings were constructed with it. The Rio de Flag river runs through the city.

Originally settled by the pre-Columbian native Sinagua people, the area of Flagstaff has fertile land from volcanic ash after eruptions in the 11th century. It was first settled as the present-day city in 1876. Local businessmen lobbied for Route 66 to pass through the city, which it did, turning the local industry from lumber to tourism and developing downtown Flagstaff. In 1930, Pluto was discovered from Flagstaff. The city developed further through to the end of the 1960s, with various observatories also used to choose Moon landing sites for the Apollo missions. Through the 1970s and '80s, downtown fell into disrepair, but was revitalized with a major cultural heritage project in the 1990s.

The city remains an important distribution hub for companies such as Nestlé Purina PetCare, and is home to the U.S. Naval Observatory Flagstaff Station, the United States Geological Survey Flagstaff Station, and Northern Arizona University. Flagstaff has a strong tourism sector, due to its proximity to Grand Canyon National Park, Oak Creek Canyon, the Arizona Snowbowl, Meteor Crater, and Historic Route 66.

PSANSDISL #LWDISP either without gas or seeing cupidic arroz in "thank you" or "allta, wild" ...

pps: a magnanimous decision ...

I stand here on the brink of what appears to be total destruction; at least of everything I had hoped and dreamed for ... for the last decade in my life which appears literally to span thousands of years if not more in the eyes of some other beholder. I spent several months in Kentucky telling a story of a post apocalyptic and post-cataclysmic delusion; some world where I was walking around in a "fake plane" something like a holodeck built and constructed around me as I "took a walk around the world" to ... it did anything but ease my troubled mind.

Recently a few weeks in Las Vegas, and a similar story; telling as I walked penniless down the streets filled with casino's and anachronistic taxi-cabs ... some kind of vision of the entirety of the heavens or the Earth or the "choir of angels" I think of when I echo the words Elohim and Aesir from mythology ... there with me in one small city in superposition; seeing what was a very well put together and interesting story about a "star port" Nirvane ... a place that could build cities into the face of mountains and half working monorails appearing in the sky---literally right before my eyes.

I suppose this is the place "post cataclysm" though I still have trouble understanding what it is that's actually about ... in my mind it connects to the words "we are losing habeas" echo'ed from the streets of Los Angeles in a more clear and more military voice than usual--as I walked block by block trying to evade a series of events that would eventually somehow connect all the way to the "outskirts of Orlando, Florida" in a place called Alhambra.

Apparently the name of a castle; though I wasn't aware of that until much later.

It doesn't feel at all like a "cataclysm" to me; I see no great rift--only a world filled with silent liars, people who collectively believe themselves to have stolen something--something gigantic--at least that's the best interpretation of the throws and impetus behind the thing that I and mythology together call Jormungandr. With an eye for "mythological connections" you could clearly see that name of the Great Serpent of Revelation connects to something like the Unseelie; the faeries of Gaelic lore. To me though this world seems still somewhat fluid, it's my entire life--moving from Plantation to a place where the whole of it might be Bethlehem and to "clear my throat" it's not hard to see here how that land of "coughs" connects to the Biblical land of Nod and to the "Adamically sieved" Snifleheim ... from just a little twist on the ancient Norse land most probably as close to Hel as anyone ever gets--or so I dream and hope---still today. It all looks so real and so fake at the same time; planned for thousands of generations, the culmination of some grand masterpiece story that certainly ties history and myth and reality into a twisted heap of "one big nothing, one big nothing at all."

I've tried to convey to the world how important I believe this place and this time to be--not by some choice of my own ... but through an understanding of the import of our history and the impact of having it be so obviously tuned and geared towards this specific time ... many thousands of years literally all focused on a single moment, on one day or one hour or even just a few years where all of that gets thrown down on the table as if some trump card has been played--and whether or not you fathom the same magnanimous statement or situation or position ... to me, I think it depends on whether or not you grew up in the same kind of way, believing our history to be so fixed and so difficult to change. I don't particularly feel like that's the "zeitgeist" of today; I feel like the children believe it to be some kind of game, and that it is such as easy thing to "sed" away or switch and turn into something else--another story, another purpose ... anyone's personal fantasy land come true.

I don't think that's the case at all, it's clearly a personal nightmare; and it's clearly one we've seen time and time again--though not myself--the Jesus Christ that is the same yesterday, today; and once again perhaps echoing "no tomorrow" never remembers or believes that we've "seen it all before" or that we've ever really gotten the point; the thing you present to me as "factual reality" is a sickness, it disgusts me; and I'd do anything to go back to the world "where I was so young, and so innocent" and so filled with starry-eyed hope that we were at the foot of something grand and amazing that would become an empire turned republic of the heavens; filling the stars ... with the kind of love for kindness and fairness that I once associated very strongly with the thing I still believe to be the American Spirit.

---

"Suddenly it changes, violently it changes" ... another song echoes through the ages--like the "words of the prophets dancing ((as light)) through the air" ... and I no longer even have a glimmer of hope that the thing I called the American People still exist; I feel we've been replaced by some broken container of minds, that the sky itself has become corrupt to the point that there's no hope of turning around this thing that I once believed with all my heart and all my mind was so obviously a "designed downward spiral" one that was---again--so obviously something of a joke, intended to be easy to bounce off a false bottom and springboard beyond "escape velocity" and beyond the dark waters of "nearest habitable star systems (being so very far away)" into a place where new words and new ideas would "soar" and "take flight."

Here though; I am filled with a kind of lonely sadness ... staring at what appears to be the same mistake(s) happening over and over again; something I've come to call "skipping stones in the pond of reality" and really do liken it to this thing that appears to be the new meaning of "days" and ... a civilization that spends absolutely no love or lust to enter a once sacred and holy place and tarnish it with their sick beliefs and their disgusting desires. You all ... you appear to be some kind of springboard to "bunt" forth yet another age or era of nothingness into the space between this planet and "none worth reaching" and thank God, out of grasp. Today, I'd condemn the entirety of this world simply for it's lack of "oathkeepers" and understanding of what the once hallowed words of Hippocrates meant to ... to the people charged and dharmically required to heal rather than harm.

It appears the place and time that was once ... at least destined to be the beginning of Heaven ... has become a "recurring stump" of some future unplanned and tarnished by many previous failed efforts and attempts to overcome this same "lack of conversation or care" for what it meant to be "humane" in a world where that was clearly set high aloft and above "humanity" in the place where they--where we were the best nature had to offer, the sanest, the kindest; the shining last best hope.

---

Today I write almost every day ... secretly thanking "my God" for the disappearance of my tears and the still small but bright hope that "Tearran" will one day connect the Boston Tea Party and the idea that "render to Caesar" and Robin of Loxley ... all have something to do with a re-ordering of society and the worth and import of "money" ... to a place that cares more for freedom from murder than it does ... "freedom from having to allow others to hear me speak." I hold back tears and emotions; not by conscious choice or ability but ... still with that strange kind of lucky awkward smile; and secretly not so far below the surface it's the hope of "a swift death" that ... that really scares me more than the automatons and mechanical responses I see in the faces of many drivers as they pass me on the street--the imagery of connecting it to the serpentine monster of the movie Beetlejuice ... something I just "assume" the world understands and ... doesn't seem to fear (either); as if Churchill had gotten it all wrong and backwards--the only thing you have to fear, is the loss of fear of "loss."

---

Here my crossroads---halfway between the city my son lives in and the city my parents live in--it's on making a decision on whether I should continue at all, or personally work on some kind of software project I've been writing about, or whether I should focus on writing about a "revolution" in government and society that clearly is ... "somewhat underway." In my mind it's obvious these things are all connected; that the software and the governance and the care of whether or not "Babylon" is remembered as a city of great laws and great change or a city of demons and depravity ... that these thi]ngs all hinge and congeal around a change in your hearts; hoping you will chose to be the beginning of a renaissance of "society and civilization" rather than the kings and queens of a sick virtual anarchy ... believing yourselves to have stolen "a throne of God" rather than to literally be the devastating and demoralizing depreciation of "lords and fiefdoms" to something more closely resembled by the time of the Four Horsemen depicted in Highlander.

These words intended to be a "forward" to yet another compliment of a ((nother installment of a partial)) chain of emails; whimsically once half-joking ... I called it the Great Chain of Revelation. The software too; part of the great chain, this "idea" that the blockchain revolution will eventually create a distributed and equal governance structure, and a rekindling of monetary value focused on "free and open collaboration" rather than "survival of the most unfit"--something society and civilization seem to have turned the "call of life" from and to ... literally just in the last few years as we were so very close to ... reaching beyond the Heaven(s).

I don't think its hard to imagine how a "new set of ground rules" could significantly change the "face of a place" -- make it something shiny and new or even on the other side of the coin, decayed or depraved. It's not hard to connect the kind of change I'm hoping for with "collision protection" and "automatic laws" to the (perhaps new, perhaps ... ancient) Norse creation story of the brothers of Odin: Vili and Ve.

It might be hard to see today how a new "kind of spiritual interaction" might be only a few "mouse clicks" away though--how it could change everything literally in a flash of overnight sensation ... or how it might take something like a literal flash of stardom (or ... on the other hand, something like totalitarian or authoritarian "iron fisting") to make a change like this "ubiquitious" or ... something like the (imagined in my mind as ... messianic) "ED" of storming through the cosmos or the heavens and turning something that might appear to be "free and perfect feeling" today into a universe "civlized overnight" and then ...

I wonder how long it would take to laud a change like that; for it to be something of a voluntary "reunderstanding" of a process ... to change the meaning of every word or every thought that connects to the process of "civilization" to recognize that something so great and so powerful has happened as to literally change the meaning of the word, to turn a process of civilization into something that had a ... "signta-lamcla☮" of forboding and then a magical staff struck into the heart of a sea and then ... and then the word itself literally changes to introduce a new "mid term" or "halfway point" in which a great singularity or enlightenment or change in perspective or understanding sort of acknowledges ...

that some "clear outside" force not only intervened on the behalf of the future and the people of our world but that it was uniquely involved in the whole of--

"waking up" tio a nu def of #Neopoliteran.

^Like the previous notation; the below text comes from an email previously sent; and while i stand behind things like my sanity, my words; and my continued and faithful attempt to speak and convey both a useful and helpful truth to the world---sometimes just a single day can make all the difference in the world.

Sometimes it's just a single moment; a flash or a comment about ^th@ blink of an eye" ... and I've literally just "thought up/had/experienced/transitioned thru" that exact moment. The lies standing between "communication" and either "cooperation" or .... some other kind of action have become more defined. More obvious. Because of this clarification; like a kind of "ins^tant* gnosis"

... search high and lo ... the depths all the way to above the heavens ...\ \ for a festive divorce ceremonial ritual ... that looks something like a bachelor party ':;]

--- @amrs@koyu.SPACe ... @suzq@rettiwtkcuf.social (@yitsheyzeus) May 22, 2020

I ... TERON;

Gjall are painting me into a corner here; and I don't see around it anymore--I don't see the light, and I don't see the point. I was a happy-go-lucky little kid in my mind; that's not "what I wanted to be" or what I wanted to present, it's who I was. I saw "Ashkenazi" and ... know I am one of those ... and I kind of understood that something horrible might have happened, or might happen here--and I kind of understand that crying smashing feeling of "to ash" that echoes through the ages in the potpourri songs about pockets full of Parker Posey .. and ancient Psalms about "from the ashes of Edom" we have come--and from that you can see the cyclical sickness of this ... place so sure it's "East of Eden" and yet gung-ho on barrelling down the same old path towards ash and towards Edom and towards ... more of Dave's "ashes to ashes dust to dust" and his "smoke clouds roll and symphony of death..." and few words of solace in a song called Recently that I imagine was fleeting and has recently come and gone--people stare, I can't ignore the sick I see.

I can't ignore his "... and tomorrow back to being friends" and all but wonder who among us doesn't realize it's "ash" and "gone" and "no memory of today" that's the night between now and ... a "tomorrow with friends" not just for me--but for all of you--for this place that snickers and pantomimes some kind of ... anything but "I'm not done yet" and "there's more ... vendetta ... and retribution to be had, Adam ... please come back in a few more of our faux-days." This is sickness; and happy-go-lucky Himodaveroshalayim really doesn't do much but complain about that word, the "sickle" and the tragic unavoidable ... ash of it all ... these days--you'd think we could "pull out" of this mess, turn another way; smile another day, but it seems there's only one way to get to that avenu in the mind of ... "he who must not know or be me."

---

I have to admit I found some joy in the epiphany that the hidden city of Zion and it's fusion with the Namayim' version of how that "Ha" gels and jives with the name Abraham and the Manna from Heaven and the bath salt and the tina and the "am in e" of amphetamine--maybe a glimmer or a shimmer or a glow of hope at the moment "Nazion" clicked ... and I said ... "no, not me ... I'm nothing like a king, no dreams of authoritarianism at all in the heart of Kish@r;" even as I wrote words that in the spirit of the moment were something of a "tis of a'we" that connected to my country and the first sing-songy "tisME" that I linked to trying to talk in the rhyming spirit of some "first Christ" that probably just like me was one limmerick away from the end of the rainbow and one "Four Non Blondes" song away from tying "or whatever that means" and this land crowned with "brotherhood" (to some personal "of the Bell, and of the bell towers so tall and Crestian") to just one Hopp skip and jump away from the heart of the obvious echoes of a bridge between haiku and Heroku... a few more gears shift into place, a click and and a mechanical turn of the face of the clock's ku-ku striking ... it was the word "Earthene" that was the last "Jesusism" around the post Cimmerian time linking Dionysus and Seuss to that same "su-s" that's belonging to a moment in the city of Uranus--codified and etched in stone as "MCO"--not just for its saucer and warp nacelles and "deflector dish" but for it's underground caverns and it's above ground "Space Mountain" and that great golf ball in the heart of it all.

The gears of time and the dawns of civilizequey.org query the missing "here" in our true understanding of what "in the beginning, to hear; to here ... to rue the loss of the Maize from Monoceros to the VEGA system and the tri-galactic origin of ... "some imaginary universal ... Earthene pax" to have dropped the ball and lost it all somewhere between "Avenu Malkaynu" and melaleuca trees--or Yggrasil and Snifleheim--or simply to miss the point and "rue brickell" because of bricks rather than having any kind of love or nostalgia linking to a once cobblestone roadway to the city in the Emerald skies paved in golden "do not return" signs ... to have lost Avenues well after not realizing it was "Heaven'es that were long gone far before I stepped foot on this road once called too Holy for sandals" in a place where that Promised Land and this place of "K'nanites" just loses it's grip on reality when it comes to mentioning the possibility that the original source and story of Ca'anan was literally designed to rid the world of ... "bad nanites" and the mentality of ... vindictiveness that I see behind every smirk.

The final hundred nanoseconds on our clock towards doom and gloom cause another bird to fly; another snake to curl up and listen again to the songs designed to charm it into oblivion; whether that's about a club in South Beach or a place not so far from our new "here..." all remains to be seen in my innocent eyes wondering what it truly is that stands between what you are ... and finding "forgiveness not needed--innocent child writes to the mass" ... and the long arm of the minute hand and the short finger of the hour for one brief moment reconcile and move towards "midnight" together; and it's simply idyllic, the Nazarene corner between nil and null you've relegated the history of Terran poast futures into ... "foreves mas" or so they (or you) think.

---

I'm still so far from "Five Finger Death Punch" though; and so far from Rammstein and so far from any kind of sick events that could stand between me and "the eternal" and change my still "casual alternative rock" loving heart to something more death metal; I rue whatever lies between me and there being any kind of Heaven that thinks there could exist a "righteous side" of Hell and it... simultaneously.

---

I still see light here in admonishing the masses and the angels standing against the story and the message God brings us in our history. I still see sparks in siding with the "causticness" of "no holodecks in sight" and the hunger and the pain of simulating ... "the hells of reality" over the story of decades or centuries of silence refusing to see "holography" and "simulated" in the word Holocaust and the horrors of this place that simply doesn't seem to fathom or understand the moments of hunger pangs and the fear of "dark Earth pits" or towers of "it's not Nintendo-DS" linking the Man in the High Castle to an Iron Mask.

I rally against being what I clearly am raised high on some pedestal by some force beyond my comprehension and probably beyond that of the "perfect storm in time" that refuses to itself acknowledge what it means to gaze at such an unfathomable loss of innocence at the cost of a "happy and serene future" or even at the glimmer of the Never-Never-Land I'd hoped we would all cherish and love and share ... the games and the newfound freedom that comes not just from "seeing Holodeck" turn into "no bullets" and "no cages" but into a world that grows and flourishes into something that's so far beyond my capability to understand that I'm stuck here; dumbfounded; staring at you refusing to stop car accidents and school shootings ... because "pedestal." For the "fire and the glory" of some night you refuse to see is this one--this place where morality rekindles from ... from what appears tobe one small candle, but truly--if it's not in your heart, and it's not coming from some great force of goodness--fear today and a world of "forever what else may come."

---

Here in a place the Bible calls Penuel at the crossing of a River Jordan ... the Angel of the Lord notes the parallels in time and space between the Potomac and the Rhine--stories of superposition and cities and nation-states that are nothing more than a history of a history of things like the Monoceros "arroz" linking not just to the constellation Orion but to Sagittarius and to Cupid and of course to the Hunter you know so well--

Searching for a Saturday; a sabbath to be made Holy once more ... "at the Rubycon"

The Einstein-Rosen Wormhole and the Marshall-Bush-JFKjr Tunnel

The waters are called narah, (for) the waters are, indeed, the offspring of Nara; as they were his first residence (ayana), he thence is named Narayana.

--- Chapter 1, Verse 10[3]

In a semi-fit of shameless arexua-self recognition i'm going to mention Amazon's new series "Upload" and connect it to the PKD work that my Martian-in-simulcrum-ciricculum-vitae on "colonization education" ... tying together Transcendance, Total Recall and ... well; to be honest it actually gave me another "uptick" in the upbeat ... maybe i'll stick around until I'm sure there's at least one more copy of me in the ivrtual-invverse ... oh, that reminds me ... Farmer)'s Lord of Opium also touches on this same "mind of God in the computer" subject (which of course leads to Ghost in the Shell and Lucy--thanks Scarlette :).

While I'm listing Matrix-intersected pieces of the puzzle to No Jack City, Elon Musk's neuralace and Anderson's Feed are also worth a mention. Also the first link in this paragraph is titled ... "the city of the name of time never spoken after time woke up and stfu'd" (which of course is the primary subject of this ... update to the city Aerosol).

The ... "actual original typed dream" included a sort of "roller coaster ride" through space all the way to Mars; where the real purpose of "the thing" I am calling the "Mars Hall" was to display previous victories and failures ... and the introduction of "older or future" culture's suggestions for "the right way" to colonize a new habitat. If it were Epcot Center, this would be something like SpaceMountain taking you to to the foture of "Epcot Countries" as if moving from "countries" to planets were as easy as simply ... "reading backwards."

THE SOFTWARE, SINGERS, AND SHIELD(S)

OF

HEIROSOLYMITHONEYY

Thinking just a little bit ahead of myself, but I'm on "Unreal Object/Map Editor within the VR Server" and calling it something like "faux-wet-ware" ... which then of course leads to a similar onomonopeia of "weapons and ..." where-with-all to find a better singer's name to connect the road of "sword" to a Wo'riordan ... but I think that fusion of warrior and woman probably does actually say ... enough of it all; on this road to the living Bright Water that the diety in my son's middle name defines well here, as "waking up," stretching it's tributaries and it's winding wonders and wistfully ....

Narayana (Sanskrit: नारायण, IAST: Nārāyaṇa) is known as one who is in yogic slumber on the celestial waters, referring to Lord Maha Vishnu. He is also known as the "Purusha" and is considered the Supreme being in Vaishnavism.

andromedic; the ports of call ... to the mediterranean (literally) from the gulf coast;

... ho engages in the creation of 14 worlds within the universe as Brahma when he deliberately accepts rajas guna, himself sustains, maintains and preserves the universe as Vishnu by accepting sattva guna. Narayana himself annihilates the universe at the end of maha-kalp ...

.

there's no place like home. there's no place like home. there's no place like home.

and so it begins ... "f:

r e l i g i o n

find out what it means to me. faucet, ever single one, stream of purity ...

from Fort Myers ... f ... flicks ... Flint.

"

^this notation will from this email forward in linear time denote some form of contact method or information related to the context of the message you are reading. This particular one sends me an encrypted email. 5if there is an "@" symbol involved in the "anchor's hypertext reference" (technically an "a href=" in HTML4) your browser should attempt to open an email client to send a message over an anonymous SMTP relay. Understand that "anonymous" in this case may or may not mean your sending email address is hidden or obvuscated--so if you want to receive a reply you must include it in the DATA of your SMTP transmission defined by the RFC5321 attached. In most cases "anonymous" also means that you will not have the recipients direct contact information unless they have made it public---additionally the exact server/system/relay used may or may not be the "Sbroken Berkman Perl Script" linked to in the "hypertext reference" specifically anchored to the words "an anonymous SMTP relay" above.

A simple "hat character" (^) and the letter "t" as you see beginning the above paragraph will denote a contact method or form that works over the internet using an HTTP protocol defined in a series of RFC's including (but not limited to) RFC's numbered as 2616, 7230, 7235, 2068 and use a simple language which is based on a definition suggested or proposed currently by an organization called the "W3C Consortium"

---and ... previously set and defined by an organiza^tion located at html.spec.whatwg.org; which appears (to me, for the first time as I write these words) to follow the conceptual spirit of the "living document" defined by the several "Continental Congresses, et alia." I personally now conjoin this document in my head to a procession of patrilineal or matrilnear predecessors to the actual event .... still to be defined ... but related to this specific email, this mailing list; its contributors and readers as well as actual members of the organization (still to be created, defined, or named) that creates a "round table" of members that is open to the public, to all voters educated enough to understand the specific issue being voted on (up to a standard that; in this place and time appears to be unset and unmet but materially related to reawching the age of 18 years old; growing up in or being born in the United States of America (related spec. to the Constitution of the United States of America which is officially "self-defined" through a process which includes all three branches of the government which it also "self-defines" and purports to be "of, for, and by the people"--though the general population is only able to contribute through an indirect process (read:the people cannot directly contribute to the constitution without either running for office (like a senator) or being appointed to a specific government position (like a judge or executive branch public servant).

The current state of American representative democracy is the highest standard to which I am currently knowledgable of "extant"--and it is specifically substandard, inferior, and "just not good enough" as a comparison to the process required to vote in the organization being "self-defined" through this process*. It is my sincere and clear hope that "this process" will result in a legal and moral amendment to the document shown in the previous link and presented by the Legislative Branch of the United States here. It is my current and faithful belief that anything else would also be significantly below the standards morally required by "this process" which of course includes over 200 years of American citizenship and (other international relations; i.e., e.g, for "iv" example, id est, exemplia gratia) as well as the Sons of Liberty and prior to that contributions from the Crown and the "Parliament and Crown" of the United Kingdom; among others et alea's ifndef: 'swikipedia/et_al..

To note specifically because of lack of personal knowledge and public notoriety (assuming all other requiremnant* achem requirements)

alas, babylon.

i listened to a man yesterday who was talking about "true heroes" ... he of course noted jesus christ and superman together, suggesting the first was one, and the second just a fiction. he also talked about people like ghandi and "leaders who use non-violent means to "change the world." i at least agree with him on the third, ghandi is a good prototype for some kind of hero. staring at this ... "to be completed" work on tales of two cities, whether from sodom and gomorrah all the way to athens and sparta and perhaps even london and paris--and this particular city, babylon; it stands out as one which truly has no equal or even "mirror" in the history of the world. i suppose i'd add "alexandria" and suggest the library and the laws; something that are fundamental to the ethos of the planet i call "athens."

i imagine he did not know "hammurabi's" name; and even today in this place where i ask and do not receive answers; i imagine you still don't connect muhammad or amsterdam ... to this king who in our history is set apart and lifted high on a pedestal of having "codified and written down" laws ... for the very first time. it's almost comical, it took me a paragraph and a sentence to connect "the king and i" to this mirror world, where the bible and the people have most assuredly decided "babylon" is a negative thing or a depraved place.

"fallen, fallen, is [the city of] babylon the great"

... just a quote from one of my favorite movies; which of course is re-quoting "dante" and/or "the bible"

"a dwelling place [of] (the) demons (say), it has become."

• www.icann.org/news/blog/the-problem-with-the-seven-keys

• kauri on IPFS: has-abaslom-and-the-ethos-of-arcadia

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

Learn more at Review Commons

---

Reply to the reviewers

We would like to thank the Review Commons editor and three reviewers for their enthusiastic response, including their constructive suggestions and appreciation of the high impact and originality of our study. We have completed the revisions and new analyses suggested by the reviewers, and we thank the reviewers for their suggestions to increase the impact and interest in this work and for guiding us towards this much improved manuscript.

In this response letter, we present the response to each reviewer comment and associated revisions made to the text and figures as bullet points below the reviewers' text (black text).

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

Summary:

Yang et al. took advantage of recently published long-read-based genomic sequences of nearly homozygous genomes from complete hydatidiform moles to retrieve allelic sequences of LINE-1, the currently only active and autonomous retrotransposon of the human genome, and produced the repertoire of intact LINE-1 in a genome. The authors performed cell-culture-based retrotransposition assays measurements and in vivo fitness estimations of all identified intact LINE-1 to infer evolutionary dynamics. In this article, the authors further validate the major contribution of polymorphic LINE-1 to the de novo retrotransposition events in the human genome. They also described, at unprecedented resolution, allelic variations among LINE-1 loci and the potential impact of these variations to the interpretation of mutagenic potential of each LINE-1 locus.

Major comments:

1 - The key conclusions of the article are mostly convincing. However, it would be a substantial improvement to consolidate the data of the article with information about known active LINE-1s in germ cells or in cancer by using data from recent publications of the Devine and Tubio labs (for example PMID: 34772701, 32024998, 25082706). Across the article, no mention is made of the transductions generated during LINE-1 de novo retrotransposition, which is instrumental to monitor in vivo activity of a group of LINE-1 active copies. It would be of particular interest to make a link between in vitro activity from this study with LINE-1 classification based on their observed activity in cancer (PMID: 32024998, Figure 3b).

• We thank this and the other reviewers for this suggestion. We agree that a more explicit comparison to the often-reported counts of 3’ transductions would be a valuable addition to our analyses. We have added the 3’ transduction counts from PMID:34772701, PMID:32024998 and PMID:25082706 to Table S2 (column Y, Z and AA), and made a comparison between these data and our Hamming-distance-based in vivo activity, as the new Figure S5. We found correlations between the two measurements in a significant proportion of LINE-1s, but some interesting exceptions exist which likely reflects the fact that most catalogued 3’ transductions come from cancer genomes, and cancer and germline cells represent distinct cellular environments in which distinct sets of LINE-1s are able to replicate (and leave 3’ transductions). In addition to the new figure (Figure S5), we have added a discussion paragraph focused on this interesting comparison.

2 - The use of CHM1 BAC library Sanger sequencing validation and comparison with CHM13 and hg38 sequences is instrumental to support the building of LINE-1 repertoire in CHM1 genome, which is a valuable contribution of the article. The use of a distance-based metric to infer fitness of a LINE-1 is an interesting approach and allow to group LINE-1 copies based on their in vivo activity potential. Again, it would be beneficial to correlate the inferred fitness and retrotransposition activity of copies/alleles, when known, from the above-mentioned literature.

• The sequence validation of LINE-1 sequences in CHM1 is an important point which we have addressed in the edited manuscript. Specifically, we used three forms of sequence validation including end-sequencing of one clone of each LINE-1 after it was cloned into the retrotransposition vector and whole-plasmid sequencing of select LINE-1s with discrepant activity amongst the three clones we assayed. In addition, we sequenced the entire LINE-1 sequence for four LINE-1s which had the largest number of mutations relative to their allelic counterpart in CHM13. Please see the above response to ‘Major comment 1’ for details of our new analysis comparing the previous literature to our data.

3 - Some aspects of the writing of the article should be improved to better support the conclusions.

• We thank the reviewer for providing these examples of parts of the text that were particularly difficult to read and comprehend. We have deeply streamlined and improved the text throughout the manuscript based upon detailed editing for readability and clarity by two experienced scientific writers. Below, we detail how we revised the particular sections presented by the reviewer, but we think the entire manuscript is now more succinct and clearer.

• In general, the descriptions are dense, and details could be provided in a more direct way to lighten the results section. Several redundancies in the discussion can be combined to increase clarity.

• We have spent considerable time tightening up the text, including removing several overlapping sections from the discussion which can be seen in the included version with changes tracked.

• There is a lack of clarity in the description of how was handled each pair of alleles for which retrotransposition measurements vary between the study and the literature (last paragraph of the "Comprehensive measurement of LINE-1 in vitro activity in a human genome" section). It is not completely clear how the analysis was done and the way the data is presented in File S3 is not helping to support the conclusion. It could be useful to include some illustrative examples in a panel of Figure 2.

• We agree that this description was hard to parse, and we have rewritten this and accompanying methods to simplify our explanation of these results. In addition, we have revised Figure 2 to show the data in much more detail. To further aid the logic flow related to this section, we moved the previous Figure 5B to Figure 2B, updated it with more suitable examples and edited the associated descriptions.

• Regarding inferred in vivo activity, the text contains alternative description with the use of "fit" / "unfit", in vivo "active" / "inactive" or "no closely related LINE-1s" terms. The authors should find a way to clearly define and systematically use one set of terms to enhance clarity along the article. To parallel with in vitro active/inactive, it would be useful to use in vivo fit/unfit.

• We thank the reviewer for this suggestion and agree with their suggested unified use of ‘in vivo fit/unfit’. To clarify and simplify these terms as much as possible, we added detailed explanations of in vivo / in vitro activity and systematically defined in vitro "active/inactive" (page 5, right column, line 50) and in vivo "fit/unfit" (page 8, left column, line 26) at their first appearance in the article, and we changed most instances of "in vivo activity" to "in vivo fitness" when context permits.

4 - The authors suggest that in vitro activity can be predicted by integration of population frequency and in vivo activity (/fitness) (second paragraph of the "An analysis of LINE-1 evolutionary history [...] and in vivo activity" section). It would be beneficial to strengthen the writing of this section and ultimately validate/test the model by including data from some of the previous studies (e.g. Brouha 2003, Lutz 2003, Seleme 2006, Beck 2010, Rodriguez-Martin 2020, Chuang 2021).

• We have thoroughly revised this section of the results (see response to ‘Major comment 3’ above), per the reviewers suggestion, to increase reader comprehension of this important data. In addition, we greatly appreciate the reviewer’s suggestion of a very interesting experimental direction – moving beyond a single long-read-based genome to many diverse genomes, and ultimately calculating the in vivo fitness of the LINE-1s from these diverse genomes. For a long time this has not been possible, but the recent publication of the Human Pangenome presents an opportunity to study this interesting question. Though beyond the scope of this paper, our lab is actively working on this fascinating question, and we appreciate the reviewer’s shared interest in this question.

5 - The identification of adaptive mutations is only partially described and not strongly supported by experimental or analytical data. It would be interesting to explore the role of phylogenetically informative sites described in Figure 5B/C by testing non CHM1 alleles in retrotransposition assay (by introducing amino acid changes into the cloned CHM1 LINE-1 alleles) or by positioning the sites in ORF1p or ORF2p structure and/or domains to infer impact on functionality.

• The reviewer rightly points out that this is one of the most interesting and novel findings of our manuscript. However, the testing of potentially adaptive mutations is potentially complicated and nuanced. Specifically, we don’t know the mechanism by which these mutations might be adaptive. It is possible that they simply increase in vivo germline retrotransposition activity and this increase would be reflected by an increase of in vitro retrotransposition activity. However, another possibility is that these adaptive phenotypes only show themselves in vivo or in the context of the host restriction factors expressed in the germline. We strongly agree with the reviewer that experimental and analytical data on the phylogenetic informative sites associated with the Figure 5 phylogeny is the key to finding out the mechanisms for these changes to affect LINE-1 activity/fitness, and we are, indeed, exploring this very question in the lab now with related projects. We respectfully suggest that these (extremely cool) experiments are beyond scope of this paper, but we have also added some more detailed description and analyses of the potentially adaptive LINE-1 variations from Figure 5 (from page 9, right column, line 50 to page 10, left column, line 5).

Minor comments:

1 - Regarding the in vitro retrotransposition assay, it would be beneficial to provide more data. The current Figure 2 could be enriched by the addition of data related to the variation in the replicates of the experiment (technical but mostly biological with the three clones per LINE-1 tested). Figure 2 could include a dashed line for 100% L1RP and 5% (since it is used as a threshold). It would be useful to provide an additional panel in Figure 2 to illustrate alleles of LINE-1 that are active in this study and compare the values obtained previously in other studies. Similarly, a supplemental table or alignment could be provided to document amino acid changes in the two alleles of each pair (see comment above in the Major Comment 5). The L1Hs subfamilies could also be included in the graph of Figure 2 to support the conclusions of remaining active old L1Hs at allelic forms in the human genome.

• Upon consideration of this helpful comment, we now augment the presentation of our in vitro activity data with a remade Figure 2 with boxplots to show the variation of the data, as well as a horizontal dashed line showing the active-cutoffs and star signs showing which LINE-1s belong to L1Hs or L1PA2.

2 - Also, the validation of cloning is not well described. The choice of PCR validation must be supported by more technical details on the design of the primers used to validate each copy. The authors should clearly state that the strategy chosen for retrotransposition assay does not rely on the transcription from LINE-1 5UTR but from an upstream strong promoter, ruling out the role of potential mutations in LINE-1 promoter.

• As detailed above in the response to ‘Major Comment 1’, we used a combination of end sequencing, whole plasmid sequencing, and multi-read Sanger sequencing to validate the sequences of each LINE-1 cloned from a CHM1 clone. When cloning each LINE-1, we used a specific set of primers designed for the ends of the UTRs for each LINE-1. We have updated the methods and text to clarify this cloning step, and the sequences of these oligos are included in Table S2.
• To clarify the fact that our retrotransposition assays use a common, strong promoter, we added text in several places stating this setup and discussing (paragraph that starts at page 11, right column, line 18) how 5'UTRs and other non-ORF factors can affect the rate of LINE-1 in vitro activity.

3 - There are discrepancies with the reported numbers of LINE-1s between Figure 1A and Table S1: 154 vs. 151 in CHM1, 144 vs. 143 in CHM13, respectively.

• We thank the reviewer for spotting this error on our part. The numbers in Figure 1 and the main text were correct, and we have revised Table S1 to reflect this data.

4 - The choice of colors in Figure 3 is not perfectly clear and sometimes not as reported in the text (green highlight and orange highlight). Part of the Figure 3 legend is missing. It should include a description of the color code chosen for the right histogram.

• We thank the reviewer for bringing this inconsistency to our attention. Based upon feedback from all reviewers, we have simplified the color scheme in Figure 3 and Figure 5 to focus on the core conclusions of these two figures. Specifically, in Figure 3, we have removed the quadrant shading and more clearly presented the cutoffs of ‘polymorphic/high frequency’ and ‘in vitro active/inactive’ as dashed lines in the scatter plot. In Figure 5, we have simplified to two colors – black for in vivo unfit and orange to show the in vivo fit LINE-1s which is also used in Figure 4 to show the definition of in vivo activity. These updated colors are now defined in the figure legends and main text, and we have made references to these colors consistent throughout.

5 - For Figure 4, it would be useful to define in the legends the color code for the top histogram. To better read the scatter plot, the words "fit" and "unfit" could be added on each side of the vertical dashed line.

• We thank the reviewer again for suggestions to improve the clarity of our figures. As mentioned above in ‘Minor comment 1’, we have removed unnecessary colors including the gradient of the histograms in Figure 3 and Figure 4, since the boundaries of each bin are already defined by the axis labels and tics. As suggested, we have also added ‘fit’ and ‘unfit’ labels to the dashed cutoff line in Figure 4 to clarify the meaning of this line.

6 - In panel B of Figure 5, it seems that the color code and hot/cold description is not fully formatted.

• This formatting error has been corrected.

Reviewer #1 (Significance (Required)):

In this article, Yang and colleagues present an unprecedented view of the allelic diversity of young LINE-1 copies related to variable retrotransposition activity in an individual genome. One key aspect of their work is the description of the presence of young active LINE-1 alleles that are absent or non-intact in other genome assemblies, while described at a lower scale in initial work from the Kazazian and Moran labs, cited in the manuscript. The work of Yang et al. demonstrates the requirement of multiple approaches and long-read-based sequencing of individual genomes to fully infer the mutagenesis risk of LINE-1 activity.

The data and methods provided by the authors open the door to a more systematic analysis of mutations and rare allelic forms to understand both mechanistic aspects and evolution of LINE-1 retrotransposition in the human genome. The identification of rare allelic forms of old LINE-1 that retain activity despite previously being considered as inactive is particularly interesting in the light of LINE-1 evolution in the human genome. The authors also describe allelic diversity inside of the Ta1d subfamily, suggesting further diversification and emergence of LINE-1 subgroups. Together with the identification of nucleotide polymorphism among LINE-1 copies, these findings strengthen the notion of individual genomes with individual set of potentially mutagenic LINE-1 alleles.

The findings and methods described in this article are of great interest to a wide audience including the fields of research focusing on human genome evolution, transposable elements, genomic instability, human genetic variation, and personalized medical diagnostic.

Aurélien J. Doucet CNRS - Université Côte d'Azur

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

This manuscript is an interesting and well-crafted study of LINE-1 activity at the single genome human genome level using long read-based haploid assemblies. The manuscript has some real gems and address critical aspects of LINE- biology that are typically not rigorously examined. The authors are to be commended for undertaking this exercise and for providing interesting perspectives that challenge the dogma that dominates the field in several areas. Despite the noted strengths of the contributions, the manuscript ignores the clear limitations inherent to the approaches taken and at times appears as dogmatic as the dogma that they themselves are trying to challenge. These deficiencies should be addressed before this manuscript is published.

• We thank Reviewer 2 for their enthusiastic appreciation of the value and innovation of our manuscript. We also thank the reviewer for encouraging us to make careful consideration of the missing references relevant to our findings. We have had two researchers with experience in relevant fields edit our text for both readability, clarity, and proper inclusion of relevant references. We have added these throughout and taken careful effort to replace ‘dogmatic’ statements with clear presentations of the data and thorough referencing of the relevant literature.

Several major and minor points to consider during revision include:

Major:

1. Several strategies have been published in the past that have confidently assign LINE-1s to specific loci despite use of shorter reads. These works should be acknowledged, even if as stated in the manuscript, use of longer reads will only continue to add confidence and validity to future assignments.

2. We thank the reviewer for this suggestion, and we apologize for the omission of these important publications. As noted above, we have added numerous relevant references (reference 17-27 in the revised text) throughout the text including previous work that used short reads to confidently assign polymorphic/non-reference LINE-1s to specific loci. For example, we now cite the MELT pipeline to detect de novo L1 insertions with short reads (PMID: 28855259), and Iskow et al. 2010, which detects LINE-1s with junction fragment sequencing (PMID: 20603005). We have also added additional text to clarify that short reads are, indeed, often sufficient to place new LINE-1 insertions, while long reads are especially useful for resolving the sequence and location of these insertions. The new text (page 2, left column, line 22-30) presents the advantages/disadvantages of both short reads and long reads.

3. One of the important requirements for precise quantification of LINE-1 activity and predicted risk scores cited in the manuscript was the need to predict activity based on sequence and location. This requirement, as posited in the manuscript, ignores the critical role of epigenetic control in the regulation of LINE-1 activity. As such, a discussion that acknowledges the critical roles of histone and DNA covalent modifications, and that integrates epigenomic insight into predictions of LINE-1 activity must be included in the manuscript.

4. We thank the reviewer for suggesting this important discussion point. In response, we have expanded our discussion of this topic to place our data in the context of other literature on the effects of epigenomic regulation on in vivo LINE-1 activity, including histone and DNA modifications, as well as the effects of post transcriptional restriction factors (paragraph starting at page 11, right column, line 42).

5. The limitations associated with the use of the CHMI were not addressed in the manuscript. While CHMI contain a paternal only genome, with no maternal contribution, the moles may arise from fertilization of an anuclear empty ovum by a haploid 23,X sperm or fertilization by two sperm giving rise to 46,XX or 46,XY karyotype. As such, generalizable conclusions about CHMI genetics should be carefully made given that the loss of maternal epigenetic imprinting and gain of paternally imprinted expression may result in abnormal gene expression, including that of LINE-1s. These variances will in turn impact LINE-1 activity profiles.

6. We thank the reviewer for pointing out this confusingly written section of our manuscript, and we agree with the reviewer that LINE-1 activity measurements could be complicated in the CHM cell lines; however, all of our retrotransposition assays were carried out in the common background of 293T cells (chosen because of their low expression of know LINE-1 restriction factors (PMID: 25182477). We have modified the text (page 11, right column, line 52) to clarify these points.

Minor

1. Important citations of previously published work are not properly referenced throughout the manuscript. These are too numerous to identify individually, but the authors should carefully read the manuscript to ensure that proper documentation and reference to previous work is duly acknowledged.

2. Please see our above response to ‘Major point 1’.

There are several typos and missing prepositions that should be corrected. For instance, on page 7, the word "great" should be "greater".

• Please see our above response to ‘Major point 1’ and Reviewer 1’s ‘Major comment 3’ for details on our in depth editing of the manuscript.

Reviewer #2 (Significance (Required)):

The contribution is highly significant as it challenges previously held concepts and advances our understanding of critical structure and function relationships of Line-1s.

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

Yang et al. perform an in-depth analysis of potentially mobile source L1 alleles in a single human genome (CHM1) previously subjected to Pacbio whole genome sequencing. The retrotransposition efficiencies of source L1 alleles with intact ORFs were tested in vitro, and these efficiencies compared to a model of in vivo activity based on Hamming distance to other ORF-intact L1 alleles. Comparisons of CHM1 L1 alleles are made to CHM13 (used for the recent T2T reference assembly), and also to population-scale sequencing efforts to establish how widespread each source L1 allele is. These data showcase the advantages of being able to resolve L1 alleles with long-read sequencing, allowing the field to make much more accurate predictions of retrotransposition potential in a given genome. The core analyses appear robust and for the most part enough detail is provided to follow what was done.

• We thank Reviewer 3 for their in depth reading and analysis of our manuscript and data, and for their enthusiasm about the importance of this work in the context of foundational research from their lab and many others in the field. We have carefully considered each comment and completed several new analyses of our data and related data from other publications. We feel that our manuscript is much improved with this new data, as detailed below. Comments:

1) The text overlooks the potential importance of L1 5'UTR mutations in L1 activity and evolution, as per PMID:25274305, PMID:1701022, and other studies, as well as the impact of genomic context on source L1 activity, as per PMID:27016617, PMID: 33186547 etc. L1 promoter evolution is arguably a major driver of L1 lineage emergence.

• We thank the reviewer for suggesting these important additions. To present the relevance of 5'UTR mutations on LINE-1 activity and evolution, we added a discussion paragraph (paragraph starting at page 11, right column, line 16) to address how 5'UTRs and other non-ORF factors can affect the rate of LINE-1 in vitro activity. Several key references have been added and discussed in the paragraph: PMID:25274305 reported the regulation of human LINE-1 by the evolution of its 5'UTR; PMID:1701022 was one of the earliest papers that found the effect the 5'UTR promoters on human LINE-1 retrotransposition; PMID: 27016617 and PMID: 33186547 reported specific L1 loci regulated by different promoters and was included in the discussion; PMID:9430649 was one of the examples of non-human LINE-1 lineages emerging because of different promoters and was cited in the added discussion paragraph. We have also added discussion points to make clear that genomic content has a clear role in the activity of source LINE-1s (paragraph starting at page 11, right column, line 42).

2) The way the retrotransposition assay is done here (I think) removes parts of the UTRs as part of introducing L1s into retrotransposition vectors, meaning that the assay tests the biochemical activity of the ORFs. It would be helpful to readers to have a more detailed method for this assay, including the origins of the reporter plasmids, whether there is a CMVp boosting the L1 promoter etc, and some clarity about how much of each L1 was cloned into the assay.

• We have added relevant details to the results (page 6, left column, line 5), discussion (page 11, right column, line 52), and methods (page 13, right column, line 16 and 30) sections to clarify the reviewer’s important points. The LINE-1s tested for in vitro activity were cloned in their entirety (UTRs and ORFs) but driven by both their native promoters in the 5'UTR as well as an upstream CMV promoter. Also, please see our response to Reviewer 1 ‘Minor comment 2’ above.

3) Pacbio long-read sequencing has been used previously to locate and characterise L1 alleles in human DNA. The Introduction states: "These represent the first scalable methods to catalog LINE-1 locations and sequences in individual human genomes". The "first" here is questionable. Citations to PMID:31853540 and PMID:34772701 should be included. The latter is particularly relevant at it not only resolves source L1 sequences with PacBio sequencing but also summarises their retrotransposition efficiencies in vitro and population frequencies.

• We apologize for leaving out these and other important references, and we agree that the “first” claim is unnecessary. We have added the references suggested for the reviewer as well as several other important references as detailed in the above response to Reviewer 2 ‘Major point 1’. In addition, we have revised the adjacent text and deleted any references to our work as the “first” in these approaches.

4) I am very interested in the two source L1s (on chr7 and chr9) that were found here to be more active in vitro than L1RP (to my knowledge the most active such element isolated to date, or close to it). Is there anything unusual about these two L1s? A quick look at the supplemental suggested the chr9 element was 5' truncated, was it tested as such in vitro? Also I think it would be worth contrasting the assay (all in HEKs) used here to test efficiency with the assay used by Brouha ... I feel readers may be surprised to find two L1s more mobile than L1RP in one genome.

• To provide more details about the two active L1s (chr7 and chr9), we investigated key changes that could be related to the in vitro activity of these elements and now show them in Figure 2B and File S3. In the process of this updated analysis and suggested modifications to Figure 2 by this reviewer and Reviewer 1, we saw that the chr7 L1, mentioned here, had one very high activity measurement pulling its activity above L1RP. As such, we decided to more rigorously normalize our data by using the positive and negative controls across all plates of each day instead of normalizing to the controls of individual plates, as we had previously done. In addition, for any L1 with discrepant activity among the three clones we assayed, we used whole plasmid sequencing to confirm the identity and consistency of all three clones. In three cases, we found that one or two of the three clones was the wrong L1, and hence excluded them for the in vitro activity calculation. After this validation and testing of additional clones, all clones from the same L1 have consistent in vitro activity (see updated Figure 2). The updated in vitro activity of the chr7 L1 is at 86.7% L1RP, and the chr9 L1 is at 261.4% L1RP in addition to the chr17 LINE-1 with 117% L1RP and two additional LINE-1s that have near-L1RP activity levels (Table S2, column S). These changes in L1 activity were updated in the text, figures, and supplemental materials. Also, we note that the chr9 element is 6019bp in length and was tested as such in vitro. Current work in the lab is attempting to understand the mechanisms of increased LINE-1 in vitro and in vivo activity, as described in detail in response to Reviewer 1’s ‘Major comment 5’.

5) In several places it is mentioned how L1 alleles may differ from sequences provided in reference assemblies, and may therefore explain discrepancies between assay results here and in other studies (e.g. Brouha). The Seleme and Lutz papers are correctly mentioned here, but arguably the most complete demonstration of this concept, from PMID:31230816, is overlooked. This study reports a chr13 source L1 that was previously found to be inactive by Brouha, and with broken ORFs in the reference genome, has both mobile and immobile alleles in the human population. This L1 is actually in CHM13, but not CHM1, and is "hot" in some individuals and not others. There are several places in the manuscript where this earlier study is very relevant and it would be fair to ask it to be mentioned, especially as the results are concordant. The same concept is reinforced by an even more recent paper (PMID:35728967), except in macaque, showing that this is a general consideration for primate L1 lineages, and actually that source L1 is relatively old and yet jumps extremely well in vitro, which fits an observation made in the present study. Mutually supporting observations like these really add confidence that what is reported in the present study is robust.

• We thank the reviewer for their suggestion to include these highly relevant and important papers; we apologize for this initial omission. We have now added several sentences to the introduction and discussion (top left paragraph page 11) in addition to citations of these papers.

6) Hamming distance between ORF-intact source L1 alleles is used to assess in vivo activity. This seems reasonable. However, in other works, transductions have been used to identify families of very closely related L1s. I realise that many highly mobile source L1s will rarely generate insertions carrying transductions, and yet I wonder if any of the youngest L1s in the present study form transduction families, and whether estimates of in vivo activity based on transductions found in population-scale data would reconcile better with in vitro retrotransposition assay data.

• We thank the reviewer for pointing out our exclusion of data on 3' transductions, the most commonly used surrogates of in vivo activity, while also acknowledging that only a small percent of new L1 retrotranspositions carry 3' transduction. Please see our above response to Reviewer 1’s ‘Major comment 1’ for details on our newly added comparison of our in vivo activity data to the 3' transduction-based somatic LINE-1 retrotransposition landscape of those reported in PMID:34772701, PMID:32024998 and PMID:25082706.

7) In the Introduction, it is stated that L1 only transmits vertically. It may be prudent to mildly qualify this position, based on PMID:29983116.

• The referenced text in the introduction has been changed from "LINE-1s only transmit vertically" to "LINE-1s generally transmit vertically with few exceptions", with the addition of the suggested citation.

8) A column in Table S2 looks mislabelled: Column R should be CHM1 not CHM13?

• We thank the reviewer for seeing this error. Column P (Column R in the previous version) of Table S2 is now correctly labeled as "CHM1 L1 intactness".

Geoff Faulkner (University of Queensland)

Reviewer #3 (Significance (Required)):

This is a well-executed study of considerable interest to the mobile DNA field, and anyone working with long-read DNA sequencing. Its strengths are the genomic and bioinformatic analysis, leveraging the PacBio long-read data and BAC library available for CHM1 to full effect. One limitation (in current form) is its near-exclusive focus on ORFs to encapsulate how mobile a given L1 allele is, when genomic context and L1 promoter mutations could also contribute heavily. Although I liked the manuscript very much and enjoyed reviewing it, some of the conceptual advances are encroached upon by other work (including some very relevant and yet uncited literature). These issues can very likely be addressed via a revision, additional analyses may be required but not new experiments.

Geoff Faulkner (University of Queensland)

Author Response

We would like to thank the reviewers for their positive and constructive comments on the manuscript.

We are planning the following revisions to both DGRPool and the corresponding manuscript to address the reviewers’ comments:

1) We agree with reviewer #1 that normalizing the data could potentially improve the GWAS results. Thus, we plan to explore the implementation of this option and assess its impact on the overall results. We will also investigate replacing the ANOVA test with a KRUSKAL test. Instead of upfront data normalization, we will consider using the PLINK –pheno-quantile-normalize option. Both options will be compared on a set of phenotypes where we can analyze the output (i.e., for phenotypes where we expect to find specific variants), to determine whether these strategies enhance the detection power.

2) We also agree with both reviewers that gene expression information is of interest. However, we recognize that incorporating such information would entail substantial work (as elaborated in our response to comments below). We feel that this extensive work is beyond the current scope of this paper, which primarily focuses on phenotypes and genotype-phenotype associations. Nonetheless, we are committed to enhancing user experience by including more gene-level outlinks to Flybase. Additionally, we will link variants and gene results to Flybase's online genome browser, JBrowse. By following the reviewers' suggestions, we aim to guide DGRPool users to potentially informative genes.

3) In agreement with reviewer #2, we acknowledge that additional tools could enhance DGRPool's functionality and facilitate meta-analyses for users. Therefore, we are in the process of developing a gene-centric tool that will allow users to query the database based on gene names. Moreover, we intend to integrate ortholog databases into the GWAS results. This feature will enable users to extend Drosophila gene associations to other species if necessary.

4) Finally, we also concur with both reviewers about making minor edits to the manuscript to address their feedback.

Reviewer #1 (Public Review):

This is a technically sound paper focused on a useful resource around the DRGP phenotypes which the authors have curated, pooled, and provided a user-friendly website. This is aimed to be a crowd-sourced resource for this in the future.

The authors should make sure they coordinate as well as possible with the NC datasets and community and broader fly community. It looks reasonable to me but I am not from that community.

We thank the reviewer for the positive comments. We are relatively well-connected to the D. melanogaster community and aim to leverage this connection to render the resource as valuable as possible. DGRPool in fact already reflects the input of many potential users and was also inspired by key tools on the DGRP2 website. Furthermore, it also rationalizes why we are often bridging our results with other resources, such as linking out to Flybase, which is the main resource for the Drosophila community at large.

I have only one major concern which in a more traditional review setting I would be flagging to the editor to insist the authors did on resubmission. I also have some scene setting and coordination suggestions and some minor textual / analysis considerations.

The major concern is that the authors do not comment on the distribution of the phenotypes; it is assumed it is a continuous metric and well-behaved - broad gaussian. This is likely to be more true of means and medians per line than individual measurements, but not guaranteed, and there could easily be categorical data in the future. The application of ANOVA tests (of the "covariates") is for example fragile for this.

The simplest recommendation is in the interface to ensure there is an inverse normalisation (rank and then project on a gaussian) function, and also to comment on this for the existing phenotypes in the analysis (presumably the authors are happy). An alternative is to offer a kruskal test (almost the same thing) on covariates, but note PLINK will also work most robustly on a normalised dataset.

We thank the reviewer for raising this interesting point. Indeed, we did not comment on the distribution of individual phenotypes due to the underlying variability from one phenotype to another, as suggested by the reviewer. Some distributions appear normal, while others are clearly not normally distributed. This information is 'visible' to users by clicking on any phenotype; DGRPool automatically displays its global distribution if the values are continuous/quantitative. We acknowledge the reviewer's concerns regarding the use of ANOVA tests. However, we consider it acceptable to perform linear regression (including ANOVA tests) on non-normally distributed data, as only the prediction errors need to follow a normal distribution.

Furthermore, the ANOVA test is solely conducted to assess whether any of the potential covariates (such as well-established inversions and symbiont infection status) are associated with the phenotype of interest. PLINK2 automatically corrects for the effects of these covariates during GWAS by considering them as part of the regression model.

Nevertheless, we concur with the reviewer that normalizing the data could potentially enhance GWAS results. Consequently, we commit to exploring the impact of data normalization on the overall outcomes. Additionally, we will consider replacing the ANOVA test with a KRUSKAL test, and using the PLINK –pheno-quantile-normalize option. We intend to compare both approaches using a set of phenotypes where we can compare the output (i.e., where specific variants are expected to be identified). This comparison will help us determine if either method enhances the detection power.

Minor points:

On the introduction, I think the authors would find the extensive set of human GWAS/PheWAS resources useful; widespread examples include the GWAS Catalog, Open Targets PheWAS, MR-base, and the FinnGen portal. The GWAS Catalog also has summary statistics submission guidelines, and I think where possible meta-data harmonisation should be similar (not a big thing). Of course, DRGP has a very different structure (line and individuals) and of course, raw data can be freely shown, so this is not a one-to-one mapping.

Thank you for the suggestion. We will cite these resources in the Introduction and check the GWAS catalog submission guidelines to compare to the ones we are proposing in this paper.

For some authors coming from a human genetics background, they will be interpreting correlations of phenotypes more in the genetic variant space (eg LD score regression), rather than a more straightforward correlation between DRGP lines of different individuals. I would encourage explaining this difference somewhere.

We appreciate this potential issue and we will make this distinction clearer in the manuscript to avoid any confusion.

This leads to an interesting point that the inbred nature of the DRGP allows for both traditional genetic approaches and leveraging the inbred replication; there is something about looking at phenotype correlations through both these lenses, but this is for another paper I suspect that this harmonised pool of data can help.

We agree with the reviewer and hope that more meta-analyses will be made possible by leveraging the harmonized data that are made available through DGRPool.

I was surprised the authors did not crunch the number of transcript/gene expression phenotypes and have them in. Is this because this was better done in other datasets? Or too big and annoying on normalisation? I'd explain the rationale to leave these out.

This is a very good point raised by the reviewer, and this is in fact something that we initially wanted to do. However, to render the analysis fair and robust, it would require processing all datasets in the same way. This implies cataloging all existing datasets and processing them through the same pipeline. Then, it also requires adding a “cell type” or “tissue” layer, because gene expression data from whole flies is obviously not directly comparable to gene expression data from specific tissues or even specific conditions. This would be key information as phenotypes are often tissue-dependent. So, as implied by the reviewer, we deemed this too big of a challenge beyond the scope of the current paper. Nevertheless, we plan to continue investigating this avenue, especially given the strong transcriptomics background of our lab, in a potential follow-up paper.

I think 25% FDR is dangerously close to "random chance of being wrong". I'd just redo this section at a higher FDR, even if it makes the results less 'exciting'. This is not the point of the paper anyway.

We agree with the reviewer that this threshold implies a higher risk of false positive results. However, this is not an uncommonly used threshold (Li et al., PLoS biology, 2008; Bevers et al., Nature Metabolism, 2019; Hwangbo et al, Elife, 2023), and one that seems robust enough in our analysis since similar phenotypes are significant in different studies. Nevertheless, we will revisit these results and explore how a more stringent threshold may impact the results.

I didn't buy the extreme line piece as being informative. Something has to be on the top and bottom of the ranks; the phenotypes are an opportunity for collection and probably have known (as you show) and cryptic correlations. I think you don't need this section at all for the paper and worry it gives an idea of "super normals" or "true wild types" which ... I just don't think is helpful.

This section of the paper was intended to investigate anecdotal evidence suggesting that certain DGRP lines consistently rank at the top or bottom when examining fitness-related traits. If accurate, this observation could imply that inbreeding might have made these lines generally weaker, potentially introducing bias into studies aimed at uncovering the genetic basis of complex traits. However, as per the analyses presented, we did not discover support for this phenomenon. Nevertheless, we consider this message important to convey. In response to the reviewer's feedback, we intend to provide a clearer explanation of the reasoning behind this section of the paper and its main conclusion.

I'd say "well-established inversion genotypes and symbiot levels" rather than generic covariates. Covariates could mean anything. You have specific "covariates" which might actually be the causal thing.

Thank you. We will update the manuscript accordingly.

I wouldn't use the adjective tedious about curation. It's a bit of a value judgement and probably places the role of curation in the wrong way. Time-consuming due to lack of standards and best practice?

Thank you. We will update the manuscript accordingly.

Reviewer #2 (Public Review):

Summary:

In the present study, Gardeux et al provide a web-based tool for curated association mapping results from DRP studies. The tool lets users view association results for phenotypes and compare mean phenotype ~ phenotype correlations between studies. In the manuscript, the authors provide several example utilities associated with this new resource, including pan-study summary statistics for sex, traits, and loci. They highlight cross-trait correlations by comparing studies focused on longevity with phenotypes such as oxphos and activity.

Strengths:

-Considerable efforts were dedicated toward curating the many DRG studies provided.

-Available tools to query large DRP studies are sparse and so new tools present appeal

Weaknesses:

The creation of a tool to query these studies for a more detailed understanding of physiologic outcomes seems underdeveloped. These could be improved by enabling usages such as more comprehensive queries of meta-analyses, molecular information to investigate given genes or pathways, and links to other information such as in mouse rat or human associations.

We appreciate the reviewer's kind comments.

Regarding the tools, we concur with the reviewer that incorporating additional tools could enhance DGRPool and facilitate users in conducting meta-analyses. Therefore, we intend to introduce a gene-centric tool that enables users to query the database based on gene names. Additionally, we will establish links to ortholog databases within the GWAS results, thereby allowing users to extend fly gene associations to other species, if required.

Furthermore, we have plans to link out to a 'genome browser-like' view (Flybase’s JBrowse tool) of the GWAS results centered around the affected variants/genes. We are considering integrating this feature into the new gene-centric tool as well.

Another potential downstream analysis we are considering is gene-set enrichment. This analysis would involve assessing the enrichment of genes in Gene Ontology or other pathway databases directly from the GWAS results page.

Author Response

The following is the authors’ response to the original reviews.

We greatly appreciate the positive feedback of the reviewers and have modified the manuscript to address their comments, including changes to the text, figures, and methods. We believe that these revisions have strengthened and improved the manuscript. Reviewers’ comments in blue and detailed responses in black are below.

Reviewer #1 Weaknesses:

• Is "function" of the ISNs to balance "nutrient need" or osmolarity? Balancing hemolymph osmolarity for physiological homeostasis is conceptually different from balancing thirst and hunger.

We have added the following text to the introduction to address this: “Thus, the ISNs sense both AKH and hemolymph osmolality, arguing that they balance internal osmolality fluctuations and nutrient need (Jourjine, Mullaney et al., 2016).” (ln 80-82).

• The final schematic nicely sums up how the different peptidergic pathways might work together, but it is unclear which connections are empirically-validated or speculative. It would be informative to show which parts of the model are speculative versus validated. For example, does FAFB volume synapse = functional connectivity and not just anatomical proximity? A bulk of the current manuscript relies on "synapses of relatively high confidence" (according to Materials and methods: line 522). I recommend distinguishing empirically tested & predicted connections in the final schematic, and maybe reword/clarify throughout the manuscript as "predicted synaptic partners"

We modified the schematic to clarify EM based connections versus functionally validated connections. We also clarified the EM predicted synaptic partners, using “predicted synaptic partners” throughout the manuscript.

Reviewer #2 Areas for further development:

• Does BIT inhibit all of the IPCs or some of them? I think it is critical to indicate the ROIs used for each neuron in the methods. Which part of the neuron is used for imaging experiments? Dendrites, cell bodies, or synaptic terminals?

ROIs used for quantification are described in the figure legends: “ArcLight response of BiT soma…” (Fig 2, Fig S2), “Calcium responses of CCHa2R-RA neurites in SEZ…” (Fig 4), “Calcium response of CCHa2R-RA SEZ neurites…” (Fig S4), “Calcium response of CCAP neurites…” (Fig 5, Fig S5), “Calcium response of all IPC somas…” (Fig S3). We have added ROIs used for quantification to the ‘In vivo calcium imaging’ and the ‘In vivo voltage imaging’ methods sections (ln 493-494).

• The discussion section is not giving big picture explanation of how these neurons work together to regulate sugar and water ingestion. Silencing and activation experiments are good, but without showing the innate activity of these neural groups during ingestion, it is not clear what their functions are in terms of regulating fly behavior.

We agree that how these peptidergic neurons coordinately regulate feeding is unclear. As peptide signals may act at a distance and may cause long-lasting neural activity state changes, studying their integration over space and time is challenging. Acute imaging during feeding would only in part address this challenge, as cumulative changes in nutrient need signals may impart circuit changes that are not apparent by monitoring the acute activity of peptidergic neurons. We modified a paragraph in the discussion to address this (ln 434-443).

“Overall, our work sheds light on neural circuit mechanisms that translate internal nutrient abundance cues into the coordinated regulation of sugar and water ingestion. We show that the hunger and thirst signals detected by the ISNs influence a network of peptidergic neurons that act in concert to prioritize ingestion of specific nutrients based on internal needs. We hypothesize that multiple internal state signals are integrated in higher brain regions such that combinations of peptides and their actions signify specific needs to drive ingestion of appropriate nutrients. As peptide signals may act at a distance and may cause long-lasting neural activity state changes, studying their integration over space and time is a future challenge to further illuminate homeostatic feeding regulation.”

Reviewer #1 (Recommendations For The Authors):

• For the final schematic figure, it may be informative to include nanchung and AKHR in the schematic.

We now include this (Fig 6).

• For the ingestion duration with optogenetic activation, I don't think the right way to represent the data is by normalizing them to the no LED control. I think it should show raw ingestion time. I understand that the normalized data make the figure "cleaner" (no need to show +/- LED separately) but I think visualization of the raw data is important.

We now include this in a new Supplemental Figure (Fig S6).

• Methods for ingestion with optogenetic activation should be detailed in the Methods section.

We expanded upon this in the ‘Temporal consumption assay (TCA)’ methods section. (ln 461-466).

Reviewer #2 (Recommendations For The Authors):

1) I think the authors are not following the recommendations of the Flywire community which recommends that people who contributed to the tracing of neurons are offered authorship in the published papers. I see the authors are thanking other lab members who have done tracing for the neurons described in this study, but I would like them to clarify whether they are following the guidelines provided by Flywire.

We followed the Flywire guidelines and contacted all Flywire users contributing more that 10% to neuron edits for permission to publish with acknowledgements. (see Flywire guidelines https://docs.google.com/document/d/1bUkOB5JnT3u__JDvAoVDHJ3zr5NXQtV_63yx2w6Tcc/edit).

2) The method section for voltage imaging is missing.

We now include a section on voltage imaging (ln 496-498).

3) ROIs for imaging are not indicated in the methods or in the figures. It is hard to judge what is the origin of neural activity plotted in the figures; are they imaging cell bodies, dendrites, or axons?

ROIs used for quantification are described in the figure legends: “ArcLight response of BiT soma…” (Fig 2, Fig S2), “Calcium responses of CCHa2R-RA neurites in SEZ…” (Fig 4), “Calcium response of CCHa2R-RA SEZ neurites…” (Fig S4), “Calcium response of CCAP neurites…” (Fig 5, Fig S5), “Calcium response of all IPC somas…” (Fig S3). We have added ROIs used for quantification to the ‘In vivo calcium imaging’ and the ‘In vivo voltage imaging’ methods sections (ln 493-494).

Author Response

The following is the authors’ response to the original reviews.

We would first like to thank the reviewers and the editor for their insightful comments and suggestions. We are particularly glad to read that our so<ware package constitutes a set of “well-written analysis routines” which have “the potential to become very valuable and foundational tools for the analysis of neurophysiological data”. We have updated the manuscript to address their remarks where appropriate.

Additionally, we would like to stress that this kind of tools is in continual development. As such, the manuscript offered a snapshot of the package at one point during this process, which in this case was several months ago at initial submission. Since then, several improvements were implemented. The manuscript has been further updated to reflect these more recent changes.

From the Reviewing Editor:

The reviewers identified a number of fundamental weaknesses in the paper.

1) For a paper demonstrating a toolbox, it seems that some example analyses showing the value of the approach (and potentially the advantage in simplification, etc over previous or other approaches) are really important to demonstrate.

As noted by the first reviewer, the online repository (i.e. GitHub page) conveys a better sense of the toolboxes’ contribution to the field than the present manuscript. This is a fair remark but at the same time, it is unclear how to illustrate this in a journal article without dedicating a great deal of page space to presenting raw code, while online tools offer an easier and clearer way to do this. As a work-around, our strategy was to illustrate some examples of data analysis in Figures 4&5 by comparing each illustrated processing step to the corresponding command line used by the Pynapple package. Each step requires a single line of code, meaning that one only needs to write three lines of code to decode a feature from population activity using a Bayesian decoder (Fig. 4a), compute a cross-correlograms of two neurons during specific stimulus presentation (Fig. 4b) or compute the average firing rate of two neurons around a specific time of the experimental task (Fig. 4c). We believe that these visual aides make it unnecessary to add code in the main text of this manuscript. However, to aid reader understanding, we now provide clear references to online Jupyter notebooks which show how each figure was generated in figure legends as well as in the “Code Availability” section.

https://github.com/pynapple-org/pynapple-paper-2023

Furthermore, we have opted-in for the “Executable Research Articles” feature at eLife, which will make it possible to include live scripts and figures in the manuscript once it is accepted for publication. We do not know at this stage what it entails exactly, but we hope that Figures 4&5 will become live with this feature. The readers will have the possibility to see and edit the code directly within the online version of the manuscript.

2) The manuscript's claims about not having dependencies seem confusing.

We agree that this claim was somewhat unfounded. There are virtually no Python packages that do not have dependencies. Our intention was to say that the package had no dependencies outside the most common ones, which are Numpy, Scipy, and Pandas. Too many packages in the field tend to have long list of dependencies making long-term back-compatibility quite challenging. By keeping depencies minimal, we hope to maximise the package’'s long term back-compatibility. We have rephrased this statement in the manuscript in the following sections:

Figure 1, legend.

“These methods depend only on a few, commonly used, external packages.”

Section Foundational data processing: “they are for the most part built-in and only depend on a few widely-used external packages. This ensures that the package can be used in a near stand-alone fashion, without relying on packages that are at risk of not being maintained or of not being compatible in the near future.”

3) Given its significant relevance, it seems important to cite the FMATool and describe connections between it (or analyses based on it) and the presented work.

Indeed, although we had already cited other toolboxes (including a review covering the topic comprehensively), we should have included this one in the original manuscript. Unfortunately, to the best of our knowledge, this toolbox is not citable (there is no companion paper). We have added a reference to it in plain text.

4) Some discussion of integration between Pynapple and the rest of a full experimental data pipeline should be discussed with regard to reproducibility.

This is an interesting point, and the third paragraph of the discussion somewhat broached this issue. Pynapple was not originally designed to pre-process data. However, it can, in theory, load any type of data streams a<er the necessary pre-processing steps. Overall, modularity is a key aspect of the Pynapple framework, and this is also the case for the integration with data pre-processing pipelines, for example spike sorting in electrophysiology and detection of region of interest in calcium imaging. We do not think there should be an integrated solution to the problem but, instead, to make it possible that any piece of code can be used for data irrespective of their origin. This is why we focused on making data loading straightforward and easy to adapt to any particular situation. To expand on this point and make it clear that Pynapple is not meant to pre-process data but can, in theory, load any type of data streams a<er the necessary pre-processing steps, we have added the following sentences to the aforementioned paragraph:

“Data in neuroscience vary widely in their structure, size, and need for pre-processing. Pynapple is built around the idea that raw data have already been pre-processed (for example, spike sorting and detection of ROIs).”

5) Relatedly, a description of how data are stored a<er processing (i.e., how precisely are processed data stored in NWB format).

We agree that this is a critical issue. NWB is not necessarily the best option as it is not possible to overwrite in a NWB file. This would require the creation of a new NWB file each time, which is computationally expensive and time consuming. It also further increases the odds of writing error. Theoretically, users who needs to store intermediate results in a flexible way could use any methods they prefer, writing their own data files and wrappers to reload these data into Pynapple objects. Indeed, it is not easy to properly store data in an object-specific manner. This is a long-standing issue and one we are currently working to resolve.

To do so, we are developing I/O methods for each Pynapple core objects. We aim to provide an output format that is simple to read and backward compatible in future Pynapple releases. This feature will be available in the coming weeks. To note, while NWB may not be the central data format of Pynapple in future releases, it has become a central node in the neuroscience ecosystem of so<ware. Therefore, we aim to facilitate the interaction of users with reading and writing for this format by developing a set of simple standalone functions.

Reviewer #1 (Public Review):

A typical path from preprocessed data to findings in systems neuroscience o<en includes a set of analyses that o<en share common components. For example, an investigator might want to generate plots that relate one time series (e.g., a set of spike times) to another (measurements of a behavioral parameter such as pupil diameter or running speed). In most cases, each individual scientist writes their own code to carry out these analyses, and thus the same basic analysis is coded repeatedly. This is problematic for several reasons, including the waste of time, the potential for errors, and the greater difficulty inherent in sharing highly customized code.

This paper presents Pynapple, a python package that aims to address those problems.

Strengths:

The authors have identified a key need in the community - well-written analysis routines that carry out a core set of functions and can import data from multiple formats. In addition, they recognized that there are some common elements of many analyses, particularly those involving timeseries, and their object- oriented architecture takes advantage of those commonalities to simplify the overall analysis process.

The package is separated into a core set of applications and another with more advanced applications, with the goal of both providing a streamlined base for analyses and allowing for implementations/inclusion of more experimental approaches.

Weaknesses:

There are two main weaknesses of the paper in its present form.

First, the claims relating to the value of the library in everyday use are not demonstrated clearly. There are no comparisons of, for example, the number of lines of code required to carry out a specific analysis with and without Pynapple or Pynacollada. Similarly, the paper does not give the reader a good sense of how analyses are carried out and how the object-oriented architecture provides a simplified user interaction experience. This contrasts with their GitHub page and associated notebooks which do a better job of showing the package in action.

As noted in the response to the Reviewing Editor and response to the reviewer’s recommendation to the authors below, we have now included links to Jupyter notebooks that highlight how panels of Figures 4 and 5 were generated (https://github.com/pynapple-org/pynapple-paper-2023). However, we believe that including more code in the manuscript than what is currently shown (I.e. abbreviated call to methods on top of panels in Figs 4&5) would decrease the readability of the manuscript.

Second, the paper makes several claims about the values of object-oriented programming and the overall design strategy that are not entirely accurate. For example, object-oriented programming does not inherently reduce coding errors, although it can be part of good so<ware engineering. Similarly, there is a claim that the design strategy "ensures stability" when it would be much more accurate to say that these strategies make it easier to maintain the stability of the code. And the authors state that the package has no dependencies, which is not true in the codebase. These and other claims are made without a clear definition of the properties that good scientific analysis so<ware should have (e.g., stability, extensibility, testing infrastructure, etc.).

Following thFMAe reviewer’s comment, we have rephrased and clarified these claims. We provide detailed response to these remarks in the recommendations to authors below.

There is also a minor issue - these packages address an important need for high-level analysis tools but do not provide associated tools for preprocessing (e.g., spike sorting) or for creating reproducible pipelines for these analyses. This is entirely reasonable, in that no one package can be expected to do everything, but a bit deeper account of the process that takes raw data and produces scientific results would be helpful. In addition, some discussion of how this package could be combined with other tools (e.g., DataJoint, Code Ocean) would help provide context for where Pynapple and Pynacollada could fit into a robust and reliable data analysis ecosystem.

We agree the better explaining how Pynapple is integrated within data preprocessing pipelines is essential. We have clarified this aspect in the manuscript and provide more details below.

Reviewer #1 (Recommendations For The Authors):

Page 1

• Title

The authors should note that the application name- "Pynapple" could be confused with something from Apple. Users may search for "Pyapple" as many python applications contain "py" like "Numpy". "Pyapple" indeed is a Python Apple that works with Apple products. They could consider "NeuroFrame", "NeuroSeries" or "NeuroPandas" to help users realize this is not an apple product.

We thank the referee for this interesting comment. However, we are not willing to make such change at this point. The community of users has been growing in the last year and it seems too late to change the name. To note, it is the first time such comment is made to us and it does not seem that users and collaborators are confused with any Apple products.

• Abstract

The authors mentioned that the Pynapple is "fully open source". It may be better to simply say it is "open source".

We agree, corrected.

Assuming the authors keep the name, it would be helpful if the full meaning of Pynapple - Python Neural Analysis Package was presented as early as possible.

Corrected in the abstract.

• Highlight

An application being lightweight and standalone does not imply nor ensure backward compatibility. In general, it would be useful if the authors identified a set of desirable code characteristics, defined them clearly in the introduction, and then describe their so<ware in terms of those characteristics.

Thank you for your comment. We agree that being lightweight and standalone does not necessarily imply backward compatibility. Our intention was to emphasize that Pynapple is designed to be as simple and flexible as possible, with a focus on providing a consistent interface for users across different versions. However, we understand that this may not be enough to ensure long-term stability, which is why we are committed to regular updates and maintenance to ensure that the code remains functional as the underlying code base (Python versions, etc.) changes.

Regarding your suggestion to identify a set of desirable code characteristics, we believe this is an excellent idea. In the introduction, we briefly touch upon some of the core principles that guided our development of Pynapple: a lightweight, stable, and simple package. However, we acknowledge that providing a more detailed discussion of these characteristics and how they relate to the design of our so<ware would be useful for readers. We have added this paragraph in the discussion:

“Pynapple was developed to be lightweight, stable, and simple. As simplicity does not necessarily imply backward compatibility (i.e. long-term stability of the code), Pynapple main objects and their properties will remain the same for the foreseeable future, even if the code in the backend may eventually change (e.g. not relying on Pandas in future version). The small number of external dependencies also decrease the need to adapt the code to new versions of external packages. This approach favors long-term backward compatibility.”

Page 2

• The authors wrote -

"Despite this rapid progress, data analysis o<en relies on custom-made, lab-specific code, which is susceptible to error and can be difficult to compare across research groups."

It would be helpful to add that custom-made, lab-specific code can lead to a violation of FAIR principles (https://en.wikipedia.org/wiki/FAIR_datadata). More generally, any package can have errors, so it would be helpful to explain any testing regiments or other approach the authors have taken to ensure that their code is error-free.

We understand the importance of the FAIR principles for data sharing. However, Pynapple was not designed to handle data through their pre-processing. The only aspect that is somehow covered by the FAIR principles is the interoperability, but again, it is a requirement for the data to interoperate with different storage and analysis pipelines, not of the analysis framework itself. Unlike custom-made code, Pynapple will make interoperability easier, as, in theory, once the required data loaders are available, any analysis could be run on any dataset. We have added the following sentence to the discussion:

“Data in neuroscience vary widely in their structure, size, and need for pre-processing. Pynapple is built around the idea that raw data has already been pre-processed (for example, spike sorting and ROI detection). According to the FAIR principles, pre-processed data should interoperate across different analysis pipelines. Pynapple makes this interoperability possible as, once the data are loaded in the Pynapple framework, the same code can be used to analyze different datasets”

• The authors wrote -

"While several toolboxes are available to perform neuronal data analysis ti–11,2ti (see ref. 29 for review), most of these programs focus on producing high-level analysis from specified types of data and do not offer the versatility required for rapidly-changing analytical methods and experimental methods."

Here it would be helpful if the authors could give a more specific example or explain why this is problematic enough to be a concern. Users may not see a problem with high-level analysis or using specific data types.

Again, we apologize for not fully elaborating upon our goals here. Our intention was to point out that toolboxes o<en focus on one particular case of high-level analysis. In many cases, such packages lack low level analysis features or the flexibility to derive new analysis pipelines quickly and effortlessly. Users can decide to use low-level packages such as Pandas, but in that case, the learning curve can be steep for users with low, if any, computational background. The simplicity of Pynapple, and the set of examples and notebooks, make it possible for individuals who start coding to be quickly able to analyze their data.

As we do not want to be too specific at this point of the manuscript (second paragraph of the intro) and as we have clarified many of the aspects of the toolbox in the new revised version, we have only added the following sentence to the paragraph:

“Users can decide to use low-level data manipulation packages such as Pandas, but in that case, the learning curve can be steep for users with low, if any, computational background.”

• The authors wrote -

"To meet these needs, a general toolbox for data analysis must be designed with a few principles in mind"

Toolboxes based on many different principles can solve problems. It is likely more accurate to say that the authors designed their toolbox with a particular set of principles in mind. A clear description of those principles (as mentioned in the comment above) would help the reader understand why the specific choices made are beneficial.

We agree that these are not “universal” principles and clearly more the principles we had in mind when we designed the package. We have clarified these principles and made clear that these are personal point of views.

We have rephrased the following paragraph:

“To meet these needs, we designed Pynapple, a general toolbox for data analysis in systems Neuroscience with a few principles in mind.“

• The authors wrote -

"The first property of such a toolbox is that it should be object-oriented, organizing so<ware around data."

What facts make this true? For example, React is a web development library. A common approach to using this library is to use Hooks (essentially a collection of functions). This is becoming more popular than the previous approach of using Components (a collection of classes). This is an example of how Object-oriented programming is not always the best solution. In some cases, for example, object- oriented coding can cause problems (e.g. it can be hard to find the place where a given function is defined and to figure out which version is being used given complex inheritance structures.)

In general, key selling points of object-oriented programming are extension, inheritance, and encapsulation. If the authors want to retain this text (which would be entirely reasonable), it would be helpful if they explained clearly how an object-oriented approach enables these functions and why they are critical for this application in particular.

The referee makes a particularly important point. We are aware of the limits of OOP, especially when these objects become over-complex, and that the inheritance become unclear.

We have clarified our goal here. We believe that in our case, OOP is powerful and, overall, is less error- prone that a collection of functions. The reasons are the following:

An object-oriented approach facilitates better interactions between objects. By encapsulating data and behavior within objects, object-oriented programming promotes clear and well-defined interfaces between objects. This results in more structured and manageable code, as objects communicate with each other through these well-defined interfaces. Such improved interactions lead to increased code reliability.

Inheritance, a key concept in object-oriented programming, allows for the inheritance of properties. One important example of how inheritance is crucial in the Pynapple framework is the time support of Pynapple objects. It determines the valid epoch on which the object is defined. This property needs to be carried over during different manipulations of the object. Without OOP, this property could easily be forgotten, resulting in erroneous conclusions for many types of analysis. The simplest case is the average rate of a TS object: the rate must be computed on the time support ( a property of TS objects), not the beginning to the end of the recording (or of a specific epoch, independent of the TS). Finally, it is easier to access and manipulate the meta information of a Pynapple object than without using objects.

• The authors wrote -

"drastically diminishing the odds of a coding error"

This seems a bit strong here. Perhaps "reducing the odds" would be more accurate.

We agree. Now changed.

Page 3

• The authors wrote -

". Another property of an efficient toolbox is that as much data as possible should be captured by only a small number of objects This ensures that the same code can be used for various datasets and eliminates the need of adapting the structure"

It may be better to write something like - "Objects have a collection of preset variables/values that are well suited for general use and are very flexible." Capturing "as much data as possible" may be confusing, because it's not the amount that this helps with but rather the variety.

We thank the referee for this remark. We have rephrased this sentence as follows:

“Another property of an efficient toolbox is that a small number of objects could virtually represents all possible data streams in neuroscience, instead of objects made for specific physiological processes (e.g. spike trains).”

• The authors wrote -

"The properties listed above ensure the long-term stability of a toolbox, a crucial aspect for maintaining the code repository. Toolboxes built around these principles will be maximally flexible and will have the most general application"

There are two issues with this statement. First, ensuring long-term stability is only possible with a long- term commitment of time and resources to ensure that that code remains functional as the underlying code base (python versions, etc.) changes. If that is something you are commisng to, it would be great to make that clear. If not, these statements need to be less firm.

Second, it is not clear how these properties were arrived at in the first place. There are things like the FAIR Principles which could provide an organizing framework, ideally when combined with good so<ware engineering practices, and if some more systematic discussion of these properties and their justification could be added, it would help the field think about this issue more clearly.

The referee makes a valid point that ensuring long-term stability requires a long-term commitment of time and resources to maintain the code as the underlying technology evolves. While we cannot make guarantees about the future of Pynapple, we believe that one of the best ways to ensure long-term stability is by fostering a strong community of users and contributors who can provide ongoing support and development. By promoting open-source collaboration and encouraging community involvement, we hope to create a sustainable ecosystem around Pynapple that can adapt to changes in technology and scientific practices over time. Ultimately, the longevity of any scientific tool depends on its adoption and use by the research community, and we hope that Pynapple can provide value to neuroscience researchers and continue to evolve and improve as the field progresses.

It is noteworthy that the first author, and main developer of the package, has now been hired as a data scientist at the Center for Computational Neuroscience, Flatiron Institute, to explicitly continue the development of the tool and build a community of users and contributors.

• The authors wrote -

"each with a limited number of methods..."

This may give the impression that the functionality is limited, so rephrasing may be helpful.

Indeed! We have now rephrased this sentence:

“The core of Pynapple is five versatile timeseries objects, whose methods make it possible to intuitively manipulate and analyze the data.”

• The authors wrote that object-oriented coding

"limits the chances of coding error"

This is not always the case, but if it is the case here, it would be helpful if the authors explain exactly how it helps to use object-oriented approaches for this package.

We agree with the referee that it is not always the case. As we explained above, we believe it is less error-prone that a collection of functions. Quite o<en, it also makes it easier to debug. We have changed this sentence with the following one:

“Because objects are designed to be self-contained and interact with each other through well-defined methods, users are less likely to make errors when using them. This is because objects can enforce their own internal consistency, reducing the chances of data inconsistencies or unexpected behavior. Overall, OOP is a powerful tool for managing complexity and reducing errors in scientific programming.”

• Fig 1

In object-oriented programming, a class is a blueprint for the classes that inherit it. Instantiating that<br /> class creates an object. An object contains any or all of these - data, methods, and events. The figure could be improved if it maintained these organizational principles as figure properties.

We agree with the referee’s remark regarding the logic of objects instantiation but how this could be incorporated in Fig. 1 without making it too complex is unclear. Here, objects are instantiated from the first to the second column. We have not provided details about the parent objects, as we believe these details are not important for reader comprehension. In its present form, the objects are inherited from Pandas objects, but it is possible that a future version is based on something else. For the users, this will be transparent as the toolbox is designed in such a way that only the methods that are specific to Pynapple are needed to do most computation, while only expert programmers may be interested in using Pandas functionalities.

• The authors wrote that Pynapple does -

"not depend on any external package"

As mentioned above, this is not true. It depends on Numpy and likely other packages, and this should be explained. It is perfectly reasonable to say that it depends on only a few other packages.

As said above, we have now clarified this claim.

Page 5.

• The authors wrote -

"represent arrays of Ts and Tsd"

For a knowledgeable reader's reference, it would be helpful to refer to these either as Numpy arrays (at least at first when they are defined) or as lists if they are native python objects.

Indeed, using the word “arrays” here could be confusing because of Numpy arrays. We have changed this term with “groups”.

• The authors wrote -

"Pynapple is built with objects from the Pandas library ... Pynapple objects inherit the computational stability and flexibility"

Here a definition of stability would be useful. Is it the case that by stability you mean "does not change o<en"? Or is some other meaning of stability implied?

Yes, this is exactly what we meant when referring to the stability of Pandas. We have added the following precision:

“As such, Pynapple objects inherit the long-term consistency of the code and the computational flexibility computational stability and flexibility from this widely used package.”

Page 6

• Fig 2

In Fig 2 A and B, the illustrations are good. It would also be very helpful to use toy code examples to illustrate how Pynapple will be used to carry out on a sample analysis-problem so that potential users can see what would need to be done.

We appreciate the kind works. Regarding the toy code, this is what we tried to do in Fig. 4. Instead of including the code directly in the paper, which does not seem a modern way of doing this, we now refer to the online notebooks that reproduce all panels of Figure 4.

• The authors wrote -

"While these objects and methods are relatively few"

In object-oriented programming, objects contain methods. If a method is not in an object, it is not technically a method but a function. It would be helpful if the authors made sure their terminology is accurate, perhaps by saying something like "While there are relatively few objects, and while each object has relatively few methods ... "

We agree with the referee, we have changed the sentence accordingly.

• The authors wrote -

"if not implemented correctly, they can be both computationally intensive and highly susceptible to user error"

Here the authors are using "correctly" to refer to two things - "accuracy" - gesng the right answer, and "efficiency" - gesng to that answer with relatively less computation. It would be clearer if they split out those two concepts in the phrasing.

Indeed, we used the term to cover both aspects of the problem, leading to the two possible issues cited in the second part of the sentence. We have changed the sentence following the referee’s advice:

“While there are relatively few objects, and while each object has relatively few methods, they are the foundation of almost any analysis in systems neuroscience. However, if not implemented efficiently, they can be computationally intensive and if not implemented accurately, they are highly susceptible to user error.”

• In the next sentence the authors wrote -

"Pynapple addresses this concern."

This statement would benefit from just additional text explaining how the concern is addressed.

We thank the referee for the suggestion. We have changed the sentence to this one: “The implementation of core features in Pynapple addresses the concerns of efficiency and accuracy”

Page 9

• The authors wrote -

This is implemented via a set of specialized object subclasses of the BaseLoader class. To avoid code redundancy, these I/O classes inherit the properties of the BaseLoader class. "

From a programming perspective, the point of a base class is to avoid redundancy, so it might be better to just mention that this avoids the need to redefine I/O operations in each class.

We have rephrased the sentence as follows:

“This is implemented via a set of specialized object subclasses of the BaseLoader class, avoiding the need to redefine I/O operations in each subclass"

• The authors wrote -

"classes are unique and independent from each other, ensuring stability"

How do classes being unique and independent ensure stability? Perhaps here again the misunderstanding is due to the lack of a definition of stability.

We thank the referee for the remark. We first changed “stability” for “long-term backward compatibility”. We further added the following sentence to clarify this claim. “For instance, if the spike sorting tool Phy changes its output in the future, this would not affect the “Neurosuite” IO class as they are independent of each other. This allows each tool to be updated or modified independently, without requiring changes to the other tool or the overall data format.”

• The authors wrote -

"Using preexisting code to load data in a specific manner instead of rewriting already existing functions avoids preprocessing errors"

Here it might be helpful to use the lingo of Object-oriented programming. (e.g. inheritance and polymorphism). Defining these terms for a neuroscience audience would be useful as well.

We do not think it is necessary to use too much technical term in this manuscript. However, this sentence was indeed confusing. We have now simplified it:

“[…], users can develop their own custom I/O using available template classes. Pynapple already includes several of such templates and we expect this collection to grow in the future.”

Page 10

• The authors wrote -

"These analyses are powerful because they are able to describe the relationships between time series objects while requiring the fewest number of parameters to be set by the user."

It is not clear that this makes for a powerful analysis as opposed to an easy-to-use analysis.

We have changed “powerful” with “easy to use".

Page 12

"they are built-in and thus do not have any external dependencies"

If the authors want to retain this, it would be helpful to explain (perhaps in the introduction) why having fewer external dependencies is useful. And is it true that these functions use only base python classes?

We have rephrased this sentence as follows:

“they are for the most part built-in and only depend on a few common external packages, ensuring that they can be used stand-alone without relying on packages that are at risk of not being maintained or of not being compatible in the near future.”

Other comments:

• It would be helpful, as mentioned in the public review, to frame this work in the broader context of what is needed to go from data to scientific results so that people understand what this package does and does not provide.

We have added the following sentence to the discussion to make sure readers understand:

“The path from data collection to reliable results involves a number of critical steps: exploratory data analysis, development of an analysis pipeline that can involve custom-made developed processing steps, and ideally the use of that pipeline and others to replicate the results. Pynapple provides a platform for these steps.”

• It would also be helpful to describe the Pynapple so<ware ecosystem as something that readers could contribute to. Note here that GNU may not be a good license. Technically, GNU requires any changes users make to Pynapple for their internal needs to be offered back to the Pynapple team. Some labs may find that burdensome or unacceptable. A workaround would be to have GNU and MIT licenses.

The main restriction of the GPL license is that if the code is changed by others and released, a similar license should be used, so that it cannot become proprietary. We therefore stick to this choice of license.

We would be more than happy to receive contributions from the community. To note, several users outside the lab have already contributed. We have added the following sentence in the introduction:

“As all users are also invited to contribute to the Pynapple ecosystem, this framework also provides a foundation upon which novel analyses can be shared and collectively built by the neuroscience community.”

• This so<ware shares some similarities with the nelpy package, and some mention of that package would be appropriate.

While we acknowledge the reviewer's observation that Nelpy is a similar package to Pynapple, there are several important differences between the two.

First, Nelpy includes predefined objects such as SpikeTrain, BinnedSpikeTrain, and AnalogSignal, whereas Pynapple would use only Ts and Tsd for those. This design choice was made to provide greater flexibility and allow users to define their own data structures as needed.

Second, Nelpy is primarily focused on electrophysiology data, whereas Pynapple is designed to handle a wider range of data types, including calcium imaging and behavioral data. This reflects our belief that the NWB format should be able to accommodate diverse experimental paradigms and modalities.

Finally, while Nelpy offers visualization and high-level analysis tools tailored to electrophysiology, Pynapple takes a more general-purpose approach. We believe that users should be free to choose their own visualization and analysis tools based on their specific needs and preferences.

The package has now been cited.

Reviewer #2 (Public Review):

Pynapple and Pynacollada have the potential to become very valuable and foundational tools for the analysis of neurophysiological data. NWB still has a steep learning curve and Pynapple offers a user- friendly toolset that can also serve as a wrapper for NWB.

The scope of the manuscript is not clear to me, and the authors could help clarify if Pynacollada and other toolsets in the making become a future aspect of this paper (and Pynapple), or are the authors planning on building these as separate publications.

The author writes that Pynapple can be used without the I/O layer, but the author should clarify how or if Pynapple may work outside NWB.

Absolutely. Pynapple can be used for generic data analysis, with no requirement of specific inputs nor NWB data. For example, the lab is currently using it for a computational project in which the data are loaded from simple files (and not from full I/O functions as provided in the toolbox) for further analysis and figure generation.

This was already noted in the manuscript, last paragraph of the section “Importing data from common and custom pipelines”

“Third, users can still use Pynapple without using the I/O layer of Pynapple.”.

We have added the following sentence in the discussion

“To note, Pynapple can be used without the I/O layer and independent of NWB for generic, on-the-fly analysis of data.”

This brings us to an important fundamental question. What are the advantages of the current approach, where data is imported into the Ts objects, compared to doing the data import into NWB files directly, and then making Pynapple secondary objects loaded from the NWB file? Does NWB natively have the ability to store the 5 object types or are they initialized on every load call?

NWB and Pynapple are complimentary but not interdependent. NWB is meant to ensure long-term storage of data and as such contains a as much information as possible to describe the experiment. Pynapple does not use NWB to directly store the objects, however it can read from NWB to organize the data in Pynapple objects. Since the original version of this manuscript was submitted, new methods address this. Specifically, in the current beta version, each object now has a “save” method. Obviously, we are developing functions to load these objects as well. This does not depend on NWB but on npz, a Numpy specific file format. However, we believe it is a bit too premature to include these recent developments in the manuscript and prefer not to discuss this for now.

Many of these functions and objects have a long history in MATLAB - which documents their usefulness, and I believe it would be fisng to put further stress on this aspect - what aspects already existed in MATLAB and what is completely novel. A widely used MATLAB toolset, the FMA toolbox (the Freely moving animal toolbox) has not been cited, which I believe is a mistake.

We agree that the FMA toolbox should have been cited. This ha now been corrected.

Pynapple was first developed in Matlab (it was then called TSToolbox). The first advantage is of course that Python is more accessible than Matlab. It has also been adopted by a large community of developers in data analysis and signal processing, which has become without a doubt much larger than the Matlab community, making it possible to find solutions online for virtually any problem one can have. Furthermore, in our experience, trainees are now unwilling to get training in Matlab.

Yet, Python has drawbacks, which we are fully aware of. Matlab can be very computationally efficient, and old code can usually run without any change, even many years later.

A limitation in using NWB files is its standardization with limited built-in options for derived data and additional metadata. How are derived data stored in the NWB files?

NWB has predetermined a certain number of data containers, which are most common in systems neuroscience. It is theoretically possible to store any kind of data and associated metadata in NWB but this is difficult for a non-expert user. In addition, NWB does not allow data replacement, making is necessary to rewrite a whole new NWB file each time derived data are changed and stored. Therefore, we are currently addressing this issue as described above. Derived data and metadata will soon be easy to store and read.

How is Pynapple handling an existing NWB dataset, where spikes, behavioral traces, and other data types have already been imported?

This is an interesting point. In theory, Pynapple should be able to open a NWB file automatically, without providing much information. In fact, it is challenging to open a NWB file without knowing what to look for exactly and how the data were preprocessed. This would require adapting a I/O function for a specific NWB file. Unfortunately, we do not believe there is a universal solution to this problem. There are solutions being developed by others, for example NWB Widgets (NWB Widgets). We will keep an eye on this and see whether this could be adapted to create a universal NWB loader for Pynapple.

Reviewer #2 (Recommendations For The Authors):

Other tools and solutions are being developed by the NWB community. How will you make sure that these tools can take advantage of Pynapple and vice versa?

We recognize the importance of collaboration within the NWB community and are committed to making sure that our tools can integrate seamlessly with other tools and solutions developed by the community.

Regarding Pynapple specifically, we are designing it to be modular and flexible, with clear APIs and documentation, so that other tools can easily interface with it. One important thing is that we want to make sure Pynapple is not too dependent of another package or file format such as NWB. Ideally, Pynapple should be designed so that it is independent of the underlying data storage pipeline.

Most of the tools that have been developed in the NWB community so far were designed for data visualisation and data conversion, something that Pynapple does not currently address. Multiple packages for behavioral analysis and exploration of electro/optophysiological datasets are compatible with the NWB format but do not provide additional solutions per se. They are complementary to Pynapple.

Reviewer #1 (Public Review):

The authors developed an extension to the pairwise sequentially Markov coalecent model that allows to simultaneously analyze multiple types of polymorphism data. In this paper, they focus on SNPs and DNA methylation data. Since methylation markers mutate at a much faster rate than SNPs, this potentially gives the method better power to infer size history in the recent past. Additionally, they explored a model where there are both local and regional epimutational processes.

Integrating additional types of heritable markers into SMC is a nice idea which I like in principle. However, a major caveat to this approach seems to be a strong dependence on knowing the epimutation rate. In Fig. 6 it is seen that, when the epimutation rate is known, inferences do indeed look better; but this is not necessarily true when the rate is not known. A roughly similar pattern emerges in Supp. Figs. 4-7; in general, results when the rates have to be estimated don't seem that much better than when focusing on SNPs alone. This carries over to the real data analysis too: the interpretation in Fig. 7 appears to hinge on whether the rates are known or estimated, and the estimated rates differ by a large amount from earlier published ones.

Overall, this is an interesting research direction, and I think the method may hold more promise as we get more and better epigenetic data, and in particular better knowledge of the epigenetic mutational process. At the same time, I would be careful about placing too much emphasis on new findings that emerge solely by switching to SNP+SMP analysis.

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

Learn more at Review Commons

---

Reply to the reviewers

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

Summary: This manuscript by Lan et al. addresses the still incompletely resolved question as to how branching morphogenesis of the embryonic mammary epithelium is regulated at the molecular and cellular level. Using (combinatorial) primary explant cultures of wildtype and genetically engineered mouse embryos, in which the authors have developed a unique expertise over many years, together with imaging and RNAseq analyses, they (i) show that the timing of epithelial branching is dictated by the biological age of the epithelium, but that an epithelial-mesenchymal interaction is required to bestow branching ability on the mammary epithelium somewhere between E13.5 and E16.5, (ii) seek to determine if and how lineage and cell proliferation affect branching, (iii) show that while salivary mesenchyme can promote growth (i.e. branching density) of the E16.5 mammary epithelium, the mode of branching (i.e. lateral branching vs tip-clefting) is an intrinsic property of the mammary epithelium, (iv) use transcriptomics to identify genes that are likely to control either mammary- or salivary gland specific growth and/or branching patterns, (v) hypothesize that low levels of WNT signaling in the mammary gland mesenchyme (due to relatively high expression of WNT signaling inhibitors) are responsible for mammary specific branching, (vi) show that hyperactivation of WNT/CTNNB1 signaling in the mesenchyme indeed induces hyperbranching, (vii) identify Eda and Igf1 as putative mediators and paracrine signaling factors that regulate branching of the mammary epithelium upon secretion from the mesenchyme downstream of WNT/CTNNB1 signaling and (viii) show that mammary gland branching is impaired in Igfr1 null embryos.

Major comments: 1. Overall, this is a solid study that is well controlled and technically of high quality. The materials and methods should allow follow up and replication by others and the transcriptomic data have been made available via NCBI GEO. I think the authors convincingly demonstrate points (i), (iii), (iv) and (vi) and (viii). I have some questions regarding (ii), (v) and (vii) and (viii) that I will pose below.

Our response:

We thank the reviewer for the careful assessment and recognition of our work. In the subsequent sections, we have tried to address all the concerns raised by the reviewer.

Re: (ii): The authors try to study the link between basal cell fate and branching. They use position of the cells (which they describe clearly and which is a good choice), since they cannot use specific markers due to the fact that the basal and luminal linages have not yet segregated at this point. This part of the manuscript is not the most straightforward to follow. The most obvious experiment would have been to focus on the location of the cells and their associated cell cycle profile - but the authors themselves have just recently published a pre-print (their REF #54, now also out in JCB) that is an in-depth study of the link between cell proliferation + cell motility and branching, but this only becomes apparent in the discussion. In that sense, Fig2 of the current manuscript is less novel, although it is nice to see that it holds up in a slightly different analysis.

Our response:

We thank the reviewer for acknowledging our recently published work, which is focusing on the active branching phase during late embryogenesis/around birth. In the current proliferation analysis, however, our focus was on a different aspect of embryonic mammary gland development: understanding the mechanism underlying the ability to acquire competence to branch, i.e. how the epithelium changes between late bud and sprout stages. Our data obtained from tissue recombination and 3D culture experiments suggest that heterotypic mesenchymes or mesenchyme-free 3D organoid culture conditions do not provide sufficient signals to support branching of mammary epithelia before E16.5. We have rephrased the text to better emphasize this point.

Instead of focusing on the cell cycle markers, the authors turn to a K14-Eda mouse model - which shows precocious branching and a temporary reduction in K8 expression. They also analyze Eda-KO embryos. Quite frankly, I find the authors' reasoning difficult to follow here and I cannot deduce how these experiments really address the question at hand (i.e. how lineage and cell proliferation affect branching), so I hope they can rewrite this section of the paper to make the arguments more clear and easy to follow for the reader who, at this point, knows little about Eda. For example, the authors present the argument that K14-Eda mice show a transient reduction in K8 expression - but we don't know if that also really means a (temporary?) change in (future?) luminal cell fate. In fact, since Eda later also makes an appearance as a candidate factor to be secreted by the mesenchyme together with Igf1, I wonder if their K14-Eda data would not be better suited to underscore that point instead and if the authors should perhaps eliminate this section altogether and just refer to their prior work in REF #45. If the authors think the current data add something more, than they need to be more explicit about this (and then also introduce the link to REF #45 in the results section).

__Our response: __

We agree with all the reviewers in that this part of the manuscript was not mature enough and provided only indirect evidence on the potential link between lineage segregation and branching ability. This is an important question in the field that merits a study of its own and should be addressed with better tools than those available to us at present. As suggested by reviewers #1 and #3, we have omitted this part in the revised manuscript.

Re: (v): Do the authors have any WNT/CTNNB1 target genes that they can include in their transcriptomics analysis to show that the WNT/CTNNB1 signaling levels are indeed lower in the mammary mesenchyme? Axin2 comes to mind, but there are some other negative feedback targets that are often induced across tissues, e.g. Rnf43 and/or Znrf3 and/or Sp5?E.g. to include in FIg6E?

__Our response: __

In the original manuscript (lines 339-342), we have performed the GSVA analysis comparing the KEGG database, and the significantly altered pathways comparing different mammary mesenchymes with salivary gland mesenchyme have been pooled and displayed as heatmap in Supplementary Fig 4b. The WNT signaling pathway is lower in the mammary mesenchyme, especially at E16.5.

As suggested by the reviewer, we have analyzed Axin2, the most commonly used readout of WNT/CTNNB1 signaling activity in our RNA-seq data that we include as a __new Supplemental Fig. 4c __in the preliminary revised manuscript. Axin2 data indicate that Wnt/β-catenin signaling activity is lower in the E16.5 fat pad, where branching takes place, compared to younger stages of mammary gland and the salivary gland.

Plan for the final revision:

Additionally, we will provide expression data of a transgenic Wnt reporter from the same developmental stages and tissues that were used to generate the RNA-seq data.

Re: (vii) and (viii): The authors convincingly show the phenotype of the Igfr1 KO mice, but I hope the authors concur that an epithelial only Igfr1 KO (or alternatively a mesenchymal only Igf1 KO, or epithelial/mesenchymal recombination experiments with WT vs IGFR1 null or IGF1 null tissue, or experiments with small molecule inhibitors of IGF1/IGFR1 signaling) would have given more solid mechanistic evidence regarding the presumed paracrine effect of IGF1 signaling. I am not asking the authors to perform another mouse experiment or even generate or use these conditional strains, but if the authors agree, then I do think this would merit some attention in the discussion section. See also my comments regarding Eda in point 1.

Our response:

As shown in the current manuscript, Igf1 is expressed in the mammary and salivary gland mesenchyme. This finding is in line with E14 in situ expression data available in Genepaint (https://gp3.mpg.de/results/Igf1) showing that overall in embryonic tissues, Igf1 is mainly produced in mesenchymal tissues. Of note, in Genepaint, a clear signal can be detected in the salivary gland mesenchyme, not the epithelium. Published E16 and E18 datasets indicate low level of Igf1 expression in the mammary epithelium (https://wahl-lab-salk.shinyapps.io/Mammary_snATAC/). Hence, we conclude that Igf1 is mainly produced by mesenchymal cells. Instead, Igf1r appears to be rather ubiquitously expressed.

A previous study assessed BrdU incorporation in Igf1r-/- mammary buds at E14.5, and reported a specific proliferation defect in the epithelium, while no difference was detected in the mesenchyme (Fig. 9, Heckman et al., 2007; PMID:17662267). However, we cannot exclude the possibility of autocrine, mesenchymal Igf1/Igf1r signaling, which in turn could lead to upregulation of a paracrine factor to regulate epithelial growth.

We agree with the reviewer in that novel conditional mouse models are beyond the scope of the current study. However, we do not think that small molecule drugs could be used to block Igf1r activity in a tissue-specific manner neither.

Plan for the final revision:

To further delineate the paracrine and/or autocrine role of Igf1/Igf1r pathway during mammary epithelial growth and branching, we will perform tissue recombination experiments between Igf1r-/- and control mammary epithelium and mesenchyme, as suggested by the reviewer.

Minor comments: - A few minor spelling/grammar errors, including a couple of "the"s missing (first line of the abstract, and also preceding "Majority" in line 148.

Our response:

We apologize for these slips. They have been corrected in the revised manuscript.

• Line 517-518: please also include the details for the Eda mice.

Our response:

We apologize for missing this important information in materials and methods. We have included a short introduction of the K14-Eda mice, a new reference for the original publication producing them, as well as the Jackson Laboratories strain number for Eda-/- (a.k.a. Tabby) mice in the revised manuscript.

• 1f spelling error: separation

Our response:

The spelling error has been corrected in the revised manuscript.

**Referees cross-commenting**

Having read all three review reports I think they are pretty much in agreement, with shared questions about the inclusion/meaning/discussion of the lineage specification data and also agreement about the overall technical solidity of the data and this approach.

I gather that reviewer #2 asks for more controls than myself or reviewer #3 and while I think all of their points are valid, in principle, I don't think all of these are required. I should add that I am inclined to trust the authors on their ability to separate mesenchyme and epithelium as they have been developing and optimising this system over many years.

Our response:

We are grateful to the reviewer for the reliance on the technical aspect of our experiments. We do routinely monitor tissue purity in the recombinants (for more details, see our response to reviewer #2). To demonstrate this, we have included new data in new Supplementary Fig. 1a,b and new Supplementary Fig. 3. We believe these additions will further enhance the validity of our findings and effectively address the concerns raised by reviewer 2.

Reviewer #1 (Significance (Required)):

General assessment: This is a carefully executed study in which an impressive amount of (combinatorial) embryonic mammary tissue explant experiments are combined with quantitative imaging and transcriptomics analysis.

The main limitations of the work lie in the fact that the investigation of a potential link between branching and the cell cycle is not entirely novel, as the authors themselves recently published an nice pre-print (now also out in JCB) describing similar analyses. In addition, the mechanistic link between WNT/CTNNB1 signaling in the mesenchyme and the paracrine signaling activities of the presumed downstream effectors EDA and IGF, while plausible, is not yet complete. The work also does not yet addresses what exactly the branching identity is that is bestowed upon the mammary epithelium between E13.5 and E16.5 and how this then becomes an intrinsic (epigenetic?) feature of the mammary gland.

Advance: This work provides more insight into the embryonic branching of the mammary gland - a stage of mammary gland development that is still poorly understood and that is, in general, understudied. In part, the work confirms prior work in the literature (their REF #19) regarding mammary and salivary gland tissue recombination experiments. It supplements this with a more elaborate time series of heterochronic and heterologous epithelium/mesenchyme explant cultures, using genetically engineered (and fluorescently labeled) mouse tissues to allow better and quantitative imaging. The transcriptomic analysis of different mesenchyme populations is also informative and allows the researchers to propose a putative mechanism for why the mammary gland branches differently from the salivary gland. The advance is both technical and functional, as well as conceptual, with some advance in terms of mechanism.

Audience: This works should appeal to mammary gland biologists interested in the molecular and cellular mechanisms of (early) mammary gland development, as well as to a broader community of developmental biologists studying branching morphogenesis in tissues such as lung, kidney and salivary gland.

My expertise: WNT signaling and mammary gland biology, at the intersection of developmental, stem cell and cancer biology

__Reviewer #2 (Evidence, reproducibility and clarity (Required)): __

The mammary gland is a branched structure that consists of a bilayered epithelium embedded in a specialized mesenchyme. In mice, at 11,5 days of embryogenesis, the ectoderm thickens forming 5 pairs of peculiar structures called placodes. During the following days, the placodes will grow and invaginate into the surrounding mammary mesenchyme and they will finally start to branch by the end of embryogenesis (E16). It has been suggested that the bidirectional communication between the growing mammary gland and the surrounding mesenchyme plays a pivotal role in the determination of each step of mammary gland development (placode formation, mammary bud invagination, gland outgrowth, branching). The role of different signalling molecules has already been shown, particularly for the placode growth and mammary bud invagination. Nevertheless, the pathways regulating embryonic mammary gland branching are still incompletely understood. In this manuscript, Lan and colleagues aim to decipher the correlation between different stages of mammary gland development such as proliferation, lineage segregation and ductal branching. Furthermore, they want to define which stage of mammary development is intrinsically determined by the epithelium and which one requires the supportive guidance of the mesenchyme. Lastly, they aim to discover the key signal for the growth and branching of mammary epithelium. To these purposes, they used an ex vivo model of heterochronic epithelial-mesenchymal recombination. In particular, they micro-dissected the epithelium and/or the mesenchyme from murine mammary glands at different stages of embryonic development (i.e. at E13,5 for the quiescent phase or 16,5 for branching phase) and explanted them together in different combinations using fluorescent reporters. To assess the role of the mesenchyme they also cultured the epithelium in a mesenchyme free 3d structure. Through this model they demonstrated that the presence of the mesenchyme is necessary for the priming of mammary epithelium for branching, since only E16,5 epithelial cells were able to grow and branch in a mesenchyme free 3D experiment. Nevertheless, intrinsic properties of the epithelium are necessary for the timing of branching, since E16,5 mesenchyme was not able to accelerate the outgrowth of E13,5 epithelia. In order to determine which epithelial properties are important, the authors correlated the beginning of cell proliferation in the embryonic mammary gland to the beginning of the branching phase. They indeed used the Fucci2a mouse model to carefully characterise the timing of mammary cells proliferation at different stages of embryonic development, concluding that the great majority of proliferating cells reside in the inner part of the mammary bud until E14,5, while in the external part at later stages. Regarding the importance of cell proliferation, Lan and colleagues claim that the beginning of the branching phase is not its direct consequence, thanks to the use of the K14Cre- Eda mouse model, known to have anticipated mammary gland development. Using this and the Eda-/- models, the authors also sustain that the branching occurs independently of the lineage specification of the epithelium. The use of salivary mesenchyme instead the mammary one was able to increase the number of branching of E16,5 mammary epithelium. Nevertheless, this model demonstrated that the branching pattern (side branching vs tip bifurcation) is an intrinsic feature of the epithelium. Lan and colleagues also defined the transcriptomic profiles of the mammary and salivary mesenchymes at different stages. In particular, they observed an increased expression of negative regulators of Wnt pathway in the mammary mesenchyme compared to the salivary mesenchyme. Moreover, using a mouse model where B-catenin is stabilised, they observed increased tip production in the mammary gland epithelium. They also showed that IGF1 production is increased after Wnt pathway activation and they tested its function, both treating their ex vivo cultures with exogenous IGF1 and using Igf1r-/- mouse models.

Major comments 1- The great majority of the results of the manuscript are based on an ex vivo model of heterochronic epithelial-mesenchymal recombination. Since the authors are studying the effect of the mesenchyme of different stages on the epithelium (and vice versa), the purity of the two compartments after the dissection is particularly important. Although they said that the purity is evaluated (line 112), it would be important to show a control staining in which they use known markers of the mesenchyme with no colocalization with the fluorescent reporter of the epithelium.

Our response:

We agree with the reviewer that the purity of the separated tissues is very important for our conclusions. This is why we have used genetically labeled tissues in all recombination experiments: the epithelium and the mesenchyme were always isolated from embryos ubiquitously expressing GFP or tdTomato. We find this the most reliable way to assess the origin and purity of the isolated tissues. If there was any carry-over mesenchyme isolated with the GFP+ epithelium, this would be revealed as GFP+ mesenchymal cells in the recombinants consisting of otherwise tdTomato+ mesenchyme. And vice versa: any carry-over tdTomato+ epithelium isolated with the mesenchyme would be revealed as tdTomato+ epithelial cells in the recombinants. We apologize for not making this clear enough in the original manuscript. In the revised manuscript, we now provide confocal high-resolution images of the recombinant explants (new Supplementary Fig. 1a,b). The explants have been co-stained with the epithelial marker EpCAM, revealing a robust colocalization between the ubiquitously expressed florescent labels in the designated epithelial tissues and the EpCAM.

2- Another important point for understanding the quality and impact of these findings is to assess the similarities and differences, if there are, between the in vivo mesenchyme and the ex vivo one. Indeed, once explanted and put in culture, mesenchymal cells could change their transcriptomic profile and consequently change their signals to the epithelium. The authors should assess the expression of the genes and pathways studied during embryonic development in vivo.

Our response:

The reviewer is correct in that the transcriptomes will likely undergo some changes when organs are cultured ex vivo. This is why RNA-seq was done on freshly isolated tissues. Regarding the potential changes taking place ex vivo, however, we do not consider them relevant with respect to the questions we are addressing in this study. The reason is (as reported in the manuscript) that all control recombinations (homochronic recombinations such as E13 epithelium + E13 mesenchyme, E16 epithelium + E16 mesenchyme etc.) branched essentially as in vivo. Therefore, we find the results and conclusions made from the tissue recombination experiments solid.

3- The authors clearly showed that E16,5 epithelium is able to branch in a mesenchyme free 3D culture model, while epithelia from earlier stages don't. This led to the conclusion that mesenchyme is necessary for acquiring the branching ability. Nevertheless, the authors also said that early stages epithelia scarcely grow in the mesenchyme free 3D culture. Therefore, the lack of branching may be due to the lack of growth, if not the increase of death, of epithelial cells. The authors should quantify the size and the cell death of the epithelia in the different culture conditions and discuss better this point.

Our response:

The reviewer is correct in that one of the key functions of the mammary mesenchyme up to E16.5 may be to provide survival signals for the epithelium, and this might explain why epithelia younger than E16.5 fail to grow/branch when recombined with salivary gland mesenchyme and in mesenchyme-free organoid culture.

Plan for the final revision:

To address this issue, we will assess apoptosis in mammary epithelia cultured in the mesenchyme-free 3D culture organoid set-up.

4- The Fucci2a model allowed to assess the proliferation of embryonic mammary epithelium, showing that the great majority of proliferating cells are basal, at late stages of development (line 182). As it has already been shown, lineage specification is a late process during mammary gland development. The fact that the proliferating cells reside at the external part of the bud does not mean that they are basal cells yet. A p63/K8 staining could be important to understand if the increased proliferation occured in already specified basal cells or not.

__Our response: __

Indeed, mammary lineage specification is a later process. As pointed out in the manuscript and by reviewer #1, the widely used basal and luminal lineage markers have not yet segregated to separate compartments at the developmental stages analyzed in our study, and therefore cannot be used as tools for this purpose. We would like to emphasize that in the manuscript, we analyzed the cells based on their position, and have used the term basal to indicate the basal position, not the prospective lineage. Accordingly, we used the term inner instead of luminal cells to indicate their location, not lineage. We have further clarified this point in the preliminary revised manuscript.

5- The use of Fucci2a model showed that 20% of epithelial cells are proliferative at E13,5. This phase is considered as "quiescent" by the authors (line 120), but the moderate proliferation rate shown in this experiment demonstrated that it is not. A change of the nomenclature is needed.

__Our response: __

We have removed the word “quiescent” from the text.

6- Through the use of K14-Eda and Eda-/- models, the authors claimed that the lineage specification is not a prerequisite for ductal branching. To support this point, they showed that the K14-Eda mice have an anticipated branching although the expression of K8 in the inner part of the bud is transitorily decreased. The authors link the K8 downregulation to a transient suppression of the luminal lineage, but this is clearly overclaimed. Although K8 is a known marker of luminal lineage, the downregulation of one marker is not sufficient to support their thesis. They should first check more markers and in particular critical regulators of luminal lineage as Notch1, Foxa1 and Elf5. Lately, the use of different models that drive embryonic epithelial cells to a forced lineage commitment (Notch1 or Δnp63 overexpression) would support more their claim. As additional evidence, the authors showed that Eda is able to promote basal cell signature. Firstly, the authors should better explain why this point would support their thesis. Secondly, the supplementary figure 2b does not show which genes are taken into account to define the basal signature. A list of these genes would be helpful, as well as staining for some representative proteins.

Our response:

We thank the reviewer for these constructive suggestions. We agree with all reviewers in that this part of the manuscript was not mature enough and provided only indirect evidence on the potential link between lineage segregation and branching ability. This is an important question in the field that merits a study of its own to be addressed with better tools than those available to us at present. As suggested by reviewers #1 and #3, we have omitted this part in the revised manuscript.

7- The authors used the same mouse models to assess the importance of proliferation in the determination of ductal branching and they claimed that proliferation is not a sufficient feature. This conclusion was supported by two observations. The first one is the fact that the K14-Eda model shows an increased cell proliferation at early stages compared to wt, coupled with anticipated branching. Secondly, although having smaller glands compared to wt and showing a delay in ductal branching, Eda-/- mice have an epithelial proliferation rate very similar to wt. Again, the conclusion that proliferation is not sufficient for branching is overclaimed. Firstly, the authors should explain how the buds in wt and Eda-/- mice have different sizes although the similar proliferation (increased cell death?, cellular volume?). Secondly, to support the thesis that proliferation is not sufficient for branching, functional experiments should be performed (see point 12). For instance, the short-time treatments with inhibitors or promotors of proliferation may help to understand the effective role of proliferation in the determination of branching.

Our response:

We show that there is no direct link between onset of proliferation and acquisition of branching ability. However, we are not claiming that proliferation is not important for branching, as obviously new cells are needed as building blocks of growing tissues. In a recently published paper, we have assessed the role of proliferation in branch point formation in embryonic mammary glands. Using mitomycin C to block proliferation, we showed that initiation of new branches occurs even when proliferation is blocked (Myllymäki et al., JCB2023, PMID: 37367826).

The reviewer was also asking why Eda-/- mammary primordia are smaller at E15.5-E16.5 despite similar proliferation rates. In the revised manuscript, we have quantified the volume of E13.5 Eda-/- and control mammary buds and show that Eda-/- buds are ~25% smaller (3.5 ± 0.8 x 105 µm3 in Eda-/- vs. 4.6 ± 0.7 x 105 µm3 in control, mean ± SD) already at the bud stage (new Supplementary Fig. 2c,d).

We have also quantified the cellular size in Eda-/- and control mammary glands at E13.5 and E15.5 and found that mammary epithelial cells in Eda-/- embryos are ~15% smaller (new Supplementary Fig. 2e,f). Together, these data indicate that the smaller size of E15.5-E16.5 Eda-/- mammary glands is a combinatorial effect the smaller mammary anlage at E13.5 and smaller cell size. These findings, while interesting on their own, do not challenge our conclusions regarding the link between onset of proliferation and acquisition of branching ability.

8- The heterotypic epithelial-mesenchymal recombination using the salivary gland is interesting. Nevertheless, some stainings to assess the purity of their systems are again required (e.g., marker of salivary epithelium to verify the purity of the mesenchyme and vice versa).

__Our response: __

As mentioned above, all tissue recombination experiments were performed so that the epithelium and the mesenchyme originated from genetically labelled embryos expressing different fluorescent proteins. In the revised manuscript, we provide confocal images of the salivary-mammary tissue recombinants (new Supplementary Fig. 3), confirming the purity of the tissue compartments used in these experiments.

This model clearly showed that the mammary epithelium can form more branching when combined with the salivary mesenchyme. Moreover, the salivary epithelium preferentially branches through tip bifurcation, while mammary epithelium combined with the salivary mesenchyme has a mixed pattern of tip bifurcation and side branching (typical of the mammary gland). The authors thus concluded that the branching pattern is an intrinsic feature of the epithelium. However, a comparison between the percentage of tip bifurcation and side branching in the heterotypic combination and the homotypic combination between mammary epithelium and mammary mesenchyme is crucial to understand this point. Indeed, these results are not sufficient to exclude that the branching pattern is partially determined by intrinsic features and partially by extrinsic signals. The authors should carefully quantify the branching pattern in the homotypic combination and compare that to the heterotypic one. If the percentage of tip bifurcation do not change, their conclusion is correct; if this percentage increases in the heterotypic combination, it would be a sign of a partial effect of the signals of the mesenchyme.

Our response:

We thank the reviewer for raising this question. We have independently generated data on the type of mammary gland branching events in two papers with somewhat different culture and imaging conditions (Lindström et al., BiorXiv 2022 and Myllymäki et al., JCB, 2023, PMID: 37367826). Both analyses showed that in embryonic mammary glands, the majority of branching events (~70%) occurs by side-branching. These data are in line with the current study that we have now complemented to include also the mammary-mammary recombination experiments (revised Supplementary Video 1, revised Fig. 4b). Quantification of branching events revealed no significant difference in the type of branching events of mammary epithelia grown with salivary or mammary gland mesenchyme (revised Fig. 4c), further supporting our initial conclusions.

9- Through the analysis of their transcriptomic data, Lan and colleagues found that the mammary mesenchyme expresses higher levels of negative regulators of Wnt pathway compared to the salivary mesenchyme. To demonstrate the value of their findings, they should confirm this in vivo, through staining of known Wnt proteins on the salivary and mammary mesenchymes at the embryonic stage.

Our response:

In mammals, there are 19 Wnt ligands, over a dozen secreted Wnt inhibitors, 10 Frizzled receptors, two Lrp co-receptors, and numerous other pathway modifiers that contribute to the net Wnt signaling activity in a complex manner. Furthermore, it has been “notoriously difficult to generate useful antibodies to vertebrate Wnt proteins...In general, these sera do not detect endogenous Wnt proteins in cell extracts, nor do they detect Wnt proteins in tissues by staining techniques. Hence, there are few data on Wnt protein distribution in intact vertebrate animals.” This is a direct citation from the Wnt Homepage, maintained by the Nusse Lab; https://web.stanford.edu/group/nusselab/cgi-bin/wnt/reagents#antibod.

For all these reasons, we do not find this approach feasible nor informative.

Instead, in the revised manuscript, we report the expression levels of Axin2, the most commonly used transcriptional readout of canonical Wnt activity in our RNA-seq data (new Supplementary Fig. 4c). Axin2 levels are lowest in the E16 fat pad where mammary branching takes place, much lower than in any other tissues analyzed in the study.

Plan for the final revision:

To complement these findings, we will additionally provide expression data of a transgenic Wnt reporter from the same developmental stages and tissues that were used to generate the RNA-seq data.

10- Since the ability of the salivary mesenchyme to promote a higher rate of branching in the mammary epithelium, the authors wanted to assess what could be the role of Wnt signalling. To do so, they used a mouse model where B-catenin is stabilised, allowing an increased Wnt signalling in the mammary mesenchyme. As a result, they observed increased branching in the mammary epithelium. They also found that IGF1 is a ligand regulated by Wnt pathway in the mesenchyme. Therefore, the use of exogenous IGF1 in their ex vivo model was able to increase the branching of the mammary epithelium. Moreover, Igf1r-/- embryos showed a significant decrease of mammary gland branching. The conclusion based on these experiments was that the Wnt-Igf1-Igf1r axis plays a pivotal role in the promotion of mammary gland branching during embryogenesis. This conclusion is overclaimed for different reasons. Firstly, the normalization of the ductal branching to the body weight is insufficient to exclude that the impact of the Igf1r knockout may have severe consequences on the mammary gland formation, upstream of the ductal branching. Another parameter for this normalization is required (e.g., size of the bud before branching, proliferation status, etc).

Our response:

We agree with the reviewer in that Igf1r knockout may affect mammary gland formation in multiple ways, and also prior to onset of branching, as already indicated in the original manuscript: “…apart from one study reporting the smaller size of the E14 mammary bud in IGF-1R deficient embryos …” (line 398-399 in the revised version) and ‘…mammary gland 3 that was consistently absent.’ (line 414-415 in the revised version).

To assess whether the reduced size and branching of E16.5/E18.5 Igf1r-/- mammary glands is merely a consequence of the smaller anlage, the revised manuscript includes new data reporting quantification of the volume of mammary gland 2 of Igf1r-/- and wild type littermate embryos at E13.5, E16.5, and E18.5 from 3D confocal images of whole mount EpCAM stained mammary glands. As can be seen from the new Fig. 7g-h, at E13.5, the mutant mammary buds are about 60% of the size of the controls, at E16.5, 25% and at E18.5 only 20 % revealing a progressive defect, indicative of a specific defect at the outgrowth and branching stage. This conclusion was validated by normalization to the body weight: at E13.5 the size of Igf1r-/- mammary anlage did not differ from that of the wild type embryos (p = 0.11), at E16.5 the sprouts were smaller in the mutants, though the difference did not reach statistical significance (p = 0.08), while at E18.5, the Igf1r-/- mammary glands were significantly smaller (p = 0.000021) (new Fig. 7i). We find these data compelling evidence for a specific role for Igf1r in outgrowth and branching of the embryonic mammary gland.

The use of alternative models to specifically knockout the receptor in the epithelium or the ligand in the mesenchyme (e.g. viruses) would be even more useful to specifically focus on the role of this pathway for ductal branching excluding side effects.

Our response:

We thank the reviewer for this suggestion. Unfortunately, based on our experience, viral shRNA delivery is not sufficiently efficient for effective gene silencing, unlike Cre delivery for a gain-of-function approach (used in the current study to flox out exon 3 of beta-catenin) in case where the endogenous pathway activity is very low and therefore, targeting even a subset of cells is sufficient for upregulation of paracrine factors.

Plan for the final revision:

To address the question on the autocrine or paracrine role of Igf1r, we will perform tissue recombination experiments between Igf1r-/- and control mammary epithelium and mesenchyme.

Another limit of this model is the fact that Igfr1 can be bound by Igf2 as well and we cannot exclude that this has an impact too (except if Igf2 is not expressed at this stage). A quantification of Igf2 expression may be useful.

Our response:

Indeed, we cannot exclude the possibility that Igf2 could also play a role (Igf2 expression was similar to Igf1 in our RNA-seq dataset, see Supplementary Fig. 5), but the connection of mesenchymal Wnt signaling activity was to Igf1, not Igf2 – in fact Igf2 was somewhat downregulated in Wnt3A treated sample reported by Wang et al. (Wang et al., 2021) (highlighted by an arrow in the revised Fig. 6). We have also clarified this point in the Discussion of the preliminary revised manuscript.

11- From the experiments presented in this section it is clear that Wnt-Igf1-Igf1r axis has to be finely regulated to have the correct amount of ductal branching in the embryonic mammary epithelium. Nevertheless, the author just showed the RNA levels of Igf1 in the different compartments they have analysed. Stainings to see the effective presence of the ligand on the tissue is mandatory to clarify the role of this axis in the ductal branching in vivo.

Our response:

Igf1-Igf1r signaling plays a critical growth promoting function during embryonic and postnatal development. The expression of Igf1 at RNA and protein level has been detected in almost all tissues in humans (Daughaday et al., Endocr. Rev., 1989; PMID: 2666112). Given that Igf1 is a secreted protein and multiple Igf binding proteins (Igfpbs) (that regulate the bioactivity of Igf1 by sequestering it) are expressed in the mammary and salivary gland mesenchyme (Supplementary Fig. 5), we find it unlikely that Igf1 staining would provide any additional information to the current study, as they cannot be used to assess the source of Igf1, nor the location of the signaling activity.

Furthermore, as underlined by the authors, this axis is specifically important and upregulated in the salivary gland. Due the limit of the Igf1R-/- model, we cannot exclude that, although Wnt-Igf1-Igf1r axis is able to increase the branching ability of mammary epithelium, the normal branching rate observed in wt mice is due to other pathways.

Our response:

We agree with the reviewer in that other pathways are also important in regulating normal mammary gland branching, for example, Eda/NF-κB and FGF pathways as we described in the Introduction. Our results do not exclude the possibility that also pathways other than Wnt regulate Igf1 expression. The reviewer is correct that if a paracrine factor is expressed in the salivary gland but not in the mammary mesenchyme, its physiological effect may be limited to the salivary gland. Indeed, cluster 5 identified by the mFuzzy analysis (Fig. 5f) is likely to include some genes like that. This is why we decided to focus on cluster 6 genes like Igf1. In the revised manuscript, we have better highlighted the difference between cluster 5 and 6 genes.

Unfortunately, with the currently available tools, we cannot test the importance of the endogenous mesenchymal Wnt signaling activity by inactivating Wnt signaling activity specifically in the mesenchyme at the time point when branching begins. This would require an inducible mesenchymal Cre line (mesenchymal β-catenin is essential for the early fate specification of the primary mammary mesenchyme; Hiremath et al., 2012, PMID: 23034629), and conditional β-catenin null mouse. We do not have such mice available and we find that these experiments are beyond the scope of the current study.

12- Lastly, once claimed to have found the key factor necessary for ductal branching promotion, the authors should also test if the proliferation and lineage segregations are unaffected in this context, confirming their dispensable role claimed in the initial part of the manuscript.

__Our response: __

Igf1/Igf1r is well-known for its growth promoting function via cell proliferation. We have no reasons to think that this would not be the case also in the mammary gland, and it was not our intention to give the impression that proliferation was not affected. In fact, Hiremath et al. (2012) already reported a defect in epithelial cell proliferation in Igf1rmammary buds at E14. Our key finding is that compared to other organs, the mammary gland is particularly sensitive to loss of Igf1r during branching morphogenesis. Finally, as pointed out earlier, better tools will be needed to assess the potential link between lineage segregation and onset of branching, a topic that we hope to address in the future.

Minor comments: 1- An important paper on mammary gland ductal branching was published on Nature in 2017 by Scheele and colleagues and should be presented in the introduction, even though it is at later stages (after birth).

Our response:

We thank the reviewer for the suggestion. In the revised manuscript, we have added the findings from Scheele et al. 2017 in the introduction.

2- In line 136 and 139 the authors referred to Fig 2 but it should be Fig 1

Our response:

We apologize for these slips. They have been corrected in the revised manuscript.

3- The sentence on line 142 should be rephrased, since "advanced developmental stages" may be referred to pubertal development. The authors should specify that they are talking about embryonic development.

Our response: We apologize for the potential misunderstanding. In the revised manuscript, we have used the phrase “advanced embryonic developmental stage” to describe our conclusion more precisely.

Reviewer #2 (Significance (Required)):

Overall, the authors concluded that embryonic mammary gland development and branching are extremely sensitive to the loss of IGF1, normally produced by the mesenchyme. The topic of the paper is interesting, the experimental approaches are well conceived, the data are convincing and the findings are of interest to developmental biologists. Nevertheless, there are some significant points that need to be further investigated before considering the manuscript suitable for publication:

Our response:

We thank the reviewer for the careful assessment and positive feedback of our manuscript. We have already addressed most of the points raised and most remaining ones will be addressed in the final revised manuscript.

__Reviewer #3 (Evidence, reproducibility and clarity (Required)): __

Here the authors use classical embryonic tissue recombination and pharmacological manipulation of explants in conjunction with cutting edge 3D imaging of tissue derived from highly sophisticated reporter and knock-out mouse models and state of the art transcriptomic analysis to masterfully delineate and dissect regulatory pathways critical for embryonic mammary development. Specifically, they set out to parse regulation of proliferation from that of branch patterning.

While it has long been established that epithelial-mesenchymal interaction is necessary for mammary branching this work shows by heterochronic recombination that initiation mammary branching is not advanced by mesenchymal stage. By examining Fucci2a embryos the authors demonstrate that branching is preceded by a significant increase in basal cell-biased proliferation but, through further analysis of Eda gain and loss of function mice, conclude that proliferation per se does not cause branching. They show by heterotypic recombination with salivary tissue that early mammary epithelia rudiments require their own mesenchyme for survival and that although later E16.5 rudiments expand more robustly when in contact with salivary mesenchyme they nevertheless retain their characteristic mammary branch pattern. Thus, they establish that initiation and patterning are intrinsic properties of the epithelium but that early survival and later expansion/proliferation is regulated by the mesenchymal context. By transcriptomic comparison of mammary and salivary mesenchyme they reveal that genes encoding canonical Wnt attenuators and antagonists are highly expressed in early mammary mesenchyme and drop as branching ensues. The low expression of these negative regulators of Wnt signaling in salivary mesenchyme is proposed as an explanation for its growth and branch stimulating capability. In keeping with these observations, the authors show that experimental activation of mammary mesenchymal Wnt signaling augments both growth and branching. Lastly, they identify transcriptomic changes in IGF1 coincident with the initiation of mammary branching and confirm its role by extending analyses of the effects of gain and loss of function of IGF1 on embryonic mammary development.

This is a thorough, well-constructed paper that adds new knowledge and important conceptual nuance and mechanistic insight to classical findings on branch patterning. This work is a technical tour de force and backed by solid quantitative and statistical analysis throughout. Their experimental approach is superb and the conclusions are sound. Their findings will be of great interest to the community of mammary gland biologists and to the wider field of embryologists focused on early development of a broad range of ectodermal appendages.

I have some minor criticisms that I believe can be quickly remedied in a minor rewrite and suggestions for the authors consideration to improve the manuscript discussion as follows:

Minor issues Abstract, line 37: The authors misuse the word "decompose" - it should be "deconstruct"

__Our response: __ We thank the reviewer for pointing out our mistake, which we have corrected in the revised manuscript.

Results, p7 line 48: Add "The" to the sentence: "The majority...."

__Our response: __ Corrected it in the revised manuscript.

P8 line 173 This sentence refers to Figure 2G which is a quantitative plot. I would suggest replacing the word "cluster" which implies a spatial organization with the word "subset" or "significant fraction" The spatial data in Fig 2d support basal bias but do NOT to my eye show any clustering - in fact the proliferative basal cells appear to be evenly dispersed within the basal layer.

Our response:

We thank the reviewer for highlighting this aspect. We agree that “significant fraction” is a more suitable term than “cluster”.

P9 line 188: The statement on basal cell lineage specification needs a reference.

__Our response: __

Following the suggestions from reviewers #1 and 3, we have removed the content about lineage segregation in Results, together with this sentence.

P10 line 201-216 I found the section on lineage specification (fig S2) weaker than the rest and a distraction from the main thrust of the paper making it difficult for the reader to focus. I suggest omitting this section and supplemental figures associated with it altogether.

__Our response: __

We agree with all reviewers in that this part of the manuscript was not mature enough and provided only indirect evidence on the potential link between lineage segregation and branching ability. This is an important question in the field that merits a study of its own that should be addressed with better tools than those available to us at present. As suggested by reviewers #1 and #3, we have omitted this part in the revised manuscript.

P9 line 190: "displays precocious onset of branching" it is sufficient to say: displays precocious branching - the use of both "precocious" and "onset' is redundant.

P10 line 229 Similarly, delete "the onset of branching was delayed" it is sufficient to say: branching was delayed.

__Our response: __ Both sentences have been corrected it in the revised manuscript.

P11 line 243: Delete "on the regulation of the" and substitute the word "to" in the sentence: "Next, we shifted our focus on the regulation of the branching pattern, which is thought to be determined by mesenchymal cues."

__Our response: __ Corrected it in the revised manuscript.

P11 line 241 subtitle and Figure 4 title: The disparity in titles here is jarring for the reader: Results text subtitle: "Salivary gland mesenchyme is rich in growth-promoting cues, but does not alter the mode of branch point formation of the mammary epithelium". Figure 4 Title: "Mammary mesenchyme is indispensable for the branching ability of the mammary gland". I suggest to the authors divide the figure as well as the text to make the two points indicated by their disparate titles separately.

__Our response: __ We thank reviewer for the suggestion to clarify the Results part of the manuscript. As suggested, we have split the data under two separate subtitles, but due to limitations in figure numbers, we prefer to report these data in one figure panel.

P12 line 279 From here on out the manuscript has a tendency to use the term "growth" ambiguously - in many instances it is unclear do they mean expansion, proliferation, increased branch number/ morphology?? Please try to clarify.

__Our response: __

Our aim is to use the term growth to mean tissue growth (expansion). We hope that this is clearer in the revised manuscript.

P16 line 341 use word "prompted" instead of word "promoted"

__Our response: __ We thank reviewer for spotting out the slip, which we have corrected in the revised manuscript.

P16 line 382: include word "embryonic" before "mammary development"

__Our response: __ We have modified the text in the revised manuscript.

Discussion P18 line 416: Add the words "later stage (E16.5)" to the sentence: "Importantly, we demonstrate that salivary gland mesenchyme could only promote the growth of later stage (E16.5) mammary epithelium"

__Our response: __ We thank reviewer for the suggestion. We have modified the text in the revised manuscript.

P19 line 437: Given the authors statement "Instead, cell motility is critical for branch point formation in the mammary gland" they should consider a brief sentence mentioning their transcriptomic findings on cadherin 11 and Tenascin.

__Our response: __ We thank the reviewer for appreciation of our transcriptomic data. In the revised manuscript, we have added the following text in discussion: “Accordingly, we observed significantly increased expression of cell migration promoting genes such as Cdh11 (encoding Cadherin 11), and Tnc (encoding Tenascin C) 60,61 in the E16.5 mesenchyme compared to E13.5 (Supplementary Table 2).”

P19 line 451: Similarly, given their statement "This observation suggests that mammary epithelium itself carries the instructions dictating the mode of branching" they could consider their transcriptomic data on Ltbp1 in "mammary specific" clusters 7,8,9 as a matrix molecule initially expressed by mammary mesenchyme but which becomes expressed by luminal epithelial cells at precisely the time they acquire lineage specification and intrinsic branching capability.

__Our response: __ This is an excellent suggestion. We have added following text in discussion: “It is worth noting that certain mesenchymal factors, such as Ltbp1, began transitioning towards epithelium-specific expression around E16.5 69. Exploring the potential impact of these factors on the self-instructed branching capacity of the mammary epithelium could yield valuable insights.”

P20 lines 462-470 The authors should address their theory of Wnt suppression in the mammary mesenchyme in the context, albeit conflictingly, of earlier studies showing expression of Wnt signaling reporters, in either epithelial or mesenchymal locations during early stages.

Our response: __ We thank reviewer for the suggestion. In the preliminary revised manuscript, we report Axin2 expression data as __new Supplementary Fig. 4c. Axin2 expression data suggest that Wnt/β-catenin activity is lowest in the E16.5 fat pad (where branching takes place) compared to all other tissues analyzed in the study.

Plan for the final revision:

For the final revised manuscript, we will additionally generate transgenic Wnt reporter expression data (see also our response to point 3 of Reviewer #1). These results will be discussed in light of the published Wnt reported literature in the final revised manuscript.

Reviewer #3 (Significance (Required)):

Here the authors use classical embryonic tissue recombination and pharmacological manipulation of explants in conjunction with cutting edge 3D imaging of tissue derived from highly sophisticated reporter and knock-out mouse models and state of the art transcriptomic analysis to masterfully delineate and dissect regulatory pathways critical for embryonic mammary development. Specifically, they set out to parse regulation of proliferation from that of branch patterning.

This is a thorough, well-constructed paper that adds new knowledge and important conceptual nuance and mechanistic insight to classical findings on branch patterning. This work is a technical tour de force and backed by solid quantitative and statistical analysis throughout. Their experimental approach is superb and the conclusions are sound. Their findings will be of great interest to the community of mammary gland biologists and to the wider field of embryologists focused on early development of a broad range of ectodermal appendages.

Our response:

We much appreciate the positive evaluation of our manuscript. We have addressed all the feedback provided by the reviewer 3 in the preliminary revised manuscript, except the last point, which will be included in the final revision along with the new data on the Wnt reporter expression.

Field of expertise: Embryonic and adult mammary development, Wnt signaling, cell adhesion

Author Response

eLife assessment

This useful paper examines changes (or lack thereof) in birds' fear response to humans as a result of COVID-19 lockdowns. The evidence supporting the primary conclusion is currently inadequate, because the model used does not properly account for many potentially confounding factors that could influence the study's outcomes. If the analytic approach were improved, the findings would be of interest to urban ecologists, behavioral biologists and ecologists, and researchers interested in understanding the effects of COVID-19 lockdowns on animals.

Many thanks for these supportive words. We did our best to improve our manuscript according to the reviewers and editor comments. Importantly, we regret being unclear in the Methods, as our models already controlled for most of the confounds (see below) discussed by the reviewers.

For example, given that a single observer collected the data at most sites, site as a random intercept in the models controls also for the observer effects (which is one of the reasons why site is in the model). We added details to Methods (L352-356, see also “Statistical analyses” in the main text).

The first reviewer asked us to use “some measure of urbanity (e.g. Human Footprint Index) that varies across the cities included here”. Our main results are now based on country-specific models and hence, the use of a single value predictor for each city is not appropriate. Please, see also below.

The second reviewer is concerned about multicollinearity in our models because of the 0.95 correlation between Period and Stringency Index. However, these are key predictor variables of interest that have never been used within the same model as predictors. We now clearly explain this in the Methods (L458-538, 548-550) and within legend of Figure S2.

The third reviewer suggested that our models would benefit from controlling for day in the species-specific breeding cycle. Although we don’t have precise city-specific information on the timing of breeding stages in the sampled populations of birds, we partly control for these effects by including a random intercept of day within each year and species. This random factor explained most of the variance (see Table S1-S2) – something that could have been expected. In other words, we do control for what the third reviewer asked for. Similarly, we account for habitat features that may influence escape distance by including site in the models. Site usually refers to a specific park (we assume that within-park heterogeneity is lower than between park variation) and hence partly addresses the reviewer’s concern. Again, we highlight this within the Methods (L466-476).

Reviewer #1 (Public Review):

This paper uses a series of flight initiation "challenges" conducted both prior to and during COVID-19-related restrictions on human movement to estimate the degree to which avian escape responses to humans changed during the "anthropause". This technique is suitable for understanding avian behavioral responses with a high degree of repeatability. The study collects an impressive dataset over multiple years across five cities on two continents. Overall the study finds no effect of lockdown on avian escape distance (the distance at which the "target" individual flees the approaching observer). The study considers the variable of interest as both binary (during lockdown or prior to lockdown) and continuous, using the Oxford Stringency Index (with neither apparently affecting escape distance). Overall this paper presents interesting results which may suggest that behavioral responses to humans are rather inflexible over "short" (~2 year) timespans. The anthropause represents a unique opportunity to disentangle the mechanistic drivers of myriad hypothesized impacts humans have on the behavior, distribution, and abundance of animals. Indeed, this finding would provide important context to the larger body of literature aimed at these ends.

Thank you very much for your positive feedback.

However, the paper could do more to carefully fit this finding into the broader literature and, in so doing, be a bit more careful about the conclusions they are able to draw given the study design and the measures used. Taking some of these points (in no particular order):

Thank you. We did our best in addressing your comments (see below and updated Methods, Results and Discussion sections).

1) Oxford Stringency Index is a useful measure of governmental responses to the pandemic and it's true that in some scenarios (including the (Geng et al. 2021) study cited by this paper) it can correlate with human mobility. However, it is far from a direct measure of human mobility (even in the Geng study, to my reading, the index only explained a minority of the variation). Moreover, particular sub-components of the index are wholly unrelated to human mobility (e.g. would changes to a country's public information campaign lead to concomitant changes in urban human mobility?). Finally, compliance with government restrictions can vary geographically and over time (i.e. we might expect lower compliance in 2021 than in 2020) and the index is calculated at the scale of entire countries and may not be very reflective of local conditions. Overall this paper could do more to address the potential shortcomings of the Oxford Stringency Index as a measure of human mobility including attempting to validate the effect on human mobility using other datasets (e.g. the google dataset and/or those discussed in (Noi et al. 2022). This is of critical importance since the fundamental logic of the experimental design relies on the assumption that stringency ~ mobility.

Thank you for this comment. First, Oxford Stringency Index seemed to us as the best available index for our purposes, i.e to estimate people's mobility during the shutdown because restrictions surely influenced the possibility that people would be outside, and because the index is a country-specific estimate. However, in addition, we now checked all indices mentioned in Noi et al. 2022 and found useful only the Google Mobility Reports, which we now use, because (a) it is publicly available, (b) it is available also for territories outside US, and (c) provides data for each city included in our dataset as well as for urban parks where most of our data were collected. Note that some platforms are no longer providing their mobility data (e.g. Apple).

However, Google Mobility provides day-to-day variation in human mobility, whereas we are interested in overall increase/decrease in human mobility. Nevertheless, we correlated the Google mobility index with the Stringency index and found that human mobility generally decreases with the strength of the anti-pandemic measures adopted in sampled countries (albeit the effect for some countries, e.g. Poland, is small; Fig. 5).

Moreover, we also added analysis using # of humans collected directly in the field during escape trials (e.g. Fig. 6 and S6) and found that the link between # of humans and Stringency index or Google Mobility was weak and noise, 95%CIs widely crossing zero (Fig. 6).

Importantly, if we use Google Mobility and # of humans, respectively, as predictors of escape distance, the results are qualitatively very similar to results based on Oxford Stringency Index (Fig. S6), or Period, with tiny effect sizes for both (95%CIs for Google Mobility -0.3 – 0.06, Table S5, for # of humans -0.12 – 0.02, Table S6) supporting our previous conclusions.

Note that Google Mobility and the number of humans have their limitations (see our comment to the editor and the Methods section in the main manuscript, e.g. L418-433). The lack of Google Mobility data for years before the COVID-19 pandemic does not allow us to fully explore whether overall human activity decreased during COVID-19 or not (our test for period prior and during COVID-19). If the year 2022 reflects a return to “normal” (which is to be disputed due to COVID-19-driven rise in home office use) the 2020 and 2021 had on average lower levels of human activity (Fig. 4). Whether such a difference is biologically meaningful to birds is unclear given the immense day-to-day change in human mobility and presence (Fig 4). Moreover, the number of humans capture within- and between-day variation rather than long-term changes in human presence.

We added details on the new analysis into the method and results sections (e.g. Fig. 4-6; L142-165, 418-438, 495-535) and Supplementary Information (Figs. S5-S9 and associated Tables) and discuss the problematic accordingly. Moreover, to enhance clarity about country specific effect (or their lack), we also add country specific estimates to the Results (Fig. 1 and Fig. S6 and respective Tables). Finally, our statistical design and random structure of the model allowed us to control for spatial and temporal variation in compliance with government restrictions.

2) The interpretation of the primary finding (that behavioral responses to humans are inflexible) could use a bit more contextualization within the literature. Specifically, the study offers three potential explanations for the observed invariance in escape response: 1) these behaviors are consistent within individuals and this study provides evidence that there was no population turnover as a result of lockdowns; 2) escape response is linked to other urban adaptations such that to be an urban-dwelling species dictates escape response; and/or 3) these populations already exhibit maximum habituation and the reduction in human mobility would only have increased that habituation but that trait is already at a boundary condition. Some comments on each of these respectively:

Thank for these comments. We incorporated them in the main text (L293-329). Your point 1) corresponds to our point (i): “Most urban bird species in our sample may be relatively inflexible in their escape responses because the species may be already adapted to human presence” (L293-306); your point 2) to our point (ii): “Urban environment might filter for bold individuals (Carrete and Tella, 2013, 2010; Sprau and Dingemanse, 2017). Thus, the lack of consistent change in escape behaviour of urban birds during the COVID-19 shutdowns may indicate an absence (or low influx) of generally shy, less tolerant individuals and species from rural or less disturbed areas into the cities…” (L307-314); your point 3) to our point (iii): “Urban birds might have been already habituated to or tolerant of variation in human presence, irrespective of the potential changes in human activity patterns” (L315-329). To distinguish between (ii) and (iii) or the two from (i), individually-marked birds and comprehensive genetic analyses are needed, which we now note in the Discussion (L330-348). Importantly, we also discuss that the lack of response might be due to relatively small changes in human activity (L253-292), which we unfortunately could not fully quantify.

a) Even had these populations turned over as a result of a massive rural-to-urban dispersal event, it's not clear that the escape distance in those individuals would be different because this paper does not establish that these hypothetical rural birds have a different behavioral response which would be constant following dispersal. Thus the evidence gathered here is insufficient to tell us about possible relocations of the focal species.

Thank you for this point. We address this point in the Introduction and Discussion (L92-101, 307-314). Rural bird populations/individuals are on average less tolerant of humans than urban birds (e.g. Díaz et al. 2013, PloS One 8:e64634; Tryjanowski et al. 2020, J Tropic Ecol 36:1-5; Mikula et al. 2023, Nat Commun 14:2146) and at the same time, bird individuals seem consistent in their escape responses (Carrete & Tella 2010, Biol Lett 23:167–170; Carrete & Tella 2013, Sci Rep 3:1–7).

Additionally, the paper cites several papers that found no changes in abundance or movements of animals in response to lockdowns but ignore others that do. For example: (Wilmers et al. 2021), (Warrington et al. 2022) (though this may have been published after this was submitted...), and (Schrimpf et al. 2021).

We added the papers (L89-91). Thank you!

There is a missed opportunity to consider the drivers of some of these results - the findings in this paper are interesting in light of studies that did observe changes in space use or abundance - i.e. changes in space use could arise precisely because responses to humans are non-plastic but the distribution and activities of humans changed.

Thank you. Indeed, we now address this in the Discussion (L303-306): “However, some studies reported changes in the space use by wildlife (Schrimpf et al., 2021; Warrington et al., 2022; Wilmers et al., 2021). and these could arise, as our results indicate, from fixed and non-plastic animal responses to humans who changed their activities”.

To wit, the primary finding here would imply that the reaction norm to human presence is apparently fixed over such timescales - however, and critically, the putative reduction in human activity/mobility combined with fixed responses at the individual level might then imply changes in avian abundance/movement/etc.

Unfortunately, we have not measured changes in avian abundance or movements. But, please, note that the change in human mobility in sampled cities might be not as dramatic as initially thought and we consider this scenario to be most plausible in explaining no significant differences in avian escape responses before and during the COVID-19 shutdowns (see Fig. 4). Nevertheless, we add your point into the Discussion: If our findings imply that in birds the reaction norm to human presence is fixed over the studied temporal scale, the putative changes in human presence might then imply changes in avian abundance or movement (L293 and text below it).

b) If this were the case, wouldn't this be then measurable as a function of some measure of urbanity (e.g. Human Footprint Index) that varies across the cities included here? Site accounted for ~15% of the total variation in escape distance but was treated as a random effect - perhaps controlling for the nature of the urban environment using some e.g. remotely sensed variable would provide additional context here.

Urbanity mirrors the long-term level of human presence in cities whereas we were interested mainly in the rather short-term effects of potential changes of human presence on bird behaviour. Thus, we are not sure how adding such variable will help elucidating the current results. Please, also note that we added the country-specific analysis. Site indeed accounted for considerable amount the total variance in escape distance and that is why it was included as random intercept, which controls for non-independents of data points from each city. This could partly help us to control for difference in habitat type (e.g. urbanization level) within cities.

c) Because it's not clear the extent to which the populations tested had turned over between years, the paper could do with a bit more caution in interpreting these results as behavioral. This study spans several years so any response (or non-response) is not necessarily a measure of behavioral change because the sample at each time point could (likely does) represent different individuals. In fact, there may be an opportunity here to leverage the one site where pre-pandemic measures were taken several years prior to the pandemic. How much variance in the change in escape distance is observed when the gap between time points far exceeds the lifetime of the focal taxa versus measures taken close in time?

We believe the initial Fig S4, now Figure 2, addresses this point. The between years temporal variation in FIDs exceeds the variation due to lockdowns. This is true both for measures taken in consecutive years, as well as for measures taken far apart.

d) Finally, I think there are a few other potential explanations not sufficiently accounted for here:

i) These behaviors might indeed be plastic, but not over the timescales observed here.

We agree and have added this point (L301-303). Thank you.

ii) Time of year - this study took place during the breeding season. The focal behavior here varies with the time of year, for example, escape distance for many of these species could be tied up in nest defense behaviors, tradeoffs between self-preservation and e.g. nest provisioning, etc.

Please, note that we controlled for the date in our analyses. Date was used as a proxy for the progress in the breeding season (L463-464 and Fig. 1 caption). Note that we collected data only from foraging or resting individuals, and data were neither collected near the nest sites nor from individuals showing warning behaviours, which we now note (L400-401).

iii) Escape behaviors from humans are adaptively evolved, strongly heritable, and not context dependent - thus we would only expect these behaviors to change on evolutionary timescales.

We discussed this at L307-308 and 381-383. Escape behaviors from humans are highly consistent for individuals, populations, and species (Carrete & Tella 2010, Biol Lett 23:167–170; Díaz et al. 2013, PloS One 8:e64634; Mikula et al. 2023, Nat Commun 14:2146). Whether such behavior is consistent across contexts is less clear (e.g. Diamant et al. 2023, Proc Royal Soc B, in press; but see, e.g. Radkovic et al. 2019, J Ecotourism 18:100-106; Gnanapragasam et al. 2021, Am Nat 198:653-659). Escape distance is often not measured simultaneously, for example, with human presence. In other words, whereas general level of human presence may have no effect on escape distance, the day-to-day or hour-to-hour variations might. We need studies on fine temporal scales (day-to-day or hour-to-hour) using marked individual to elucidate this phenomenon.

iv) See point one above - it's possible that the lockdown didn't modify human activity sufficiently to trigger a behavioral response or that the reaction norm to human behavior is non-linear (e.g. a threshold effect).

We agree, now use also Google Mobility Reports and # of humans data to elucidated this phenomenon and have added such interpretations to L253-292 and, e.g. Fig. 4.

LITERATURE CITED Geng DC, Innes J, Wu W, Wang G. 2021. Impacts of COVID-19 pandemic on urban park visitation: a global analysis. J For Res 32:553-567. doi:10.1007/s11676-020-01249-w

Noi E, Rudolph A, Dodge S. 2022. Assessing COVID-induced changes in spatiotemporal structure of mobility in the United States in 2020: a multi-source analytical framework. Int J Geogr Inf Sci.

Schrimpf MB, Des Brisay PG, Johnston A, Smith AC, Sánchez-Jasso J, Robinson BG, Warrington MH, Mahony NA, Horn AG, Strimas-Mackey M, Fahrig L, Koper N. 2021. Reduced human activity during COVID-19 alters avian land use across North America. Sci Adv 7:eabf5073. doi:10.1126/sciadv.abf5073

Warrington MH, Schrimpf MB, Des Brisay P, Taylor ME, Koper N. 2022. Avian behaviour changes in response to human activity during the COVID-19 lockdown in the United Kingdom. Proc Biol Sci 289:20212740. doi:10.1098/rspb.2021.2740

Wilmers CC, Nisi AC, Ranc N. 2021. COVID-19 suppression of human mobility releases mountain lions from a landscape of fear. Curr Biol 31:3952-3955.e3. doi:10.1016/j.cub.2021.06.050

Reviewer #2 (Public Review):

Mikula et al. have a large experience studying the escape distances of birds as a proxy of behavioral adaptation to urban environments. They profited from the exceptional conditions of social distance and reduced mobility during the covid-19 pandemic to continue sampling urban populations of birds under exceptional circumstances of low human disturbance. Their aim was to compare these new data with data from previous "normal" years and check whether bird behavior shifted or not as a consequence of people's lockdown. Therefore, this study would add to the growing body of literature assessing the effect of the covid-19 shutdown on animals. In this sense, this is not a novel study. However, the authors provide an interesting conclusion: birds have not changed their behavior during the pandemic shutdown. This lack of effects disagrees with most of the previously published studies on the topic. I think that the authors cannot claim that urban birds were unaffected by the covid-19 shutdown. I think that the authors should claim that they did not find evidence of covid-19-shutdown effects. This point of view is based on some concerns about data collection and analyses, as well as on evolutionary and ecological rationale used by the authors both in their hypotheses and results interpretation. I will explain my criticisms point by point:

We are grateful for your positive appraisal of our manuscript, as well as for your helpful critical comments. We toned down the discussion to claim, as suggested by you, that we did not find evidence for effects of covid-19-shutdowns on escape behaviour of birds in urban settings (see Results and Discussion sections). In general, we attempted to provide a more nuanced discussion and reporting of our findings. We also changed the manuscript title to “Urban birds' tolerance towards humans was largely unaffected by the COVID-19 shutdowns” and added validation using Google Mobility Reports (Fig. 5 & S6, Table S3a and S5) and the actual number of humans (Fig. 6 and S6; Table S3b-e and S6). Note however that there is only a single robust study on the topic of shutdown and animal escape distances (Diamant et al. 2023, Proc Royal Soc B, in press), i.e. the topic is largely unexplored (e.g. L99-101), whereas we discuss our finding in light of shutdown influences on other behaviours (L293-329).

1) The authors used ambivalent, sometimes contradictory, reasoning in their predictions and results interpretation. Some examples:

We tried to clarify our reasoning and increased consistency in our claims in the Introduction. Please, note that we simplified the Introduction and now provide one main expectation: FIDs of urban birds should increase with decreased human presence. This pattern is robustly empirically documented, regardless of the mechanism involved (e.g. Díaz et al. 2013, PloS One 8:e64634; Tryjanowski et al. 2020, J Tropic Ecol 36:1-5; Mikula et al. 2023, Nat Commun 14:2146). Please, see our revised Discussion for a more comprehensive discussion of mechanisms which could explain the patterns described in our study.

1.1) The authors claimed that urban birds perceive humans as harmless (L224), but birds actually escape from us, when we approach them... Furthermore, they escape usually 5 to 20 m away. This is more distance that would be necessary just to be not trampled.

We agree and have deleted mentions that humans are perceived as harmless.

1.2) If we are harmless, why birds should spend time monitoring us as a potential threat (L102)? Indeed, I disagree with the second prediction of the authors. I could argue that reduced human activity should increase animal vigilance because real bird predators (e.g. raptors) may increase their occurrence or activity in empty cities. If birds should increase their vigilance because the invisible shield of human fear of their predators is no longer available, then I would expect longer escape distances.

Thank you for this comment. We deleted this prediction and largely rewrote Introduction based on your comments and comments from the other reviewers.

1.3) To justify the same escape behavior shown by birds in pre- and pandemic conditions from an adaptive point of view, the authors argued a lack of plasticity and a strong genetic determination of such behavior. This contravenes the plasticity proposed in the previous point or the expected effect of the stringency index (L112).

We now attempted to write this more clearly while incorporating your suggestions. In the Discussion, we now propose various hypothesis that can, but need not be mutually exclusive. Please, note that we simplified the Introduction and now provide one main hypothesis: FIDs of urban birds should increase with decreased human presence.

In my opinion, some degree of plasticity in the escape behavior would be really favorable for individuals from an adaptive perspective, as they may face quite different fear landscapes during their lives. Looking at the figures, one can see notable differences in the escape distance of the same species between sites in the same city. As I can hardly imagine great genetic differences between birds sampled in a park or a cemetery in Rovaniemi, for instance, I would expect a major role of plasticity to explain the observed variability. Furthermore, if escape behavior would not be plastic, I would not expect date or hour effects. By including them in their models, the authors are accepting implicitly some degree of plasticity.

We regret being unclear. We do accept some degree of plasticity. Yet, our study design prohibits the assessment of the degree of individual plasticity because sampled birds were not individually marked and approached repeatedly. We tried to soften the statements in our Discussion to not fully dismiss a possibility that urban birds have some degree of plasticity in their antipredator behaviour (L293-329). Note however, that while our data collection was not designed to test how hour-to-hour changes in human numbers influence escape distance, the effect of the number of humans (i.e. hour-to-hour variation in human numbers) in our sample was tiny.

The date and hour effect simply control for the particularities of the given day and hour (e.g. warm vs cold times or the time until sunset). In other words, the within species differences (even from the same park) may have little to do with individual plasticity, but instead may reflect between individual differences. We now add this issue to Methods (L471-476): “This approach enabled us to control for spatial and temporal heterogeneity and specificity in escape behaviour of birds (e.g. species-specific responses, changes in escape distances with the progress in the breeding season, spatial and temporal variation in compliance with government restrictions or particularities of the given day and hour)....”

2) Looking at the figures I do not see the immense stochasticity (L156, Fig. S3, S5) claimed by the authors. Instead, I can see that some species showed an obvious behavioral change during the shutdown. For instance, Motacilla alba, Larus ridibundus, or Passer domesticus clearly reduced their escape distances, while others like the Dendrocopos major, Passer montanus, or Turdus merula tended to increase it.

At L138-141 and 327-329 we discussed the within and between genera and cross-country variation and stochasticity in response to the shutdowns (Fig. 2). The reference to species-specific plots was perhaps a little bit misleading. We think that the essential figure, that we now reference at this point, is Figure 2 that shows the temporal trends and/or stochasticity that seem to have little in common with lockdowns. Please, also look at Figure 3 and S3-S4. These show that in all selected genera/species, the trends did not significantly deviate from central regression line which indicates no change in FID before and during the COVID-19 shutdowns.

On the other hand, birds in Poland tended to have larger escape distances during the shutdown for most species, while in Rovaniemi there was an apparent reduction of escape distances in most cases. The multispecies and multisite approach is a strength of this study, but it is an Achilles' heel at the same time. The huge heterogeneity in bird responses among species and sites counterbalanced and as a result, there was an apparent lack of shutdown effects overall. Furthermore, as most data comes from a few (European) species (i.e. Columba, Passer, Parus, Pica, Turdus, Motacilla) I would say that the overall results are heavily influenced (or biased) by them. The authors realize that results are often area- or species-specific (L203), therefore, does a whole approach make sense?

We are grateful for this valuable comment. We believe the general approach makes sense as there is a general expectation about how birds should respond to changes in human presence. That is why we control for non-independence of data points in our sample. Thus, although lots of data come from a few European species, this is corrected for by the model. Note that given the sheer number of sampled species, some site- or species-specific trends may have occurred by chance. Importantly, we believe that Figure 2, with species-site specific temporal trends, reveals that the between year stochasticity in escape distances seems greater that any effects of lockdowns. Nevertheless, we have further dealt with this issue in the revised manuscript by running country-specific models which again clearly showed no significant effect of Period on escape behaviour of birds (including, no effects in Poland and Finland).

3) The previous point is worsened by the heterogeneity of cities and periods sampled. For instance:

3.1) I can hardly imagine any common feature between a small city in northern Finland (Rovaniemi) and a megacity in Australia (Melbourne). Thus, I would not be surprised to find different results between them.

3.2) Prague baseline data was for 2014 and 2018, while for the rest of the study sites were for 2018 and 2019. If study sites used a different starting point, you cannot compare differences at the final point.

We are slightly confused by these comments.

3.1) The cities are expected to be different but (i) the difference may be smaller than imagined (e.g. park structures, managed grass cover, few shrubs and deciduous-dominated tree species) and (ii) we expect the effects of lockdowns to be similar across cities. Whether we have no people in Rovaniemi parks (which despite Rovaniemi’s small size are usually extremely well-visited) or no people in Melbourne parks should not make a difference in principle. Note however, that to avoid overconfident conclusions, we allow for different reaction norms within cities. Please, also note that we are now providing country-specific results which should identify whether shutdowns lead to different reaction in sampled countries. We found no strong effect of shutdowns in any of sampled countries/cities.

3.2) Because of the possible between site differences at the starting point, we use study site as random intercept and control for the between site reaction norms by including the random slope of the period. In other words, such possible differences do not influence outcomes of our models. Regardless, our a priori expectation is that the human activity levels in a given park was similar prior to covid and hence in 2014, 2018, and 2019. Again, we are now providing country-specific results which identify whether shutdowns led to different reactions in sampled countries, which they mostly did not

3.3) Due to the obvious seasonal differences between the northern and southern hemispheres, data collection in Australia began five months later than in the rest of the sites (Aug vs Mar 2020). There, urban birds faced already too many months of reduced human disturbances, while European birds were sampled just at the beginning of the lockdown.

We agree that each city or even park within the city has its specific environmental conditions (here including the time point of lockdown). That is why we control for city and park location in the random structure of the model (see Method section). We now add results per country that shows no clear differences (e.g. Fig. 1).

However, the aim of our study was to test for general, global effects of lockdowns, which are minimal. Note that we now specifically test for country-specific effects in separate models on each country (e.g. Fig. 1, Fig S6) but all country-specific effects are small and still centre around zero.

3.4) Some cities were sampled by a single observer, while others by many of them. Even if all of them are skilled birders, they represent different observers from a statistical point of view and consequently, observer identity was an extra source of noise in your data that you did not account for.

We agree. In Finland and Hungary, data were collected by two closely cooperating observers. In Poland, all data were collected by a single observer. In the Czech Republic and Australia, a single observer (P.M. and M.W., respectively) sampled 46 sites out of 56 and 32 sites out of 37, respectively. Each site was sampled by the same observer both before and during the shutdowns. We now clearly state it in the Methods (L352-356). In other words, our models already largely control for the possible observer confound by having site as a random intercept. Moreover, previous study showed that FID estimates do not vary significantly between trained observers (Guay et al. 2013, Wildlife Research, 40, 289-293).

4) Although I liked the stringency index as a variable, I am not sure if it captured effectively the actual human activity every day. Even if restrictive measures were similar between countries, their actual accomplishment greatly depended on people's commitment and authorities' control and sanctions. I would suggest using a more realistic measure of human activity, such as google mobility reports.

Thank you for this comment. We now validate the use of the stringency index with the Google Mobility Reports, showing that human mobility generally (albeit in some countries relatively weakly) decreases with the strength of governmental antipandemic measures. Please, note that our main research question is related to the general change in human outdoor activity and not to week-to-week, day-to-day or hour-to-hour changes captured by stringency index, Google Mobility or the number of humans during an escape trial data. Nevertheless, using Google Mobility and the number of humans as predictors led to the similar results as for stringency index and Period (Fig. 1 and S6). Please, see extended discussion on this topic in our manuscript (L270-292).

5) The authors used escape trials from birds on the ground and perched birds. I think that they are not comparable, as birds on the ground probably perceive a greater risk than those placed some meters above the ground, i.e. I would expect shorter escape distances for perched birds. As this can be strongly dependent on the species preferences or sampling site (i.e, more or less available perches), I wonder how this mixture of observations from birds on the ground and perched birds could be affecting the results.

We now added information that most birds were sampled when on the ground (79%). Importantly, previous studies have found that perch height has a minimum effect on FIDs (e.g. Bjørvik et al. 2015. J Ornithol 156:239–246; Kalb et al. 2019, Ethology 125:430-438; Ncube & Tarakini 2022, Afr J Ecol 60:533– 543; Sreekar et al. 2015,. Tropic Conserv Sci 8:505-512). We added this information to the Method section (L394-395).

6) The authors did not sample the same location in the same breeding season to avoid repeated sampling of the same individuals (L331). This precaution may help, but it does not guarantee a lack of pseudoreplication. Birds are highly mobile organisms and the same individuals may be found in different places in the same city. This pseudoreplication seems particularly plausible for Rovaniemi, where sampling points must be necessarily close due to the modest size of this city.

We appreciate your concern. We cannot fully exclude the possibility of sampling some individuals twice. However, we sampled during the breeding season within which most birds are territorial, active in the areas around the nests and hence an individual switching parks is unlikely. Also, most sampled birds in our study are passerines which have small territories (typically few hundred square meters). Some larger birds may have larger territories and move larger distance to forage (e.g. kestrels which often forage outside cities) but these birds represent a minority of our records and we have not sampled outside the cities.

7) An intriguing result was that the authors collected data for 135 species during the shutdown, while they collected data only for 68 species before the pandemic. Such a two-fold increase in bird richness would not be expected with a 36% increase in sampling effort during 2020-21. I wonder if this could be reflecting an actual increase in bird richness in urban areas as a positive result of the shutdown and reduced human presence.

There were 141 unique day-years during before COVID and 161 during COVID. So, the sampling effort as calculated by days does not explain the difference in species numbers. Whether the actual effort, which was 381 vs 463 h of sampling, explains the difference is unclear, which we now note in the Methods (L476-483). If not, your proposition is possible, but we would like to avoid any speculations on this topic in the manuscript as it is difficult to infer species diversity from FID sampling.

8) The authors dismissed the multicollinearity problem of explanatory variables unjustifiably (L383). However, looking at fig. S1, I can see strong correlations between some of them. For instance, period and stringency index were virtually identical (r=0.95), while temperature and date were also strongly correlated.

We are confused by this comment and think this reflects a misunderstanding. Period and stringency index are explanatory variables of interest that were never included in the same model and hence their correlation does not contribute to the within a model multicollinearity. To avoid further confusion, we note this within (Fig. S2) legend. However, we must be cautious when interpreting the results from the models on period, Google Mobility, # of humans and stringency index, as the four measure are similar.

We discuss multicollinearity of explanatory variables within the manuscript (L458-538, 548-550) and noted that, with the exception of temperature and day within the breeding season (r = 0.48), the correlations among explanatory variables were minimal. We thus used only temperature as an explanatory variable (i.e. fixed factor; also because temperature reflects both season and variation in temperature across a season) whereas the day was included as a random intercept to control for pseudoreplication within day. Collinearity between all other predictors was low (|r| <0.36).

9) The random structure of the models is a key element of the statistical analyses but those random factors are poorly explained and justified. I needed to look up the supplementary tables to fully understand the complex architecture of the random part of the models. To the best of my knowledge, random variables aim to account for undesirable correlations in the covariance matrix, which is expected in hierarchical designs, such as the present one. However, the theoretical violation of data independence may happen or not. As the random structure is usually of little interest, you should keep it as simple as necessary, otherwise random factors may be catching part of data variability that you would like to explain by fixed variables. I think that this is what is happening (at least, in part) here, as the authors included a too-complex random structure. For instance, if you include the year as a random factor, I think that you are leaving little room for the period effect. The authors simplified the random structure of the models (L387), but they did not explain how. Nevertheless, this model selection was not important at all, as the authors showed the results for several models. I assume, consequently, that the authors are considering all these models equally valid. This approach seems quite contradictory.

The random structure of the model controls for possible pseudoreplication in the data, that is for the cases where we have multiple data points that may not be independent and hence technically represent one. Apart from that, random structure tells us about where the variance in the data lies. This is often of interest and your previous questions about city, site or species specificities can be answered with the random part of the model. To follow up on your example, year is included in the model because data from a single year are not independent (for example because of delayed breeding season in one year vs. in another).

We regret being unclear about the model specification and have attempted to clarify the methods (L466-476). We first specified a model with an ideal random structure that necessarily was complex (perhaps too complex). We then showed that using models with simpler random structures did not influence the outcomes. We now use a simpler model within the main text, but do keep the alternative models to show that the results are not dependent on the random structure of the model (Fig. S1 and Table S2).

Reviewer #3 (Public Review):

This study examined the changes in fear response, as measured by the flight initiation distances (FID), of birds living in urban areas. The authors examined the FIDs of birds during the pandemic (COVID-19 lockdown restrictions) compared to FIDs measured before the pandemic (mostly in 2018 & 2019). The main study justification was that human presence changed drastically during the pandemic lockdowns and the change in human presence might have influenced the fear response of birds as a result of changing the "landscape of fear". Human presence was quantified using a 'stringency' index (government-mandated restrictions). Urban areas were selected from within five different cities, which included four European cities (Czech Republic - Prague, Finland - Rovaniemi, Hungary - Budapest, Poland - Poznan), and one city in the global south (Australia - Melbourne). Using 6369 flight initiation distances across 147 different bird species, the authors found that FIDs were not significantly different before the pandemic versus during the pandemic, nor was the variation in FID explained by the level of 'stringency'.

Major strengths: There are several strengths to this study that allows for understanding the variety of factors that influence a bird's response to fear (measured as flight initiation distances). This study also demonstrates that FIDs are highly variable between species and regions.

Specifically,

1) One of the major strengths of this paper is the focus on birds living in urban areas, a habitat type that is hypothesized to have changed drastically in the 'landscape of fear' experienced by animals during the pandemic lockdown restrictions (due to the presumed decrease in human presence and densities). Maintaining the focus on urban birds allowed for a deeper examination of the effect of human behaviour changes on bird behaviour in urban habitats, which are at the interface of human-wildlife interactions.

2) This study accounted for several variables that are predicted to influence flight initiation distances in birds including species, genus, region (country), variability between years, pandemic year (pre- versus during), the strictness of government-mandated lockdown measures, and ecological factors such as the human observer starting distance, flock size, species-specific body size, ambient air temperature (also a proxy of the timing during the breeding season), time of day, date of data collection (timing within the regional [Europe or Australia] breeding season), and categorization of urban site type (e.g. park, cemetery, city centre).

3) This study examined FIDs in two years previous to the pandemic (mostly 2018 and 2019, one site was 2014) which would account for some of the within- and between-year FID variation exhibited prior to the pandemic.

4) This study uses strong statistical approaches (mixed effect models) which allows for repeat sampling, and a post hoc analysis testing for a phylogenetic signal.

Thank you for your supportive and positive comments.

Major weaknesses: The authors used government 'stringency' as a proxy for human presence and densities, however, this may not have been an accurate measure of actual human presence at the study sites and during measurements of FIDs. Furthermore, although the authors accounted for many factors that are predicted to influence fear response and FIDs in birds, there are several other factors that may have contributed to the high level of variation and patterns in FIDS observed during this study, thus resulting in the authors' conclusion that FIDs did not vary between pre- and during pandemic years.

Thank you for your suggestions. We agree. To capture the general human presence in parks, we now incorporated an analysis using Google Mobility Reports (Fig S6b) that directly measures human mobility in each of sampled cities and specifically in urban parks where most our data were collected, and also address your further concerns that you detail below. Albeit not the main interest of our study, we now also incorporated an analysis using actual # of humans during an escape trial (Fig. S6c).

Moreover, we think that including further possible confounds should not influence our conclusions. In other words, including further confounds will decrease the variance that can be explained by shutdowns and thus such shutdown effects (if any) would be tiny and hence likely not biologically meaningful.

Specifically,

1) The authors used "government stringency" as a measure of change in human activity, which makes the assumption that the higher the level of 'stringency', the fewer humans in urban areas where birds are living. However, the association between "stringency" and actual human presence at the study sites was not measured, nor was 'stringency' compared to other measures of human presence such as human mobility.

Thank you for this essential comment. Initially, we viewed Oxford Stringency Index as the best available index for our purposes. However, we now further acknowledge its limitations (L) and validate the Oxford Stringency Index with the Google Mobility Reports data, showing that both indices are generally negatively (albeit sometimes weakly) correlated across sampled cities (i.e. human mobility decreases with the increasing stringency index). Although other human presence indices were used in the past, e.g. Cuebiq, Descartes Labs and Maryland Uni index, Apple (see Noi et al. 2022, Int J Geograph Info Sci, 36, 585-616), we used only the Google Mobility index because (a) it is publicly available, (b) is available also for territories outside US, and (c) provides data for urban parks within each city included in our dataset. Note however that Google Mobility data are inappropriate to answer our primary question, i.e. whether changes in human presence outdoors due to the COVID-19 shutdowns had any effect on avian tolerance towards humans. First, Google Mobility was available only for 2020-22, i.e. the baseline pre-COVID-19 data for 2018-2019 were unavailable. Thus, there was no way to check whether the human activity levels really changed during the COVID-19 years. Second, Google Mobility data are calculated as a change from 2020 January–February baseline for each day of the week for each city and its location (here we used parks). In other words, the data are not comparable between days and cities, albeit we attempted to correct for this within the random structure of the mixed model. Also, the data may be influenced by extreme events within the 2020 Jan–Feb baseline period (see here). Third, the Google Mobility varies greatly between days and across season (see Fig 4 & S5 or the first figure in these responses), likely more than the possible change due to shutdowns. Nevertheless, we found that results based on Google Mobility are qualitatively very similar to results based on stringency index. Moreover, we showed that the relationships between # of humans and both Google Mobility or Stringency index (Figure 6) are weak and noise with 95%CIs widely overlapping zero (Table S3b-e). Also, similarly to other predictors of human presence, # of humans only poorly predicted changes in avian escape distances. We added details on the new analysis into the Methods and Results and Supplement (L134-165 and associated figures and tables, L415-535).

2) There was considerable variation in FID measurements, which can be seen in the figures, indicating that most of the variation in FID was not accounted for in the authors' models.

We are confused by this statement. The fact that the FIDs varied does not translate directly to that our models did not account for the variation. Nevertheless, we do control for most of the discussed confounds (see further answers below). Importantly, it is unclear how including further possible confounds should influence our conclusions, unless the lockdowns effects are tiny, in which case those might not be biologically meaningful.

Factors that may have contributed to variation in FIDs that were not accounted for in this study are as follows:

a. The authors accounted for the date of data collection using the 'day' since the start of the general region's breeding season (Europe: Day 1 = 1 April; Australia: Day 1 = 15 August). Using 'day' since the breeding season started probably was an attempt to quantify the effect of the breeding stage (e.g. territory establishment, nest young, fledgling) on FIDs. However, breeding stages vary both within- and between species, as well as between sub-regions (e.g. Finland vs. Hungary). As different species respond to predation or human presence differently depending on the stage during their breeding cycle, more specificity in the breeding cycle stage may allow for explaining the observed variation and patterns in FID.

We agree. Although we don’t have a precise city-specific information on the timing of breeding stages in sampled populations of birds, we partly control for these effects by including a random intercept of day within each year and species. This random factor explained relatively high portion of the variance in our data (see Table S1 and S2) - perhaps something you expected.

b. Variation in species-specific FIDs may also vary with habitat features within urban sites, such as the proximity of trees and other protective structures (e.g. perches and cover), the openness of the area, and the level of stressors present (e.g. noise pollution, distance to roads). Perhaps accounting for this habitat heterogeneity would account for the FID variation measured in this study.

We agree. We don’t have such fine-scale data, but we included site identity (typically within a particular park or cemetery) which should account for the habitat heterogeneity among localities. Depending on the model, site explained relatively little variance (1-6%), indicating low heterogeneity between localities in these undescribed characteristics. Also note that park structure may be quite similar both within and between cities, i.e. managed green grass areas, with only a few shrubs and deciduous trees. Therefore, the possible minor habitat heterogeneity should not have any great impacts on our results.

c. The authors accounted for species and genus within their models, however, FIDs may vary with other species-specific (or even specific populations of a species) characteristics such as whether the species/population is neophobic versus neophilic, precocial versus altricial, and the level of behavioural plasticity exhibited. These variables were not accounted for in the analysis.

We agree that FIDs can be correlated with many possible factors. Here, we were interested in general patterns, while controlling for FID differences between species, as well as for possible species-specific reaction norms to lockdowns. Whether neophobic vs neophilic population or precocial versus altricial species react differently to lockdowns might be of interest, but it is beyond the scope of this study. However, that population and population specific reaction norms explain little variation (Table S2a, 0-6% of variation) so such a confound should not substantially influence our conclusion much. We do not have fine-scale data on the level of neophobia, but the effects of lockdowns seem similar for precocial (see Anas, Larus, Cygnus) and altricial (the remaining, mostly passerine) species in our dataset (see Fig. 3 and S3-S4). Please, note that we sampled mainly adults (L386). Moreover, the effects for clades, which may differ in their cognitive skills, are also similar (e.g. Corvids vs. Anas or Cygnus; Fig. 3).

d. Three different methods of measuring the distances between flight and the observer location were used, and FIDs were only measured once per bird, such that there were no measures of repeatability for a test subject. Thus, variation surrounding the measurement of FIDs would have contributed to the variation in FIDs seen during this study.

While all observers were trained, the three methods may add some noise to the FID estimates. However, the FID estimates from a single method may still slightly differ between observers (so do well standardized morphology measurements; Wang, et al. 2019, PLoS Biology, 17, e3000156). Importantly, FID estimates are highly replicable among skilled observers (Guay et al. 2013, Wildlife Research 40:289-293), and we previously validated this approach and showed that distance measured by counting steps did not differ from distance measured by a rangefinder (Mikula 2014, Ardea 102:53-60), which we now explicitly state (L391-394). Importantly, we control for observer bias by specifying locality as a random intercept (see further details in our response to the Editor). Moreover, each site was sampled by the same observer both before and during the shutdowns.

3) The sample design of this study may have influenced the FID variability associated with specific species, and specific populations of species. A different number of species were sampled across the time periods of interest; 68 species were sampled before the pandemic versus 135 species after the pandemic. However, the authors do not appear to have directly compared the FIDs for the same species before the pandemic compared to during the pandemic (e.g. the FIDs of Eurasian blackbirds before the pandemic versus during the pandemic). Furthermore, within the same country-city, it is unclear whether the species observed before the pandemic were observed at the same location (e.g. same habitat type such as the same park) during the pandemic. As a species' FID response may be influenced by population characteristics and features specific to each site (e.g. habitat openness), these factors may have influenced the variability in FID measurements in this study.

We regret being unclear in our methods. Our full model uses all data, but alternative models (see e.g. Fig. S1) used data with ≥5 as well as ≥10 observations before and during lockdowns for a given species. Importantly, Figure 2 and 3 depict data for species sampled at specific sites. We now clarify this within the Methods (L460-483) and the Results (L125-133 and associated figures) and in the figure legends (Fig. S1).

4) The models in this study accounted for many factors predicted to affect FIDs (see the section on major strengths), however, the number of fixed and random factors are large in number compared to the total sample size (N =6369), such that models may have been over-extended.

The number of predictors and random effects is well within the limits for the given sample size (Korner-Nievergelt et al. 2015. Bayesian Data Analysis in Ecology Using Linear Models with R, BUGS, and Stan). Importantly, simpler models give similar results as the more complex ones (Fig. S1) and the visual (model free) representations of our raw and aggregated data confirm our model results. This, we suggest, makes our findings robust and convincing.

Overarching main conclusion

Overall, this study examines factors influencing FIDs in a variety of bird species and concludes that FIDs did not differ during the pandemic lockdowns compared to before the pandemic (2019 and earlier). Furthermore, FIDs were not influenced by the strictness of government-mandated restrictions. Although the authors accounted for many factors influencing the measurement of FIDs in birds, the authors did not achieve their aim of disentangling the effects of pandemic-specific ecological effects from ecological effects unrelated to the pandemic (such as habitat heterogeneity).

We find this statement confusing. We accounted for most relevant confounding factors and found little evidence for the strong effect of pandemic. Moreover, we now added country-specific analyses that confirm the lack of evidence, highlight the Figure 3 that shows no clear shutdown effect and also explore how levels of human presence changed over and within the years. Adding more possible confounds (albeit note that not many are left to add) might only further reduce the variation that could be explained by pandemic and hence such hypothetical effects of pandemic will be if anything small and thus likely not biologically meaningful.

Their findings indicate that FIDs are highly variable both within- and between- species, but do not strongly support the conclusion that FIDs did not change in urban species during the pandemic lockdown. Therefore, this study is of limited impact on our understanding of how a drastic change in human behaviour may impact bird behaviour in urban habitats.

It is unclear why you think our study lacks support for the conclusion that FIDs changed little during pandemic, if all results show no such effects. However, we toned down our Discussion and highlighted also potential issues linked to our approach (e.g. that sampled individuals were not marked and hence we cannot distinguish between various mechanisms that might explain the described pattern (L293-329) or that human presence may not have changed (L253-269). For further details see our previous response.

Overall, the study demonstrates the challenges in using FIDs as a general fear response in birds, even during a pandemic lockdown when fewer humans are presumably present, and this study illustrates the large degree of variation in FIDs in response to a human observer.

We appreciate and agree that our study demonstrates the challenges in quantifying human activity to understand bird escape distance and we added a paragraph on this topic to the discussion (L270-292).

Nevertheless, we hope that our above responses clarify and address most of the issues you had with our manuscript. We tried to show that (a) most of your proposed controls are indeed included in our study design, models, and visualisations, and that (b) multiple evidence (from models and visualisation of raw and aggregated data) support the no overall effect conclusion. We further emphasize the temporal and between- and within-species variability in FIDs in the Results and now specifically indicate that lockdowns did not influenced FIDs above such variability (Fig. 2-3, Fig. S3). In other words, the natural (e.g. temporal) variation in FIDs seems far greater that potential effects of lockdowns (Fig. 2). We believe that even if lockdowns would have tiny effects that could have been detected with more. stringent experimental design (e.g. individually tagged birds) or even more complex models, such effects would be far from being biologically meaningful.

Reviewer #1 (Public Review):

This paper uses a series of flight initiation "challenges" conducted both prior to and during COVID-19-related restrictions on human movement to estimate the degree to which avian escape responses to humans changed during the "anthropause". This technique is suitable for understanding avian behavioral responses with a high degree of repeatability. The study collects an impressive dataset over multiple years across five cities on two continents. Overall the study finds no effect of lockdown on avian escape distance (the distance at which the "target" individual flees the approaching observer). The study considers the variable of interest as both binary (during lockdown or prior to lockdown) and continuous, using the Oxford Stringency Index (with neither apparently affecting escape distance).

Overall this paper presents interesting results which may suggest that behavioral responses to humans are rather inflexible over "short" (~2 year) timespans. The anthropause represents a unique opportunity to disentangle the mechanistic drivers of myriad hypothesized impacts humans have on the behavior, distribution, and abundance of animals. Indeed, this finding would provide important context to the larger body of literature aimed at these ends. However, the paper could do more to carefully fit this finding into the broader literature and, in so doing, be a bit more careful about the conclusions they are able to draw given the study design and the measures used. Taking some of these points (in no particular order):

1) Oxford Stringency Index is a useful measure of governmental responses to the pandemic and it's true that in some scenarios (including the (Geng et al. 2021) study cited by this paper) it can correlate with human mobility. However, it is far from a direct measure of human mobility (even in the Geng study, to my reading, the index only explained a minority of the variation). Moreover, particular sub-components of the index are wholly unrelated to human mobility (e.g., would changes to a country's public information campaign lead to concomitant changes in urban human mobility?). Finally, compliance with government restrictions can vary geographically and over time (i.e., we might expect lower compliance in 2021 than in 2020) and the index is calculated at the scale of entire countries and may not be very reflective of local conditions. Overall this paper could do more to address the potential shortcomings of the Oxford Stringency Index as a measure of human mobility including attempting to validate the effect on human mobility using other datasets (e.g., the google dataset and/or those discussed in (Noi et al. 2022). This is of critical importance since the fundamental logic of the experimental design relies on the assumption that stringency ~ mobility.

2) The interpretation of the primary finding (that behavioral responses to humans are inflexible) could use a bit more contextualization within the literature. Specifically, the study offers three potential explanations for the observed invariance in escape response: 1) these behaviors are consistent within individuals and this study provides evidence that there was no population turnover as a result of lockdowns; 2) escape response is linked to other urban adaptations such that to be an urban-dwelling species dictates escape response; and/or 3) these populations already exhibit maximum habituation and the reduction in human mobility would only have increased that habituation but that trait is already at a boundary condition. Some comments on each of these respectively:

a) Even had these populations turned over as a result of a massive rural-to-urban dispersal event, it's not clear that the escape distance in those individuals would be different because this paper does not establish that these hypothetical rural birds have a different behavioral response which would be constant following dispersal. Thus the evidence gathered here is insufficient to tell us about possible relocations of the focal species. Additionally, the paper cites several papers that found no changes in abundance or movements of animals in response to lockdowns but ignore others that do. For example: (Wilmers et al. 2021), (Warrington et al. 2022) (though this may have been published after this was submitted...), and (Schrimpf et al. 2021). There is a missed opportunity to consider the drivers of some of these results - the findings in this paper are interesting in light of studies that *did* observe changes in space use or abundance - i.e., changes in space use could arise precisely *because* responses to humans are non-plastic but the distribution and activities of humans changed. To wit, the primary finding here would imply that the reaction norm to human presence is apparently fixed over such timescales - however, and critically, the putative reduction in human activity/mobility combined with fixed responses at the individual level might then imply changes in avian abundance/movement/etc.

b) If this were the case, wouldn't this be then measurable as a function of some measure of urbanity (e.g. Human Footprint Index) that varies across the cities included here? Site accounted for ~15% of the total variation in escape distance but was treated as a random effect - perhaps controlling for the nature of the urban environment using some e.g., remotely sensed variable would provide additional context here.

c) Because it's not clear the extent to which the populations tested had turned over between years, the paper could do with a bit more caution in interpreting these results as behavioral. This study spans several years so any response (or non-response) is not necessarily a measure of behavioral change because the sample at each time point could (likely does) represent different individuals. In fact, there may be an opportunity here to leverage the one site where pre-pandemic measures were taken several years prior to the pandemic. How much variance in the change in escape distance is observed when the gap between time points far exceeds the lifetime of the focal taxa versus measures taken close in time?

d) Finally, I think there are a few other potential explanations not sufficiently accounted for here:

i) These behaviors might indeed be plastic, but not over the timescales observed here.<br /> ii) Time of year - this study took place during the breeding season. The focal behavior here varies with the time of year, for example, escape distance for many of these species could be tied up in nest defense behaviors, tradeoffs between self-preservation and e.g., nest provisioning, etc.<br /> iii) Escape behaviors from humans are adaptively evolved, strongly heritable, and not context dependent - thus we would only expect these behaviors to change on evolutionary timescales.<br /> iv) See point one above - it's possible that the lockdown didn't modify human activity sufficiently to trigger a behavioral response or that the reaction norm to human behavior is non-linear (e.g. a threshold effect).

LITERATURE CITED<br /> Geng DC, Innes J, Wu W, Wang G. 2021. Impacts of COVID-19 pandemic on urban park visitation: a global analysis. J For Res 32:553-567. doi:10.1007/s11676-020-01249-w

Noi E, Rudolph A, Dodge S. 2022. Assessing COVID-induced changes in spatiotemporal structure of mobility in the United States in 2020: a multi-source analytical framework. Int J Geogr Inf Sci.

Schrimpf MB, Des Brisay PG, Johnston A, Smith AC, Sánchez-Jasso J, Robinson BG, Warrington MH, Mahony NA, Horn AG, Strimas-Mackey M, Fahrig L, Koper N. 2021. Reduced human activity during COVID-19 alters avian land use across North America. Sci Adv 7:eabf5073. doi:10.1126/sciadv.abf5073

Warrington MH, Schrimpf MB, Des Brisay P, Taylor ME, Koper N. 2022. Avian behaviour changes in response to human activity during the COVID-19 lockdown in the United Kingdom. Proc Biol Sci 289:20212740. doi:10.1098/rspb.2021.2740

Wilmers CC, Nisi AC, Ranc N. 2021. COVID-19 suppression of human mobility releases mountain lions from a landscape of fear. Curr Biol 31:3952-3955.e3. doi:10.1016/j.cub.2021.06.050

Reviewer #2 (Public Review):

Mikula et al. have a large experience studying the escape distances of birds as a proxy of behavioral adaptation to urban environments. They profited from the exceptional conditions of social distance and reduced mobility during the covid-19 pandemic to continue sampling urban populations of birds under exceptional circumstances of low human disturbance. Their aim was to compare these new data with data from previous "normal" years and check whether bird behavior shifted or not as a consequence of people's lockdown. Therefore, this study would add to the growing body of literature assessing the effect of the covid-19 shutdown on animals. In this sense, this is not a novel study. However, the authors provide an interesting conclusion: birds have not changed their behavior during the pandemic shutdown. This lack of effects disagrees with most of the previously published studies on the topic. I think that the authors cannot claim that urban birds were unaffected by the covid-19 shutdown. I think that the authors should claim that they did not find evidence of covid-19-shutdown effects. This point of view is based on some concerns about data collection and analyses, as well as on evolutionary and ecological rationale used by the authors both in their hypotheses and results interpretation. I will explain my criticisms point by point:

1) The authors used ambivalent, sometimes contradictory, reasoning in their predictions and results interpretation. Some examples:<br /> 1.1) The authors claimed that urban birds perceive humans as harmless (L224), but birds actually escape from us, when we approach them... Furthermore, they escape usually 5 to 20 m away. This is more distance that would be necessary just to be not trampled.<br /> 1.2) If we are harmless, why birds should spend time monitoring us as a potential threat (L102)? Indeed, I disagree with the second prediction of the authors. I could argue that reduced human activity should increase animal vigilance because real bird predators (e.g., raptors) may increase their occurrence or activity in empty cities. If birds should increase their vigilance because the invisible shield of human fear of their predators is no longer available, then I would expect longer escape distances.<br /> 1.3) To justify the same escape behavior shown by birds in pre- and pandemic conditions from an adaptive point of view, the authors argued a lack of plasticity and a strong genetic determination of such behavior. This contravenes the plasticity proposed in the previous point or the expected effect of the stringency index (L112). In my opinion, some degree of plasticity in the escape behavior would be really favorable for individuals from an adaptive perspective, as they may face quite different fear landscapes during their lives. Looking at the figures, one can see notable differences in the escape distance of the same species between sites in the same city. As I can hardly imagine great genetic differences between birds sampled in a park or a cemetery in Rovaniemi, for instance, I would expect a major role of plasticity to explain the observed variability. Furthermore, if escape behavior would not be plastic, I would not expect date or hour effects. By including them in their models, the authors are accepting implicitly some degree of plasticity.

2) Looking at the figures I do not see the immense stochasticity (L156, Fig. S3, S5) claimed by the authors. Instead, I can see that some species showed an obvious behavioral change during the shutdown. For instance, Motacilla alba, Larus ridibundus, or Passer domesticus clearly reduced their escape distances, while others like the Dendrocopos major, Passer montanus, or Turdus merula tended to increase it. On the other hand, birds in Poland tended to have larger escape distances during the shutdown for most species, while in Rovaniemi there was an apparent reduction of escape distances in most cases. The multispecies and multisite approach is a strength of this study, but it is an Achilles' heel at the same time. The huge heterogeneity in bird responses among species and sites counterbalanced and as a result, there was an apparent lack of shutdown effects overall. Furthermore, as most data comes from a few (European) species (i.e., Columba, Passer, Parus, Pica, Turdus, Motacilla) I would say that the overall results are heavily influenced (or biased) by them. The authors realize that results are often area- or species-specific (L203), therefore, does a whole approach make sense?

3) The previous point is worsened by the heterogeneity of cities and periods sampled. For instance:<br /> 3.1) I can hardly imagine any common feature between a small city in northern Finland (Rovaniemi) and a megacity in Australia (Melbourne). Thus, I would not be surprised to find different results between them.<br /> 3.2) Prague baseline data was for 2014 and 2018, while for the rest of the study sites were for 2018 and 2019. If study sites used a different starting point, you cannot compare differences at the final point.<br /> 3.3) Due to the obvious seasonal differences between the northern and southern hemispheres, data collection in Australia began five months later than in the rest of the sites (Aug vs Mar 2020). There, urban birds faced already too many months of reduced human disturbances, while European birds were sampled just at the beginning of the lockdown.<br /> 3.4) Some cities were sampled by a single observer, while others by many of them. Even if all of them are skilled birders, they represent different observers from a statistical point of view and consequently, observer identity was an extra source of noise in your data that you did not account for.

4) Although I liked the stringency index as a variable, I am not sure if it captured effectively the actual human activity every day. Even if restrictive measures were similar between countries, their actual accomplishment greatly depended on people's commitment and authorities' control and sanctions. I would suggest using a more realistic measure of human activity, such as google mobility reports.

5) The authors used escape trials from birds on the ground and perched birds. I think that they are not comparable, as birds on the ground probably perceive a greater risk than those placed some meters above the ground, i.e. I would expect shorter escape distances for perched birds. As this can be strongly dependent on the species preferences or sampling site (i.e, more or less available perches), I wonder how this mixture of observations from birds on the ground and perched birds could be affecting the results.

6) The authors did not sample the same location in the same breeding season to avoid repeated sampling of the same individuals (L331). This precaution may help, but it does not guarantee a lack of pseudoreplication. Birds are highly mobile organisms and the same individuals may be found in different places in the same city. This pseudoreplication seems particularly plausible for Rovaniemi, where sampling points must be necessarily close due to the modest size of this city.

7) An intriguing result was that the authors collected data for 135 species during the shutdown, while they collected data only for 68 species before the pandemic. Such a two-fold increase in bird richness would not be expected with a 36% increase in sampling effort during 2020-21. I wonder if this could be reflecting an actual increase in bird richness in urban areas as a positive result of the shutdown and reduced human presence.

8) The authors dismissed the multicollinearity problem of explanatory variables unjustifiably (L383). However, looking at fig. S1, I can see strong correlations between some of them. For instance, period and stringency index were virtually identical (r=0.95), while temperature and date were also strongly correlated.

9) The random structure of the models is a key element of the statistical analyses but those random factors are poorly explained and justified. I needed to look up the supplementary tables to fully understand the complex architecture of the random part of the models. To the best of my knowledge, random variables aim to account for undesirable correlations in the covariance matrix, which is expected in hierarchical designs, such as the present one. However, the theoretical violation of data independence may happen or not. As the random structure is usually of little interest, you should keep it as simple as necessary, otherwise random factors may be catching part of data variability that you would like to explain by fixed variables. I think that this is what is happening (at least, in part) here, as the authors included a too-complex random structure. For instance, if you include the year as a random factor, I think that you are leaving little room for the period effect. The authors simplified the random structure of the models (L387), but they did not explain how. Nevertheless, this model selection was not important at all, as the authors showed the results for several models. I assume, consequently, that the authors are considering all these models equally valid. This approach seems quite contradictory.

This work has been published in GigaByte Journal under a CC-BY 4.0 license (https://doi.org/10.1093/gigascience/giad055) and has published the reviews under the same license. These are as follows.

**Reviewer 1. Thomas Brüls **

The authors present a snakemake-based workflow to automate and chain the main computational ingredients (assembly and binning) of genome-centric metagenomics; the authors developed a technically sound tool for this purpose, and by itself it is certainly valuable to the research community and worth of publication. however, even if the article is casted as a technical note -hence with an emphasis on the design, implementation and assessment of the tool-, I feel that a more thorough discussion of both its abilities and inabilities (e.g. strain resolution, detection of low abundance organisms, identification of virus bins, etc) would be worth for a more general audience. On the same token, a more deep discussion of some of the results obtained with their tool (see below) would be of interest and would also illustrate useful use cases. I would suggest the following modifications/additions:-the experiments with the strain madness dataset suggest that the genomes (or fragments thereof, i.e. the bins) resolved should be viewed as "species" genomes, or composite genomes possibly originating from multiple strains. if so, do the authors think this represents a hard limit to the assembly + binning approach, or could further existing tools (e.g. performing variant detection on top of cross-assembly before the binning step) be integrated or developed in the future for strain-resolution (i.e. to identify strains not dominant in any sample)? -related, a simple summary of the number of individual strains recovered in individual bins for the strain madness experiment would be interesting.-another issue that would be worth discussing in my opinion is the impact of genome abundance on the recovery of corresponding bins and their quality. the platform developed by the authors appears to be well suited for such kind of analyses and the results would be of both theoretical and practical interest. to put it simply, what is the minimal initial coverage of genomes required in order for them to be recovered in bins of a given size and quality?-rem: theses two issues (strain-level diversity and individual strain genome abundances) likely interact to limit bin resolution, and this could be mentioned by the authors.-the data presented by the authors suggest that the metabat binning engine significantly outperforms the other two tools (concoct and vamb, which are both widely used), see e.g Figure 2; what would account for that, and do the authors think this is a general observation (i.e. beyond the specific CACB setting or marine metagenome shown in Fig 2)? -a bin refinement step (based on the DAS tool and dereplication) is frequently mentioned but should be more detailed (including a precise definition of the bin quality metric used).

further rather minor comments: -in the abstract, when mentioning "technical challenges associated with...", it would be worth mentioning that algorithmic challenges are present as well. -in the introduction, "It is hypothesised that pooled assembly and binning may lead to improved results when analysing communities with high genetic diversity, and to poorer results when there is a high level of intraspecies/strain-level diversity". I would assume there are many instances in the real world that are both, i.e. that present both high inter-species and intra-species genetic diversity, what then?-in the future directions, the authors mention the identification of eukaryotic and viral contigs and bins, and could shortly elaborate how this could be done properly. -the sentence "In summary, our assessment of ..." at the end of the ms appears to have a syntactic problem.

**Reviewer 2. Paolo Provero **

In this work Himmelstein and collaborators introduce a statistically controlled way of extracting significant node pairs in heterogeneous networks (hetnets) without relying on a ground truth and related training. The method "explains" why two nodes are significantly connected by extracting the metapaths most responsible for the enrichment. This is based on computing a null distribution of the DWPC, which allows assigning a P-value to each metapath joining two nodes, and then to visualize the individual paths responsible for the enrichment. The method is novel and significant, and can be in principle be applied to many hetnets, in life sciences and beyond, when a ground truth is not available or not desirable as it would introduce bias. The software tools developed appear to be readily available to other researchers.

Major comment: If I understand correctly, given two nodes (say "Alzheimer disease" and "Circadian rhythm") the method extracts, in a statistically controlled way, the most significant metapaths joining the two nodes, and then the individual paths responsible for the enrichment. But this is not the most obvious question a life scientist would ask the network, which would be instead something like "Which are the pathways most significantly connected to "Alzheimer disease"? Indeed this type of question would be the one to ask when aiming for drug repurposing (possibly replacing "pathways" with "compounds" or "pharmacologic classes"). Based on Fig. 4A, the pathways are presented, or "suggested," in decreasing order of number of metapaths, but this is hardly a ranking by significance. Would it be possible to summarize the results in such a way as to rank the pathway nodes connected to a given disease node by significance (or more generally to rank the nodes of a certain type by the significance of their connection to a given node of another type)? This should be discussed.

I also have several minor concerns. (1) The authors introduce and compute a null distribution of the DWPC which takes into account node degree in a statistically controlled way when evaluating the connectivity between two nodes. However, the DWPC itself does take into account node degree, as the name implies, and contains a tunable parameter that can be optimized, at least when a ground truth is available (as in Ref 39 by the same first author). I understand such tuning is not possible when, as in the present case, no ground truth is available, but the authors should make this point more clearly. (2) I find Fig. 1B a bit confusing: according to the legend, the top rows are known treatments, which should have higher than expected connectivity. However, based on the colors as explained by the legend, the bottom treatment/disease pairs seem to have higher connectivity (3) The acronym DWPC is defined after it has been used several times (4) The legend of Figure 2 should specify that these results apply to the nodes "Alzheimer disease" and "Circadian rhythm", although this becomes clear in Fig. 4 (5) I don't think Figure 3, representing the home page of the web site, is especially useful (6) I found Fig. 4 confusing: the sum of the path counts for the selected metapaths in panel B is way larger than the 425 results shown in Panel C. As far as I understand no path can belong to more than one metapaths, so is there some further selection here? (7) The "Frontend" section of the Methods seems a bit too detailed for the Gigascience audience.

Re-review: The authors have addressed all my comments in a satisfactory way.

Author Response

Reviewer #2 (Public Review):

This work attempts to connect the diet of a mother to the physiology and feeding behaviors of multiple generations of her offspring. Using genetic and molecular biology approaches in the fruit fly model, the authors argue that this Lamarckian inheritance is mediated by germline-inherited chromatin and is regulated by the general activity of a histone methylase. However, many of the measured effects are small and variable, the statistical tests to prove their significance are missing or poorly described, and some experiments are inadequately described and lack important controls.

1) The authors claim that the diet of a mother can influence the physiology of her progeny for several generations. However, the observed effects of maternal diet on later generations were small and variable for most assays (see Fig1C, S1.1A, B, D). Additionally, the effect size between F0 HSD to ND was often larger than the effect size between the progeny of F0 parents and ND. To put it another way, if the authors were to compare the F1, F2, etc. to the F0 HSD flies, they would conclude that the majority of the response to diet is not maternally transmitted, and is directly controlled by the diet of the individual being measured.

We agree with the reviewer that the effect size of acute HSD exposure (in HSD-F0 flies) was stronger than that of transgenerational inheritance (in HSD-F1/2/3/4 flies). Similar observations were also made in other studies, see Klosin et al., Science, 2017, Bozler et al., eLife, 2019. We would argue this difference in effect size was as expected and with clear biological relevance.

For all living organisms, acute environmental changes (diet change included) have direct and profound influences on their survival and reproduction, and therefore need robust and immediate responses. In comparison, ancestral environmental changes may only provide some vague and indirect indications of the current living environment of the offspring. Such information may be beneficial for the survival and reproduction of the offspring, but the effect size is expected to be much smaller, or at least smaller than that of acute environmental changes.

Studies on Dutch Famine offers a good example. Human individuals who were prenatally exposed to famine were found to be associated with greater risk in metabolic diseases (Ravelli et al., NEJM, 1976). But nevertheless, direct high-fat diet exposure was still the much stronger risk factor for obesity and metabolic disorders (Bray et al., Am J Clin Nutr, 1998, Jéquier et al., Int J Obes Relat Metab Disord, 2002).

We have added additional discussions in the manuscript for clarification.

Furthermore, since our current study aimed to investigate the mechanism of behavioral transgenerational inheritance, we focused on the comparison between HSD-F1 flies (and their progeny) vs. ND-fed flies. As the ancestors of HSD-F1/2/3/4 flies were exposed to HSD, whereas HSD-F1/2/3/4 flies themselves were never exposed to HSD, any difference we observed between the two groups could be solely attributed to transgenerational inheritance of ancestral HSD exposure. With that saying, to better distinguish the effects of acute HSD exposure vs. transgenerational inheritance upon ancestral HSD exposure, we re-analysed and presented the comparisons among ND, HSD-F0, and HSD-F1 data in the manuscript (Figure 1. B-E, Figure 1-figure supplement 1. A-E, Figure 1-figure supplement 2. A-D, Figure 3. D-E, Figure 3-figure supplement 1. B-D, Figure 3-figure supplement 2 and 3. A-B).

2) The authors chose to study PER, which had the largest average effect sizes between conditions. However, PER was highly variable in the averaged data, with some individuals showing large effects and others having no effects. A better characterization of transgenerational PER may increase the robustness of this assay and confidence in its results. For example, the authors could measure PER in lineages derived from individual flies to determine when transgenerational effects on PER decline or disappear. This form of data collection could help to explain the high variation in the averaged data presented in the paper.

We acknowledged that PER in general was quite a variable behavioural trait (probably as to most if not all behavioural measures). It was not surprising since animal behaviours, as complex traits, could be influenced by numerous intrinsic and extrinsic factors, such as genetic background, developmental environment, diet, population density, environmental conditions, etc. Numerous PER studies have exhibited similar variability (Masek et al., PNAS, 2010, Marella et al., Neuron, 2012, Charlu et al., Nature Communication, 2013, Wang et al., Cell Metabolism, 2016, Wang et al., Cell Reports, 2020).

Nevertheless, in our current study we were able to identify statistically significant behavioural difference between ND-fed flies and HSD-F1/2/3 flies, demonstrating that ancestral HSD exposure imposed transgenerational inheritance on sweet sensitivity. To further increase the robustness of the study as suggested by the reviewer, we have conducted additional repetitions of many PER experiments and further confirmed the phenotype with less variability and more statistical power (Figure 1. G-I, Figure 3. D-E, Figure 3-figure supplement 1. B-D, Figure 3-figure supplement 2 and 3. A-B). The reviewer also suggested the use of isogenic flies, which might help to minimize the variations of genetic background. However, we think that demonstrating the behavioural difference in genetically diverse fly populations is a more credible way to show that such transgenerational inheritance is a reliable and generalizable phenomenon.

3) What do the error bars represent on any figure? There are many examples where the data is highly variable and lies completely outside of the error bars. What is the statistical test for significance that is carried out in each figure? The brief comment about statistics in the methods section is inadequate. The authors should also supply the raw data used to generate the figures so that readers can perform their own statistical tests.

Data in the manuscript were represented as means ± SEM (standard error of the mean) in all of our figures, which is a standard practice in the field (Masek et al., PNAS, 2010, Charlu et al., Nature Comm, 2013, Wang et al., Cell Metabolism, 2016). We have provided detailed explanations of the statistical tests in the manuscript. We have also prepared raw data files as suggested by the reviewer.

The model that global H3K27me3 is regulated by ancestral diet is unconvincing without further experimental validation and explanation. Points 4-10 address specific issues.

4) The authors performed ChIP on cycle 11 embryos. This stage is extremely short (11 min) and contains roughly 10 times less chromatin than embryos only 30 minutes older. These features make it very difficult to collect large numbers of precisely staged embryos without significant contamination. It is also debatable whether early cell cycles (including and preceding cycle 11) are slow enough to deposit and propagate histone marks in the presence of new histone incorporation. See the opposing arguments in Zenk et al 2017 and Li et al 2014. The authors could perform ChIP on older embryos to avoid this controversy.

We thank the reviewer for the clarification. Our embryo collection protocol involved allowing flies to lay eggs freely in a cage for 30 minutes followed by 50 minutes of incubation on a juice plate, and then completing the embryo sorting within 30 minutes. Therefore, to describe it in a more stringent way, our embryos should be in the stage between cycle 10-12. We have corrected this information in the manuscript (Figure 2. A).

Since all the embryos were sorted using the same morphological criteria within the same time frame, their developmental stages should be comparable (i.e. all from cycle 10-12). In several references we consulted, a broader range (cycle 9-13) was used for ChIP-seq sequencing analysis (for example, see Zenk et al., Science, 2017).

Surely any maternally inherited information will also be present in cycle 14 or 15 embryos if it is to influence the development or physiology of the brain. The observed differences in global H3K27me3 levels in F1 vs ND flies could be explained by slightly different aged embryo collections or technical variations in the ChIP protocol. The authors could strengthen their conclusion by performing more ChIP replicates. Alternatively, the authors could use orthogonal approaches like antibody staining or western blots to measure global H3K27me3 levels in precisely staged embryos.

We chose to use cycle 10-12 embryos because we aimed to identify epigenetic modulations directly transmitted through the maternal germline. Embryos in cycle 14-15 might reveal more profound changes, but since embryos in that stage had entered the zygotic phase and started the remodeling of histone modifications, we think it might mask the maternally transmitted changes we sought to identify.

In addition, we conducted two biological replicates for each group for the ChIP-seq analysis, which was a standard in the field (Zenk et al., Nature, 2021, Ing-Simmon et al., Nature Genetics, 2021). In the current study we further verified the genes identified in the ChIP-seq analysis in RNA-seq and qPCR analysis.

We further verified the ChIP-seq results by using western blot, which showed a ~2 folds increase in H3K27me3 modification in HSD-F1 early embryos vs. ND-fed embryos, in line with the ChIP-seq data (Figure 2-figure supplement 1. B). We have also provided immunofluorescence results for embryos at cycle 13 and cycle 14, which clearly showed a significant increase in H3K27me3 modifications in HSD-F1 embryos (Figure 2-figure supplement 1. C).

5) The authors measure PRC2 subunit mRNA levels in adult fly heads to attempt to explain the observed differences in inherited H3K27me3 levels in fly embryos. The authors should examine PRC2 components in germ cells and early embryos to understand how germ cells and early embryos generate H3K27me3 patterns.

We have now shown that Pcl and E(z) mRNA expression in HSD-F0 flies were not significantly changed vs. ND-fed flies (Figure 2-figure supplement 2. D-G). Meanwhile, H3K27me3 demethylase UTX and H3K27ac acetyltransferase Cbp showed significant decrease (Figure 2-figure supplement 2. H). Therefore, HSD exposure imposed complex epigenetic modifications in HSD-F0 flies, which then led to transmission of epigenetic marks to their progeny. Given the main scope of this study was to understand which epigenetic program mediated the behavioral transgenerational inheritance upon ancestral HSD exposure (but not that mediated acute HSD exposure), we focused our effect on H3K27me3 which was significantly changed between HSD-F1 flies vs. ND-fed flies.

6) The RNAi experiment targeting PRC2 components in embryos is uninterpretable without appropriate controls and an explanation of the genotypes used in the experimental paradigm. Are the authors crossing nosNGT mothers to UAS-RNAi fathers and assaying the progeny? What is the genotype of the F1 flies and how does it compare to the genotype of the ND flies? The authors should also note that the Gal4 drivers they use are not necessarily restricted to the ovary, and could directly affect other tissues controlling PER like neurons and muscle. Additionally, the authors should supply the appropriate controls to verify that their experimental paradigm has the intended effect. PRC2 proteins are presumably loaded into embryos and would be immune to zygotic-expressed RNAi. The authors could validate when PRC2 RNAi is effective by staining embryos for H3K27me3.

We have now added schematic diagrams and detailed explanations in our revised manuscript to better explain the RNAi experiments (Figure 3-figure supplement 1. A). As shown in the diagram, we compared each RNAi treatment group to appropriate genetic controls. We have also noted in the manuscript that the GAL4 drivers we used were not restricted to the ovary.

We have now verified the effect of PRC2 knockdown to reduce H3K27me3 in female germline by both western blot and immunofluorescence staining (Figure 3. B-C).

7) Although the authors do not note this, nosNGT>RNAi affects the PER of ND flies (compare Gal4>RNAi to just RNAi or just Gal4 in ND columns in Fig3A-D). This could be due to RNAi expression in neurons or muscles or some other indirect effect. Regardless of the mechanism, this result makes it difficult to interpret how RNAi treatments affect the transgenerational inheritance of PER if there is an equivalently strong nontransgenerational effect.

Although nosNGT>RNAi appeared to slightly affect PER response of ND-fed flies, there was no statistically significant difference (Figure 3-figure supplement 1. B and D, Figure 3-figure supplement 2. A-B). Rather, the effect of E(z) knockdown was evident in HSD-F1 flies (Figure 3-figure supplement 1. B), further confirming the involvement of H3K27me3 in transgenerational inheritance of PER reduction.

8) The matalpha gal4 experiment is inadequately explained in the text or methods. Are the authors expressing RNAi in the ovaries of the F0 flies that are fed an HSD? Does the ovary influence their PER somehow? Similar to point 8, there appears to be a nontransgenerational component to the RNAi phenotype that clouds the interpretation of the transgenerational effect (compare F0 in S3.1A-C).

We have now added a schematic diagram and detailed explanations in our revised manuscript to better explain the RNAi experiments (Figure 3. A). As shown in the diagram, we compared each RNAi treatment group to appropriate genetic controls.

Similar to point 7, although Mat-tub-GAL4>RNAi might seem to affect PER responses of ND-fed flies, there was no statistically significant difference (Figure 3. D-E). Rather, the effect of E(z) knockdown was evident in HSD-F1 flies (Figure 3. D), further confirming the involvement of H3K27me3 in transgenerational inheritance of PER reduction.

9) For the EED inhibitor experiments (both PER and calcium imaging), it is unclear whether the authors fed the mothers or their adult progeny the EED inhibitor. If adult progeny were fed, what tissues were affected? The authors should stain various tissues with an H3K27me3 antibody to verify the effectiveness of their inhibitor. Finally, the effect of the EED inhibitor on calcium imaging was not convincing because the variation was so large.

We have added a new schematic diagram and provided more detailed explanations in the manuscript for pharmacological interventions (Figure 4. A-D). To verify the effect of the drug treatment, we showed that compared to the control group fed with DMSO, flies fed with the inhibitor showed a significant decrease in H3K27me3 levels, demonstrating the effectiveness of the inhibitor (Figure 4-figure supplement 1. A).

We acknowledged the unsatisfactory quality of our calcium imaging experiments in our initial submission. We have now improved our experimental procedures to reach better data quality, while the conclusions remained consistent (Figure 4. E).

10) In all of the PRC2 RNAi and inhibitor experiments, are there any other phenotypes that would suggest that the treatments are working? There are many published PRC2 loss-offunction phenotypes (molecular and developmental) in different tissues. The authors could assure the reader that their treatments are working as expected by doing these controls.

As discussed above, we have now used western blot and immunofluorescence staining to validate the efficiency of PRC2 RNAi in female germline (Figure 3. B-C).

11) The authors propose that a transgenerationally inherited state of the caudal gene is responsible for the transgenerationally inherited PER. However, the experiments investigating the methylation state and expression level of caudal are unconvincing. Cad mRNA abundance varied immensely in the ND RNAseq samples. When the authors compared cad levels across generations, the effect size was small. A single outlier in the ND sample in both the RNAseq and the RTPCR experiments appears to drive up its mean and effect size. The H3K27me3 ChIP on cad is very similar in the F1 and ND samples and the acetylation peak on its promoter appears unchanged. The authors could vastly improve the caudal experiments in this paper by simply using cad antibodies to stain the relevant tissues that contribute to PER. For example, the authors could stain GR5a neurons for cad expression in different generations that inherit (or don't inherit) maternal PER to more accurately determine if cad levels are indeed transgenerationally regulated. The authors could also perform more ChIP experiments at a less variable stage to convincingly correlate epigenetic marks on cad with its expression level.

As discussed above, we conducted two biological replicates for each condition of the ChIP-seq analysis, which was a standard in the field (Zenk et al., Nature, 2021, IngSimmon et al., Nature Genetics, 2021). We have also performed western blot and immunofluorescence for H3K27me3 in ND vs. HSD-F1 embryos to further validate our ChIP-seq data (Figure 2-figure supplement 1. B-C).

As for Cad gene, H3K27m3 signals showed a statistically significant difference between ND-fed and HSD-F1 flies (Figure 5. D). We have also conducted additional qPCR experiments to verify the gene expression changes of the Cad gene (Figure 5. F, right), which was in line with the ChIP-seq data and further supported its validity.

It was worth noting that during the developmental time window of our ChIP-seq analysis, the acetylation signals in the promoter region of cad were very low (Figure 5. D), making it impossible to make a comparison.

Reviewer #3 (Public Review):

Jie Yang et al. investigated the transgenerational behavioral modification of a high-sugar diet (HSD) in Drosophila and revealed the underlying molecular and neural mechanisms. It has been reported that HSD exposure decreases sweet sensitivity in gustatory sensory neurons, resulting in reduced sugar response (Proboscis extension reflex, PER) in flies. The current study reports that this effect can be transmitted across generations through the maternal germline. Furthermore, the authors show that H3K27me3 modification is enhanced in the first-generation progenies of HSD-treated flies (F1), and genetical or pharmacological disruption of PCL-PRC2 complex blocks the behavioral change and restores the sweet sensitivity in the Gr5a+ sweet sensory neurons. The authors further analyze the differentially expressed genes in the F1 flies. Among H3K27me3 hypermethylated regions, they focus on homeobox genes and find a transcription factor Caudal (Cad), which shows decreased expression in the F1 flies. Knocking down Cad in Gr5a+ neurons results in decreased PER response to sucrose.

Transgenerational changes in physiology and metabolism have been broadly studied, while inherited changes at the behavioral level are much less investigated. This work provides convincing evidence for transgenerational modification of feeding behavior and digs out the underlying molecular and neural mechanisms. However, there still are several concerns that need to be clarified.

1) The epigenetic regulator PCR2 has been found to play an essential role in the 7d-HSDinduced modification of the PER response. In this study, it's important to clarify for the transgenerational change, whether epigenetic modification is required in the flies exposed to HSD (F0), the progenies (F1), or both. It would be very helpful for better interpretation if the procedures of HSD treatment in RNAi experiments and the drug treatments were stated in more detail. In addition, the F0 flies should be examined as the control.

In this current study our main scope was to understand the transgenerational influence of HSD exposure on the progeny. To this aim, we chose to study the physiological and behavioral differences between ND-fed flies vs. HSD-F1 flies (and their progeny on ND). HSD-F1 flies (and their progeny) were not exposed to HSD in their whole life cycle and therefore the physiological and behavioral changes we observed vs. ND-fed flies could be solely attributed to epigenetic modifications transmitted via germline cells from HSD-F0 flies. Therefore ND-fed flies were used as the main control.

As for HSD-F0 flies, the acute effects of HSD exposure could be more complex. Epigenetic factor was likely involved, as evident in Figure 3-figure supplement 1. C, Figure 3-figure supplement 3. A-B and Figure 4. C. In addition, HSD exposure might also directly affect gene expression and multiple signaling pathways in HSD-F0 flies (see Chen et al., Science China Life Sciences, 2020). Therefore, we did not aim to investigate how HSD exposure affected HSD-F0 flies in this current study. We have added additional discussions in the manuscript for clarification.

With that saying, we still added more HSD-F0 flies as controls when needed (Figure 2-figure supplement 2. D-G, Figure 3-figure supplement 1. C, Figure 4. C, Figure 5. F， left).

We have also modified the schematic diagrams and added more detailed explanations in the manuscript, in order to provide a clearer illustration of the experimental procedures (Figure 3. A, Figure 3-figure supplement 1. A, Figure 4. A, B and D). Specifically, we employed two different RNAi approaches. Firstly, we used genetic methods to obtain homozygous Mat-tub-gal4>UAS-gene X RNAi fly lines on chromosomes Ⅱ and Ⅲ for germline-specific knockdown (Figure 3, Figure 3-figure supplement 3). Secondly, we used heterozygous nosNGT-gal4>UAS-gene X RNAi flies for embryo-specific knockdown (Figure 3-figure supplement 1 and 2). Our drug experiments involved both treating the flies and measuring their PER (Figure 4. A-C) and treating the parental flies and measuring the PER of their progeny (Figure 4. D).

2) The information on the drug treatment period is also missing for imaging experiments (Fig.4C). Moreover, the response curve is very different from those recorded in the same neurons in previous studies. What’s the reason? Please also provide a representative image showing which part of the Gr5a neurons is recorded.

The experimental procedures of drug treatments were shown in Figure 4. A now. We fed adult flies with specific compounds for five days after eclosion, then measuring the calcium signals of Gr5a+ neurons when flies were fed with sucrose.

As suggested by the reviewer, we have now conducted calcium imaging experiments more carefully and thoroughly. We have now added the new data into the revised manuscript and the conclusions remained consistent (Figure 4. E). We recorded the calcium signal in the axons of Gr5a+ neurons in the SEZ.

3) It's unclear whether the decreased Cad expression upon HSD treatment specifically occurred in Gr5a+ neurons or a lot of cells. If the change in gene expression is significant in the qPCR test, it should occur in a large number of cells, most likely including different types of gustatory sensory neurons. If lower cad expression led to lower neural response and thereby lower behavioral response, how to specifically decrease the PER response to sucrose but not to other tastes? -whether HSD-induced desensitization is specific to sucrose in the offspring?

We agree that Cad expression might decrease in a lot of cells including Gr5a+ neurons in the proboscis. In order to investigate whether taste perception other than sweet sensing was also affected, we conducted PER experiments with fatty acids, which was another type of appetitive taste cues like sugars. Perception of fatty acids is mediated by ionotropic receptors such as ir25a, ir76b, and ir56b (Ahn, et al., eLife, 2017, Brown., et al, eLife, 2021).

Our results indicate that PER of fatty acid in HSD-F0 and HSD-F1 was not significantly reduced compared to the ND-fed controls (Figure 1-figure supplement 2. E-F). This suggests that the impact of Cad on gustatory sensory neurons might be specific to sweet sensitivity of Gr5a+ neurons.

4) In Fig.2D, data are sorted for genomic regions showing an up-regulated modification of H3K27me. It's unclear whether similar sorting was performed in panel C. This needs to be clarified.

The analysis shown in Figure 2C and 2D were linked. As for 2C, we identified genomic loci with enriched H3K27me3, H3K9me3, and H3K27ac peaks, and found that H3K27me3 peaks showed the most robust changes between ND-fed and HSD-F1 flies. Therefore we concentrated on these loci where H3K27me3 modifications were significantly changed between the two groups, and further analyzed their difference. As shown in Figure 2D, within these loci, H3K27ac modifications, which was functionally antagonizing to H3K27me3, were significantly reduced; whereas H3K9me3 signals within these loci remained unchanged. Such results confirmed that ancestral HSD exposure induced robust H3K27me3 modifications in certain genomic loci.

**Reviewer 2. Jianxin Wang **

In this manuscript, the authors present MuLan-Methyl, a deep-learning framework for predicting 6mA, 4mC, and 5hmC sites. They use DNA sequence and taxonomic identity as features, and implement five popular transformer-based language models in MuLan-Methyl. MuLan-Methyl is open-sourced, and a web server is also provided for users to access it. Overall, I think the methodology of MuLan-Methyl is clear and innovative, and the experiments seem comprehensive. However, I do have several concerns that I believe should be addressed before the paper is accepted by GigaScience.

Major 1. One major concern is that, in my opinion, DNA methylation is dynamic. Cytosines in the same position of the DNA sequence may have different methylation status in different samples, different cells, or even in different development stages of a cell. So, how can we predict the methylation status of a site based on only its sequence (and taxonomic identity)? -- The authors should clarify that in what cases, MuLan-Methyl (as well as other methods that use only DNA sequence) can be used to study DNA methylation, in Introduction or Discussion section. -- The authors discuss motifs in Fig. 3, but only for positive samples. How about the motif distribution in the negative samples? Can I understand that this method is actually for discovering motifs (or sequence structures) that are highly correlated with methylation? -- How is the performance of MuLan-Methyl without taxonomic identity? 2. The authors compared MuLan-Methyl against iDNA-ABF and iDNA-ABT, especially on the independent test set (Fig. 2E). I think the authors should clarify that whether they trained the models of the three methods using the same training datasets. If not, the authors should clarify the reason. 3. I'm curious about the computational efficiency of MuLan-Methyl. How many parameters in its model? Does MuLan-Methyl have advantages over other methods in terms of computational efficiency?

Minor 1. I don't understand why the references were not ordered from 1 in the main text. 2. I suggest that the authors re-organize the Introduction section. There are too many small paragraphs in this section. 3. At the end of Page 2, "The type 4mC type is present in 4 species" should be corrected.

Re-review:

The authors have addressed most of my concerns. However, I still have one minor concern about the computational efficiency. The response of the authors is not convincing by only saying "The number of models that MuLan-Methyl need to train and test on is less than the others, thus it has better computational efficiency than other models to some extent". If possible, I strongly suggest that the authors show some data to compare how much time and resources (GPU/CPU/RAM) needed by each method. The authors have addressed most of my concerns. However, I still have one minor concern about the computational efficiency. The response of the authors is not convincing by only saying "The number of models that MuLan-Methyl need to train and test on is less than the others, thus it has better computational efficiency than other models to some extent". If possible, I strongly suggest that the authors show some data to compare how much time and resources (GPU/CPU/RAM) needed by each method.

Author Response

Reviewer #1 (Public Review):

This paper falls in a long tradition of studies on the costs of reproduction in birds and its contribution to understanding individual variation in life histories. Unfortunately, the meta-analyses only confirm what we know already, and the simulations based on the outcome of the meta-analysis have shortcomings that prevent the inferences on optimal clutch size, in contrast to the claims made in the paper.

There was no information that I could find on the effect sizes used in the meta-analyses other than a figure listing the species included. In fact, there is more information on studies that were not included. This made it impossible to evaluate the data-set. This is a serious omission, because it is not uncommon for there to be serious errors in meta-analysis data sets. Moreover, in the long run the main contribution of a meta-analysis is to build a data set that can be included in further studies.

It is disappointing that two referees comment on data availability, as we supplied a link to our full dataset and the code we used in Dryad with our submitted manuscript. We were also asked to supply our data during the review process and we again supplied a link to our dataset and code, along with a folder containing the data and code itself. We received confirmation that the reviewers had been given our data and code. We support open science and it was our intention that our dataset should be fully available to reviewers and readers. Our data and code are at https://doi.org/10.5061/dryad.q83bk3jnk.

The main finding of the meta-analysis of the brood size manipulation studies is that the survival costs of enlarging brood size are modest, as previously reported by Santos & Nakagawa on what I suspect to be mostly the same data set.

We disagree that the main finding of our paper is the small survival cost of manipulated brood size. The major finding of the paper, in our opinion, is that the effect sizes for experimental and observational studies are in opposite directions, therefore providing the first quantitative evidence to support the influential theoretical framework put forward by van Noordwijk and de Jong (1986), that individuals differ in their optimal clutch size and are constrained to reproducing at this level due to a trade-off with survival. We show that while the manipulation experiments have been widely accepted to be informative, they are not in fact an effective test of whether within-species variation in clutch size is the result of a trade-off between reproduction and survival.

The comment that we are reporting the same finding as Santos & Nakagawa (2012) is a misrepresentation of both that study and our own. Santos & Nakagawa found an effect of parental effort on survival only in males who had their clutch size increased – but no effect for males who had their clutch size reduced and no survival effect on females for either increasing or reducing parental effort. However, we found an overall reduction in survival for birds who had brood sizes manipulated to make them larger (for both sexes and mixed sex studies combined). In our supplementary information, we demonstrate the overall survival effect of a change in reproductive effort to be close to zero for males, negative (though non-significant) for females and significantly negative for mixed sexes (which are not included in the Santos & Nakagawa study).

The paper does a very poor job of critically discussing whether we should take this at face value or whether instead there may be short-comings in the general experimental approach. A major reason why survival cost estimates are barely significantly different from zero may well be that parents do not fully adjust their parental effort to the manipulated brood size, either because of time/energy constraints, because it is too costly and therefore not optimal, or because parents do not register increased offspring needs. Whatever the reason, as a consequence, there is usually a strong effect of brood size manipulation on offspring growth and thereby presumably their fitness prospects. In the simulations (Fig.4), the consequences of the survival costs of reproduction for optimal clutch size were investigated without considering brood size manipulation effects on the offspring. Effects on offspring are briefly acknowledged in the discussion, but otherwise ignored. Assuming that the survival costs of reproduction are indeed difficult to discern because the offspring bear the brunt of the increase in brood size, a simulation that ignores the latter effect is unlikely to yield any insight in optimal clutch size. It is not clear therefore what we learn from these calculations.

The reviewer’s comment is somewhat of a paradox. We take the best studied example of the trade-off between reproductive effort and parental survival, a key theme in life-history and the biology of ageing, and subject this to a meta-analysis. The reviewer suggests we should interpret our finding as if there must be something wrong with the method or studies we included, rather than maybe considering the original hypothesis could be false or inflated in importance. The reviewer’s inclination to question the premise of the data in favor of a held hypothesis we consider not necessarily the best scientific approach here. In many places in our manuscript do we question and address issues in the underlying data and interpretation (L101-105, L149-150, 182-185 and L229-233). Moreover, we make it clear that we focus on the trade-off between current reproductive effort and subsequent parental survival and we are aware that other trade-offs could counter-balance or explain our findings, discussed on L189-191 & L246-253. Note that it is also problematic, when you do not find the expected response, to search for an alternative that has not been measured. In the case here, with trade-offs, there are endless possiblilities of where a trade-off might be incurred between traits. We purposfully focus on the one well-studied and theorised trade-off. We clearly acknowledge though that when all possible trade-offs are taken into account a trade-off on the fitness level can occur and cite two famous studies (Daan et al., 1990 and Verhulst & Tinbergen 1991) that have done just that (L250-253).

So whilst, we agree with the reviewer that the offspring may incur costs themselves, rather than costs being incurred by the parents, the aim of our study was to test for a generalised trend across species in the survival costs of reproductive effort. It is unrealistic to suggest that incorporating offspring growth into our simulations would add insight, as a change in offspring number rarely affects all offspring in the nest equally and there can even be quite stark differences; for example this will be most evident in species that produce sacrificial offspring. This effect will be further confounded by catch-up growth, for example, and so it is likely that increased sibling competition from added chicks alters offspring growth trajectories, rather than absolute growth as the reviewer suggests. There are mixed results in the literature on the effect of altering clutch size on offspring survival, with an increased clutch size through manipulation often increasing the number of recruits from a nest.

There are other reasons why brood size manipulations may not reveal the costs of reproduction animals would incur when opting for a larger brood size than they produced spontaneously themselves. Firstly, the manipulations do not affect the effort incurred in laying eggs (which also biases your comparison with natural variation in clutch size). Secondly, the studies by Boonekamp et al on Jackdaws found that while there was no effect of brood size manipulation on parental survival after one year of manipulation, there was a strong effect when the same individuals were manipulated in the same direction in multiple years. This could be taken to mean that costs are not immediate but delayed, explaining why single year manipulations generally show little effect on survival. It would also mean that most estimates of the fitness costs of manipulated brood size are not fit for purpose, because typically restricted to survival over a single year.

First, our results did show a survival cost of reproduction for brood manipulations. We agree that there could be longer-term costs, and so our estimate of the survival cost for manipulated birds is likely to be an underestimate, meaning that our interpretation still holds – the cost to reproduce prevents individuals from laying beyond their optimal level. Note, however, that much theory is build on the immediate costs of reproduction and as such these costs are likely overinterpreted.

We agree with the reviewer that lifetime manipulations could be even more informative than single-year manipulations. Unfortunately, there are currently too few studies available to be able to draw generalisable conclusions across species for lifetime manipulations. This is, however, the reason we used lifetime change in clutch size in our fitness projections, which the reviewer seems to have missed – please see methods line 360-362, where we explicitly state that this is lifetime enlargement. Of course such interpretations do not include an accumulation of costs that is greater than the annual cost, but currently there is no clear evidence that such an assumption is valid. Such a conclusion can also not be drawn from the study on jackdaws by Boonekamp et al (2014) as the treatments were life-long and, therefore, cannot separate annual from accrued (multiplicative) costs that are more than the sum of annual costs incurred.

Details of how the analyses were carried out were opaque in places, but as I understood the analysis of the brood size manipulation studies, manipulation was coded as a covariate, with negative values for brood size reductions and positive values for brood size enlargements (and then variably scaled or not to control brood or clutch size). This approach implicitly assumes that the trade-off between current brood size (manipulation) and parental survival is linear, which contrasts with the general expectation that this trade-off is not linear. This assumption reduces the value of the analysis, and contrasts with the approach of Santos & Nakagawa.

We thank the reviewer for highlighting a lack of clarity in places in our methods. We will add additional detail to this section in our revised manuscript.

For clarity in our response, each effect size was extracted by performing a logistic regression with survival as a binary response variable and clutch size was the absolute value of offspring in the nest (i.e., for a bird who laid a clutch size of 5 but was manipulated to have -1 egg, we used a clutch size value of 4). The clutch size was also standardised and, separately, expressed as a proportion of the species mean.

We disagree that our approach reduces the value of our analysis. First, our approach allows a direct comparison between experimental and observational studies, which is the novelty of our study. Our approach does differ from Santos & Nakagawa but we disagree that it contrasts. Our approach allows us to take into consideration the severity of the change in clutch size, which Santos & Nakagawa do not. Therefore, we do not agree that our approach is worse at accounting for non-linearity of trade-offs than the approach used by Santos & Nakagawa.

Our analysis, alongside a plethora of other ecological studies, does assume that the response to our predictor variable is linear. However, it is common knowledge that there are very few (if any) truly linear relationships. We use linear relationships because they serve a good approximation of the trend and provide a more rigorous test for an underlying relationship than would fitting nonlinear models. For many datasets there is not a range of chicks added for which a non-linear relationship could be estimated. The question also remains of what the shape of this non-linear relationship should be and is hard to determine a priori. We will address non-linear effects in our revised manuscript.

The observational study selection is not complete and apparently no attempt was made to make it complete. This is a missed opportunity - it would be interesting to learn more about interspecific variation in the association between natural variation in clutch size and parental survival.

We clearly state in our manuscript that we deliberately made a tailored selection of studies that matched the manipulation studies (L279-282). We paired species extracted for observational studies with those extracted in experimental studies to facilitate a direct comparison between observational and experimental studies, and to ensure that the respective datasets were comparable. The reviewer’s focus in this review seems to be solely on the experimental dataset. This comment dismisses the observational component of our analysis and thereby fails to acknowledge the question being addressed in this study.

Reviewer #2 (Public Review):

I have read with great interest the manuscript entitled "The optimal clutch size revisited: separating individual quality from the costs of reproduction" by LA Winder and colleagues. The paper consists in a meta-analysis comparing survival rates from studies providing clutch sizes of species that are unmanipulated and from studies where the clutch sizes are manipulated, in order to better understand the effects of differences in individual quality and of the costs of reproduction. I find the idea of the manuscript very interesting. However, I am not sure the methodology used allows to reach the conclusions provided by the authors (mainly that there is no cost of reproduction, and that the entire variation in clutch size among individuals of a population is driven by "individual quality").

We would like to highlight that we do not conclude that there is no cost of reproduction. Please see lines 258–260, where we state that our lack of evidence for trade-offs driving within-species variation in clutch size does not necessarily mean the costs of reproduction are non-existent. We conclude that individuals are constrained to their optima by the survival cost of reproduction. It is also an over-statement of our conclusion to say that we believe that variation in clutch size is only driven by quality. Our results show that unmanipulated birds who have larger clutch sizes also live longer, and we suggest this is evidence that some individuals are “better” than others, but we do not say, nor imply, that no other factors affect variation in clutch size.

I write that I am not sure, because in its current form, the manuscript does not contain a single equation, making it impossible to assess. It would need at least a set of mathematical descriptions for the statistical analysis and for the mechanistic model that the authors infer from it.

We appreciate this comment, but this is the first time we have been asked to put equations in a manuscript rather than explain them in terms that are accessible to a wider audience. Note however that our meta-analysis is standard and based on logistic regression and standard meta-analytic practices. We do not think we need to repeat such equations and we cite the relevant data. For the simulation, we simply simulated the resulting effects and this is not something that we feel is captured more accurately in equations rather than in text and the associated graphs. We of course supplied our code for this along with our manuscript (https://doi.org/10.5061/dryad.q83bk3jnk), though as we mentioned above, we believe this was not shared with the reviewers despite us making this available for the review process. We therefore understand the reviewer feels the simulations were not explained thoroughly. We will revise our text to see if we can add additional explanation where relevant in our revision.

The texts mixes concepts of individual vs population statistics, of within individual vs among-individuals measures, of allocation trade-offs and fitness trade-offs, etc ....which means it would also require a glossary of the definitions the authors use for these various terms, in order to be evaluated.

We would like to thank the reviewer for highlighting this lack of clarity in our text. We will simplify the terminology and define terms in our revised manuscript.

This problem is emphasised by the following sentence to be found in the discussion "The effect of birds having naturally larger clutches was significantly opposite to the result of increasing clutch size through brood manipulation". The "effect" is defined as the survival rate (see Fig 1). While it is relatively easy to intuitively understand what the "effect" is for the unmanipulated studies: the sensitivity of survival to clutch size at the population level, this should be mentioned and detailed in a formula. Moreover, the concept of effect size is not at all obvious for the manipulated ones (effect of the manipulation? or survival rate whatever the manipulation (then how could it measure a trade-off ?)? at the population level? at the individual level ?) despite a whole appendix dedicated to it. This absolutely needs to be described properly in the manuscript.

We would like to thank the reviewer for bringing to our attention the lack of clarity on the details of our methodology. We will make this more clear in our revised manuscript.

For clarity, the effect size for both manipulated and unmanipulated nests was survival, given the brood size raised. We performed a logistic regression with survival as a binary response variable (i.e., number of individuals that survived and number of individuals that died after each breeding season), and clutch size was the absolute value of offspring in the nest (i.e., for a bird who laid a clutch size of 5 but was manipulated to have -1 egg, we used a clutch size value of 4). This allows for direct comparison of the effect size (survival given clutch size raised) between manipulated and unmanipulated birds.

Despite the lack of information about the underlying mechanistic model tested and the statistical model used, my impression is still that the interpretation in the introduction and discussion is not granted by the outputs of the figures and tables. Let's use a model similar to that of (van Noordwijk and de Jong, 1986): imagine that the mechanism at the population level is

a.c_(i,q)+b.s_(i,q)=E_q

Where c_(i,q) are s_(i,q) are respectively the clutch size for individual i which is of quality q, and E_q is the level of "energy" that an individual of quality q has available during the given time-step (and a and b are constants turning the clutch size and survival rate into energy cost of reproduction and energy cost of survival, and there are both quite "high" so that an extra egg (c_(i,q) is increased by 1) at the current time-step, decreases s_(i,q) markedly (E_q is independent of the number of eggs produced), that is, we have strong individual costs of reproduction). Imagine now that the variance of c_(i,q) (when the population is not manipulated) among individuals of the same quality group, is very small (and therefore the variance of s_(i,q) is very small also) and that the expectation of both are proportional to E_q. Then, in the unmanipulated population, the variance in clutch size is mainly due to the variance in quality. And therefore, the larger the clutch size c_(i,q) the higher E_q, and the higher the survival s_(i,q).

In the manipulated populations however, because of the large a and b, an artificial increase in clutch size, for a given E_q, will lead to a lower survival s_(i,q). And the "effect size" at the population level may vary according to a,b and the variances mentioned above. In other words, the costs of reproduction may be strong, but be hidden by the data, when there is variance in quality; however there are actually strong costs of reproduction (so strong actually that they are deterministic and that the probability to survive is a direct function of the number of eggs produced)

We would like to thank the reviewer for these comments. Please note that our simulations only take the experimental effect of brood size on parental survival into account. Our model does not incorporate quality effects. The reviewer is right that the relationship between quality and the effects exposed by manipulating brood size can take many forms and this is a very interesting topic, but not one we aimed to tackle in our manuscript. In terms of quality we make two points: 1) overall quality effects connecting reproduction and parental survival are present 2) these effects are opposite in direction to the effects when reproduction is manipulated and similar in magnitude. We do not go further than that in interpreting our results. The reviewer is right however that we do suggest and repeat suggestions by others that quality can also mask the trade-off in some individuals or circumstances (L63-65, L85-88 & L237-240), but we do not quantify this as this is dependent on the unknown relationships between quality and the response to the manipulation. A focussed set of experiments in that context would be interesting and there is some data that could get at this, i.e. the relationship between produced clutch size and the relative effect of the manipulation. Such information is however not available for all studies and although we explored also analyzing this, currently this is not possible to do with sufficient confidence. We will include this rationale in our revision.

Moreover, it seems to me that the costs of reproduction are a concept closely related to generation time. Looking beyond the individual allocative (and other individual components of the trade-off) cost of reproduction and towards a populational negative relationship between survival and reproduction, we have to consider the intra-population slow fast continuum (some types of individuals survive more and reproduce less (are slower) than other (which are faster)). This continuum is associated with a metric: the generation time. Some individuals will produce more eggs and survive less in a given time-period because this time-period corresponds to a higher ratio of their generation time (Gaillard and Yoccoz, 2003; Gaillard et al., 2005). It seems therefore important to me, to control for generation time and in general to account for the time-step used for each population studied when analysing costs of reproduction. The data used in this manuscript is not just clutch size and survival rates, but clutch size per year (or another time step) and annual (or other) survival rates.

The reviewer is right that this is interesting. There has been unexplained difference in temperate (seasonal) and tropical reproduction strategies. Most of our data come from seasonal breeders however. Although there is some variation in second brooding and such often these species only produce one brood. We do agree that a wider consideration here is relevant, but we are not trying to explain all of life-history in our paper. It is clearly the case that other factors will operate and the opportunity for trade-offs will vary among species according to their respective life histories. However, our study focuses on the two most fundamental components of fitness – longevity and reproduction – to test a major hypothesis in the field, and we uncover new relationships that contrast with previous influential studies, and cast doubt on previous conclusions. We question the assumed trade-off between reproduction and annual survival. We show quality is important and that the effect we find in experimental studies, is so small that it can only explain between-species patterns but is unlikely to be the selective force that constrains reproduction within-species. We do agree that there is a lot more work that can be done in this area. We hope we contribute to this, by questioning this central trade-off. We will try and incorporate some of these suggestions in the revision where possible.

Finally, it is important to relate any study of the costs of reproduction in a context of individual heterogeneity (in quality for instance), to the general problem of the detection of effects of individual differences on survival (see, e.g., Fay et al., 2021). Without an understanding of the very particular statistical behaviour of survival, associated to an event that by definition occurs only once per life history trajectory (by contrast to many other traits, even demographic, where the corresponding event (production of eggs for reproduction, for example) can be measured several times for a given individual during its life history trajectory).

Thank you for raising this point. The reviewer is right that heterogeneity can dampen or augment selection. Note that by estimating the effect of quality here we give an example of how heterogeneity can possibly do exactly this. We thank the reviewer for raising that we should possibly link this to wider effects of heterogeneity and we aim to do so in the revision.

References:

Fay, R. et al. (2021) 'Quantifying fixed individual heterogeneity in demographic parameters: Performance of correlated random effects for Bernoulli variables', Methods in Ecology and Evolution, 2021(August), pp. 1-14. doi: 10.1111/2041-210x.13728.

Gaillard, J.-M. et al. (2005) 'Generation time: a reliable metric to measure life-history variation among mammalian populations.', The American naturalist, 166(1), pp. 119-123; discussion 124-128. doi: 10.1086/430330.

Gaillard, J.-M. and Yoccoz, N. G. (2003) 'Temporal Variation in Survival of Mammals: a Case of Environmental Canalization?', Ecology, 84(12), pp. 3294-3306. doi: 10.1890/02-0409.

van Noordwijk, A. J. and de Jong, G. (1986) 'Acquisition and Allocation of Resources: Their Influence on Variation in Life History Tactics', American Naturalist, p. 137. doi: 10.1086/284547.

Reviewer #3 (Public Review):

The authors present here a comparative meta-analysis analysis designed to detect evidence for a reproduction/ survival trade-off, central to expectations from life history theory. They present variation in clutch size within species as an observation in conflict with expectations of optimisation of clutch size and suggest that this may be accounted for from weak selection on clutch size. The results of their analyses support this explanation - they found little evidence of a reproduction - survival trade-off across birds. They extrapolated from this result to show in a mathematical model that the fitness consequences of enlarged clutch sizes would only be expected to have a significant effect on fitness in extreme cases, outside of normal species' clutch size ranges. Given the centrality of the reproduction-survival trade-off, the authors suggest that this result should encourage us to take a more cautious approach to applying concepts the trade-off in life history theory and optimisation in behavioural ecology more generally. While many of the findings are interesting, I don't think the argument for a major re-think of life history theory and the role of trade-offs in fitness maximisation is justified.

The interest of the paper, for me, comes from highlighting the complexities of the link between clutch size and fitness, and the challenges facing biologists who want to detect evidence for life history trade-offs. Their results highlight apparently contradictory results from observational and experimental studies on the reproduction-survival trade-off and show that species with smaller clutch sizes are under stronger selection to limit clutch size.

Unfortunately, the authors interpret the failure to detect a life history trade-off as evidence that there isn't one. The construction of a mathematical model based on this interpretation serves to give this possible conclusion perhaps more weight than is merited on the basis of the results, of this necessarily quite simple, meta-analysis. There are several potential complicating factors that could explain the lack of detection of a trade-off in these studies, which are mentioned and dismissed as unimportant (lines 248-250) without any helpful, rigorous discussion. I list below just a selection of complexities which perhaps deserve more careful consideration by the authors to help readers understand the implications of their results:

We would like to thank the reviewer for their thoughtful response and summary of the findings we also agree are central to our study. The reviewer also highlights areas where our manuscript could benefit from a deeper discussion and we will add detail to our discussion in our revised manuscript.

We would like to highlight that we do not interpret the failure to detect a trade-off as evidence that there isn’t one. First, and importantly, we do find a trade-off but show this is only incurred when individuals lay beyond their optimal level. Secondly, we also state on lines 258-260 that the lack of evidence to support trade-offs being strong enough to drive variation in clutch size does not necessarily mean there are no costs of reproduction.

The statement that we have constructed a mathematical model based on the interpretation that we have not found a trade-off is, again, factually incorrect. We ran these simulations because the opposite is true – we did find a trade-off. There is a significant effect of clutch size when manipulated on annual parental survival. To appreciate whether this effect alone can explain why reproduction is constrained, we ran the simulations. From these simulations we find that this effect size is too small to explain the constraint so something else must be going on and we do spend a considerable amount of text discussing the possible explanations (L182-194). Note the possibly most parsimonious conclusion here is that costs of reproduction are not there so we also give that explanation some thought (L201-205 and L247-253).

We are disappointed by the suggestion that we have dismissed complicating factors which could prevent detection of a trade-off, as this was not our intention. We were aiming to highlight that what we have demonstrated to be an apparent trade-off can be explained through other mechanisms, and that the trade-off between clutch size and survival is not as strong in driving within-species variation in clutch size as previously assumed. We will add further discussion to our revised manuscript to make this clear and give readers a better understanding of the complexity of factors associated with life-history theory. Although we do feel we have addressed this (L248-255).

• Reproductive output is optimised for lifetime reproductive success and so the consequences of being pushed off the optimum for one breeding attempt are not necessarily detectable in survival but in future reproductive success (and, therefore, lifetime reproductive success).

We agree this is a valid point, which is mentioned in our manuscript in terms of alternative stages where the costs of reproduction might be manifested (L248-250). We would also like to highlight that in our simulations, the change in clutch size (and subsequent survival cost) was assumed for the lifetime of the individual, for this very reason.

• The analyses include some species that hatch broods simultaneously and some that hatch sequentially (although this information is not explicitly provided (see below)). This is potentially relevant because species which have been favoured by selection to set up a size asymmetry among their broods often don't even try to raise their whole broods but only feed the biggest chicks until they are sated; any added chicks face a high probability of starvation. The first point this observation raises is that the expectation of more chicks= more cost, doesn't hold for all species. The second more general point is that the very existence of the sequential hatching strategy to produce size asymmetry in a brood is very difficult to explain if you reject the notion of a trade-off.

We agree with the reviewer that the costs of reproduction can be absorbed by the offspring themselves, and may not be equal across offspring (we also highlight this at L249 in the manuscript). However, we disagree that for some species the addition of more chicks does not equate to an increase in cost, though we do accept this might be less for some species. This is, however, difficult to incorporate into a sensible model as the impacts will vary among species and some species do also exhibit catch-up growth. So without a priori knowledge on this we kept our model simple. To test whether the effect on parental survival (often assumed to be a strong cost) can explain the constraint on reproductive effort, and we conclude it does not.

We would also like to make clear that we are not rejecting the notion of a trade-off. Our study shows evidence that a trade-off between survival and reproductive effort likely does not drive within-species variation in clutch size. We do explicitly say this throughout our manuscript, and also provide suggestions of other areas where a trade-off may exist (L246-250). The point of our study is not whether trade-offs exist or not, it is whether there is a generalisable across-species trend for a trade-off between reproductive effort and survival – the most fundamental trade-off in our field but for which there is a lack of conclusive evidence within species.

• For your standard, pair-breeding passerine, there is an expectation that costs of raising chicks will increase linearly with clutch size. Each chick requires X feeding visits to reach the required fledge weight. But this is not the case for species which lay precocious chicks which are relatively independent and able to feed themselves straight after hatching - so again the relationship of care and survival is unlikely to be detectable by looking at the effect of clutch size but again, it doesn't mean there isn't a trade-off between breeding and survival.

Precocial birds still provide a level of parental care, such as protection from predators. Though we agree that the level of parental care in provisioning food (and in some cases in all parental care given) is lower in precocial than altricial birds, this would only make our reported effect size for manipulated birds to be an underestimate. Again, we would like to draw the reviewer’s attention to the fact we did detect a trade-off in manipulated birds and we do not suggest that trade-offs do not exist. The argument the reviewer suggests here does not hold for unmanipulated birds, as we found that birds that naturally lay larger clutch sizes have higher survival.

• The costs of raising a brood to adulthood for your standard pair-breeding passerine is bound to be extreme, simply by dint of the energy expenditure required. In fact, it was shown that the basal metabolic rate of breeding passerines was at the very edge of what is physiologically possible, the human equivalent being cycling the Tour de France (Nagy et al. 1990). If birds are at the very edge of what is physiologically possible, is it likely that clutch size is under weak selection?

If birds are at the very edge of what is physiologically possible, then indeed it would necessarily follow that if they increase the resource allocated in one area then expenditure in another area must be reduced. In many studies however, the overall brood mass is increased when chicks are added and cared for in an experimental setting, suggesting that birds are not operating at their limit all the time. Our simulations show that if individuals increase their clutch size, the survival cost of reproduction counterbalances the fitness gained by increasing clutch size and so there is no overall fitness gain to producing more offspring. Therefore, selection on clutch size is constrained to the within-species level. We do not say in our manuscript that clutch size is under weak selection – we only ask why variation in clutch size is maintained if selection always favours high-producing birds.

• Variation in clutch size is presented by the authors as inconsistent with the assumption that birds are under selection to lay the Lack clutch. Of course, this is absurd and makes me think that I have misunderstood the authors' intended point here. At any rate, the paper would benefit from more clarity about how variable clutch size has to be before it becomes a problem for optimality in the authors' view (lines 84-85; line 246). See Perrins (1965) for an exquisite example of how beautifully great tits optimise clutch size on average, despite laying between 5-12 eggs.

We woud like to thank the reviewer for highlighting that our manuscript may be misleading in places, however, we are unsure which part of our conclusions the author is referring to here.The question we pose is “why all birds don’t lay at the population optimum?”, and is central to the decades-long field of life-history theory. Why is variation maintained at such a level? As the reviewer outlines it ranges massively with some birds laying half of what other birds lay.

I completely agree with the following quote, I feel like this relates to the education system a lot for 60+ years when POC started to receive education from schools. I believe there have been positive changes resulting in them attending school. But I also believe that the education system in the US negatively continues to fail them since POC are the minorities in America and before were the poorest people on the planet. And still to his day the education system fails to help under privileged students succeed in these social institutions like in schools. Especially because I feel like the education system has not been changed for years and is outdated and only very recently there has been few changes to change it. But I think we need to reform the policies to help under privileged students attending schools in America by first acknowledging insutions failed POC. And secondly, reform the information. And lastly, create more welcome in groups at schools to help them succeed socially and many other things as well to help students in the future.

If I am understanding this right - I think the potential dilemma that arises from professional versus local archaeology is interesting. Local archeology efforts could provide insight into the past that would've gone unresearched otherwise, but with lower budgets and potentially greater mistakes (due to it not being 'professional quality' (?)) could harm the items in the dig site. Professional archaeology affords research in key places, motivated by economics, politics, or culture, and allows for the use of advanced techniques such as carbon dating. Unfortunately, this may remove the discovery that many of the locals of the area would've likely enjoyed performing (since it is their roots). How are we to weigh preservation, quality, and ethics together to form the idea of 'just' archeology?

• for: memex, Vannevar Bush, Indyweb, Ted Nelson

In the health and wellness world, a fascinating trend has emerged with the rise of hydrogen infused water machine. These innovative devices promise to deliver a refreshing beverage beyond ordinary hydration – hydrogen-rich water. Packed with potential health benefits, the science behind these machines is captivating and sheds new light on how we think about water consumption and its impact on our well-being.

Hydrogen: The Unsung Hero Of Molecules

Before delving into the science of hydrogen-rich water machines, it's essential to understand the pivotal role of hydrogen itself. Hydrogen is the lightest and simplest element on the periodic table, consisting of a single proton and an electron. While hydrogen is generally known for its explosive nature, it has recently garnered attention for its potential health benefits when dissolved in water.

The Power Of Hydrogen-Infused Water

Hydrogen-infused water, often called hydrogen-rich water, is created when molecular hydrogen gas (H2) is dissolved into plain water. This process typically involves using advanced technologies found in hydrogen-rich water machines. The resulting beverage is touted for its potential antioxidant properties, which could contribute to various health improvements.

Antioxidant Action: Hydrogen's Hidden Potential

Antioxidants are essential for neutralizing dangerous chemicals known as free radicals, which may damage cells and contribute to a variety of health problems such as chronic illnesses and ageing. Molecular hydrogen is thought to have antioxidant characteristics that are more effective than well-known antioxidants such as vitamins C and E.

Hydrogen's unique antioxidant potential lies in its ability to easily penetrate cell membranes and access cellular compartments, including the nucleus and mitochondria. This attribute gives hydrogen an edge in protecting cellular components from oxidative stress, potentially reducing the risk of oxidative damage.

The Mechanism: How Hydrogen Works Its Magic

The exact mechanism behind hydrogen's antioxidant effects is still an area of ongoing research, but several theories have been proposed. One prominent theory suggests that hydrogen is a selective scavenger of harmful free radicals, targeting the most reactive and damaging ones without affecting beneficial molecules like oxygen or nitric oxide.

Another theory is that hydrogen has the power to modify signalling pathways within cells. By altering these pathways, hydrogen may elicit preventive responses that boost the body's natural defence systems against oxidative stress and inflammation.

Hydrogen-Rich Water Machines: The Technology

Hydrogen-rich water machines are designed to harness the power of molecular hydrogen by infusing it into plain drinking water. These devices commonly use electrolysis, which involves sending an electric current through water to divide it into hydrogen and oxygen gases. The hydrogen gas is subsequently dissolved in water, yielding a beverage high in this beneficial chemical.

These machines are equipped with advanced membranes that allow only hydrogen molecules to pass through while preventing the escape of potentially harmful byproducts like ozone. This ensures the purity and safety of the resulting hydrogen-infused water.

Potential Health Benefits

While research on the health benefits of hydrogen-rich water is still in its infancy, preliminary studies have shown promising results. Some of the potential benefits include the following:

Antioxidant Defense: Hydrogen-rich water's antioxidant properties could help reduce oxidative stress and associated health risks. Anti-Inflammatory Effects: Hydrogen may have anti-inflammatory effects that could benefit conditions like arthritis and other inflammatory disorders. Cellular Health: Hydrogen might contribute to overall cellular health and function by protecting cellular components. Exercise Performance: Some research suggests that hydrogen-rich water might enhance exercise performance and reduce muscle fatigue. Conclusion: A Glimpse Into The Future Of Hydration

Hydrogen-rich water machines are ushering in a new era of hydration, where molecular hydrogen's benefits are harnessed to enhance our well-being potentially. While more research is needed to understand the extent of these benefits fully, the early findings are exciting and have sparked interest among health-conscious individuals.

As technology advances, we can anticipate more refined hydrogen-infused water machines and a deeper understanding of how molecular hydrogen interacts with our bodies. Whether you're an early adopter or a cautious observer, the science behind these machines invites us to explore the intriguing potential of hydrogen-infused water and its impact on our health.

Reviewer #1 (Public Review):

Summary:<br /> This paper by Schommartz and colleagues investigates the neural basis of memory reinstatement as a function of both how recently the memory was formed (recent, remote) and its development (children, young adults). The core question is whether memory consolidation processes as well as the specificity of memory reinstatement differ with development. A number of brain regions showed a greater activation difference for recent vs. remote memories at the long versus shorter delay specifically in adults (cerebellum, parahippocampal gyrus, LOC). A different set showed decreases in the same comparison, but only in children (precuneus, RSC). The authors also used neural pattern similarity analysis to characterize reinstatement, though I have substantive concerns about how this analysis was performed and as such will not summarize the results. Broadly, the behavioural and univariate findings are consistent with the idea that memory consolidation differs between children and adults in important ways, and takes a step towards characterizing how.

Strengths:<br /> The topic and goals of this paper are very interesting. As the authors note, there is little work on memory consolidation over development, and as such this will be an important data point in helping us begin to understand these important differences. The sample size is great, particularly given this is an onerous, multi-day experiment; the authors are to be commended for that. The task design is also generally well controlled, for example as the authors include new recently learned pairs during each session.

Weaknesses:<br /> As noted above, the pattern similarity analysis for both item and category-level reinstatement was performed in a way that is not interpretable given concerns about temporal autocorrelation within the scanning run. Below, I focus my review on this analytic issue, though I also outline additional concerns.

1. The pattern similarity analyses were not done correctly, rendering the results uninterpretable (assuming my understanding of the authors' approach is correct).

a. First, the scene-specific reinstatement index: The authors have correlated a neural pattern during a fixation cross (delay period) with a neural pattern associated with viewing a scene as their measure of reinstatement. The main issue with this is that these events always occurred back-to-back in time. As such, the two patterns will be similar due simply to the temporal autocorrelation in the BOLD signal. Because of the issues with temporal autocorrelation within the scanning run, it is always recommended to perform such correlations only across different runs. In this case, the authors always correlated patterns extracted from the same run, which moreover have temporal lags that are perfectly confounded with their comparison of interest (i.e., from Fig 4A, the "scene-specific" comparisons will always be back-to-back, having a very short temporal lag; "set-based" comparisons will be dispersed across the run, and therefore have a much higher lag). The authors' within-run correlation approach also yields correlation values that are extremely high - much higher than would be expected if this analysis was done appropriately. The way to fix this would be to restrict the analysis to only cross-run comparisons, but I don't believe this is possible unfortunately given the authors' design; I believe the target (presumably reinstated) scene only appears once during scanning, so there is no separate neural pattern during the presentation of this picture that they can use. For these reasons, any evidence for "significant scene-specific reinstatement" and the like is completely uninterpretable and would need to be removed from the paper.

b. From a theoretical standpoint, I believe the way this analysis was performed considering the fixation and the immediately following scene also means that the differences between recent and remote could have to do with either the reactivation (processes happening during the fixation, presumably) or differences in the processing of the stimulus itself (happening during the scene presentation). For example, people might be more engaged with the more novel scenes (recent) and therefore process those scenes more; such a difference would be interpreted in this analysis as having to do with reinstatement, but in fact could be just related to the differential scene processing/recognition, etc. It would be important when comparing scene-specific neural patterns as templates for reinstatement across conditions that, at the time of scene presentation itself, the two conditions are equal (e.g., no difference in familiarity and so on); otherwise, we do not know which trial period (and therefore which underlying process) is driving the differences.

c. For the category-based neural reinstatement: (1) This suffers from the same issue of correlations being performed within the run. Again, to correct this the authors would need to restrict comparisons to only across runs (i.e., patterns from run 1 correlated with patterns for run 2 and so on). With this restriction, it may or may not be possible to perform this analysis, depending upon how the same-category scenes are distributed across runs. However, there are other issues with this analysis, as well. (2) This analysis uses a different approach of comparing fixations to one another, rather than fixations to scenes. The authors do not motivate the reason for this switch. Please provide reasoning as to why fixation-fixation is more appropriate than fixation-scene similarity for category-level reinstatement, particularly given the opposite was used for item-level reinstatement. Even if the analyses were done properly, it would remain hard to compare them given this difference in approach. (3) I believe the fixation cross with itself is included in the "within category" score. Is this not a single neural pattern correlated with itself, which will yield maximal similarity (pearson r=1) or minimal dissimilarity (1-pearson r=0)? Including these comparisons in the averages for the within-category score will inflate the difference between the "within-category" and "between-category" comparisons. These (e.g., forest1-forest1) should not be included in the within-category comparisons considered; rather, they should be excluded, so the fixations are always different but sometimes the comparisons are two retrievals of the same scene type (forest1-forest2), and other times different scene types (forest1-field1). (4) It is troubling that the results from the category reinstatement metric do not seem to conceptually align with past work; for example, a lot of work has shown category-level reinstatement in adults. Here the authors do not show any category-level reinstatement in adults (yet they do in children), which generally seems extremely unexpected given past work and I would guess has to do with the operationalization of the metric.

2. I did not see any compelling statistical evidence for the claim of less robust consolidation in children. Specifically in terms of the behavioural results of retention of the remote items at 1 vs 14 days, shown in Figure 2B, the authors conclude that memory consolidation is less robust in children (line 246). Yet they do not report statistical evidence for this point, as there was no interaction of this effect with the age group. Children had worse memory than adults overall (in terms of a main effect - i.e. across recent and remote items). If it were consolidation-specific, one would expect that the age differences are bigger for the remote items, and perhaps even most exaggerated for the 14-day-old memories. Yet this does not appear to be the case based on the data the authors report. Therefore, the behavioural differences in retention do not seem to be consolidation specific, and therefore might have more to do with differences in encoding fidelity or retrieval processes more generally across the groups. This should be taken into account when interpreting the findings.

3. Please clarify which analyses were restricted to correct retrievals only. The univariate analyses states that correct and incorrect trials were modelled separately, but does not say which were considered in the main contrast (I assume correct only?). The item specific reinstatement analysis states that only correct trials were considered, but the category-level reinstatement analysis does not say. Please include this detail.

4. To what extent could performance differences be impacting the differences observed across age groups? I think (see prior comment) that the analyses were probably limited to correct trials, which is helpful, but still yields pretty big differences across groups in terms of the amount of data going into each analysis. In general, children showed more attenuated neural effects (e.g., recent/remote or session effects); could this be explained by their weaker memory? Specifically, if only correct trials are considered that means that fewer trials would be going into the analysis for kids, especially for the 14-day remote memories, and perhaps pushing the remove > recent difference for this condition towards 0. The authors might be able to address this analytically; for example, does the remote > recent difference in the univariate data at day 14 correlate with day 14 memory?

5. Some of the univariate results reporting is a bit strange, as they are relying upon differences between retrieval of 1- vs. 14-day memories in terms of the recent vs. report difference, and yet don't report whether the regions are differently active for recent and remote retrieval. For example in Figure 3A, neither anterior nor posterior hippocampus seem to be differentially active for recent vs. remote memories for either age group (i.e., all data is around 0). This difference from zero or lack thereof seems important to the message - is that correct? If so, can the authors incorporate descriptions of these findings?

6. Please provide more details about the choices available for locations in the 3AFC task. (1) Were they different each time, or always the same? If they are always the same, could this be a motor or stimulus/response learning task? (2) Do the options in the 3AFC always come from the same area - in which case the participant is given a clue as to the gist of the location/memory? Or are they sometimes randomly scattered across the image (in which case gist memory, like at a delay, would be sufficient for picking the right option)? Please clarify these points and discuss the logic/impact of these choices on the interpretation of the results.

7. Often p values are provided but test statistics, effect sizes, etc. are not - please include this information. It is at times hard to tell whether the authors are reporting main effects, interactions, pairwise comparisons, etc.

8. There are not enough methodological details in the main paper to make sense of the results. For example, it is not clear from reading the text that there are new object-location pairs learned each day.

9. The retrieval task does not seem to require retrieval of the scene itself, and as such it would be helpful for the authors to both explain their reasoning for this task to measure reinstatement. Strictly speaking, participants could just remember the location of the object on the screen. Was it verified that children and adults were recalling the actual scene rather than just the location (e.g. via self-report)? It's possible that there may be developmental differences in the tendency to reinstate the scene depending on e.g., their strategy.

10. In general I found the Introduction a bit difficult to follow. Below are a few specific questions I had.

a. At points findings are presented but the broader picture or take-home point is not expressed directly. For example, lines 112-127, these findings can all be conceptualized within many theories of consolidation, and yet those overarching frameworks are not directly discussed (e.g., that memory traces go from being more reliant on the hippocampus to more on the neocortex). Making these connections directly would likely be helpful for many readers.

b. Lines 143-153 - The comparison of the Tompary & Davachi (2017) paper with the Oedekoven et al. (2017) reads like the two analyses are directly comparable, but the authors were looking at different things. The Tompary paper is looking at organization (not reinstatement); while the Oedekoven et al. paper is measuring reinstatement (not organization). The authors should clarify how to reconcile these findings.

c. Line 195-6: I was confused by the prediction of "stable involvement of HC over time" given the work reviewed in the Introduction that HC contribution to memory tends to decrease with consolidation. Please clarify or rephrase.

d. Lines 200-202: I was a bit confused about this prediction. Firstly, please clarify whether immediate reinstatement has been characterized in this way for kids versus adults. Secondly, don't adults retain gist more over long delays (with specific information getting lost), at least behaviourally? This prediction seems to go against that; please clarify.

Author Response

We thank the reviewers for their work and their thoughtfulness. However, it seems to us that much (but not all) of the critique reflects a misunderstanding of the goals and methods of computational modeling. Details are below. We are grateful for the opportunity to include our views about this in the context of our replies to the Public Critiques of our paper. The comments of the reviewers were very helpful in allowing us to see what might not be clear to our readers.

eLife assessment

This useful modeling study explores how the biophysical properties of interneuron subtypes in the basolateral amygdala enable them to produce nested oscillations whose interactions facilitate functions such as spike-timing-dependent plasticity. The strength of evidence is currently viewed as incomplete because the relevance to plasticity induced by fear conditioning is viewed as insufficiently grounded in existing training protocols and prior experimental results, and alternative explanations are not sufficiently considered. This work will be of interest to investigators studying circuit mechanisms of fear conditioning as well as rhythms in the basolateral amygdala.

Most of our comments below are intended to rebut the sentence: “The strength of evidence is currently viewed as incomplete because the relevance to plasticity induced by fear conditioning is viewed as insufficiently grounded in existing training protocols and prior experimental results, and alternative explanations are not sufficiently considered”. Details are below in the answer to reviewers.

We believe this work will be interesting to investigators interested in dynamics associated with plasticity, which goes beyond fear learning. It will also be of interest because of its emphasis on the interactions of multiple kinds of interneurons that produce dynamics used in plasticity, in the cortex (which has similar interneurons) as well as BLA.

We note that the model has sufficiently detailed physiology to make many predictions that can be tested experimentally. In the revision, we will be more explicit about this.

We thank Reviewer #1 for stressing our work's important contribution to providing concrete hypotheses that can be tested in vivo and highlighting the importance of examining in the future the synergistic role of the interneurons in the BLA in fear learning in the BLA. The weaknesses reported by the Reviewer concern deviations of the model compared to the experimental literature. We describe below why we think those differences are minor in the context of the aims of our model. Specifically,

1) Some connections among neurons in the BLA reported by (Krabbe et al., 2019) have not been taken into account in the model. Some connections between cell types were excluded without adequate justification (e.g. SOM+ to PV+).

In order to constrain our model, we focused on what is reported in (Krabbe et al., 2019) in terms of functional connectivity instead of structural connectivity. Thus, we included only those connections for which there was strong functional connectivity. For example, the SOM+ to PV+ connection is shown to be small (Supp. Fig. 4, panel t). We also omitted PV+ to SOM+, PV+ to VIP+, SOM+ to VIP+, VIP+ to excitatory projection neurons; all of these are shown in (Krabbe et al. 2019, Fig. 3 (panel l), and Supp. Fig. 4 (panels m,t)) to have weak functional connectivity, at least in the context of fear conditioning. See below for comments on modeling strategies. We will explain this better in our revision.

2) The construction of the afferent drive to the network does not reflect the stimulus presentations that are given in fear conditioning tasks. For instance, the authors only used a single training trial, the conditioning stimulus was tonic instead of pulsed, the unconditioned stimulus duration was artificially extended in time, and its delivery overlapped with the neutral stimulus, instead of following its offset. These deviations undercut the applicability of their findings.

Regarding the use of a single long presentation of US rather than multiple presentations (i.e., multiple trials): in early versions of this paper, we did indeed use multiple presentations. We were told by experimental colleagues that the learning could be achieved in a single trial. We note that, if there are multiple presentations in our modeling, nothing changes; once the association between CS and US is learned, the conductance of the synapse is stable. Also, our model does not need a long period of US if there are multiple presentations. This point will be made clearer in our revision.

We agree that, in order to implement the fear conditioning paradigm in our in-silico network, we made several assumptions about the nature of the CS and US inputs affecting the neurons in the BLA and the duration of these inputs. A Poisson spike train to the BLA is a signal that contains no structure that could influence the timing of the BLA output; hence, we used this as our CS input signal. We also note that the CS input can be of many forms in general fear conditioning (e.g., tone, light, odor), and we wished to de-emphasize the specific nature of the CS. The reference mentioned in the Recommendations for authors, (Quirk, Armony, and LeDoux 1997), uses pulses 2 seconds long. At the end of fear conditioning, the response to those pulses is brief. However, in the early stages of conditioning, the response goes on for as long as the figure shows. The authors do show the number of cells responding decreases from early to late training, which perhaps reflects increasing specificity over training. This feature is not currently in our model, but we look forward to thinking about how it might be incorporated. Regarding the CS pulsed protocol used in (Krabbe et al., 2019), it has been shown that intense inputs (6kHz and 12 kHz inputs) can lead to metabotropic effects that last much longer than the actual input (200 ms duration) (Whittington et al., Nature, 1995). Thus, the effective input to the BLA may indeed be more like Poisson.

Our model requires the effect of the CS and US inputs on the BLA neuron activity to overlap in time in order to instantiate fear learning. Despite paradigms involving both overlapping (delay conditioning, where US coterminates with CS (Lindquist et al., 2004), or immediately follows CS (e.g., Krabbe et al., 2019)) and non-overlapping (trace conditioning) CS/US inputs existing in the literature, we hypothesized that concomitant activity in CS- and US-encoding neuron activity should be crucial in both cases. This may be mediated by the memory effect, as suggested in the Discussion of our paper, or by metabotropic effects as suggested above, or by the contribution from other brain regions. We will emphasize in our revision that the overlap in time, however instantiated, is a hypothesis of our model. It is hard to see how plasticity can occur without some memory trace of US. This is a consequence of our larger hypothesis that fear learning uses spike-timing-dependent plasticity; such a hypothesis about plasticity is common in the modeling literature. We will discuss these points in more detail in our revision.

We thank Reviewer #2 for their comments. Below, we reply to each of them:

1) Gamma oscillations are generated locally; thus, it is appropriate to model in any cortical structure. However, the generation of theta rhythms is based on the interplay of many brain areas therefore local circuits may not be sufficient to model these oscillations. Moreover, to generate the classical theta, a laminal structure arrangement is needed (where neurons form layers like in the hippocampus and cortex)(Buzsaki, 2002), which is clearly not present in the BLA. To date, I am not aware of any study which has demonstrated that theta is generated in the BLA. All studies that recorded theta in the BLA performed the recordings referenced to a ground electrode far away from the BLA, an approach that can easily pick up volume conducted theta rhythm generated e.g., in the hippocampus or other layered cortical structure. To clarify whether theta rhythm can be generated locally, one should have conducted recordings referenced to a local channel (see Lalla et al., 2017 eNeuro). In summary, at present, there is no evidence that theta can be generated locally within the BLA. Though, there can be BLA neurons, firing of which shows theta rhythmicity, e.g., driven by hippocampal afferents at theta rhythm, this does not mean that theta rhythm per se can be generated within the BLA as the structure of the BLA does not support generation of rhythmic current dipoles. This questions the rationale of using theta as a proxy for BLA network function which does not necessarily reflect the population activity of local principal neurons in contrast to that seen in the hippocampus.

In both modeling and experiments, a laminar structure does not seem to be needed to produce a theta rhythm. A recent experimental paper, (Antonoudiou et al. 2021), suggests that the BLA can intrinsically generate theta oscillations (3-12 Hz) detectable by LFP recordings under certain conditions, such as reduced inhibitory tone. The authors draw this conclusion by looking at mice ex vivo slices. The currents that generate these rhythms are in the BLA, since the hippocampus was removed to eliminate hippocampal volume conduction and other nearby brain structures did not display any oscillatory activity. Also, in the modeling literature, there are multiple examples of the production of theta rhythms in small networks not involving layers; these papers explain the mechanisms producing theta from non-laminated structures (Dudman et al., 2009, Kispersky et al., 2010, Chartove et al. 2020). We are not aware of any model description of the mechanisms of theta that do require layers.

2) The authors distinguished low and high theta. This may be misleading, as the low theta they refer to is basically a respiratory-driven rhythm typically present during an attentive state (Karalis and Sirota, 2022; Bagur et al., 2021, etc.). Thus, it would be more appropriate to use breathing-driven oscillations instead of low theta. Again, this rhythm is not generated by the BLA circuits, but by volume conducted into this region. Yet, the firing of BLA neurons can still be entrained by this oscillation. I think it is important to emphasize the difference.

Many rhythms of the nervous system can be generated in multiple parts of the brain by multiple mechanisms. We do not dispute that low theta appears in the context of respiration; however, this does not mean that other rhythms with the same frequencies are driven by respiration. Indeed, in the above answer we showed that theta can appear in the BLA without inputs from other regions. In our paper, the low theta is generated in the BLA by VIP+ neurons. Using intrinsic currents known to exist in VIP+ neurons (Porter et al., 1998), modeling has shown that such neurons can intrinsically produce a low theta rhythm. This is also shown in the current paper. This example is part of a substantial literature showing that there are multiple mechanisms for any given frequency band. We will emphasize these points in our revision; we note that, for any individual case, such as this one, the mechanism needs to be tested experimentally.

3) The authors implemented three interneuron types in their model, ignoring a large fraction of GABAergic cells present in the BLA (Vereczki et al., 2021). Recently, the microcircuit organization of the BLA has been more thoroughly uncovered, including connectivity details for PV+ interneurons, firing features of neurochemically identified interneurons (instead of mRNA expression-based identification, Sosulina et al., 2010), synaptic properties between distinct interneuron types as well as principal cells and interneurons using paired recordings. These recent findings would be vital to incorporate into the model instead of using results obtained in the hippocampus and neocortex. I am not sure that a realistic model can be achieved by excluding many interneuron types.

The interneurons and connectivity that we used were inspired by the functional connectivity reported in (Krabbe et al., 2019) (see above answer to Reviewer #1). As reported in (Vereczki et al., 2021), there are multiple categories and subcategories of interneurons; that paper does not report on which ones are essential for fear conditioning. We did use all the highly represented categories of the interneurons, except NPY-containing neurogliaform cells.

The Reviewer says “I am not sure that a realistic model can be achieved by excluding many interneuron types”. We agree with the Reviewer that discarding the introduction of other interneurons subtypes and the description of more specific connectivity (soma-, dendrite-, and axon-targeting connections) may limit the ability of our model to describe all the details in the BLA. However, this work represents a first effort towards a biophysically detailed description of the BLA rhythms and their function. As in any modeling approach, assumptions about what to describe and test are determined by the scientific question; details postulated to be less relevant are omitted to obtain clarity. The interneuron subtypes we modeled, especially VIP+ and PV+, have been reported to have a crucial role in fear conditioning (Krabbe et al., 2019). Other interneurons, e.g. cholecystokinin and SOM+, have been suggested as essential in fear extinction. Thus, in the follow-up of this work to explain fear extinction, we will introduce other cell types and connectivity. In the current work, we have achieved our goals of explaining the origin of the experimentally found rhythms and their roles in the production of plasticity underlying fear learning. Of course, a more detailed model may reveal flaws in this explanation, but this is science that has not been yet done.

4) The authors set the reversal potential of GABA-A receptor-mediated currents to -80 mV. What was the rationale for choosing this value? The reversal potential of IPSCs has been found to be -54 mV in fast-spiking (i.e., parvalbumin) interneurons and around -72 mV in principal cells (Martina et al., 2001, Veres et al., 2017).

A GABA-A reversal potential around -80 mV is common in the modeling literature (Jensen et al., 2005; Traub et al., 2005; Kumar et al., 2011; Chartove et al., 2020). Other computational works of the amygdala, e.g. (Kim et al., 2016), consider GABA-A reversal potential at -75 mV based on the cortex (Durstewitz et al., 2000). The papers cited by the reviewer have a GABA-A reversal potential of -72 mV for synapses onto pyramidal cells; this is sufficiently close to our model that it is not likely to make a difference. For synapses onto PV+ cells, the papers cited by the reviewer suggest that the GABA-A reversal potential is -54 mV; such a reversal potential would lead these synapses to be excitatory instead of inhibitory. However, it is known (Krabbe et al., 2019; Supp. Fig. 4b) that such synapses are in fact inhibitory. Thus, we wonder if the measurements of Martina and Veres were made in a condition very different from that of Krabbe. For all these reasons, we consider a GABA-A reversal potential around -80 mV in amygdala to be a reasonable assumption. We will discuss these points in our revision.

5) Proposing neuropeptide VIP as a key factor for learning is interesting. Though, it is not clear why this peptide is more important in fear learning in comparison to SST and CCK, which are also abundant in the BLA and can effectively regulate the circuit operation in cortical areas.

We do not think that VIP is necessarily more fundamental in fear learning, and certainly not for fear extinction. We will make this clear in the revision.

We thank Reviewer #3 for their comments and for recognizing that we achieved our modeling aims. We reply to the criticisms below.

Weaknesses:

The main weakness of the approach is the lack of experimental data from the BLA to constrain the biophysical models. This forces the authors to use models based on other brain regions and leaves open the question of whether the model really faithfully represents the basolateral amygdala circuitry. Furthermore, the authors chose to use model neurons without a representation of the morphology. However, given that PV+ and SOM+ cells are known to preferentially target different parts of pyramidal cells and given that the model relies on a strong inhibition form SOM to silence pyramidal cells, the question arises whether SOM inhibition at the apical dendrite in a model representing pyramidal cell morphology would still be sufficient to provide enough inhibition to silence pyramidal firing. Lastly, the fear learning relies on the presentation of the unconditioned stimulus over a long period of time (40 seconds). The authors justify this long-lasting input as reflecting not only the stimulus itself but as a memory of the US that is present over this extended time period. However, the experimental evidence for this presented in the paper is only very weak.

Many of these issues were addressed in the previous responses.

1) Our neurons were constrained by electrophysiology properties in response to hyperpolarizing currents in the BLA (Sosulina et al., 2010). We choose the specific currents, known to be present in these neurons, to replicate those responses.

2) Though a much more detailed description of BLA interneurons was given in (Vereczki et al., 2021), it is not clear that this level of detail is relevant to the questions that we were asking, especially since the experiments described were not done in the context of fear learning.

3) It is true that we did not include the morphology, which undoubtedly makes a difference to some aspects of the circuit dynamics. As we described above, modeling requires the omission of many details to bring out the significance of other details.

4) As described above, some form of memory or overlap in the activity of the excitatory projection neurons is necessary for spike-timing-dependent plasticity. In modeling, one must be specific about hypotheses, and describe why they are plausible, if not proved; indeed, modeling can explain known phenomena by showing how they are consequences of some (plausible) hypotheses, which themselves are open to experimental verification.

5) The 40 seconds is not necessary if there are multiple presentations.

Other critiques:

1) It is correct that PV+ and SOM+ preferentially target different parts of excitatory projection neurons and that the model relies on a strong inhibition from SOM+ and PV+ to silence the excitatory projection neurons. This choice of parameters comes from using simplified models: it is standard in modeling to adjust parameters to compensate for simplifications.

2) The SOM+ inhibition of the pyramidal cell firing can be seen as a hypothesis of our model. It is well known that VIP+ cells disinhibit pyramidal cells through inhibition of SOM+ and PV+ cells, which is all we are using in our model; hence this hypothesis is generally believed.

The authors achieved the aim of constructing a biophysically detailed model of the BLA not only capable of fear learning but also showing spectral signatures seen in vivo. The presented results support the conclusions with the exception of a potential alternative circuit mechanism demonstrating fear learning based on a classical Hebbian (i.e. non-depression-dominated) plasticity rule, which would not require the intricate interplay between the inhibitory interneurons. This alternative circuit is mentioned but a more detailed comparison between it and the proposed circuitry is warranted.

We agree with the reviewer that it would be good to have a more detailed comparison with the classical Hebbian rule (non-depression-dominated rule). However, we demonstrated in Supplementary Materials that the non-depression-dominated rule is less robust and only operates within a limited window of PV+ excitation. We will have a more robust discussion of plasticity in the revision.

Author Response

The following is the authors’ response to the original reviews.

Public Review:

The authors report the first use of the bacterial Tus-Ter replication block system in human cells. A single plasmid containing two divergently oriented five-fold TerB repeats was integrated on chromosome 12 of MCF7 cells. ChIP and PLA experiments convincingly demonstrate the occupancy of Tus at the Ter sites in cells. Using an elegant Single Molecule Analysis of Replicated DNA (SMARD) assay, convincing data demonstrate the replication block at Ter sites dependent on the presence of the protein. As an orthogonal method to demonstrate fork stalling, ChIP data show the accumulation of the replicative helicase component MCM3 and the repair protein FANCM around the Ter sites. It is unclear whether the Ter sites integrated by a single copy plasmid have any effect on the replication of this region but the data show that the observed effects are dependent on expression of the Tus protein. The SMARD data do not reveal what proportion of forks are arrested at Tus/Ter, or how long the fork delay is imposed. Fork stalling led to a highly localized gammaH2AX response, as monitored by ChIP using primer pairs spread along the integrated plasmid carrying the Ter sites. This response was shown to be dependent on ATR using the ATR inhibitor VE-822. This contrasts with a single Cas9-induced DSB between the two Ter sites, which causes a more spread gammaH2AX response. While this was monitored only at a single distal site, the difference between the DSB and the Tus-induced stall is very significant. Interestingly, despite evidence for ATR activation through the gammaH2AX response, no evidence for phosphorylation of ATR-T1989, CHK1-S345, or RPA2-S33 could be found under fork stalling conditions. The global replication inhibitor hydroxyurea (HU) elicited phosphorylation of ATR-T1989, CHK1-S345, or RPA2-S33. In this context, it would have been of interest to examine if a single DSB in the Ter region leads to phosphorylation of ATR-T1989, CHK1-S345, or RPA2-S33 and cell cycle arrest. It is not shown whether the replication inhibitor HU leads to the same widely spread gamma H2AX response. Overall, this is a well written manuscript, and the data provide convincing evidence that the Tus-Ter system poses a site-specific replication fork block in MCF7 cells leading to a localized ATR-dependent DNA damage checkpoint response that is distinct from the more global response to HU or DSBs.

Author response to public review:

“It is unclear whether the Ter sites integrated by a single copy plasmid have any effect on the replication of this region but the data show that the observed effects are dependent on expression of the Tus protein.”

-The lack of perturbation of the TerB sequence on fork progression has extensively been studied previously in both Willis et al, 2014 and Larsen et. al, 2014. Furthermore, as the detection of the SMARD signal at the TerB sites is dependent on the 7.5kb probe that spans the TerB sites (orange probe, Fig 2B & 2D), it would be impossible to study the effect on replication in this region, with and without the integration of the single copy plasmid.

“The SMARD data do not reveal what proportion of forks are arrested at Tus/Ter, or how long the fork delay is imposed.”

-The percentage of fork stalling at the TerB sites, with and without Tus expression, has been quantified in Figure 2E & 2F. Essentially, 36% forks stall at the TerB block, i.e. 18% of the forks stall in both the 5’ to 3’ (orange) and 3’ to 5’ (blue) direction when the Tus-TerB block is active.

“It is not shown whether the replication inhibitor HU leads to the same widely spread gamma H2AX response.”

-While we have not shown gH2AX accumulation via ChIP after HU treatment, Supplementary Figure 5A & 5B clearly show increased gH2AX foci when the cells are treated with HU, suggesting a global replication stress response that is in stark contrast to the response to Tus-TerB.

Recommendations for the authors:

Lines 78, 95: In the experimental set-up there are two divergent 5-TerB sites in the orientation that is non-permissive for the fork progression notwithstanding the direction. This raises an obvious question: How an intervening (~1kb-long) DNA segment in being replicated? Does it stay under-replicated and then break?

-The reviewers pose an important question about how the intervening sequence flanked by the two TerB sites is replicated, and if this leads to formation of anaphase bridges resulting in breaks. We think this is very plausible and this very question is part of ongoing studies in the lab with the aim to understand how the cell resolves a site-specific block. Unfortunately, this falls outside the scope of the current study.

Also, it is unclear what is meant with non-permissive orientation. This depends on the predominant replication direction. As the construct has Ter repeats in opposite orientation, any direction is non-permissive. These descriptions could be rephrased to avoid confusion

-The text has been edited to clarify this.

Fig 1A: It would be helpful to annotate the map to show the position of each primer relative to the Ter array. Why is there no signal for pp52?

-Figure 1A has the map of the locus with the annotated primer pairs and their relative positions to the TerB array.

-pp52 is positioned beyond the TerB array so binding of the Tus-His protein there is unlikely, confirming the specificity of the Tus binding to only the TerB array and not to the adjacent chromatin.

Figure 1B: Change Tus to Tus-His to make it easier to understand that the anti-His ChIP is targeting Tus. Provide information what normalization method was used in the ChIP experiments.

-Figure 1B has been edited to reflect this change

Line 113: Willis et al. 2014 also worked with chromosomal Ter sites, which should be acknowledged here.

The text has been modified to indicate this. We apologize for the oversight.

Line 126: Define pWB15 and its significance in text.

-The text has been edited to clarify this and mentions pWB15.

Figure 2E, F: Define legend (blue, orange boxes and arrow heads).

-The figure legend corresponding to Figure 2 has a detailed description of the boxes and the arrows.

Figure 3E, 4C: Add map of primers like in Figures 1 and 2.

-The map added to Figures 3 & 4 and text updated.

Figure 4: Showing that the gammaH2AX response is spread like with the single DSB would bolster the conclusion about the difference between a local and global response. Fig 4A, Lane-3: A loading control for the chromatin fraction is missing.

-Measuring gH2AX chromatin spread after global replication stress can be challenging. We have tried to address the question of global and local gH2AX response post replication stress by quantifying gH2AX foci in cells treated with and without hydroxyurea, comparing it with cells that have a functional Tus-TerB block (Supplementary Figure 5A& 5B). A single fork block seems to only elicit a local response while a global replication stress leads to gH2AX accumulation globally in the cell.

-Lamin A/C has been added to Fig 4A as a loading control for the chromatin fraction.

Figure S4: Analyzing ATR, CHK1 and RPA phosphorylation as well as cell cycle profile under single DSB condition may reveal that different localized responses exist. I mention this because it was reported in yeast that a single DSB in G1 cells leads to a similarly localized Mec1 (ATR) -dependent response that does not elicit phosphorylation of Rad53 (CHK1) and other downstream targets, but leads to H2A phosphorylation as well as phosphorylation of RPA and the Rad51 paralog Rad55 (see PMCID: PMC2853130). It might be of interest to the reader to discuss this publication and the commonalities and differences between both localized checkpoint response

-The reviewers raise an interesting question about the phosphorylation of ATR/CHK1/RPA and its effect on cell cycle after a single DSB. The aim of using the Cas9 break site in this study was merely to corroborate previously published observations pertaining to the spread of gH2AX after a DSB and to contrast that with the local response seen with Tus-TerB. Thus, while an intriguing question, we do not think this particular experiment will help in the understanding of the localized checkpoint response after a single replication fork block. However, we have included the observations previous published in the yeast system (PMC2853130) in our discussion as it helps compare and contrast fork blocks and DSBs further. It is of worth though that the yeast studies were looking at the cellular response to a DSB in G1.

Lines 256-260: In the discussion of ATRIP, unpublished data are discussed that show no increase in ssDNA. What is the effect of ATRIP depletion? Maybe delete this mention of unpublished data, if no new data can be provided. The authors are aware that this makes the mechanism of ATR activation at the 5-TerB site elusive.

-This statement has been deleted and the text has been modified.

Another possibility discussed by the authors is fork reversal. Since Tus/Ter complex block the CMG progression, fork reversal would result in a chicken foot structure with the long single-stranded 3'-overhang of an Okazaki fragment site. Such a structure should be protected by BRCA2 or RAD52 proteins from degradation. Any role for these proteins in the checkpoint activation at the TerB site?

-The reviewers suggest an interesting scenario where the Tus-TerB block induced reversed fork structure could be protected by the loading of known DNA repair proteins and this in turn could lead to a signaling mechanism and checkpoint activation. While we have not tested this hypothesis, nor studied the temporal dynamics of the formation if the reversed fork with respect to gH2AX accumulation, we think the localized gH2AX signal observed in the vicinity of the block is what initiates the downstream DDR response, promoting fork stabilization, followed either by fork reversal and restart or fork collapse. If the reversed fork was responsible for the gH2AX signaling, one would envision the spread to be more widespread, perhaps decorating the entire stretch of DNA between the block and the reversed fork. However, further studies are warranted to tease out this mechanism and the spatio-temporal dynamics.

Lines 292-294: The authors state that "unpublished work from our laboratory has demonstrated that replication forks are cleaved at or near the TerB site..." Unless the data are shown, it might be best to eliminate discussion of unpublished work, also because the occurrence of DNA ends at Ter sites was already described in Willis et al. 2017.

-The statement has been deleted and Willis et al. 2017 has been referenced.

Suppl Table 1: It would help to also show representative images of stretched fibers in addition to the summary data shown.

-Since the data is negative, the fiber images do not show any discernible differences and we do not think it adds useful information.

Suppl Fig 4. ChIP for gamma H2AX data. It would be helpful to show the distribution of the gamma H2AX signal along the chromosome for both the DSB response and the Tus/Ter response.

-The gH2AX ChIP signal at PP0-2 and PP10 has been included in Supplementary Fig4D. Though not significant for PP0-2, the data strongly suggests that there is increased spread of gH2AX along the chromosome after a DSB, strongly contrasting with the response after Tus-TerB block. The text has been modified to include both primer pairs.

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

Learn more at Review Commons

---

Reply to the reviewers

Response to Reviewer comments

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

__Summary __

The manuscript by Parker et al addresses the important question of how different organisms have evolved pre-messenger RNA systems that are either more or less complex. This question underlies the evolution of complex organisms and the genome adaptation of simple organisms to their specific environments, so is an important question to answer. This manuscript now provides the underlying molecular mechanisms of how 5' splice site sequence preference may have evolved which is both an interesting and exciting advance for the field. 

We thank the reviewer for these kind comments.

__Major comments __

__This manuscript builds on the previous work from this group where they identified the role of adenosine N6 methylation (m6A) of the U6 small nuclear RNA (snRNA) of the spliceosome by METTL16 as being important for 5' splice site selection. This work led to the speculation that loss of a METTL16 ortholog, or potentially other splicing factors, in some species could contribute to an evolutionary change in 5' splice site sequence preference. Here the authors now use the power of phylogenetics, interspecies association mapping and the available spliceosome structures to provide convincing conclusions that 5' splice site sequence preferences in the extensive number of organisms examined correlate with the presence of the U6 snRNA methyltransferase METTL16 and the splicing factor SNRNP27K. __

__An analysis of METTL16 conservation was first carried out by comparing the METTL16 methyltransferase domain (MTD) in 29 diverse eukaryotic species. All the METTL16 orthologs were found to have either one or two globular domains. Three domain types were identified and compared in detail. What was not clear from this analysis was the functional significance of orthologs having either one or two domains. __

We identified several species, including Drosophila melanogaster, whose METTL16 orthologs do not contain a VCR domain. However, in this study we do not draw specific conclusions about the functional significance of orthologs having different domain topologies.

__In addition, while this analysis provides important new information on the domain structure of METTL16 orthologs, especially where these domains had not been identified previously, the link between this section of the results and the following sections is not that apparent. __

We agree that there is a significant difference in approach between the first section of the Results and the following sections. However, we are keen to keep this part of the manuscript because it provides an orthogonal line of evidence suggesting that the ancestral role of METTL16 in eukaryotes is specifically the methylation of U6 snRNA.

__Next novel bioinformatics pipelines were developed to compare both introns and orthologous groupings of protein coding genes between 227 Sacchromycotina genomes as well as 13 well-annotated eukaryote genomes. First, the 5' splice site sequence preference was compared and clearly indicates that the +4 position has the greatest variation in preferences within the Sacchromycotina. The ability to now compare a large number of genomes has provided novel information on the evolution of the 5' splice site sequence and the conclusion that there is more complexity to the 5' splice site in fungi that previously recognized. While it is apparent why only the 5' splice site signal was investigated here, with its relationship to the U6 snRNA and METTL16, it seems a shame the other splice site sequences were not analyzed using this novel pipeline. In any case, the complexity of the 5' splice site +4 position now allows, for the first time, interesting interspecies association studies. __

We have now included the variance plots for 3’SS motifs (analogous to the 5’SS variance plots shown in Figure 2B) as Figure 2 supplementary figure 4A, and a traitgram for 3’SS -3C to U ratio as Figure 2 supplementary figure 4B. We have included a short section of text in the Results section to describe these additional findings.

__With ____the 5' splice site +4 variation identified, the next step was to determine the underlying molecular mechanisms that dictate the evolution of the various sequence preferences. Some obvious players here are the U1 and U6 snRNAs which directly interact with the 5' splice site during splicing. However, no association was found between these snRNAs and the 5' splice site +4 sequence. __

__The powerful interspecies association mapping was then used to determine whether the presence or absence of METTL16 ortholog or a splicing factor correlated with the 5' splice site +4 sequence variation. Interestingly, a clear association was found between METTL16 and the 5' splice site +4 position; METTL16 presence was associated with +4A at the 5' splice site and METTL16 absence was associated with +4U at the 5' splice site. This is an exciting and significant finding. __

We thank the reviewer for these comments on the importance of this study.

__Interestingly, the next most significant association with the 5' splice site +4 position was with SNRNP27K. This result makes sense as in the cryo-EM structure of the pre-B spliceosome complex the C-terminal domain of SNRNP27K is found near the region of the U6 snRNA that will interact with the 5' splices site. Absence of SNRNP27K was associated with an increased preference for +4U at the 5' splice site. Now the real power of the interspecies association mapping was demonstrated by investigating whether any association could be determined specifically within the C-terminus of SNRNP27K. Significantly, the methionine 141 position in SNRNP27K was found to be associated with the +4 position of the 5' splice site. This finding fits nicely with previous studies where mutation of M141 caused a shift in 5' splice site selection away from +4A 5' splice sites, to 5' splice sites without +4A. What is not clear is whether M141 is conserved or invariant between all the species that were compared? __

M141 is not completely conserved across the species that were compared for the SNRNP27K C-terminus analysis. We did not test positions with very strong sequence conservation, because without variation in both the genotype and phenotype it is not possible to test for an association. We have rephrased the relevant Results and Methods sections to make this point clearer. In addition, we have incorporated a sequence logo to illustrate the degree of conservation of each position in the SNRNP27K C-terminal domain as Figure 5 -figure supplement 1A. Finally, we have included an additional box-plot to illustrate the finding that species which have lost SNRNP27K or have only lost the Methionine equivalent to human SNRNP27K position 141, show a similar preference for +4U at 5’ SSs. This is now included as Figure 5 - figure supplement 1B.

Overall, this result reveals the power of the interspecies association approach and provides interesting and exciting information on the molecular determinants of 5' splice site evolution. 

We are grateful to the reviewer for these comments.

__The final analysis was to investigate the interaction potentials of the U5 and U6 snRNAs with the 5' splice site in the Sacchromycotina genomes and try to relate this to species with fewer introns and less alternative splicing. Species with low intron numbers and low splicing complexity were revealed to have weaker U5 and U6 anti-correlation potentials and favor +4U at the 5' splice site. On the other hand, species with high intron number and presumably higher splicing complexity featured anti-correlated U5 and U6 snRNA interaction potentials and favored +4A 5' splice sites. This extensive analysis provides novel information on the interactions and splice site properties of species with simple and complex splicing. Again, I see why there is emphasis on the 5' splice site here but a similar analysis with the U2 snRNA and the branch site could also be informative. __

We absolutely agree that inter-species association mapping could be applied to other splicing signal phenotypes including 3’ splice sites and intron branchpoints. Accordingly, we raise this subject in the final section of the Discussion. However, branchpoint sequences are challenging to predict with genomic data. Because preliminary analyses suggest independent variation in these other splicing signal phenotypes, we feel a separate focused study is required to properly explain (and substantiate) even the analytical approaches involved. We hope the reviewer would agree that incorporating U2 snRNA and branchpoint variation analyses into this manuscript as well, could detract from the clarity of the conceptual advances that we make here.

__Minor comments __

__Should the Title include SNRNP27K? __

We have included SNRNP27K into the revised title.

Should the title specify that it is the evolution of only the 5’ splice site sequence preference being studied here? 

Because apostrophes in titles can compromise some scholarly online search engines (https://insights.uksg.org/articles/10.1629/uksg.534), we would prefer not to include 5’ in the title.

Include information on intron number and 5’ splice site interaction potential of U5 and U6 snRNA in the Summary? 

We thank the reviewer for this suggestion. We have updated the Summary to include our findings on U5 and U6 interaction potential in species with reduced intron number.

__Figure 1C is not referred to in the text? __

We apologise for this oversight. We have added references to figure 1C in the appropriate Results section.

Page 8, line 5 – better to say “splicing signal phenotypes”. 

We have amended this statement on Page 8 and at other places in the text where related phrasing was made.

__What are the other points on Figure 3B? What is the next point below SNRNP27K? Is it U2A’? __

The other points on Figure 3B represent Orthofinder orthogroups which contain human orthologs that are known components of the spliceosome. The list of spliceosomal components was taken from Sales-Lee et al. 2021. The third most significant point is indeed the orthogroup containing the human ortholog of U2A’. As we state in the text, however, the correlation of U2A’ with the 5’SS+4 A to U ratio phenotype is no longer significant once METTL16 presence/absence is controlled for, indicating that the correlation of U2A’ with the +4A phenotype is likely explained by similarity in the patterns of gene loss of U2A’ and METTL16.

__The second paragraph of the Discussion is vague and lacks a reference. “we could also identify an association with a methionine residue in the conserved C-terminal domain of SNRNP27K orthologs.” There are a few methionines in the C-terminus, which one? Please reference the statement “transcriptome analysis of C. elegans SNRP-27 M141T mutants..” __

We apologise for the lower quality of writing in this section of the Discussion. We have updated the text, made the statements about the SNRNP27K C-terminus less ambiguous, and added the relevant citations as appropriate.

Reviewer #1 (Significance (Required)): 

Overall, this is a well written and clearly presented study that provides some key molecular information on the splicing factors involved in the evolution of 5’ splice sites and shows the power of interspecies association studies. Some important conceptual principles have now been defined for the field going forward. 

With thank the reviewer for this kind comment on the importance of this work.

__The question remains as to whether METTL16 and SNRNP27K are the sole determinants of 5’ splice site preference evolution at +4? __

We cannot say for certain that METTL16 and/or SNRNP27K determine the 5’SS +4 phenotype – only that they are correlated with it. In our response to reviewer 3, and in a new Discussion section, we have detailed some of the scenarios that could explain these correlations. We also cannot rule out whether there are changes in the presence/absence (or domain/sequence-level changes) of other, untested proteins that correlate with the 5’SS +4 phenotype and we allude to this in the final section of the Discussion.

One splicing factor that immediately comes to mind is Prp8 where there is extensive evidence for involvement in splice site selection and is clearly in the right location throughout splicing to be involved. This question should at least be discussed but Prp8 would also be a very interesting candidate for the interspecies association mapping. 

Prp8 is a core component of spliceosomes and is conserved throughout the Saccharomycotina. For this reason, we were unable to associate splicing phenotypes with Prp8 presence or absence variation at the level of orthogroups. However, we revisited this question posed by the reviewer. Our experience with inter-species association mapping, so far, indicates it works well with orthogroup presence/absence or when straightforward amino acid substitutions can be detected in conserved and hence alignable protein sequence domains. We analysed the conserved U6 snRNA-interacting region of the Prp8 linker domain, which maps close to the 5’ splice site in cryo-EM models, using the profile HMM PF10596 available from Pfam. We found that the majority of this domain was extremely highly conserved with variation in only a few species and positions. The strongest correlation with the +4A to U ratio phenotype was at position 58, which is conserved as a Glycine in all but 8 species (6 Dipodascaceae, 2 CUG-Ser1), that also tend to have a stronger preference for +4A. However, examination of the species contributing to this result (and to similar results at other positions) indicated that in the 6 Dipodascaceae species, this change is part of a larger deletion or replacement that makes the whole linker region align poorly to the model. Hence, the G58 position itself may not be specifically important for the +4 phenotype. Although the wholesale loss or replacement of the U6 snRNA-interacting region in these species is potentially interesting, these larger scale structural changes in a small number of species are difficult to interpret. Therefore, to maintain the focus of the manuscript and the clear links to METTL16 and SNRNP27K that have orthogonal support, we have decided not to add these results to the manuscript but present them here (Figure not available on biorXiv commenting window).

Also, as mentioned previously, only the 5’ splice site was investigated here and the manuscript could become a more substantial piece of work if the other splice sites were included in some way. 

We agree that it will be exciting to apply this approach to other splicing signal phenotypes and in other phylogenetic clades with emerging tree-of-life-scale genomics data. We have included variation in 3’ splice sites in the revised manuscript. As the first of its kind, this study should pioneer a wider use of this approach, by us and others, to understand the mechanisms and functions of molecular interactions not only in splicing but in other areas of biology too.

__The obvious audience here are those directly in the splicing field but the overall principles are relevant for evolutionary biologists and those studying organismal complexity. __

We thank the reviewer for recognising the broad importance of this work.

My expertise is in yeast and human splicing mechanisms. I do not have the expertise to critically evaluate the bioinformatic pipelines but they were clearly explained and presented. 

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

In their manuscript, Parker et al. investigate the evolutionary patterns of splice site preference, focusing on the A/U ratio at position A+4 on the 5´ splice site. Building upon prior studies in S. pombe and A. thaliana, the authors establish a strong correlation between this preference and the co-evolution of the METTL16 U6 snRNA methyltransferase. Furthermore, through inter-species association mapping, they identify the involvement of the splicing factor SNRNP27K in altered A/U ratios and highlight the significance of the residue Met-141 in SNRNP27K for this function. Overall, the paper effectively presents impactful new findings on the evolution of METTL16, U6 snRNA, and splicing.

We thank the reviewer for these kind comments on the importance of our study.

The computational analyses employed in this study are situated outside our field of expertise, preventing us from offering a comprehensive evaluation of the methodology’s appropriateness and rigor. Nonetheless, the identification of METTL16 through the authors’ methods, which aligns with previous research in S. pombe and A. thaliana, lends support to the validity of their approach. Notably, the close proximity between SNRNP27K and the methylated A43 residue in U6 snRNA within the spliceosome, particularly near Met-141, is an impressive finding. Previous studies have shown that a mutation at position M141T affects splicing at +4A introns, thus providing robust validation for their methods.

We thank the reviewer for these kind comments on our work.

The data presented in this study furnish crucial insights into the role of METTL16, U6 snRNA methylation, and splice site recognition. The authors expand upon recent observations that the “vertebrate conserved region” exists in non-vertebrates, despite the absence of primary sequence homology. These results will serve as a valuable guide for future molecular investigations into U6 snRNA methylation and its mechanisms in splicing. Furthermore, the implications of this paper extend to human evolution, as the plasticity in splicing is an essential factor in the evolution of developmental complexity.

We thank the reviewer for these kind comments.

Minor suggestions for improvement:

1. __ Given the significance of the interaction between U6 snRNA and the intron for understanding the data, it would be beneficial to include a figure illustrating the RNA-RNA base-pairing interactions between U6 snRNA and the 5´ splice site. This addition is particularly important if the paper is intended for publication in a journal with a general readership.__  We thank the reviewer for this excellent suggestion. We have included this as Figure 3A.

__ Similarly, the section on U1 snRNA would be more comprehensible with the inclusion of U1 RNA-RNA intron diagrams and improved descriptions of both the figures and the assay. Despite being negative data in the supplement, clarifying this section is essential. As currently written, it is challenging to follow.__ 

We agree that this section is difficult to follow. We have updated the text to improve the readability and included a figure of U1 snRNA:5’SS basepairing as Figure 3 – figure supplement 1A.

__ Whenever possible, consider increasing the figure and font sizes to enhance readability for readers.__ 

We agree that some of the more complex figures can be difficult to read when embedded into a Word document/pdf. We hope that providing high-resolution figures for reading online will mitigate this.

__ In the text, there is no reference to Figure 1C.__ 

We apologise for this oversight. We have resolved this issue with the appropriate references in the Results text.

__ In Figure 5B, the y-axis in the top panel is labelled “species,” but the legend only mentions U5/6p as the y-axis. Please revise the legend to include the appropriate information.__ 

We apologise for the confusion caused by our poorly written legend for this plot. We have updated the legend so that the text clearly refers to either the scatter plot or the marginal histograms.

Reviewer #2 (Significance (Required)):

The data presented in this study furnish crucial insights into the role of METTL16, U6 snRNA methylation, and splice site recognition. The authors expand upon recent observations that the “vertebrate conserved region” exists in non-vertebrates, despite the absence of primary sequence homology. These results will serve as a valuable guide for future molecular investigations into U6 snRNA methylation and its mechanisms in splicing. Furthermore, the implications of this paper extend to human evolution, as the plasticity in splicing is an essential factor in the evolution of developmental complexity.

We are grateful to the reviewer for these kind comments on the importance of this work.

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

In this manuscript, Parker et al present a nice exploration of the evolutionary and mechanistic relationships between 5′ splice site consensus sequences, intron numbers and METTL16/SNRNP27K. By performing inter-species association mapping in Saccharomycotina species, they found that a T in position +4 is strongly associated with the absence of METTL16 (and/or in some cases SNRNP27K or mutations in it). They also provide solid structural modelling data in support of this association.

In general, I think this is a very nice manuscript. I only have a few comments, which could be addressed by rewording specific parts and/or improving the current figures.

We are grateful to the reviewer for the kind comments on this work.

1) As the authors acknowledge, a key issue that cannot be fully resolved in this study is causality between the different events investigated. Overall, the authors are careful about this, but there are some exceptions that should be corrected. Probably the most important is in the abstract, where they write: “We conclude that variation in concerted processes of 5’ splice site selection by U6 snRNA is crucial to evolutionary change in splicing complexity”. I suggest they write something more open (and correct), such as: “We conclude that variation in concerted processes of 5’ splice site selection by U6 snRNA is associated with evolutionary changes in splicing complexity”. Similarly, other plausible scenarios should be discussed in the corresponding Discussion section.

We agree with the reviewer that it is not possible to infer the causal relationship between METTL16 absence and 5’SS+4 preference change from the current data. We, therefore, apologise for failing to be more careful in the Summary and Introduction. We have reworded these statements to better reflect what we can currently say about the evolutionary relationship between METTL16 and 5’SS sequence preference.

The correlation between METTL16 absence and 5'SS+4 sequence preference change could most likely be explained by one of several scenarios: (a) sudden loss of METTL16 causes a rapid necessity to change 5'SS sequence preferences. This is unlikely as such rapid change without widespread corresponding 5'SS changes would likely impose a high fitness cost. (b) Changes in 5'SS sequence preference occur first, driven by some other selective pressure, until there is no longer a benefit to retaining the METTL16 gene. (c) Gradual changes in the expression or catalytic efficiency of METTL16 reduce the stoichiometry of U6 snRNA m6A modification, which permits gradual change in 5'SS+4 sequence preference until complete loss of the METTL16 no longer imposes a major fitness cost. As we suggest in the Discussion, future work could examine this question by determining whether the METTL16 orthologs found in Zygosaccharomyces and Eremothecium species, which have altered their 5'SS+4 preference to a U, are expressed and functional. We have updated the Discussion to include a new section that addresses these scenarios.

2) I do not agree with the statement that "The extent of alternative splicing is the best genomic predictor of developmental complexity". To start with, there are many ways to quantify "extent of alternative splicing" and there are also different types of alternative splicing that might have different prevalence and biological impact. Then, this claim is usually related with exon skipping, which is tightly linked with intron length, and that is likely a better prediction of complexity (yet clearly not causative). My concern is: to what extent has this claim been formally and properly assessed by comparing splicing prevalence with other genomic features, such as intergenic region length, intron length, or average distance between enhancer-promoter interactions (arguably the most relevant predictor, in light of many other studies)? Moreover, I found it a bit misleading to frame the work presented in this study as directly related with developmental (or even splicing) complexity. The work is very interesting on its own, and I doubt their findings on +4 position preference in Saccharomycotina has anything to do with developmental complexity (as the Abstract and Introduction seem to imply). 

On reflection, we agree with the reviewer. Some of our framing of the text isn’t balanced with other studies on the scaling of alternative splicing with developmental complexity. We have edited the Summary and Introduction sections accordingly and cited other references that broaden the consideration of this subject. We are grateful to the reviewer for this suggestion because the changes we make improve the focus of the manuscript since our findings relate more to splicing simplification than to an understanding of increased developmental complexity.

__3) I found Figure 2 and its associated supplementary figure very difficult to follow. I suggest the authors try to improve it and make it clearer. Also, other trees summarizing the results might be helpful. __

We apologise for the complexity of these figures. We opted to show phylogenetic trees with phenotypes plotted on the y axis, rather than simply trait histograms or box-plots, because the underlying structure of the tree is important for demonstrating that multiple independent changes in the 5’SS phenotype have occurred in the Saccharomycotina. We have tried to improve the comprehensibility of the figures in the following ways: (a) We have added 5’SS sequence motifs to the x-axis of figure 2B to make what the plot represents clearer, (b) as suggested by the reviewer, we have created a pruned tree showing the 5’SS motifs of a selection of Saccharomycotina species, which demonstrates that the changes in 5’SS+4 position preferences seen in S. cerevisiae and C. albicans are likely to be a result of convergent evolution. We have added this tree as Figure 2 - figure supplement 3.

__4) I also found the Results section corresponding to Figure 5B a bit confusing. I would argue (as I think the authors do) that there are two main patterns here: below 500 introns, there is no association, while above 500 introns there is an increasingly negative association (correlation). I think it would help to more explicitly distinguishing these two patterns. Then, for the intron-poor species: is the correlation (or lack of) for species with a T or an A in position +4 different? __

We do indeed think that there are two patterns here, as indicated by the reviewer. In the previous version of the manuscript, we separated species into those having an overall preference for A at the +4 position, and those having +4U. By showing regression lines for these two classes, rather than for the general relationship between intron number and U5/6rho, we somewhat imply that the switch in +4 base preference might be causing the loss of correlation between U5/6rho and intron number. However, since essentially all species with a 5'SS +4U preference are intron poor, it seems more likely that these trends are the result of a loss of the negative correlation between intron number and U5/6rho in intron poor species, as suggested by the reviewer. To address this issue, we have replaced the regression lines on Figure 6B with a single loess (locally estimated scatterplot smoothing) regression line for all species and updated the text to make it clearer that we think loss of U5/6rho and +4A preference are separate traits of intron poor species. Although this is not exactly what the reviewer requested, we hope that it satisfies their issue with the analysis.

__Reviewer #3 (Significance (Required)): __

__This is a very interesting study that sheds light on an intriguing evolutionary pattern: the change in consensus sequence at position +4 of the 5' splice site. This topic is relevant since it is closely associated with intron loss and splicing efficiency and evolution. __

We thank the reviewer for the kind and constructive comments on this study.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1 (Evidence, reproducibility and clarity):

Summary

The manuscript by Parker et al addresses the important question of how different organisms have evolved pre-messenger RNA systems that are either more or less complex. This question underlies the evolution of complex organisms and the genome adaptation of simple organisms to their specific environments, so is an important question to answer. This manuscript now provides the underlying molecular mechanisms of how 5' splice site sequence preference may have evolved which is both an interesting and exciting advance for the field.

We thank the reviewer for these kind comments.

Major comments

This manuscript builds on the previous work from this group where they identified the role of adenosine N6 methylation (m6A) of the U6 small nuclear RNA (snRNA) of the spliceosome by METTL16 as being important for 5' splice site selection. This work led to the speculation that loss of a METTL16 ortholog, or potentially other splicing factors, in some species could contribute to an evolutionary change in 5' splice site sequence preference. Here the authors now use the power of phylogenetics, interspecies association mapping and the available spliceosome structures to provide convincing conclusions that 5' splice site sequence preferences in the extensive number of organisms examined correlate with the presence of the U6 snRNA methyltransferase METTL16 and the splicing factor SNRNP27K. 

An analysis of METTL16 conservation was first carried out by comparing the METTL16 methyltransferase domain (MTD) in 29 diverse eukaryotic species. All the METTL16 orthologs were found to have either one or two globular domains. Three domain types were identified and compared in detail. What was not clear from this analysis was the functional significance of orthologs having either one or two domains.

We identified several species, including Drosophila melanogaster, whose METTL16 orthologs do not contain a VCR domain. However, in this study we do not draw specific conclusions about the functional significance of orthologs having different domain topologies.

In addition, while this analysis provides important new information on the domain structure of METTL16 orthologs, especially where these domains had not been identified previously, the link between this section of the results and the following sections is not that apparent.

We agree that there is a significant difference in approach between the first section of the Results and the following sections. However, we are keen to keep this part of the manuscript because it provides an orthogonal line of evidence suggesting that the ancestral role of METTL16 in eukaryotes is specifically the methylation of U6 snRNA.

Next novel bioinformatics pipelines were developed to compare both introns and orthologous groupings of protein coding genes between 227 Sacchromycotina genomes as well as 13 well-annotated eukaryote genomes. First, the 5' splice site sequence preference was compared and clearly indicates that the +4 position has the greatest variation in preferences within the Sacchromycotina. The ability to now compare a large number of genomes has provided novel information on the evolution of the 5' splice site sequence and the conclusion that there is more complexity to the 5' splice site in fungi that previously recognized. While it is apparent why only the 5' splice site signal was investigated here, with its relationship to the U6 snRNA and METTL16, it seems a shame the other splice site sequences were not analyzed using this novel pipeline. In any case, the complexity of the 5' splice site +4 position now allows, for the first time, interesting interspecies association studies.

We have now included the variance plots for 3’SS motifs (analogous to the 5’SS variance plots shown in Figure 2B) as Figure 2 supplementary figure 4A, and a traitgram for 3’SS -3C to U ratio as Figure 2 supplementary figure 4B. We have included a short section of text in the Results section to describe these additional findings.

With the 5' splice site +4 variation identified, the next step was to determine the underlying molecular mechanisms that dictate the evolution of the various sequence preferences. Some obvious players here are the U1 and U6 snRNAs which directly interact with the 5' splice site during splicing. However, no association was found between these snRNAs and the 5' splice site +4 sequence. 

The powerful interspecies association mapping was then used to determine whether the presence or absence of METTL16 ortholog or a splicing factor correlated with the 5' splice site +4 sequence variation. Interestingly, a clear association was found between METTL16 and the 5' splice site +4 position; METTL16 presence was associated with +4A at the 5' splice site and METTL16 absence was associated with +4U at the 5' splice site. This is an exciting and significant finding.

We thank the reviewer for these comments on the importance of this study.

Interestingly, the next most significant association with the 5' splice site +4 position was with SNRNP27K. This result makes sense as in the cryo-EM structure of the pre-B spliceosome complex the C-terminal domain of SNRNP27K is found near the region of the U6 snRNA that will interact with the 5' splices site. Absence of SNRNP27K was associated with an increased preference for +4U at the 5' splice site. Now the real power of the interspecies association mapping was demonstrated by investigating whether any association could be determined specifically within the C-terminus of SNRNP27K. Significantly, the methionine 141 position in SNRNP27K was found to be associated with the +4 position of the 5' splice site. This finding fits nicely with previous studies where mutation of M141 caused a shift in 5' splice site selection away from +4A 5' splice sites, to 5' splice sites without +4A. What is not clear is whether M141 is conserved or invariant between all the species that were compared?

M141 is not completely conserved across the species that were compared for the SNRNP27K C-terminus analysis. We did not test positions with very strong sequence conservation, because without variation in both the genotype and phenotype it is not possible to test for an association. We have rephrased the relevant Results and Methods sections to make this point clearer. In addition, we have incorporated a sequence logo to illustrate the degree of conservation of each position in the SNRNP27K C-terminal domain as Figure 5 -figure supplement 1A. Finally, we have included an additional box-plot to illustrate the finding that species which have lost SNRNP27K or have only lost the Methionine equivalent to human SNRNP27K position 141, show a similar preference for +4U at 5’ SSs. This is now included as Figure 5 - figure supplement 1B.

Overall, this result reveals the power of the interspecies association approach and provides interesting and exciting information on the molecular determinants of 5' splice site evolution.

We are grateful to the reviewer for these comments.

The final analysis was to investigate the interaction potentials of the U5 and U6 snRNAs with the 5' splice site in the Sacchromycotina genomes and try to relate this to species with fewer introns and less alternative splicing. Species with low intron numbers and low splicing complexity were revealed to have weaker U5 and U6 anti-correlation potentials and favor +4U at the 5' splice site. On the other hand, species with high intron number and presumably higher splicing complexity featured anti-correlated U5 and U6 snRNA interaction potentials and favored +4A 5' splice sites. This extensive analysis provides novel information on the interactions and splice site properties of species with simple and complex splicing. Again, I see why there is emphasis on the 5' splice site here but a similar analysis with the U2 snRNA and the branch site could also be informative.

We absolutely agree that inter-species association mapping could be applied to other splicing signal phenotypes including 3’ splice sites and intron branchpoints. Accordingly, we raise this subject in the final section of the Discussion. However, branchpoint sequences are challenging to predict with genomic data. Because preliminary analyses suggest independent variation in these other splicing signal phenotypes, we feel a separate focused study is required to properly explain (and substantiate) even the analytical approaches involved. We hope the reviewer would agree that incorporating U2 snRNA and branchpoint variation analyses into this manuscript as well, could detract from the clarity of the conceptual advances that we make here.

Minor comments

Should the Title include SNRNP27K?

There is certainly a case that the title should include SNRNP27K. Our aim was to make the title as short and informative as possible without too many acronyms that need explaining. Since the clearest correlation is with METTL16 and this has broader implications for understanding the role of this enzyme not only in splicing but in possibly modifying other RNA targets too, we think not including SNRNP27K is a suitable compromise. In addition, retaining the current title simplifies the tracking of the manuscript from pre-print through to journal publication.

Should the title specify that it is the evolution of only the 5’ splice site sequence preference being studied here?

Because apostrophes in titles can compromise some scholarly online search engines (https://insights.uksg.org/articles/10.1629/uksg.534), we would prefer not to include 5’ in the title.

Include information on intron number and 5’ splice site interaction potential of U5 and U6 snRNA in the Summary?

We thank the reviewer for this suggestion. We have updated the Summary to include our findings on U5 and U6 interaction potential in species with reduced intron number.

Figure 1C is not referred to in the text?

We apologise for this oversight. We have added references to figure 1C in the appropriate Results section.

Page 8, line 5 – better to say “splicing signal phenotypes”.

We have amended this statement on Page 8 and at other places in the text where related phrasing was made.

What are the other points on Figure 3B? What is the next point below SNRNP27K? Is it U2A’? 

The other points on Figure 3B represent Orthofinder orthogroups which contain human orthologs that are known components of the spliceosome. The list of spliceosomal components was taken from Sales-Lee et al. 2021. The third most significant point is indeed the orthogroup containing the human ortholog of U2A’. As we state in the text, however, the correlation of U2A’ with the 5’SS+4 A to U ratio phenotype is no longer significant once METTL16 presence/absence is controlled for, indicating that the correlation of U2A’ with the +4A phenotype is likely explained by similarity in the patterns of gene loss of U2A’ and METTL16.

The second paragraph of the Discussion is vague and lacks a reference. “we could also identify an association with a methionine residue in the conserved C-terminal domain of SNRNP27K orthologs.” There are a few methionines in the C-terminus, which one? Please reference the statement “transcriptome analysis of C. elegans SNRP-27 M141T mutants..”

We apologise for the lower quality of writing in this section of the Discussion. We have updated the text, made the statements about the SNRNP27K C-terminus less ambiguous, and added the relevant citations as appropriate.

Reviewer #1 (Significance):

Overall, this is a well written and clearly presented study that provides some key molecular information on the splicing factors involved in the evolution of 5’ splice sites and shows the power of interspecies association studies. Some important conceptual principles have now been defined for the field going forward.

With thank the reviewer for this kind comment on the importance of this work.

The question remains as to whether METTL16 and SNRNP27K are the sole determinants of 5’ splice site preference evolution at +4?

We cannot say for certain that METTL16 and/or SNRNP27K determine the 5’SS +4 phenotype – only that they are correlated with it. In our response to reviewer 3, and in a new Discussion section, we have detailed some of the scenarios that could explain these correlations. We also cannot rule out whether there are changes in the presence/absence (or domain/sequence-level changes) of other, untested proteins that correlate with the 5’SS +4 phenotype and we allude to this in the final section of the Discussion.

One splicing factor that immediately comes to mind is Prp8 where there is extensive evidence for involvement in splice site selection and is clearly in the right location throughout splicing to be involved. This question should at least be discussed but Prp8 would also be a very interesting candidate for the interspecies association mapping.

Prp8 is a core component of spliceosomes and is conserved throughout the Saccharomycotina. For this reason, we were unable to associate splicing phenotypes with Prp8 presence or absence variation at the level of orthogroups. However, we revisited this question posed by the reviewer. Our experience with inter-species association mapping, so far, indicates it works well with orthogroup presence/absence or when straightforward amino acid substitutions can be detected in conserved and hence alignable protein sequence domains. We analysed the conserved U6 snRNA-interacting region of the Prp8 linker domain, which maps close to the 5’ splice site in cryo-EM models, using the profile HMM PF10596 available from Pfam. We found that the majority of this domain was extremely highly conserved with variation in only a few species and positions. The strongest correlation with the +4A to U ratio phenotype was at position 58, which is conserved as a Glycine in all but 8 species (6 Dipodascaceae, 2 CUG-Ser1), that also tend to have a stronger preference for +4A. However, examination of the species contributing to this result (and to similar results at other positions) indicated that in the 6 Dipodascaceae species, this change is part of a larger deletion or replacement that makes the whole linker region align poorly to the model. Hence, the G58 position itself may not be specifically important for the +4 phenotype. Although the wholesale loss or replacement of the U6 snRNA-interacting region in these species is potentially interesting, these larger scale structural changes in a small number of species are difficult to interpret. Therefore, to maintain the focus of the manuscript and the clear links to METTL16 and SNRNP27K that have orthogonal support, we have decided not to add these results to the manuscript but present them here (Figure not available on biorXiv commenting window).

Also, as mentioned previously, only the 5’ splice site was investigated here and the manuscript could become a more substantial piece of work if the other splice sites were included in some way.

We agree that it will be exciting to apply this approach to other splicing signal phenotypes and in other phylogenetic clades with emerging tree-of-life-scale genomics data. We have included variation in 3’ splice sites in the revised manuscript. As the first of its kind, this study should pioneer a wider use of this approach, by us and others, to understand the mechanisms and functions of molecular interactions not only in splicing but in other areas of biology too.

The obvious audience here are those directly in the splicing field but the overall principles are relevant for evolutionary biologists and those studying organismal complexity.

We thank the reviewer for recognising the broad importance of this work.

My expertise is in yeast and human splicing mechanisms. I do not have the expertise to critically evaluate the bioinformatic pipelines but they were clearly explained and presented.

Reviewer #2 (Evidence, reproducibility and clarity):

In their manuscript, Parker et al. investigate the evolutionary patterns of splice site preference, focusing on the A/U ratio at position A+4 on the 5´ splice site. Building upon prior studies in S. pombe and A. thaliana, the authors establish a strong correlation between this preference and the co-evolution of the METTL16 U6 snRNA methyltransferase. Furthermore, through inter-species association mapping, they identify the involvement of the splicing factor SNRNP27K in altered A/U ratios and highlight the significance of the residue Met-141 in SNRNP27K for this function. Overall, the paper effectively presents impactful new findings on the evolution of METTL16, U6 snRNA, and splicing.

We thank the reviewer for these kind comments on the importance of our study.

The computational analyses employed in this study are situated outside our field of expertise, preventing us from offering a comprehensive evaluation of the methodology’s appropriateness and rigor. Nonetheless, the identification of METTL16 through the authors’ methods, which aligns with previous research in S. pombe and A. thaliana, lends support to the validity of their approach. Notably, the close proximity between SNRNP27K and the methylated A43 residue in U6 snRNA within the spliceosome, particularly near Met-141, is an impressive finding. Previous studies have shown that a mutation at position M141T affects splicing at +4A introns, thus providing robust validation for their methods.

We thank the reviewer for these kind comments on our work.

The data presented in this study furnish crucial insights into the role of METTL16, U6 snRNA methylation, and splice site recognition. The authors expand upon recent observations that the “vertebrate conserved region” exists in non-vertebrates, despite the absence of primary sequence homology. These results will serve as a valuable guide for future molecular investigations into U6 snRNA methylation and its mechanisms in splicing. Furthermore, the implications of this paper extend to human evolution, as the plasticity in splicing is an essential factor in the evolution of developmental complexity.

We thank the reviewer for these kind comments.

Minor suggestions for improvement:

1. Given the significance of the interaction between U6 snRNA and the intron for understanding the data, it would be beneficial to include a figure illustrating the RNA-RNA base-pairing interactions between U6 snRNA and the 5´ splice site. This addition is particularly important if the paper is intended for publication in a journal with a general readership.

We thank the reviewer for this excellent suggestion. We have included this as Figure 3A.

1. Similarly, the section on U1 snRNA would be more comprehensible with the inclusion of U1 RNA-RNA intron diagrams and improved descriptions of both the figures and the assay. Despite being negative data in the supplement, clarifying this section is essential. As currently written, it is challenging to follow.

We agree that this section is difficult to follow. We have updated the text to improve the readability and included a figure of U1 snRNA:5’SS basepairing as Figure 3 – figure supplement 1A.

1. Whenever possible, consider increasing the figure and font sizes to enhance readability for readers.

We agree that some of the more complex figures can be difficult to read when embedded into a Word document/pdf. We hope that providing high-resolution figures for reading online will mitigate this.

1. In the text, there is no reference to Figure 1C.

We apologise for this oversight. We have resolved this issue with the appropriate references in the Results text.

1. In Figure 5B, the y-axis in the top panel is labelled “species,” but the legend only mentions U5/6p as the y-axis. Please revise the legend to include the appropriate information.

We apologise for the confusion caused by our poorly written legend for this plot. We have updated the legend so that the text clearly refers to either the scatter plot or the marginal histograms.

Reviewer #2 (Significance):

The data presented in this study furnish crucial insights into the role of METTL16, U6 snRNA methylation, and splice site recognition. The authors expand upon recent observations that the “vertebrate conserved region” exists in non-vertebrates, despite the absence of primary sequence homology. These results will serve as a valuable guide for future molecular investigations into U6 snRNA methylation and its mechanisms in splicing. Furthermore, the implications of this paper extend to human evolution, as the plasticity in splicing is an essential factor in the evolution of developmental complexity.

We are grateful to the reviewer for these kind comments on the importance of this work.

Reviewer #3 (Evidence, reproducibility and clarity):

In this manuscript, Parker et al present a nice exploration of the evolutionary and mechanistic relationships between 5′ splice site consensus sequences, intron numbers and METTL16/SNRNP27K. By performing inter-species association mapping in Saccharomycotina species, they found that a T in position +4 is strongly associated with the absence of METTL16 (and/or in some cases SNRNP27K or mutations in it). They also provide solid structural modelling data in support of this association.

In general, I think this is a very nice manuscript. I only have a few comments, which could be addressed by rewording specific parts and/or improving the current figures.

We are grateful to the reviewer for the kind comments on this work.

1) As the authors acknowledge, a key issue that cannot be fully resolved in this study is causality between the different events investigated. Overall, the authors are careful about this, but there are some exceptions that should be corrected. Probably the most important is in the abstract, where they write: “We conclude that variation in concerted processes of 5’ splice site selection by U6 snRNA is crucial to evolutionary change in splicing complexity”. I suggest they write something more open (and correct), such as: “We conclude that variation in concerted processes of 5’ splice site selection by U6 snRNA is associated with evolutionary changes in splicing complexity”. Similarly, other plausible scenarios should be discussed in the corresponding Discussion section.

We agree with the reviewer that it is not possible to infer the causal relationship between METTL16 absence and 5’SS+4 preference change from the current data. We, therefore, apologise for failing to be more careful in the Summary and Introduction. We have reworded these statements to better reflect what we can currently say about the evolutionary relationship between METTL16 and 5’SS sequence preference.

The correlation between METTL16 absence and 5'SS+4 sequence preference change could most likely be explained by one of several scenarios: (a) sudden loss of METTL16 causes a rapid necessity to change 5'SS sequence preferences. This is unlikely as such rapid change without widespread corresponding 5'SS changes would likely impose a high fitness cost. (b) Changes in 5'SS sequence preference occur first, driven by some other selective pressure, until there is no longer a benefit to retaining the METTL16 gene. (c) Gradual changes in the expression or catalytic efficiency of METTL16 reduce the stoichiometry of U6 snRNA m6A modification, which permits gradual change in 5'SS+4 sequence preference until complete loss of the METTL16 no longer imposes a major fitness cost. As we suggest in the Discussion, future work could examine this question by determining whether the METTL16 orthologs found in Zygosaccharomyces and Eremothecium species, which have altered their 5'SS+4 preference to a U, are expressed and functional. We have updated the Discussion to include a new section that addresses these scenarios.

2) I do not agree with the statement that "The extent of alternative splicing is the best genomic predictor of developmental complexity". To start with, there are many ways to quantify "extent of alternative splicing" and there are also different types of alternative splicing that might have different prevalence and biological impact. Then, this claim is usually related with exon skipping, which is tightly linked with intron length, and that is likely a better prediction of complexity (yet clearly not causative). My concern is: to what extent has this claim been formally and properly assessed by comparing splicing prevalence with other genomic features, such as intergenic region length, intron length, or average distance between enhancer-promoter interactions (arguably the most relevant predictor, in light of many other studies)? Moreover, I found it a bit misleading to frame the work presented in this study as directly related with developmental (or even splicing) complexity. The work is very interesting on its own, and I doubt their findings on +4 position preference in Saccharomycotina has anything to do with developmental complexity (as the Abstract and Introduction seem to imply).

On reflection, we agree with the reviewer. Some of our framing of the text isn’t balanced with other studies on the scaling of alternative splicing with developmental complexity. We have edited the Summary and Introduction sections accordingly and cited other references that broaden the consideration of this subject. We are grateful to the reviewer for this suggestion because the changes we make improve the focus of the manuscript since our findings relate more to splicing simplification than to an understanding of increased developmental complexity.

3) I found Figure 2 and its associated supplementary figure very difficult to follow. I suggest the authors try to improve it and make it clearer. Also, other trees summarizing the results might be helpful. 

We apologise for the complexity of these figures. We opted to show phylogenetic trees with phenotypes plotted on the y axis, rather than simply trait histograms or box-plots, because the underlying structure of the tree is important for demonstrating that multiple independent changes in the 5’SS phenotype have occurred in the Saccharomycotina. We have tried to improve the comprehensibility of the figures in the following ways: (a) We have added 5’SS sequence motifs to the x-axis of figure 2B to make what the plot represents clearer, (b) as suggested by the reviewer, we have created a pruned tree showing the 5’SS motifs of a selection of Saccharomycotina species, which demonstrates that the changes in 5’SS+4 position preferences seen in S. cerevisiae and C. albicans are likely to be a result of convergent evolution. We have added this tree as Figure 2 - figure supplement 3.

4) I also found the Results section corresponding to Figure 5B a bit confusing. I would argue (as I think the authors do) that there are two main patterns here: below 500 introns, there is no association, while above 500 introns there is an increasingly negative association (correlation). I think it would help to more explicitly distinguishing these two patterns. Then, for the intron-poor species: is the correlation (or lack of) for species with a T or an A in position +4 different? 

We do indeed think that there are two patterns here, as indicated by the reviewer. In the previous version of the manuscript, we separated species into those having an overall preference for A at the +4 position, and those having +4U. By showing regression lines for these two classes, rather than for the general relationship between intron number and U5/6rho, we somewhat imply that the switch in +4 base preference might be causing the loss of correlation between U5/6rho and intron number. However, since essentially all species with a 5'SS +4U preference are intron poor, it seems more likely that these trends are the result of a loss of the negative correlation between intron number and U5/6rho in intron poor species, as suggested by the reviewer. To address this issue, we have replaced the regression lines on Figure 6B with a single loess (locally estimated scatterplot smoothing) regression line for all species and updated the text to make it clearer that we think loss of U5/6rho and +4A preference are separate traits of intron poor species. Although this is not exactly what the reviewer requested, we hope that it satisfies their issue with the analysis.

Reviewer #3 (Significance):

This is a very interesting study that sheds light on an intriguing evolutionary pattern: the change in consensus sequence at position +4 of the 5' splice site. This topic is relevant since it is closely associated with intron loss and splicing efficiency and evolution. 

We thank the reviewer for the kind and constructive comments on this study.

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

Learn more at Review Commons

---

Reply to the reviewers

Reviewer #1*. This is a good paper dealing with gap of our knowledge in understanding reason of ICB failures. Subject being difficult it is expected that the design and content of such experiment will be complex.But the authors forget practicality of readers attention and making paper apear interesting. They need to organise and may be classify the varied information in such a way that reader can find a rhythm in excavating data more easily. It appears confusing at time, so they may try to make it more simple. In this way they may concentrate more on methods and classify results too. A thorough revision is suggested, to make it consize. *

__Authors’ answer: __We thank the Reviewer for his positive evaluation and constructive feedback. We appreciate the complexity of single-cell RNA-sequencing analyses. In order to simplify our manuscript, our revised manuscript now focuses on the transitional states of tumor-resident and circulating T cells found in ovarian cancer patients. Our study is timely as it is the first to report the developmental relationship of TILs in ovarian cancer. We substantially edited our manuscript to make it clear that our findings suggest a gradual acquisition of the exhaustion program initiated by effector-like cells (cluster CD8_GZMH) that eventually gives rise to more terminal states with features of tissue residency and chemotaxis (clusters CD8_CCL4, CD8_XCL1, and CD8_CXCL13). We also include new analyses revealing the presence and proportion of these T cell states in different cancer patients (New Fig. 4A-B), and how these T cell states associate with clinical responses to immune checkpoint blockade (ICB). We hope the Reviewer will find our revised manuscript easier to read.

Reviewer #2. I think the first half of the article, in which the GZMH-CD8 cluster is considered to be in an intermediate state of transition to exhaustion, is interesting, and I feel that the single-cell seq and TCR data are well analyzed to make the point. On the other hand, I feel that the latter part of the paper may not be anything more than a hypothesis. In particular, the part claiming that it is related to prognosis or applicable to the prediction of the effect of ICB is insufficient, since their gene signature is not described in detail and the contents of the Figure are not mentioned in the manuscript. In the latter part, the effects of GPR184 and 25-HC, or the effects of IL21, would require experiments to verify (to verify whether the addition of chemokine or the inhibition of the receptor changes the specific CD8 population).

Author’s answer: Thank you for discussing the limitation of the signature employed. We agree with the reviewer’s comment. Old Figure 5 has been removed from the revised manuscript.

Reviewer #2. Minor point: In particular, there is little mention of Figure 5 in the text, making it difficult to understand.

Author’s answer: Thank you for your comment. As we previously discussed, we have removed Figure 5 from the revised manuscript. The method used to generate the signature was found to be inappropriate.

Reviewer #2. The latter part is difficult to understand. To begin with, it is already known that ovarian cancer does not contribute much to ICB, so what does it mean to analyze the CD8 population, which is known as a marker of ICB response in other carcinomas, as an indicator? Especially for clinicians like us, it is hard to imagine that the results will lead to clinical trials that will attempt to sort out the population that ICB is favored in.

Author’s answer: Although immune checkpoint blockade has demonstrated limited effectiveness against ovarian cancer, subset analyses suggest superior efficacy for some patients and according to subtype. Combination anti-PD-1/CTLA-4 therapy for instance achieved response rates up to 31% (Zamarin et al., 2020), and superior benefit for single agent PD-1 blockade has been reported in clear cell ovarian cancer. Moreover, encouraging clinical results have recently been reported in studies exploring combinations with PARP and VEGF inhibitors. As example, interim analysis of the phase 3 DUO-O trial (NCT03737643) showed a statistically significant and clinically meaningful improvement in PFS in patients with newly diagnosed advanced ovarian cancer without a BRCA1/2 mutation (Harter et al., 2023).

Our study aimed to better understand how ovarian tumor-infiltrating T cells acquire their exhaustion program after migrating from the periphery and whether these mechanisms are unique or shared amongst cancer types. Recent studies in other cancer types had shown the dynamics of T cells and demonstrated the clonal replacement of intratumoral T cells after ICB and emphasized the role of peripheral clones in this process (Wu et al., 2020; Yost et al., 2019). In lung cancer, it has been proposed a transitional state between precursor and terminally differentially cells (Gueguen et al., 2021). Our study demonstrates, for the first time in ovarian cancer, the presence of similar transitional states of CD8 T cells. Our revised manuscript also now includes new data revealing that pre-effector GZMK- and intermediary GZMH-expressing CD8 cells are better biomarkers of ICB response than terminally differentiated XCL1 and CXCL13 expressing CD8 T cells (New Figure 4). Altogether, our study provides important and novel insights on the development of tumor-infiltrating T cells in ovarian cancer patients, which may serve to better select ovarian cancer patients for ICB therapy.

Reviewer #2. Since the first half of the study is very interesting, we feel that it is more important to confirm the mechanism of exhaustion from the blood via the intermediate (GZMH_CD8), including functional experiments. Also, as a clinician, we are very interested in the perspective of whether some of the fractions identified in this study are different in proportion in different patients and whether they correlate with the clinical course of the disease since the study only analyzed a sample of 5 patients.

Author’s answer: We thank the reviewer for proposing to extend our analysis. As suggested, our revised manuscript now includes new analyses which reveals the different proportions of our identified T cells states in different cancer patients (New Figure 4). We further investigated whether these T cell states associate with clinical responses and observed that pre-effector GZMK- and intermediary GZMH-expressing CD8 T cells are better biomarkers of ICB response than terminally exhausted XCL1- and CXCL13-expressing CD8 T cells (New Figure 4).

Reviewer #3. Question 1: Whether the distribution patterns of CD4+ and CD8+ T cell clusters in Figure 1B were comparable among the 5 patient samples? Whether the proportion of five types of clones in Figure 3C are comparable among the 5 patient samples?

Author’s answer: Thank you for the question. We included the results to answer these questions in the supplementary material (fig. S1C-D). For each patient, we calculated the proportion of a cluster among T cells in the blood or tumor. As observed in the boxplot (fig. S1C), the proportion of some subsets were higher in certain patients, such as the higher proportion of CD8_GZMK in the tumor of patient p09454. A recent study classified patients’ tumors based on the spatial distribution of CD8 T cells and performed scRNA-seq to identified cell subsets enriched in the groups inflamed/infiltrated (characterized by the distribution of CD8 T cells within the tumor epithelium), excluded (infiltrating CD8 T cells are restricted to the tumor stroma) or desert (T cells are not present or have low frequency) (Hornburg et al., 2021). Interestingly, this subset of CD8_GZMK cells were enriched in desert tumors, suggesting that the difference we observed in our dataset might reflect the spatial distribution of CD8 T cells in patient p09454. Regarding the TCR-seq data, the frequency of the five types of clones was different among patients. To show this data, we included a barplot (fig. S2D), showing for example, a higher proportion of tumor-expanded clones in patient p10329.

Reviewer #3. Question 2: In Figure S2C, only a very small number of cells in the CD8-GZMK K-22 population. Are these cells representative? Do they generally exist in multiple samples or only in one sample?

Author’s answer: Thank you for your comment. The subcluster k_22 indeed has a smaller number of cells compared to other subclusters. Nevertheless, the K_22 cluster was found in every patient and in every healthy donor. To clarify, we edited our revised manuscript to include a statement that cluster k_22 was composed of fewer cells compared to other clusters.

Reviewer #3. Question 3: In the Fig.S6 legend, the authors stated "Our results suggest the differentiation of cluster CD8-GZMK into the effector-like subset CD8-GZMH." However, there seems to be no corresponding analysis in the main text to support this conclusion.

Author’s answer: We appreciate your attention to this statement. We agree the results of our study doesn’t sustain this statement and so we have excluded it in the revised manuscript.

Reviewer #3____. Question 4: Is there more detailed clinical information that can be provided for the 5 patients included in the study? Per the methods all patients were receiving debulking surgery and were treatment naïve, but did they differ in stage, age, comorbidities, etc.?

Author’s answer: Thank you for your comment on this. We have included a table with clinical information on the stage, age, and menopause status of the five patients.

Reviewer #3. Question 5: Were any cells included for sequencing from adjacent 'normal' tissue uninvolved with tumor (these samples are from surgical debulking of primary tumors, which may include such areas of non-involved tissue.) While shared TCR clonotypes between blood and intratumoral T cells strongly suggests the tumor-resident populations are recruited from the blood, the degree of sharing with normal tissue-resident T cells would be of interest as well.

Author’s answer: Thank you for your comment. Samples were provided for sc-RNA-seq after pathology review and validation of tumor histology. We did not perform sc-RNA-seq on normal adjacent tissue (NAT) We agree this would be interesting as a follow up study, since in other cancer types (renal, colon and lung) it has been demonstrated that T clones expanded in the tumor and NAT are also present in peripheral blood (Wu et al., 2020).

Reviewer #3. Question 6: Very little is discussed about HGSOC itself in the main text (eg clinical background, prior literature on the composition of infiltrating immune populations and potential reasons for at best modest poor responses to IO) until the first sentence of the discussion. As the entirety of the new data produced in this study is from HGSOC tumors there should be more focus on this tumor type and conversation with the prior literature on it (mainly from prior studies on the immune environment of HGSOC). Further, how distinct do the authors suspect the cell populations found in their study to be to ovarian as opposed to other epithelial tumor types?

Author’s answer: Thank you for the suggestion. We now included more background information on immunotherapy of HGSOC. Specifically, we added the following paragraph in our introduction: “In ovarian cancer, the presence of both T and B cells improves patients' survival (Nelson, 2015; Nielsen et al, 2012). They are usually organized in lymphoid aggregates ranging from a small group of cells to a well-organized TLS (Kroeger et al, 2016). Organized TLSs correlate with better survival, such as observed in patients treated with ICB. Although immunotherapy has demonstrated limited effectiveness against ovarian cancer, subsets of patients may thus benefit from ICB. In support of this, combination anti-PD-1/CTLA-4 therapy can achieve response rates above 30% (Zamarin et al., 2020), and encouraging clinical results have recently been reported when combining ICB with with PARP and VEGF inhibitors (Harter et al., 2023)”.

Reviewer #3. Question 7: Were the signature genes used for analysis in figure 5 remove chosen in a formal, unbiased manner, or simply hand-picked as representative of the respective cell types? This information is not provided in the supplement.

Author’s answer: Another reviewer has also expressed similar concerns. The genes selected to represent cell types were chosen manually, which we acknowledge is not the best method for defining a signature. As a result, we have decided to exclude Figure 5 from the manuscript under review. We believe an unbiased approach is more suitable for characterizing the cell network proposed in our study.

Reviewer #3. Question 8: While the NicheNet analysis of potential interactions among lymphocyte populations raises some strong hypotheses, it would be interesting to extend the interaction analysis to all CD45+ populations, given the sequencing was done on CD45+ immune cells.

Author’s answer: Thank you for suggesting analysis. We have included the results of cell interaction including all CD45+ cells (fig. S3). We observed CD40L as one of the top predicted ligands highly expressed in CD4_CXCL13 subset mediating a response in subsets of antigen-presenting cells, such as B cells (cluster B), plasma cells (cluster PC_2), and plasmacytoid dendritic cells (cluster pDC). Interestingly, this result also support the hypothesis of Tfh-like cells (cluster CD4_CXCL13) coordinating the action of intratumoral immune cells involved in the antitumor immune response.

Reviewer #3. Question 9: A sample size of 5 patients is relatively small for current single cell RNAseq studies of human tumor patients.

Author’s answer: We agree with the reviewer that a sample size of 5 patients is relatively small. Thus, to validate our results in other patients, we included in the reviewed manuscript the analysis of scRNA-seq of 47 patients across10 cancer types (dataset from (Zheng et al., 2021). As demonstrated in figure 3 and figure 5, we could identify subsets of CD8 and CD4 T cells from our ovarian cancer patients in those 10 cancer types dataset.

Reviewer #3.____ Minor

*1. In lines 96-97, "CD8-GZMB" was mentioned twice in the description. *

2. In line 126, this section did not discuss residency markers, yet a conclusion about residency was made in this sentence.

Author’s answer: We appreciate you bringing these errors to our attention. We fixed them in the updated version of the manuscript.

References:

Gueguen, P., Metoikidou, C., Dupic, T., Lawand, M., Goudot, C., Baulande, S., … Amigorena, S. (2021). Contribution of resident and circulating precursors to tumor-infiltrating CD8 T cell populations in lung cancer. Science Immunology, Vol. 6, p. eabd5778. doi:10.1126/sciimmunol.abd5778

Harter, P., Trillsch, F., Okamoto, A., Reuss, A., Kim, J.-W., Rubio-Pérez, M. J., … Aghajanian, C. (2023). Durvalumab with paclitaxel/carboplatin (PC) and bevacizumab (bev), followed by maintenance durvalumab, bev, and olaparib in patients (pts) with newly diagnosed advanced ovarian cancer (AOC) without a tumor BRCA1/2 mutation (non-tBRCAm): Results from the randomized, placebo (pbo)-controlled phase III DUO-O trial. Journal of Clinical Orthodontics: JCO, 41(17_suppl), LBA5506–LBA5506.

Hornburg, M., Desbois, M., Lu, S., Guan, Y., Lo, A. A., Kaufman, S., … Wang, Y. (2021). Single-cell dissection of cellular components and interactions shaping the tumor immune phenotypes in ovarian cancer. Cancer Cell. doi:10.1016/j.ccell.2021.04.004

Wu, T. D., Madireddi, S., de Almeida, P. E., Banchereau, R., Chen, Y.-J. J., Chitre, A. S., … Grogan, J. L. (2020). Peripheral T cell expansion predicts tumour infiltration and clinical response. Nature. doi:10.1038/s41586-020-2056-8

Yost, K. E., Satpathy, A. T., Wells, D. K., Qi, Y., Wang, C., Kageyama, R., … Chang, H. Y. (2019). Clonal replacement of tumor-specific T cells following PD-1 blockade. Nature Medicine. doi:10.1038/s41591-019-0522-3

Zamarin, D., Burger, R. A., Sill, M. W., Powell, D. J., Jr, Lankes, H. A., Feldman, M. D., … Aghajanian, C. (2020). Randomized Phase II Trial of Nivolumab Versus Nivolumab and Ipilimumab for Recurrent or Persistent Ovarian Cancer: An NRG Oncology Study. Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology, 38(16), 1814–1823.

Zheng, L., Qin, S., Si, W., Wang, A., Xing, B., Gao, R., … Zhang, Z. (2021). Pan-cancer single-cell landscape of tumor-infiltrating T cells. Science, 374(6574), abe6474.

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

Summary: This study used single-cell transcriptomics and T cell receptor profiling to identify the developmental relationships of T cell populations in ovarian cancer patients. The researchers proposed a model of differentiation pathway that showed how an intermediate GZMH-expressing CD8 T cell subset progressively reinforces exhaustion and tissue residency programs towards terminally exhausted cells. Then they also focus on the nature of TPEX, dual-expanded clone, which is considered an important indicator for the efficacy of ICB, and argue that it is strongly related to GPR183, 25-OHC, and IL21. Based on the analysis of clinical samples, they argue that their proposed gene signature may also be prognostically relevant and predictive of ICB efficacy.

Major comment: I think the first half of the article, in which the GZMH-CD8 cluster is considered to be in an intermediate state of transition to exhaustion, is interesting, and I feel that the single-cell seq and TCR data are well analyzed to make the point. On the other hand, I feel that the latter part of the paper may not be anything more than a hypothesis. In particular, the part claiming that it is related to prognosis or applicable to the prediction of the effect of ICB is insufficient, since their gene signature is not described in detail and the contents of the Figure are not mentioned in the manuscript. In the latter part, the effects of GPR184 and 25-HC, or the effects of IL21, would require experiments to verify (to verify whether the addition of chemokine or the inhibition of the receptor changes the specific CD8 population).

Minor point: In particular, there is little mention of Figure 5 in the text, making it difficult to understand.

Significance

It is interesting to note that the authors simultaneously analyze immune cells in the blood and in the tumor, and examine in detail what is characteristic of the blood, what is characteristic of the tumor, and what is seen in both. And it is very interesting that they specifically proposes an intermediate group that is recruited from the blood to the tumor and is in the process of becoming exhausted. I am sure there are many studies on TILs and TLSs, but this study would be helpful to understand how they are concentrated locally (near the tumor) in comparison with immune cells in the blood as well.

However, the latter part is difficult to understand. To begin with, it is already known that ovarian cancer does not contribute much to ICB, so what does it mean to analyze the CD8 population, which is known as a marker of ICB response in other carcinomas, as an indicator? Especially for clinicians like us, it is hard to imagine that the results will lead to clinical trials that will attempt to sort out the population that ICB is favored in.

Since the first half of the study is very interesting, we feel that it is more important to confirm the mechanism of exhaustion from the blood via the intermediate state, including functional experiments. Also, as a clinician, we are very interested in the perspective of whether some of the fractions identified in this study are different in proportion in different patients and whether they correlate with the clinical course of the disease, since the study only analyzed a sample of 5 patients.

Reviewer #2 (Public Review):

Kraus, Aurora et al. investigated the potential immune response of the olfactory bulb after exposure of the infectious hematopoietic necrosis virus (IHNV), via the olfactory epithelia. Specifically, they show that a) viral-specific neuronal activation of "OSNs" (Crypt cells), b) changes in behaviour of both adult and larval zebrafish after viral exposure, c) Pituitary adenylate-cyclase-activating polypeptide (PACAP), was enriched when assayed by single cell transcriptomic profiling of cells in the OB after OSNs are exposed to IHNV

Although the paper does have strengths in principle, the weaknesses of the manuscript are that these strengths are not directly demonstrated and the referencing of the manuscript omits many references important for the understanding of the questions and the results of the study. Furthermore, the data presented are not sufficient to fully support the key claims in the manuscript. In particular:

a) Viral-specific neuronal activation of OSNs:<br /> What type of neurons? The authors are a bit elusive and do not clearly state that the neurons are crypt cells (Sepahi et al.: rainbow trout) which have a very specific axonal projection to the brain and whose response characteristics are not well characterized (see work of Korsching lab). Crypt cells are not present in the olfactory epithelia of mammals. Furthermore, in their previous work the crypt cells die; so how do they think the (inflammatory) virus response is transmitted to the olfactory bulbs in order to protect the brain?<br /> The authors state from previous work that they never detected virus in the brain, but why would they? Does INHV move trans-synaptically?<br /> The neuronal activity was monitored using a pan-neuronal marker thus these data are of limited use when trying to understand the role of neuronal activity (crypt cells) in the IHNV-triggered activity: the authors may be looking at a generalized inflammation response, and the image presented is not particularly informative it is difficult to decipher the results. The authors assume IHNV is an odorant without carefully ruling out the possibility of a generalized inflammation response.<br /> b) Changes in behaviour of both adult and larval zebrafish after viral exposure:<br /> What is the motivating question for looking at behaviour of the virus infected animals? Do we know the effects of crypt cell loss on the behaviour in any fish species? Authors need to build a better conceptual framework for the behaviour experiments.

c) Pituitary adenylate-cyclase-activating polypeptide (PACAP) was enriched when assayed by single cell transcriptomic profiling of cells in the OB after OSNs are exposed to IHNV. Authors draw many generous conclusions from limited data. Authors seem to have forgotten to cite papers previously published showing that PACAP-38 has anti-viral activities in fish (VHSV: trout) such as: Velasquez et al 2020, First in vivo evidence of pituitary adenylate cyclase-activating polypeptide antiviral activity in teleost.<br /> The histology for PACAP presented in the manuscript is not convincing. The antibody is against the human form of PACAP thus any labelling should be treated with caution (and called PACAP-38-like).

Summary: The authors need to better develop their model (perhaps a diagram would be helpful) explaining exactly which neurons are transmitting the information. Because of the elusive nature of some referencing and the skirting of important issues such as clearly stating which neurons are affected (crypt cells), what the point of the behaviour is (relate to neuronal type infected by virus), and, the lack of an antibody specific to the zebrafish protein, the model appears to be built on an unstable base.

Joint Public Review:

In this manuscript, the authors challenge the fundamental concept that all neurons are derived from ectoderm. The key points of the authors argument are as follows:

1) Roughly half of the cells in the small intestinal longitudinal muscle-myenteric plexus (LM-MP) that express a pan-neuronal marker do not, by lineage tracing, appear to be derived from the neural crest.

2) Lineage tracing and marker gene imaging suggest that these non-neural crest derived neurons originate in the mesoderm, leading to their designation as mesodermal-derived enteric neurons (MENs).

3) Single-cell sequencing of LM-MP tissues confirms the mesodermal origin of MENs.

4) MENs progressively replace neural crest derived enteric neurons as mice age, eventually representing the bulk of the EN population.

There is broad agreement among the reviewers that the identification and description of this cell population is important, and that the failure of these cells to be labeled by neural crest lineage tracers is not artifactual. The work with transgenic lines is convincing that some presumptive neurons in the enteric nervous system (ENS) likely originate from an alternative source in the postnatal intestine and that this population increases in aging mice.

There is, however, ongoing disagreement between the authors and reviewers about whether the authors' provocative and potentially paradigm-changing proposal that these are neurons of mesodermal origin has been established. While the authors believe they have addressed the reviewers' concerns in multiple rounds of review (much of this prior to submission), the reviewers remain unconvinced and continue to request additional data and analyses.

A key premise of the preprint review system is that the best interests of science are not served by endlessly litigating disagreements around papers by either compelling the authors to do extensive and expensive additional experiments that they do not believe to be necessary or by treating the authors' claim as established in the face of continued skepticism. Accordingly the editor believes it is time to present this work, which everyone agrees contains important observations and valuable data, along with the following editor's synthesis of the reviewers' concerns and author responses about the question of these cells' origins. We encourage anyone interested in the details to review the already posted reviews and authors' response.

The following key issues have been raised during review:

* Is the lineage tracing and marker gene expression data definitive as to mesodermal origin?

* Are the cells analyzed in the genomic experiments the same as those identified in the lineage tracing experiments?

* Does the genomic data establish that the sub-population of cells the authors focus on are of mesodermal origin?

* Are there alternative explanations for the lineage tracing and genomic observations than a mesodermal origin?

* Is the lineage tracing and marker gene expression data definitive as to mesodermal origin? *

The proximal evidence that the authors present for a mesodermal origin of the non-NC derived cells is presented in Figure 2, which establishes the presence, via lineage tracing of Tek+ and Mesp1+ (and therefore mesoderm derived) and Hu+ (and therefore neuronal) cells. The fraction of lineage labeled cells in each case (~50%) corresponds roughly to the fraction of cells that do not appear to be NC derived.

The reviewers raise several technical questions about the lineage tracing experiments, including issues of incomplete labeling, ectopic labeling and toxicity. The authors have addressed each of these with data and/or citations, and the editor believes they have demonstrated, subject to the broader limits of lineage tracing experiments, that there are Hu+ cells in the tissue that are derived from cells that do not express NC markers and that do express mesodermal markers.

One reviewer raised the question of whether these cells are neurons. This appears to the editor to be a valid question, in that specific neuronal activity of these cells has not been established. But the authors' argument is persuasive that their Hu+ state would have led them to be designated neurons and that changing that designation based on not being derived from NC is circular. However the possibility that, despite this accepted designation, these cells are not functionally neurons should be noted by readers.

* Are the cells analyzed in the genomic experiments the same as those identified in the lineage tracing experiments, and does this data establish mesodermal origin? *

To provide orthogonal evidence for the presence of mesodermally derived enteric neurons, the authors carried out single-cell sequencing of dissociated cells from hand-dissected longitudinal muscle - myenteric plexus (LM-MP) tissue. They use standard methods to identify clusters of cells with similar transcriptomes, and designate, based on marker gene expression, two clusters to be neural crest derived enteric neurons (NENs) and mesoderm derived enteric neurons (MENs). However the reviewers raised several issues about the designation of the cells MENs, and therefore their equation with the cells identified in lineage tracing.

While the logic behind specific choices made in the single-cell analysis is not always clear in the manuscript, such as why genes not-specific to MENs were used to identify the MEN cluster and how genes were selected for subsequent analysis (although both issues are explained better in the authors' response to reviewers), they in the end identify a single large cluster that has the characteristics of MENs (it expresses both neuronal and mesodermal markers) that is (by immunohistochemistry) broadly associated with the previously described tissue MENs.

The standard methods for the delineation of clusters in single-cell sequencing data (which the authors use) are stochastic and defy statistical interpretation, and the way these data and analyses are used is often subjective. The editor shares the reviewers' confusion about aspects of the analysis, but also finds the authors' assertions that they have described a cluster of cells that express both neuronal and mesodermal genes, and that this cluster corresponds to the tissue MENs described in lineage tracing, to be broadly sound.

The biggest weakness in the single-cell data and analysis - identified by all reviewers - is the massive overrepresentation of MENs relative to NENs. The authors' explanation - that some cells are more sensitive to manipulations required to prepare cells for sequencing - is certainly well-represented in the literature and is therefore plausible. But it isn't fully satisfactory, especially because it undermines the notion that the MENs and NENs are functionally equivalent (though one could argue in response that increased fragility of NENs is why they are progressively replaced by MENs).

There are many additional questions about the single cell analysis that are difficult to resolve with the data in hand. I think everyone would agree that an ideal analysis would have more cells, deeper sequencing, and comprehensive validation of the identity of each cluster of cells. But given the time and expense required to carry out such experiments, we cannot demand them, and must take the data for what they are rather than what they could be. And in the end, it is the editors' view that these data and analyses bolster the authors' claims, without conclusively establishing them. That is, these data should neither be dismissed nor, on their own, considered definitive.

* Are there alternative explanations for the data than that they are mesodermally derived neurons? *

As discussed above, the reviewers generally agree that the lineage tracing experiments are careful and well-executed, and the authors have provided data that demonstrates that the data are highly unlikely to be due to either incomplete or ectopic lineage marking. The reviewers raise several possible alternative hypotheses, some based on the literature and some based on the genomic data. The authors discuss each in detail in their response. The editor would note that, at this stage in the history of single-cell analysis, the criteria for using single cell sequencing data to establish cell type and cell origin is are not well established, and that neither the presence nor absence of specific sets of genes in single cells should not, for both technical and biological reasons, be considered dispositive as to identity.

* Additional aspects of paper: *

There are additional intriguing aspects of the paper, especially the increase in the number of MENs relative to NENs over time, suggesting functional replacement of one population with the other, and some evidence for and speculation about what might be regulating this evolution. However these are somewhat secondary points relative to the central question at hand of whether the authors have discovered a population of mesodermally derived neurons.

* Editor's summary and comment: *

The editor believes it is a fair summary to say that the authors believe they have gone to great lengths to provide multiple lines of evidence that support their hypothesis, but that these reviewers, while appreciating the potential importance of the authors' discovery of an unusual cell type, are not yet convinced of its origin.

In an ideal world, the authors, reviewers and editor would all ultimately agree on what claims the data presented in a paper supports, and indeed this is what the traditional journal publishing system tries to achieve. But the system fails in cases like this where no consensus between authors and reviewers can be reached, as it neither makes sense to "accept" the paper and imply that it has been endorsed by the reviewers, nor to "reject" it and keep the work in peer review limbo.

There is certainly enough here to warrant the idea and the data and arguments behind it being digested and considered by people in the field. It may very well be that the authors - who have spent years working on this problem and likely know more about this population of cells than anyone on Earth - are right that they have discovered something that changes how we think about the development of the nervous system. To the extent the reviewers are representative, people are likely to need additional data to be convinced. But it is time to put that to the test.

Reviewer #3 (Public Review):

This manuscript aims to exploit experimental measurements of the extracellular voltages produced by colliding action potentials to adjust a simplified model of action potential propagation that is then used to predict the extracellular fields at axon terminals. The overall rationale is that when solving the cable equation (which forms the substrate for models of action potential propagation in axons), the solution for a cable with a closed end can be obtained by a technique of superposition: a spatially reflected solution is added to that for an infinite cable and this ensures by symmetry that no axial current flows at the closed boundary. By this method, the authors calculate the expected extracellular fields for axon terminals in different situations. These fields are of potential interest because, according to the authors, their magnitude can be larger than that of a propagating action potential and may be involved in ephaptic signalling. The authors perform direct measurements of colliding action potentials, in the earthworm giant axon, to parameterise and test their model.

Although simplified models can be useful and the trick of exploiting the collision condition is interesting, I believe there are several significant problems with the rationale, presentation, and application, such that the validity and potential utility of the approach is not established.

Simplified model vs. Hogdkin and Huxley<br /> The authors employ a simplified model that incorporates a two-state membrane (in essence resting and excited states) and adds a recovery mechanism. This generates a propagating wave of excitation and key observables such as propagation speed and action potential width (in space) can be adjusted using a small number of parameters. However, even if a Hodgkin-Huxley model does contain a much larger number of parameters that may be less easy to adjust directly, the basic formalism is known to be accurate and typical modifications of the kinetic parameters are very well understood, even if no direct characterisations already exist or cannot be obtained. I am therefore unconvinced by the utility of abandoning the Hodgkin-Huxley version.

In several places in the manuscript, the simplified model fits the data well whereas the Hodgkin-Huxley model deviates strongly (e.g. Fig. 3CD). This is unsatisfying because it seems unlikely that the phenomenon could not be modelled accurately using the HH formulation. If the authors really wish to assert that it is "not suitable to predict the effects caused by AP [collision]" (p9) they need to provide a good deal more analysis to establish the mechanism of failure.

(In)applicability of the superposition principle<br /> The reflecting boundary at the terminal is implemented using the symmetry of the collision of action potentials. However, at a closed cable there is no reflecting boundary in the extracellular space and this implied assumption is particularly inappropriate where the extracellular field is one objective of the modelling, as here. I believe this assumption is not problematic for the calculation of the intracellular voltage, because extracellular voltage gradients can usually be neglected, but the authors need to explain how the issue was dealt with for the calculation of the extracellular fields of terminals. I assume they were calculated from the membrane currents of one-half of the collision solution, but this does not seem to be explained. It might be worth showing a spatial profile of the calculated field.

Missing demonstrations<br /> Central analytical results are stated rather brusquely, notably equations (3) and (4) and the relation between them. These merit an expanded explanation at the least. A better explanation of the need for the collision measurements in parameterising the models should also be provided.

Adjusted parameters<br /> I am uncomfortable that the parameters adjusted to fit the model are the membrane capacitance and intracellular resistance. These have a physical reality and could easily be measured or estimated quite accurately. With a variation of more than 20-fold reported between the different models in Appendix 2 we can be sure that some of the models are based upon quite unrealistic physical assumptions, which in turn undermines confidence in their generality.

p8 the values of both the extracellular (100 Ohm m) and intracellular resistivity (1 Ohm m) appear to be in error, especially the former.

(In)applicability to axon terminals<br /> The rationale of the application of the collision formalism to axon terminals is somewhat undermined by the fact that they tend not to be excitable. There is experimental evidence for this in the Calyx of Held and the cerebellar pinceau. The solution found via collision is therefore not directly applicable in these cases.

Comparison with experimental data<br /> More effort should be made to compare the modelling with the extracellular terminal fields that have been reported in the literature.

Choice of term "annihilation"<br /> The term annihilation does not seem wholly appropriate to me. The dictionary definitions are something along the lines of complete destruction by an external force or mutual destruction, for example of an electron and a positron. I don't think either applies exactly here. I suggest retaining the notion of collision which is well understood in this context.

Author Response

We thank the Editor and the Reviewers for the kind words, the helpful suggestions, and the points of critique, which have all helped us substantially strengthen the manuscript. We have made the aesthetic changes requested by Reviewer 2.

Response to Reviewer 2

We thank the Reviewer for their thorough feedback. We provide point by point responses below.

Concern 1

In paragraph 4.2, I found it unclear why the authors find it unsurprising that different experiments would correspond to different betas. I think that this point should be discussed, as beta and N appear in combination in determining the interaction strength. Otherwise, they could try to fit all distributions with the same beta, which would be more natural for me. I guess that the fits would be anyway good to the eye, though quantitatively suboptimal (which could be quantified with the distance introduced).

The reviewer raises valid concerns since as shown in Fig 3, the chosen values for beta, the additional fitting parameter introduced in the agent-based simulation, are: β = 0.18, 0.13, 0.12 and 0.64 respectively for N = 5, 10, 15, 20. We (RS, OM, and OP) find it intriguing that the optimum beta clusters around similar values for N = 5, 10, 15, while the optimum beta for N = 20 is significantly different. We acknowledge that we do not have an explanation why the fitted parameters values are what they are but note that the fitting curve is flat, implying that several beta values could possibly achieve a satisfactory fit. While further agent-based simulations could explore these findings more systematically, we believe that investigating this matter is outside the scope of this paper. Instead, we have acknowledged these points explicitly in the revised discussions.

Portion added to discussions: “As shown in Fig. 3, the chosen values for beta, the additional fitting parameter introduced in the agent-based simulation, are: β = 0.18, 0.13, 0.12 and 0.64 respectively for N = 5, 10, 15, 20. Perhaps it is intriguing that the optimum beta clusters around similar values for N = 5, 10, 15, while the optimum beta for N = 20 is significantly different. While we do not currently have an explanation for why the fitted parameter values are what they are, we note that the fitting curve is flat, implying that several beta values could possibly achieve a satisfactory fit. Further agent-based simulations could explore these findings more systematically, and provide useful insights.”

Concern 2

Citation of previous work on dynamical quorum sensing (lines 51 & 52) I think misses two important points: first these works (and others following them) deal with the appearance of collective oscillations at high density (therefore, the same general problem addressed here); second, Taylor et al. studied also a transition where the oscillators involved did not oscillate at low density, whereas above a density threshold, they display coherent collective oscillations whose period decreases with density - similar to what observed here. I do not think this takes anything away from the originality of this work, which refers to a different system, and models it with different equations, but the parallelism between integrate-and-fire dynamics with quenched noise and excitable dynamics in the presence of noise should in my opinion not be overlooked.

We have explicitly mentioned this in the revised text.

Concern 3

As the authors stress in lines 105 and 132, the analytical model shows that all that really matters in this phenomenon is the fastest frequency of the system. This could be used as an argument to say that the actual frequency distribution of individual fireflies is not all that important, as long as their fastest frequency is comparable. The assumption that they are identical would then sound less radical. Ideally, one could use the numerical simulations to check this, as well as the fact that the phenomenon does not break down when the shortest individual interburst interval Tbmin is narrowly distributed (which could also explain why having a few individuals who can flash at a higher frequency does not affect the outcome).

We thank the reviewer for these observations.

Concern 4

I still feel that the agreement between the model and observations is a bit overstated (line 120). At least, I think the authors may stress that whereas the model predicts that the frequency of the 7-14 minutes oscillations should increase a lot with N, this is not observed in the data. Maybe this mismatch would be reduced if inter-individual variability was added.

Please see the last three paragraphs of the discussion section. In reality, as the swarm size increases, we expect that swarms will no longer be all-to-all connected, and the dynamics of the system will depend upon the speed of propagation of information across the swarm. Precisely how this happens is outside of the scope of the current experimental work and theoretical description presented here.

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 #3

Evidence, reproducibility and clarity

In the manuscript entitled "Aurora A mediated new phosphorylation of RAD51 is observed in Nuclear Speckles", the authors unveil the Serine S97 as a novel phosphorylation site of the RAD51 recombinase and that this phosphorylation is mediated by the Aurora A kinase using a set of in vitro and in cellulo experiments. The authors also describe this phosphorylation being in the nucleus specifically in nuclear speckles where mRNA maturation and splicing occurs suggesting a role of RAD51 in the latter. The confocal microscopy images provided to test this hypothesis are convincing. However, using confocal images as well, the authors claim that RAD51 phosphorylated at S97 foci do not colocalize with the DNA damage marker -H2AX, hence a function not related to DNA damage, however the data provided does not fully support this statement. In this study, Alaouid et al, utilize mutants of RAD51 that alter S97 phosphorylation to further study its function and provide data that support RAD51 as an RNA binding protein. Overall, the manuscript shows some interesting observations that are worth pursuing however the in vitro and in cellulo results are not aligned, lack some controls, and many points should be reconsidered.

Major comments:

• Are the key conclusions convincing?

Not as stated.

Fig. 1A. The authors conclude that pS97-RAD51 favors RAD51 strand invasion capacity using the D-loop assay. Indeed, the S97D phosphomimic increased the D-loop activity 2.5-fold compared to WT RAD51. However, the S97A mutant, which is the non-phosphorylated form also increased the D-loop activity by 2-fold compared to WT (figure 1C). So, the phosphorylation or the absence of it seem to promote strand invasion. So, what is the role of the phosphorylation? There is no discussion about this. Besides, no representative image of the D-loop assay is shown, this is very important as these experiments need to be run with the relevant controls to be meaningful.

Fig. 1D. The polymerization rate of RAD51 is probably irrelevant for its function in the absence of DNA. What do they want to get at with this assay?

In figure 2B, the authors conclude that RAD51 phosphorylation at S97 is dynamically regulated throughout the cell cycle. Indeed, the pS97-RAD51 is well observed in asynchronous cells, and the double thymidine block time course experiment followed by PI staining shows the oscillation of the pS97-RAD51 from G1 to G2/M stage. The authors quantified the ratio of pS97-RAD51/total RAD51 to conclude this. However, it would be more accurate to also divide the above over the intensity of the loading control (tubulin) because in figure 3A for example, they quantified the ratio of pS97-RAD51/tubulin but did not consider the levels of RAD51 in their quantifications.

In figure 3B, the authors state that pS97-RAD51 is decreased after CPT treatment and that the pS97-RAD51 foci do not localize with the DNA damage marker -H2AX. The signal of gH2AX is already weird as it does not change from Ctrl to CPT conditions (especially in HCC1806 cells). A pre-extraction of soluble protein with CSK should be used to then look at the co-localization, with the pan-staining of the two signals is difficult to draw any conclusions of colocalization. Nevertheless, the signal of RAD51 seems equal in all conditions in the images shown and it does not seem to be reduced after CPT.

In figure 4A, the authors show that Aurora A is responsible for the S97-RAD51 phosphorylation in cellulo. Indeed, the use of an Aurora A inhibitor reduces the pS97-RAD51 signal, however, this is only true in one cell line (HCC1806) but no effect was observed in HeLa cells. Is this effect cell-specific?

The authors find that RAD51 binds both DNA and RNA and measure the affinities of the RAD51 bearing the S97D and S97A mutations. S97D shows the highest affinity for ssDNA and RNA in Fig. 7A, B, however the opposite is true for dsDNA in Fig 7C, D. All three forms of RAD51 bind RNA although with different affinities however no error bars are shown. The description of the results does not seem accurate. Importantly, these data should somehow correlate/be discussed with respect to the D-loop assay performed in Fig. 1. The authors conclude that the binding to RNA is reduced in S97D-RAD51 suggesting that the pRAD51 that they observe at nuclear speckles would be probably not associated with RNA at these nuclear speckles, right? this goes against their idea of this phosphorylated form being related to RNA splicing... - Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether?

The manuscript seems to be in early days and requires lots of editing, rewriting to relate the in vitro and in cell data and make a coherent story - Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation.

The authors performed chromatin fractionation to determine the correct localization of the pS97-RAD51 and looked for the phosphorylated form by western blots. But then they confirmed the finding using immunofluorescence. I think it would be more convincing and consistent if the authors do a pre-extraction before the use the antibody because as such, they would be indeed confirming the localization of the protein they are looking at that is specifically in the nucleus.

As well, in order to test the specificity of the pS97-RAD51 antibody they generated, a simple treatment of the lysates with phosphatases would be a good control for the specificity of their antibody These and the critics mentioned above need to be address. - Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments.

This manuscript is not ready for submission - Are the data and the methods presented in such a way that they can be reproduced?

Yes. However, the legends of the images are way too concise. - Are the experiments adequately replicated and statistical analysis adequate?

In Fig. 2B, the authors performed a double thymidine block followed by a time course release to track cell cycle progression of the cells and phosphorylation of RAD51 at S97. They do not indicate the biological replicates they performed. There are no error bars in the estimated KD shown in Fig.7.

Minor comments:

• Specific experimental issues that are easily addressable.

The authors conclude that the S97 is specifically phosphorylated by the Aurora A kinase. How? Have they looked at other documented kinases known to phosphorylate RAD51?

In figure 6 the authors overexpress HA-tagged RAD51 proteins corresponding to WT, S97D and S97A mutants in cells and label them for immunofluorescence. Maybe it would be better to downregulate the endogenous RAD51 to discard possible combined effects.

In figure - Are prior studies referenced appropriately?

The authors show in their manuscript that RAD51 protein CAN interact with RNA in vitro, a finding not previously described to my knowledge. However, a recent study entitled "RAD51-dependent recruitment of TERRA lncRNA to telomeres through R-loops, Nature, 2020" provides in vitro data showing the binding of RAD51 to TERRA, a LncRNA, which I think would be worth mentioning their manuscript.

The authors should mention previous contributions in the field especially when it comes to RAD51 in the HR pathway post DNA damage, which is quite documented and updated. For example, in this section of the introduction, "RAD51 is a recombinase protein implicated in the strand exchange mechanism during the DSB repair by the Homologous Recombination (HR) pathway. In the absence of DNA Damage (DD), RAD51 is predominantly cytoplasmic and translocates to the nucleus during the DNA Damage Response (DDR) to manage HR repair. As it needs the undamaged sister chromatid as a template, the HR repair pathway occurs mainly in the late S, G2 phases of the cell cycle. However, it has been documented that HR repair can also occur during G1 and early S phases, and in this case, the undamaged template used for the repair could be the homologous chromosome or an RNA transcript2". This statement is definitely worth more references.

The same problem is recurrent in the rest of the introduction; therefore, it needs to be updated and better referenced. - Are the text and figures clear and accurate?

The text needs a lot of editing to accurately describe the results, see for example: "The resulting KD evaluation shows that the S97D mutant had a dsDNA binding affinity lower to that of the WT (a KD of 2.26 μM for the S97D-RAD51 vs a KD of 0.38 μM for the WT RAD51). Concerning, the S97A mutant comparison to the WT RAD51, we observed modified association and dissociation curves that resulted in an identical affinity to dsDNA (a KD of 0.33 μM for the S97A-RAD51 vs a KD of 0.38 μM for the WT RAD51). We can conclude that in our in vitro conditions, the Ser97 phosphorylation has a high impact on RAD51 affinity for DNA by dividing its affinity by 5.8." Besides, the figures are of low quality and should be more carefully crafted and presented. Some experiments (such as the D-loop) are not represented in the figures.<br /> - Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

Using a different representation for the graphs would be a plus (also see previous comments)

Referees cross-commenting

I think the other reviewers and I have raised very important and complementary points that will help the authors improve the quality of the manuscript substantially.

Significance

• Describe the nature and significance of the advance (e.g. conceptual, technical, clinical) for the field.

The discovery of a new phosphorylation site in RAD51 (S97) by Aurora A is potentially interesting for the field of the maintenance of genome stability as it could broaden the understanding of how such an important recombinase may be regulating the maintenance of genome integrity throughout the cell cycle. Also, the idea of RAD51 being involved in splicing and mRNA maturation seems very attractive and a very important conceptual advance. However, given the premature status of the text and the figures, the manuscript falls short to show convincing evidence. - Place the work in the context of the existing literature (provide references, where appropriate).

Many works are highlighting the role of RNA binding proteins as an integral part of the DNA damage response. In addition, a wealth of evidence in the literature suggest that many DNA repair proteins are RNA binding proteins, and that RNA is an important player in the DDR. The possible finding that RAD51 interacts with RNA and localize to nuclear speckles possibly acting in splicing is very interesting and attractive. How is Aurora A involved in this, what is the trigger, and whether RAD51 is binding RNA at these sites is still unclear. - State what audience might be interested in and influenced by the reported findings.

Labs working in genome integrity mechanisms and the crosstalk between transcription and DNA repair would be interested. - Define your field of expertise with a few keywords to help the authors contextualize your point of view.

Genome Instability, homologous recombination

Author Response

The following is the authors’ response to the original reviews.

We would like to thank the Reviewers for their careful reading and the many thoughtful suggestions to improve our manuscript, as well as both the Editors and Reviewers for the generally positive evaluations and encouraging statements.

Editorial assessment:

This important work presents an interesting perspective for the generation and interpretation of phase precession in the hippocampal formation. Through numerical simula- tions and comparison to experiments, the study provides solid evidence for the role of the DG-CA3 loop in generating theta-time scale correlations and sequences, which would be reinforced through the clarification of the concepts introduced in the study, in particular the notion of intrinsic and extrinsic sequences. This study will be of interest for the hippocampus and neural coding fields.

We appreciate that our work has been considered important. In our revision we made a considerable effort to improve on the presentation of our results and the justification of our model assumptions. Particularly we aimed to clarify the meaning of intrinsic and extrinsic sequences by ad- ditional figure panels as well as fleshing out their definition via spike-timing correlations being independent or dependent on the direction of the running trajectory, respectively. To address all the requests, we added 3 new Fig- ures, multiple new Figure panels and simulated a new model variant.

Reviewer #1 in their public review assessed ”The manuscript has the potential to contribute to the way we interpret hippocampal temporal coding for navigation and memory.”

They criticized

• The findings generally relate to network models of phase precession (re- viewed in e.g., Maurer and McNaughton, 2007, Jaramillo and Kempter, 2017). An important drawback of these models with respect to explaining specific experimentally observed features of phase precession, is that they cannot straightforwardly explain phase precession upon first exposure onto a novel track. This is because, specific connectivity in network models may re- quire experience-dependent plasticity, which would not be possible upon first exposure. This is essential, given that the manuscript addresses the possible origin of phase precession in terms of network models and at minimum, this weakness should be discussed.

We agree with Reviewer # 1 (and also with Reviewer # 2, who brought up a similar point) that models based on recurrence struggle to ex- plain how the recurrent connectivity matrix should come about. While we feel that a full model of how the 2-d topology in the recurrent weights can be learned goes far beyond the scope of this paper (and to our knowledge has not been solved so far in any existing model), we added a new model variant (new Figure 6 and Supplementary Figure 1), which explains the ba- sic phenomenology of extrinsic and intrinsic sequences without the need of recurrent connections, only using feed-forward synaptic facilitation. Thus, assuming recurrent connection is not necessary for our main findings. How- ever, we would like to point out that this does not exclude the possibility that recurrent connections, if set up in an appropriate way, also contribute to phase precession and theta sequences.

• An important and perhaps essential component of the manuscript, is the distinction between extrinsic and intrinsic models. However, the main con- cepts on which this hinges, namely extrinsic and intrinsic sequences (and the related extrinsicity and intrinsicity) could be better explained and illustrated. Along these lines, the result suggested by the title, namely, hippocampal theta correlations, may be important yet incidental in light of the new concepts (e.g., extrinsicity, intrinsicity) and computational models (e.g., DG-CA3 recurrent loop) that are put forward.

We have added substantial new explanatory material to the figures, captions and text to more didactically introduce the concepts of in- trinsicity and extrinsicity. We have also completely rewritten the abstract and added a subtitle: ”extrinsic and intrinsic sequences”

• The study seems to put forward novel computational ideas related to neural coding. However, assessing novelty is challenging as this manuscript builds on previous work from the authors, including published (Leibold, 2020, Yiu et al., 2022) and unpublished (Ahmadi et al., 2022. bioRxiv) work. For example, the interpretation of intrinsic sequences in terms of landmarks had been introduced in Leibold, 2020.

We agree with the reviewer that this paper touches on many related ideas from previous papers (not only of our lab) and is supposed to tie loose ends. Thus, the novel contribution is a biologically plausible mechanistic model of how intrinsic sequences and 2-d place maps interact on the level of interconnected spiking neurons. Such a level of explanation has not yet been available in previous work. We have considerably extended the Discussion section in our revision detailing the bigger picture underlying this theory. Also our addition of the non-recurrent model variant (see above) adds considerable novelty, since it provides an account of phase precession and preplay in novel environments.

• The significance of the readout tempotron neuron could be expanded on. In particular, there is room for interpretation of the output signal of that neuron (e.g., what is the significance of other neurons downstream? Why is the rationale for this output to being theta-modulated?)

We have added an additional Figure 8 to better illustrate the inner workings of the tempotron. We also extended the discussion to better explain the potential use of the tempotron output (see above). In short, we consider the tempotron to signal a unique behaviorally important context that is independent of remapping induced by changes of sensory cues, which is a new prediction of the model. Since the context signal is resulting from DG loops it requires a stable code to also exits in the DG. Evidence for such long-term stability in DG has been found in Hainmu¨ller & Bartos (2018).

Reviewer #2 in their public review find ”this research topic to be both important and interesting” and appreciates ”the clarity of the paper.”, com- mending our ”efforts to integrate previous theories into their model and con- duct a systematic comparison”.

We are very happy about these positive remarks and sincerely would like to thank the reviewer!

Reviewer #1 made the following specific recommendations for changes:

The abstract is somewhat difficult to parse. I have identified some words and/or sections that could be improved.

• ’ ....inherently 1 dimensional’. This statement seems to be related to an a priori interpretation of the authors. On the other hand, if offline sequences are trivially 1 dimensional because they are sequences (i.e., they constitute a vector), then online sequences would be 1-dimensional as well. What is the key difference between offline and online? Is it the omnidirectional place fields in two dimensions? Perhaps more importantly, how relevant is this fact with respect to the main results of the manuscript, which concern ex- trinsic and intrinsic sequences?

We indeed meant that the sequences are trivially 1-dimensional. The main challenge that we would like to address in this paper is how a 2-d topology of place cells (and direction dependent theta sequences) and a 1-d sequence topology of intrinsic theta correlations and during (p)replay can be reconciled. We hope this has become clearer in the rewritten abstract.

• The language in lines 36-38 is overly technical. I suggest modifying the language, the language was less technical and more understandable in the body of the manuscript, which should be also reflected in the Abstract.

We would would like to apologize for making the abstract too technical. Also in response to Reviewer #2, we decided to rewrite the ab- stract entirely.

The authors use a mixture of conductance based models and Izhikevich neurons, presumably for the spiking generating mechanism. The conductance component can be readily interpreted in terms of the underlying biophysics. The Izhikhevich neuron model, however, is phenomenological. I suggest you address i) the rationale for using Izhikevich model, 2) its biophysical inter- pretation, 3) and its combination with conductance-based currents.

The reviewer is correct that spike generation is modelled using Izhikevich’s model whereas synaptic integration is included in a conductance- based manner. As suggested by the reviewer, we have added further expla- nation in the Methods part, explaining that the Izhikevich approach allows to adjust burst firing properties with only few parameters by efficiently em- ulating the bifurcation structure of spike generation in the full biophysical model (1&2) and otherwise has no effect on the integration of conductance- based synaptic currents in a subthreshold regime (3).

Line 126: when you say preferred angle, do you mean preferred (heading) direction? If so, please maintain consistency throughout.

We thank the reviewer for pointing out the inconsistency. We have added the word ”heading” throughout the manuscript whenever ap- propriate. To further improve the consistency, we have clarified the meanings of ”best” (or ”worst”) direction and reserved the use of it solely for cases when trajectory direction is compared with the preferred heading direction, namely, ”best” (”worst”) direction when trajectory is along (opposite) the preferred heading direction.

Line 174: When discussing cross-correlation, sometimes you mean a cross-correlation function between two place fields and sometimes to the his- togram of all such correlations? Please clarify.

We used histograms to empirically estimate the underlying cross-correlation function. For clarity, we have specified that it is a cross- correlation histogram in the revised manuscript whenever we refer to the empirical estimate.

Figure 3:

Understanding the difference between extrinsic and intrinsic sequences is fundamental for the manuscript. I suggest that in the section that refers to Figure 3 (or Figure 3 itself), you kindly provide an example depicting how extrinsic and intrinsic sequences can

1) coexist yet be distinctly identified

2) depend on trajectory

3) depend on DG input

By coexistence, we meant the heterogeneous population of ex- trinsic and intrinsic cell pairs and, hence, the extrinsic and intrinsic theta correlations, as shown in Figure 3J. To improve the clarity, we added the following sentence in the section that refers to Figure 3: ”In our simula- tion, extrinsically and intrinsically driven cell pairs are both present in the population (Figure 3J), indicating a coexistence of extrinsic and intrinsic sequences.”. To illustrate how extrinsic and intrinsic sequences depend on both tra- jectory and DG recurrence, we have also added annotations in Figure 3F to mark the extrinsic and intrinsic part of the sequence.

Moreover, the caption of Figure 3 refers to the directionality of the theta sequences. How does this again relate to the extrinsic/intrinsic distinction?

We hope the highlighting in panel F of Figure 3 has resolved this problem.

Figure 5:

• This is a crucial figure that should illustrate the differences between extrinsic and intrinsic sequences, as the figure caption suggests. Surprisingly, it is not at all clear where (i.e., in which panel) and how (i.e., methodologi- cally) should one distinguish one type of sequence from another. I suggest that at least one such panel is dedicated to illustrating the difference and/or detection of these sequences in time and/or from phase precession plots. Moreover, there is significant visual crowding that makes the interpretation challenging (e.g., insert a space between G and E)

We would like to apologize that in the previous version of the manuscript, we seemed to have evoked the impression that the difference between intrinsic and extrinsic sequences should be mainly illustrated in Figure 5. We hope that our revisions of Figures 1 and 3 have made it sufficiently clear to this point. The main purpose of Figure 5 was (and is) to illustrate how intrinsic sequences can lead to out-of-field firing. We have modified the figure caption (and text) accordingly. To address the visual crowding problem in Figure 5, we have inserted a space between panels and also removed repeated labels.

Tempotron neuron and Figure 6:

From the reviewer’s questions on Figure 6, we feel that our presentation caused considerable confusion about the motivation and inter- pretation of the tempotron simulations. We therefore rewrote parts of the associated text and Figure caption. We hope that the revised presentation clarifies the issues. We therefore only briefly respond to the reviewer’s points here, because we think they largely resulted from misunderstandings.

• Intuitively, and as the manuscript results suggest, late phases are asso- ciated to extrinsic mechanisms while early phases are associated to intrinsic. Why not construct a simpler classifier readout based on this fact? How does it compare to a tempotron?

Opposite to the reviewer’s comment, extrinsic mechanisms are visible at early phases (late in the field), intrinsic mechanisms at late phases (early in the field). In fact, what the tempotron does is learning to identify the intrinsic (late phase) part and to disregard the extrinsic (early phase) part.

• What is the significance of theta-modulated output of the tempotron (readout) neuron?

The theta modulation of the tempotron output is a trivial re- sult of the theta-modulation of the input, i.e., the detection of the intrinsic sequence pattern is done once every cycle.

Suggestion for Figure 6 related to Tempotron readout: Focus on ’with DG loop condition’, as the challenge and most important point here is to identify extrinsic and intrinsic sequences. The No-loop condition could be left as a supplementary figure or side panel.

The no-loop condition is the essential control showing that the tempotron only responds to the previously learned intrinsic pattern and can- not identify spatial location based on the extrinsic pattern.

Further work/predictions.

Lines 196-198. ”Since intrinsic sequences can also propagate outside the trajectory (Figure 5) and activate place cells non-locally, our model predicts direction-dependent expansion of place fields.” If remote activation is ’suffi- ciently’ remote, wouldn’t this predict two separate place fields instead of an expansion?

The reviewer is completely correct. Out of field spiking can be also affecting remote locations, if the intrinsic sequences link to remote place fields. This would lead to double fields, however, the intrinsic part would only be active at late theta phases. For simplicity, we have not added such a case in our paper, but we would like to thank the reviewer for this comment, since it leads to a nice prediction of the model, which can be experimentally tested and therefore was included to the discussion.

Lines 556-558. ”In our model, firing rate is determined by both low-phase spiking from sensory input and high-phase spike arrivals of DG-CA3 loops, both producing opposing effects on the phase distribution.” Is it possible to make a differential prediction based on lesions here, e.g., along the lines of reduced range phase precession, for either high phases or for low phases?

We thank the reviewer for this great suggestion. Lesion of DG in the model does indeed reduce the phase range and mean spike phase. This further corroborates the effect of DG-loop on theta compression and high-phase spiking. We have included a new panel D in Figure 4 and a corresponding mention in the result section.

Line 570. ”We speculate that the functional roles of intrinsic sequences may not be limited to spatial memories.”. Is there any relationship to re- play and/or sleep-dependent memory consolidation? Some speculation in the Discussion section would be welcome and appropriate.

We have added some further speculative ideas to the last section of the Discussion. We propose that replay and preplay reflects the intrinsic sequences that express the current expectation of the animal. We have not yet thought well enough about their relation to memory consolidation to phrase this in the manuscript, but would suggest that they could serve to signal multimodal context information to the neocortex where it can evoke retrieval of unimodal memory traces.

The description of the results, as stated in the public review, can be im- proved. A key component is the definition and identification of extrinsic and intrinsic sequences.

Some comments:

• I think that the words ’extrinsic’ and ’intrinsic’ are problematic as both types of sequences/models rely on external (spatial) input, hence both are in some sense ’extrinsic’. On the other hand, both are network mechanisms, thus in some sense ’intrinsic’, where the asymmetry is either programmed directly onto the weights or due to synaptic depression. To add to the con- fusion, ’intrinsic’ mechanisms very often refer to cellular mechanisms in neurophysiology. I kindly ask you to, ideally, reconsider the terminology, or at the very least, be very thorough and precise when describing the mech- anisms. For example, sometimes extrinsic (intrinsic) ’models’ are referred to, sometimes ’sequences’, sometimes ’factors’, sometimes ’pairs’, etc.

We understand and appreciate the reviewers argument, but would like to stick to the terminology, since it was already used in our prior publication. We have made considerable effort to improve the explanation and illustration of extrinsic vs. intrinsic pairs in the main text, Figure 1 and 3 to highlight our definition that is based on pair correlations: Extrin- sic pairs flip the correlation lag with reversal of running direction, intrinsic pairs don’t. This is simply a functional definition and should not be con- fused with potential microscopic mechanisms. One of those (DG-loops) is suggested in our paper.

• As discussed in the public review, network mechanisms may require experience-dependent plasticity and hence cannot easily explain phase pre- cession on the first pass. Please discuss why and/or how your model fits with this observation.

We agree that the two models under consideration both require the recurrent network be set up appropriately and there is no theory so far that would explain how. The reason we chose these two models is because they are well known in the community and relatively similar. We reasoned that comparison between an intrinsic model and an extrinsic model would make most sense if the two are a similar as possible. Nevertheless, we ex- tended the manuscript by a new set of simulations in which we do not use re- current CA3 connections and obtain phase precession solely be feed-forward synaptic facilitation (new Figure 6 and supplementary Figure S1). The new simulations show that the basic phenomenology can also be obtained with- out using recurrent CA3 connections, however, as expected when removing one mechanisms of phase precession, the range of phase range is somewhat reduced as compared to the full model.

Along a similar vein, phase precession in Figure 1E only has a range of pi/2, which is about half of the typical range of phase precession for single runs. This should be characterized as a weakness of the intrinsic model.

The precession range in spiking models is highly sensitive to a large number of parameters such that it is hard to make such definite claims (see also above response). In the original Tsodyks et al. 1996 paper the phase range went up to 270 degrees with a slightly different implementation to ours in terms of current vs. conductance-based synapses, an exponen- tial instead of a Gaussian recurrent weight function, and 1-d (original) vs 2-d (ours). We chose conductance-based synapses, and a Gaussian weight profile for better comparison with the Romani and Tsodyks (2015) model. In the original non-spiking implementation by Romani and Tsodyks (2015), the phase range was hardly 70 degrees. Our model implementation of the Romani and Tsodyks (2015) model fits the experimentally reported phase ranges of about 70 to 180 degrees in CA3 (Harris et al., 2001).

Lines 282-284: ”...since phase precession properties change in relation to running directions, nor are they solely intrinsic since reversal of correlation is still observed in most of the sequences (Huxter et al., 2008; Yiu et al., 2022).”. To which extent is this a consequence of the phase precession model (extrinsic vs intrinsic) or the fact that place fields are sometimes directional?

The reversal of sequences with reversed running direction is how we define extrinsic correlation. We hope our changes in relation to Figure 1 has clarified this point.

Figure 2: Is it i) directional input or ii) short-term facilitation that gives rise to lower phase? (or perhaps both?) Please clarify.

It’s both. This is now clarified in the revised version of the Re- sults sections related to Figure 2: higher depolarization always yields earlier phases in spiking models, however, pair correlations are not affected by ei- ther of the two mechanisms.

Line 320. ”...onset of phase precession”. Do you mean in CA3/CA1/DG?

Thank you for pointing this out. We have clarified that this statement refers to CA3.

Line 323. ”....at a different location”. Please add rationale why it has to be at a different location and a reference to the appropriate equation.

The sequence rationale as well as the equation number have been added.

Line 384. ” ... predicting that loss of DG inputs is compensated for by the increase of release probability in the spared afferent synapses from the MEC.”. It wasn’t clear whether this was a ’homeostasis prediction’, or and implementation in the model. Please clarify.

Since the model explained the experimental observations by implementing an increased probability of release, the model predicts that in animals with DG lesion the probability of release should be enhanced. We have modified the wording to avoid confusion.

Line 428 ”...and near future locations) is obvious, the potential role of the lesser expressed intrinsic sequence contributions is not straightforward.”. Similar to my comments above regarding terminology, please clarify what are both contributions and why are intrinsic sequences ’lesser expressed’.

We have rewritten this passage to avoid unclear wording.

Line 474. ”...we showed that the trajectory-independent sequences”. Do you mean ’intrinsic sequences’?

We thank the reviewer for careful reading! We have changed the wording ”intrinsic sequences” in the revision.

Line 482. ”...field pairs being extrinsic”. Please clarify, as the usage of extrinsic now refers to field pairs.

Thank you for pointing this out. We went through the whole manuscript and clarified the terms.

Line 245 (heading). Consider rewriting as ’Dependence of theta se- quences on heading directions’. Extrinsic and Intrinsic models have not yet been introduced.

Since the main purpose of the first Results section is to explain the difference between extrinsic and intrinsic sequences we kept these terms in the heading but modified it to ”Dependence of theta sequences on head- ing directions: Extrinsic and intrinsic sequences”. Additionally, we have put more emphasis on introducing the terms ”extrinsic” and ”intrinsic” in this section.

Figure 1.

• I suggest using the same font - C and D, and F and G are too close to each other, consider adding space. For example, the exponent, 10-2 makes reading cumbersome. Line 300. Phase tail means offset phase? Phase tail may be too informal. Line 325: DG loop. Do you mean CA3-DG projection?

We thank the reviewer for the suggestions. In the revised manuscript, we have ensured that the same font is used in all of the fig- ures. To improve the readability of Figure 1, we have added space between panels as suggested, removed repeated axis label and downsized the text ”10-2”. Furthermore, we have rewritten the referenced line without using the word ”tail”, and also, clarified the meaning of DG loop as the short form of CA3-DG projection.

Figure 4 caption: ”DG lesion reduces temporal correlations...”. It is more precise to say that the lesion reduces the slope of the fitted lag vs dis- tance. And how is this related to sequence compression?

In the paragraph referring to Figure 4, we have elaborated on the meaning of theta compression and its relation with the the lag-distance plot. However, we argue that ”reduces the slope of the fitted curve” is not comprehensive enough to express our summarized conclusion in a caption title. We have modified the wording to be ”DG lesion reduces theta compression”.

In addition, we have changed the slope unit to be radians per cm rather than radians per maximum pair distance, in conformity to unit standards.

General comment about terminology with regards to tuning and connec- tivity: it is not formally correct to compare connectivity with trajectories (e.g., lines 388-395, caption of Figure 5A, etc). Perhaps compare tuning to particular directions/preference or receptive field?

We have corrected the wording such that the direction of DG- loop projection is compared to the direction of trajectory.

Line 470. ’...fixed recursive loop.” Sentence is not clear, do you mean recurrent loops?

The reviewer is correct. We corrected the wording

Reviewer #2 had the following recommendations.

M1. The abstract focuses on the differences between online and offline hippocampal replays. However, the replay topic is not touched upon in the rest of the manuscript. I found this very confusing when I first read the pa- per. I suggest the authors reconsider the best way to approach the opening or at least discuss if and how their model would incorporate replay phenomena.

Also in response to reviewer #1 we have rewritten the abstract focusing on the problem of how to generate 2-d topology from 1-d sequences. In addition, also in response to Reviewer#1 we added a paragraph in the discussion detailing a hypothesis on how er think replay and intrinsic se- quences work together.

m2. On lines 89-91, the authors provide the selection of neuronal pa- rameters for excitatory pyramidal cells and inhibitory cells in the Izhikevich model. While the choice of model is reasonable, it would be helpful to clarify the source of these neuronal parameters, especially for readers who are not familiar with the model.

Again, also in response to reviewer # 1, we have added more motivation for the Izhikevich model.

M3. On lines 94-98, the model considers a 2D sheet of CA3 neurons. One of the most significant assumptions is that each 2x2 tile of place cells is considered a unit with four directional angles. What is the basis for this assumption? Is there any experimental result supporting this, or is it a completely artificial design for the model? This is important since the or- ganization of CA3 cells also affects the network architecture discussed later and impacts the realism of the model.

This comment is related to Reviewer #1’s concern on experience- dependent plasticity: How is this connectivity pattern established? We fully agree that this is an open problem for the Tsodyks et al.-type networks. The main reason for choosing them (as argued in our response to reviewer #1) is to have two published models, representing one type of sequence each, that are similar enough for comparison. In addition, we added new simulations (new Figure 6 and Supplementary Figure S1), showing that the basic phe- nomenology can also be obtained in a model without recurrent connections (see also response to Reviewer # 1)

m4. Similarly, on lines 111 and 140, the model uses 500 ms for the timescales of short facilitation and short-term synaptic depression. The choices of these two timescales are vital for producing directionality in extrin- sic and intrinsic sequences, yet their experimental sources are not clarified.

In the Methods section of the revised manuscript, we have in- cluded the sources of previous experimental data and modelling work to support our choice of the time constants.

M5. On line 126, the authors assume that the synaptic strengths be- tween CA3 cells, Wij, are given by the distances between neurons and the similarity between their directional preferences. While this assumption seems reasonable in the sensory cortex, I am unsure if this is also the case in the hippocampus, and the authors should clarify the basis for this assumption.

The distance dependence simply reflects the original Romani and Tsodyks 2015 model (see response to M3) and we share the concern of the reviewers. The increased connectivity for neurons with the same di- rectional preference was necessary to recover the direction dependent phase precession properties (Figure 2) in the realm of the Romani and Tsodyks 2015 model. Please also see our new Figure 6 showing simulations without the recurrent matrix.

More importantly, the existing connections within CA3 and DG cells completely determine the ”intrinsic” sequences. But wouldn’t this be fragile when place cells undergo global remapping, which can take place within only a few seconds? The author should comment on this in the discussion.

We would like to thank the reviewer for bringing up this inter- esting point. In our thinking, the DG-CA3 connectivity is fixed (multiple 1-d trajectories, not necessarily requiring 2-d topology), i.e., the same in- trinsic sequence should show up in multiple environments (and should not remap), although it may just not be active in some environments). This is a prediction of our model and we have added it to the Discussion.

M6. I found the setup of DG place cells unreasonable. DG place cells are found to be granule cells rather than pyramidal cells. Moreover, the model does not consider recurrent connections between DG cells (These setups are closer to CA1 place cells).

We agree with the reviewer, DG granule cells should rather be modelled as high-input resistance EIF neurons. However, the feedback loop via the dentate is not a direct one. It involves hilar mossy cells plus multiple hierarchies of feedback inhibition (this is probably what the reviewer means with recurrent connections between DG neurons, because granule cells are not recurrently connected in the non-pathological state). To our knowledge a biologically realistic model of the hilar-DG network does not exist and it would be far beyond the scope of this paper to develop one. We therefore see our DG feedback model rather as phenomenological. The discussion paragraph on the anatomy of the dentate gyrus touches on these points.

Therefore, a significant concern is: Why should it be the DG feedback projection to CA3 responsible for the ”intrinsic” sequences instead of pro- jections from other brain areas?

The reviewer is generally correct, any brain structure which im- plements fixed sequences via a loop would do. The reason why we suggest the DG to be the best candidate is purely empirical referring to papers with dentate lesions: Sasaki et al. 2018 and Ahmadi et a. 2022. We have added a similar argument to the discussion.

m7. On line 166, the authors claim that there are no connections between inhibitory cells at all. While I understand that this is for simplification of the model, the lack of recurrent inhibition between interneurons may have limited the model’s ability to produce gamma-band dynamics (referring to PING and ING mechanisms), which are robust rhythms produced in CA3. I am very curious if the model can incorporate theta-gamma coupling by in- troducing connections between CA3 inhibitory cells.

We have omitted the gamma oscillation for simplicity, because we do not have a hypothesis for a functional role in the context of dis- tinguishing extrinsic from intrinsic sequences (Occam’s razor) and, as the reviewer correctly anticipates, they unavoidably show up when inhibitory in- terneurons connect to each other (e.g. Thurley et al. 2013). Of course, one could envision situations in which gamma for intrinsic sequences my have different frequency than for extrinsic ones, by differentially manipulating the CA3 and DG basket cell networks, but, as long as there is no experimental data, it would be pure speculation and thus we have not included it in the model.

m8. The authors should clarify the source of parameters in Table 1, especially the synaptic strengths. These values are vital for extrinsic and intrinsic theta sequences.

The weight values have been chosen to allow for large theta phase precession range, coexistence of extrinsic and intrinsic sequences, and stability of the network activity. A similar statement has been added to the manuscript.

M9. I have another concern regarding the measurements of ”extrinsic- ity” and ”intrinsicity” defined on lines 185-196. Are they the best measures? To distinguish the cause of spike correlations, the ”extrinsicity” and ”intrin- sicity” of a pair of spikes should not be high at the same time. However, this is clearly not the case in the model, according to Figs 3 and 5. Moreover, in the data analysis carried out later, spike pairs are considered extrinsic or intrinsic merely by comparing the two measurements. I suggest the authors consider counterfactual methods in causal inference. For example, would a spike pair (cell1, cell2) still exist if we change the sensorimotor inputs or the DG-CA3 projections? If this is difficult to implement, the authors should at least discuss how different choices of measurements would impact the con- clusions of the paper.

The problem the reviewer has identified arises from the funda- mental symmetry of theta phase quantification: if spikes of a pair of place fields have a phase difference of 180◦ one cannot say which cell leads and which cell follows, hence, the phase difference is both intrinsic (because the peak doesn’t flip) and extrinsic (because the peak flips and ends up at the same phase). The fact that in some cases extrinsicity as well as intrinsicity are high simply means that the field pair has a correlation peak lag close to 180◦. Since in the experimental data set in (Yiu et al. 2022) only field pairs were available, we have not been able to use a different quantification then and decided to apply the same quantification in our model for comparison. Moreover, Figure 5F nicely shows that the measures are able to retrieve the ground-truth intrinsic DG-loop structure when considered on the population level.

In our model, though, we can go beyond 2-nd order statistics and derive sequence similarity measures including multiple cells, e.g., Chenani et al. 2019. However, since, we already know the ground truth by construction, we decided to not use these methods. We added a paragraph in the discus- sion elaborating on beyond 2nd order sequence quantification.

m10. The authors begin discussing ”intrinsic sequences” from line 316. However, it is not defined before that (and in the rest of the paper as well), causing confusion when reading the paper. The exact definitions of extrinsic and intrinsic sequences should come earlier.

We hope that our changes to the beginning of the results section (Figure 1), also asked for by Reviewer # 1 could clarify the confusion.

m11. On lines 345-347, the authors claim that ”the intrinsic sequences are played out backward as determined by the direction of fixed recurrence (Figure 3F),” which is vague. If such sequences are present in that panel, it should be more explicitly indicated graphically.

Also in response to Reviewer #1, we have graphically high- lighted the two types of sequences.

M12. On lines 309, 356, 484, 495, 515, and possibly other instances, the authors repeatedly claim that the model simulations are in ”quantitative agreement” with their previous experimental paper. However, no experimen- tal data or comparison with the simulations are presented in this paper. The authors should at least create one figure to demonstrate the degree of consistency between them, instead of merely asking the reader to refer back to their previous paper.

We agree with the reviewer that the experimental data of our previous paper should be presented in the manuscript. However, creating more panels or figures is likely to clutter the already crowded visuals and ob- scure our main message. We therefore decided to give numerical comparisons the previous findings in the main text whenever appropriate, specifically, in the sections referring to Figures 2, 3 and in the Discussion.

Author response

Reviewer #1 (Public Review):

This careful study reports the importance of Rab12 for Parkinson's disease associated LRRK2 kinase activity in cells. The authors carried out a targeted siRNA screen of Rab substrates and found lower pRab10 levels in cells depleted of Rab12. It has previously been reported that LLOMe treatment of cells breaks lysosomes and with time, leads to major activation of LRRK2 kinase. Here they show that LLOMe-induced kinase activation requires Rab12 and does not require Rab12 phosphorylation to show the effect.

We thank the reviewer for their comments regarding the carefulness and importance of our work and for their specific feedback which has substantially improved our revised manuscript.

1) Throughout the text, the authors claim that "Rab12 is required for LRRK2 dependent phosphorylation" (Page 4 line 78; Page 9 line 153; Page 22 line 421). This is not correct according to Figure 1 Figure Supp 1B - there is still pRab10. It is correct only in relation to the LLOMe activation. Please correct this error.

We appreciate the reviewer’s comment around the requirement of Rab12 for LRRK2-dependent phosphorylation of Rab10 and question regarding whether this is relevant under baseline conditions or only in relation to LLOMe activation. Using our MSD-based assay to quantify pT73 Rab10 levels under basal conditions, we observed a similar reduction in Rab10 phosphorylation when we knockdown Rab12 as we also observed with LRRK2 knockdown (Figure 1A). Further, we see comparable reduction in Rab10 phosphorylation in RAB12 KO cells as that observed in LRRK2 KO cells using our MSD-based assay (Figure 2A and B). Based on this data, we believe Rab12 is a key regulator of LRRK2 activation under basal conditions without additional lysosomal damage. However, as the reviewer noted, we do observe some residual Rab10 phosphorylation upon Rab12 knockdown when assessed by western blot analysis (Figure 1D and Figure 1- figure supplement 1). A similar signal is observed upon LRRK2 knockdown, which may suggest that some small amount of Rab10 phosphorylation may be mediated by another kinase in this cell model. Nevertheless, we appreciate this reviewer’s point and have therefore modified the text to remove any reference to Rab12 being required for LRRK2-dependent Rab phosphorylation and now instead refer to Rab12 as a regulator of LRRK2 activity.

As noted by the reviewer, our data does suggest that Rab12 is required for the increase in Rab10 phosphorylation observed following LLOMe treatment to elicit lysosomal damage, and we now refer to this appropriately throughout the text.

2) The authors conclude that Rab12 recruitment precedes that of LRRK2 but the rate of recruitment (slopes of curves in 3F and G) is actually faster for LRRK2 than for Rab12 with no proof that Rab12 is faster-please modify the text-it looks more like coordinated recruitment.

The reviewer raises an excellent point regarding our ability to delineate whether Rab12 recruitment precedes that of LRRK2 on lysosomes following LLOMe treatment. As noted by the reviewer, we do see both the recruitment of Rab12 and LRRK2 to lysosomes increase on a similar timescale, so we cannot truly resolve whether Rab12 recruitment precedes LRRK2 recruitment in our studies. Based on this, we have modified the text to emphasize that this data supports coordinated recruitment, as suggested, and we have further removed any mention of Rab12 preceding LRRK2. The specific change is as follows “Rab12 colocalization with LRRK2 increased over time following LLOMe treatment, supporting potential coordinated recruitment of these proteins to lysosomes upon damage (Figure 3I). Together, these data demonstrate that Rab12 and LRRK2 both associate with lysosomes following membrane rupture.” and can be found on lines 460-463 of the updated manuscript.

3) The title is misleading because the authors do not show that Rab12 promotes LRRK2 membrane association. This would require Rab12 to be sufficient to localize LRRK2 to a mislocalized Rab12. The authors DO show that Rab12 is needed for the massive LLOME activation at lysosomes. Please re-word the title.

To address the reviewer’s concern regarding the title of our manuscript, we have modified the title from “Rab12 regulates LRRK2 activity by promoting its localization to lysosomes” to “Rab12 regulates LRRK2 activity by facilitating its localization to lysosomes” to soften the language around the sufficiency of Rab12 in regulating the localization of LRRK2 to lysosomes. We show that Rab12 deletion significantly reduces LRRK2 activity (as assessed by Rab10 phosphorylation on lysosomes) and significantly increases the localization of LRRK2 to lysosomes upon lysosomal damage. The updated title better reflects the regulatory role of Rab12 in modulating LRRK2 activity, and we thank the reviewer for their suggestion to modify this accordingly.

Reviewer #2 (Public Review):

This study shows that rab12 has a role in the phosphorylation of rab10 by LRRK2. Many publications have previously focused on the phosphorylation targets of LRRK2 and the significance of many remains unclear, but the study of LRRK2 activation has mostly focused on the role of disease-associated mutations (in LRRK2 and VPS35) and rab29. The work is performed entirely in an alveolar lung cell line, limiting relevance for the nervous system. Nonetheless, the authors take advantage of this simplified system to explore the mechanism by which rab12 activates LRRK2. In general, the work is performed very carefully with appropriate controls, excluding trivial explanations for the results, but there are several serious problems with the experiments and in particular the interpretation.

We appreciate the reviewer’s comments regarding the rigor of our work and the potential impact of our studies to address a key unanswered question in the field regarding the mechanisms by which LRRK2 activation is mediated. Our studies focused on the A549 cell model given its high endogenous expression of LRRK2 and Rab10, and this cell line provided a simple system to investigate the mechanism and impact of Rab12-dependent regulation of LRRK2 activity. We agree with the reviewer that future studies are warranted to understand whether similar Rab12-dependent regulation of LRRK2 occurs in relevant CNS cell types.

First, the authors note that rab29 appears to have a smaller or no effect when knocked down in these cells. However, the quantitation (Fig1-S1A) shows a much less significant knockdown of rab29 than rab12, so it would be important to repeat this with better knockdown or preferably a KO (by CRISPR) before making this conclusion. And the relationship to rab29 is important, so if a better KD or KO shows an effect, it would be important to assess by knocking down rab12 in the rab29 KO background.

The reviewer raises a good point regarding the importance of confirming that loss of Rab29 has no effect on Rab10 phosphorylation. To address potential concerns about insufficient Rab29 knockdown, we measured the levels of pT73 Rab10 in RAB29 KO A549 cells by MSD-based analysis. RAB29 deletion had no effect on Rab10 phosphorylation, confirming findings from our RAB siRNA screen and the observations of Dario Alessi’s group reported previously (Kalogeropulou et al Biochem J 2020; PMID: 33135724). We have included this new data into our updated manuscript in Figure 1- figure supplement 1 and comment on it on page 6 in the updated Results section.

Secondly, the knockdown of rab12 generally has a strong effect on the phosphorylation of the LRRK2 substrate rab10 but I could not find an experiment that shows whether rab12 has any effect on the residual phosphorylation of rab10 in the LRRK2 KO. There is not much phosphorylation left in the absence of LRRK2 but maybe this depends on rab12 just as much as in cells with LRRK2 and rab12 is operating independently of LRRK2, either through a different kinase or simply by making rab10 more available for phosphorylation. The epistasis experiment is crucial to address this possibility. To establish the connection to LRRK2, it would also help to compare the effect of rab12 KD on the phosphorylation of selected rabs that do or do not depend on LRRK2.

The reviewer raises an interesting question regarding whether Rab12 can further reduce Rab10 phosphorylation independently of LRRK2. Using our quantitative MSD-based assay, we observe that pRab10 levels are at the lower limits of detection of the assay in LRRK2 KO A549 cells. Unfortunately, this means that we are unable to detect whether there might be any additional minor reduction in Rab10 phosphorylation with Rab12 knockdown in LRRK2 KO cells. We cannot rule out that Rab12 may play a LRRK2-independent role in regulating Rab10 phosphorylation in other cell lines, and future studies are warranted to explore whether Rab12 knockdown can further reduce Rab10 phosphorylation in other systems, including in CNS cells.

Regarding exploring the effects of RAB12 knockdown on the phosphorylation of other Rabs, we also assessed the impact of RAB12 KO on phosphorylation of another LRRK2-Rab substrate, Rab8a. We observed a strong reduction in pT72 Rab8a levels in RAB12 KO cells compared to wildtype cells, suggesting the impact of RAB12 deletion extends beyond Rab10 (see representative western blot in Author response image 1). Due to potential concerns with the selectivity of the pT72 Rab8a antibody (potentially detecting the phosphorylation of other LRRK2-Rabs), we cannot definitively demonstrate that Rab12 mediates the phosphorylation of other Rabs. This question should be revisited when additional phospho-Rab antibodies become available that enable us to selectively detect LRRK2-dependent phosphorylation of additional Rab substrates under endogenous expression conditions.

Author response image 1.

A strength of the work is the demonstration of p-rab10 recruitment to lysosomes by biochemistry and imaging. The demonstration that LRRK2 is required for this by biochemistry (Fig 4A) is very important but it would also be good to determine whether the requirement for LRRK2 extends to imaging. In support of a causal relationship, the authors also state that lysosomal accumulation of rab12 precedes LRRK2 but the data do not show this. Imaging with and without LRRK2 would provide more compelling evidence for a causative role.

We thank the reviewer for their suggestion to assess Rab12 recruitment to damaged lysosomes with and without LRRK2 using imaging-based analyses to add confidence to our findings from biochemical approaches. To address this comment, we have imaged the recruitment of mCherry-tagged Rab12 to lysosomes (as assessed using an antibody against endogenous LAMP1) and observed a significant increase in Rab12 levels on lysosomes following LLOMe treatment. This occurs to a similar extent in LRRK2 KO A549 cells, suggesting that Rab12 is an upstream regulator of LRRK2 activity. This new data has been incorporated into the revised manuscript (Figure 3E) and is presented on page 20 of the updated manuscript.

Our conclusions on this are further strengthened by new data assessing Rab12 recruitment to lysosomes using orthogonal analysis of isolated lysosomes biochemically. Using the Lyso-IP method, we observed a strong increase in the levels of Rab12 on lysosomes following LLOMe treatment that was maintained in LRRK2 KO cells. These data have been added to the updated manuscript (new data added to Figure 3- figure supplement 1).

Together, these data support our hypothesis that Rab12 recruitment to damaged lysosomes is upstream, and independent, of LRRK2.

The authors also touch base with PD mutations, showing that loss of rab12 reduces the phosphorylation of rab10. However, it is interesting that loss of rab12 has the same effect with R1441G LRRK2 and D620N VPS35 as it does in controls. This suggests that the effect of rab12 does not depend on the extent of LRRK2 activation. It is also surprising that R1441G LRRK2 does not increase p-rab10 phosphorylation (Fig 2G) as suggested in the literature and stated in the text.

We agree with the reviewer that it is quite interesting that RAB12 knockdown significantly attenuates Rab10 phosphorylation in the context of PD-linked variants in addition to that observed in wildtype cells basally and after LLOMe treatment. As noted by the reviewer, we did not observe increased levels of phospho-Rab10 in LRRK2 R1441G KI A549 cells at the whole cell level (Figure 2G). However, we observed a significant increase in Rab10 phosphorylation on isolated lysosomes from LRRK2 R1441G KI cells compared to WT cells (Figure 4B). This may suggest that the LRRK2 R1441G variant leads to a more modest increase in LRRK2 activity in this cell model. Previous studies in MEFs from LRRK2 R1441G KI mice or neutrophils from human subjects that carry the LRRK2 R1441G variant showed a 3-4 fold increase in Rab10 phosphorylation (Fan et al Acta Neuropathol 2021 PMID: 34125248 and Karaye et al Mol Cell Proteomics 2020 PMID: 32601174), supporting that this variant does lead to increased Rab10 phosphorylation and that the extent of LRRK2 activation may vary across different cell types.

Most important, the final figure suggests that PD-associated mutations in LRRK2 and VPS35 occlude the effect of lysosomal disruption on lysosomal recruitment of LRRK2 (Fig 4D) but do not impair the phosphorylation of rab10 also triggered by lysosomal disruption (4A-C). Phosphorylation of this target thus appears to be regulated independently of LRRK2 recruitment to the lysosome, suggesting another level of control (perhaps of kinase activity rather than localization) that has not been considered.

The reviewer suggests an interesting hypothesis around the existence of additional levels of control beyond the lysosomal levels of LRRK2 to lead to increased Rab10 phosphorylation of lysosomes. Given the variability we have observed in measuring endogenous LRRK2 levels on lysosomes, we performed two additional replicates to assess lysosomal LRRK2 levels in LRRK2 R1441G KI and VPS35 D620N KI cells at baseline and after treatment with LLOMe. We observed a significant increase in LRRK2 levels on lysosomes in cells expressing either PD-linked variant and a trend toward a further increase in the levels of LRRK2 on lysosomes after LLOMe treatment in these cells (Figure 4D in the updated manuscript). We have updated the text on page 24 to reflect this change, suggesting that the PD-linked variants do not fully occlude the effect of lysosomal disruption on the lysosomal recruitment of LRRK2.

LLOMe treatment leads to a stronger increase in Rab10 phosphorylation on lysosomes from LRRK2 R1441G and VPS35 D620N cells compared to the modest increase in LRRK2 levels observed. This could suggest that, as the reviewer noted, additional mechanisms beyond increased lysosomal localization of LRRK2 may be driving the robust increase in Rab10 phosphorylation observed. We have modified the results section on lines 548-551 to highlight this possibility: “Rab10 phosphorylation showed a more significant increase in response to LLOMe treatment than LRRK2 on lysosomes from LRRK2 R1441G and VPS35 D620N KI cells, suggesting that there may be more regulation beyond the enhanced proximity between LRRK2 and Rab that contribute to LRRK2 activation in response to lysosomal damage.”

Reviewer #3 (Public Review):

Increased LRRK2 kinase activity is known to confer Parkinson's disease risk. While much is known about disease-causing LRRK2 mutations that increase LRRK2 kinase activity, the normal cellular mechanisms of LRRK2 activation are less well understood. Rab GTPases are known to play a role in LRRK2 activation and to be substrates for the kinase activity of LRRK2. However, much of the data on Rabs in LRRK2 activation comes from over-expression studies and the contributions of endogenously expressed Rabs to LRRK2 activation are less clear. To address this problem, Bondar and colleagues tested the impact of systematically depleting candidate Rab GTPases on LRRK2 activity as measured by its ability to phosphorylate Rab10 in the human A549 type 2 pneumocyte cell line. This resulted in the identification of a major role for Rab12 in controlling LRRK2 activity towards Rab10 in this model system. Follow-up studies show that this role for Rab12 is of particular importance for the phosphorylation of Rab10 by LRRK2 at damaged lysosomes. Increases in LRRK2 activity in cells harboring disease-causing mutants of LRRK2 and VPS35 also depend (at least partially) on Rab12. Confidence in the role of Rab12 in supporting LRRK2 activity is strengthened by parallel experiments showing that either siRNA-mediated depletion of Rab12 or CRISPR-mediated Rab12 KO both have similar effects on LRRK2 activity. Collectively, these results demonstrate a novel role for Rab12 in supporting LRRK2 activation in A549 cells. It is likely that this effect is generalizable to other cell types. However, this remains to be established. It is also likely that lysosomes are the subcellular site where Rab12-dependent activation of LRRK2 occurs. Independent validation of these conclusions with additional experiments would strengthen this conclusion and help to address some concerns that much of the data supporting a lysosome localization for Rab12-dependent activation of LRRK2 comes from a single method (LysoIP). Furthermore, there is a discrepancy between panel 4A versus 4D in the effect of LLoMe-induced lysosome damage on LRRK2 recruitment to lysosomes that will need to be addressed to strengthen confidence in conclusions about lysosomes as sites of LRRK2 activation by Rab12.

We thank the reviewer for their comments regarding our work that identifies Rab12 as a novel regulator of LRRK2 activation and the appreciation of the parallel approaches we employed to add confidence in this effect.

As suggested by the reviewer, we have updated our manuscript to now include independent validation of our conclusions using imaging-based analyses to complement our data from biochemical analyses using the Lyso-IP method. Specifically, we have included new imaging data that confirms that Rab12 levels are increased on lysosomes following membrane permeabilization with LLOMe treatment and demonstrates that this occurs independent of LRRK2, providing additional support that Rab12 is an upstream regulator of LRRK2 activity (Figure 3E in the updated manuscript).

Regarding the reviewer’s comment on a discrepancy between our findings in Figure 4A and Figure 4D, we have performed additional independent replicates in Figure 4D to assess the impact of lysosomal damage on the lysosomal levels of LRRK2 at baseline or upon the expression of genetic variants. We observed a significant increase in LRRK2 levels on lysosomes following LLOMe treatment in our set of experiments included in Figure 4A and a non-significant trend toward an increase in LRRK2 levels on isolates lysosomes in Figure 4D. As described in more detail below (in response to the second point raised by this reviewer), we think this variability arises because of a combination of low levels of LRRK2 on lysosomes with endogenous expression and variability across experiments in the efficiency of lysosomal isolation. Our observations of increased recruitment of LRRK2 to lysosomes upon damage are further supported by parallel imaging-based studies (Figure 3F-I) and are consistent with previous studies using overexpression systems.

We thank the reviewer for all of the suggestions which have added further confidence to our conclusions and substantially improved the manuscript.

Author response

Reviewer #1 (Public Review):

The potential role of the CaMKII holoenzyme in synaptic information processing, storage, and spread has fascinated neuroscientists ever since it has been described that self-phosphorylation of CaMKII at T286 (pT286) can maintain the kinase in an activated state beyond the initial Ca2+ stimulus that induced kinase activation and pT286. The current study by Lučić et al utilizes biochemical and biophysical methods to re-examine two pT286 mechanisms and finds:

(1) that a previously proposed activation-induced subunit exchange within the holoenzyme can not provide pT286 maintenance or propagation; and

(2) that pT286 can occur not only within a holoenzyme but also between two holoenzymes, at least at sufficiently high concentrations.

For the observation regarding the subunit exchange, the authors go above and beyond to demonstrate that a previously proposed activation-induced subunit exchange does not actually occur in their hands and that the previous appearance of such a subunit exchange may instead be due to activation-induced interactions between the kinase domains of separate holoenzymes. This provides important clarification, as the imagination about the possible functions of this subunit exchange has been running wild in the literature.

By contrast, pT286 between holoenzymes at sufficiently high concentrations was largely predicted by the previously reported concentration-dependence of pT286 between monomeric truncated CaMKII (although these previous experiments did not rule out that such pT286 could have been excluded for intact full-length holoenzymes). Notably, the reaction rate reported here for pT286 between two holoenzymes is more than two orders of magnitude slower compared to the previously described rate of the pT286 reaction within a holoenzyme.

The only point on which we disagree (and we think it’s unarguable) is that the current consensus is that inter-holoenzyme phosphorylation simply doesn’t happen (whether or not monomers can phosphorylate each other). The reviewer is of course right that this view seems now less and less likely. We now performed new experiments to investigate this critical point further (see below).

The probable reason for the discrepancy in reported half-time of phosphorylation measured in earlier reports and in our paper is the fact that earlier reports (for example Bradshaw et al., 2002) measured autophosphorylation rate of wild-type CaMKII holoenzymes, at catalytically-competent enzyme concentrations of 0.1-5 µM. We are reporting the phosphorylation rate of 4 µM kinase-dead CaMKII, which is only a substrate, by 10 nM catalytically competent enzyme (CaMKII wild-type). There is up to 500 times less catalytically competent enzyme in our reactions, which is probably the reason why the reaction itself is several orders of magnitude slower.

In summary, this study contains two somewhat disparate parts: (1) one technical tour-de-force to provide evidence that argues against activation-induced subunit exchange, which was a tremendous effort that provides influential novel information, and (2) another set of experiments showing the somewhat predictable potential for pT286 between holoenzymes, but without indication for the functional relevance of this rather slow reaction. Unfortunately, in the current/initial title of the manuscript, the authors chose to emphasize the weaker part of their findings.

We agree with the reviewer that the title should be modified to emphasize both findings of our study. We also hope that our new experiments do bolster our findings with regard to pT286 between holoenzymes, as the reviewer puts it.

The seemingly slow inter-holoenzyme phosphorylation is only slow under conditions in which one of the proteins is kinase-dead. In situation in which all CaMKII holoenzymes are wild-type and therefore capable of performing phosphorylation (both intra- and inter-holoenzyme) the reaction rates for pT286 are expected to be orders of magnitudes faster, than those reported here for the phosphorylation of T286 on kinase-dead protein.

Reviewer #2 (Public Review):

This well-written manuscript provides a technical tour-de-force to provide a novel mechanism for sustaining CaMKII autophosphorylation through an interholoenzyme reaction mechanism the authors term inter-holoenzyme phosphorylation (IHP). The authors use molecular engineering to create designer molecules that permit detailed testing of the proposed interholoenzyme reaction mechanism. By catalytically inactivating one population of enzymes, they show using standard assays that the inactive enzyme can be phosphorylated by active holoenzymes. They go on to show that in cells, the inactive enzyme is phosphorylated only in the presence of co-expressed active CaMKII and that this does not appear to be due to active and inactive subunits mixing within the same holoenzyme. The authors suggest reasons for why previous experiments failed to expose IHP and in some experiments provide evidence that reproduces and then extends earlier studies. Some noted differences from earlier experiments are the reaction temperature, the time course of the reactions, and that significantly higher concentrations of the inactive (substrate) kinase in the present study amplify the IHP. These are plausible reasons for earlier studies not finding significant evidence for IHP and the presented data is well-controlled and of high quality.

The authors then take on the idea of subunit exchange employing multiple strategies. Using genetic expansion, they engineer an unnatural amino acid into the hub domain of the kinase (residue 384). In the presence of the photoactivatable crosslinker BZF and UV illumination, a ladder of subunits was generated indicating intraholoenzyme crosslinks were established. Using this cross-linked enzyme, presumably incapable of subunit exchange, the authors show significant phosphorylation of the kinase-dead mutant. This further supports that IHP is the cause of phosphorylation and not subunit exchange. Extending these experiments, they could not find evidence when CaMKIIF394BZF was mixed with the kinase-dead mutant and exposed to UV light, that there was evidence of the kinasedead subunits exchanged into CaMKIIF394 (active) enzymes.

Just a note, instead of residue 384, this should read 394.

With an entirely different approach, the authors use isotopic labeling of different pools of wt CaMKII (N14 or N15) followed by bifunctional cross-linking and mass spec to assess potential intra- and interholoenzyme contacts. Several interesting findings came of these studies detailed in Figure 4, mapped in detail in Figure 5, and extensively documented in supplementary tables. Critically, numerous crosslinks were found between different domains of the enzyme (catalytic, regulatory, hub) that are themselves a nice database of proximity measurements, but critical to the hypothesis, no heterotypic cross-links were found in the hub domains at any activated state or time point of incubation. This data supports two findings, that catalytic domains come into close proximity between holoenzymes when activated, supporting the potential for IHP, but that no subunit exchange occurs.

The authors then pursue the approach used originally to provide evidence of subunit mixing, single molecule-based fluorescence imaging. Using pools of CaMKII labeled with spectrally separable dyes, the authors reproduce the earlier findings (Stratton et al, 2016) showing that under activating conditions, but not basal conditions, colocalized spots were detected. Numerous controls were done that confirm the need for full activation (Ca2+/CaM + Mg2+/ATP) to visualize co-localized CaMKII holoenzymes. Extending these studies, the authors mix holoenzymes, fully activate them, and after sufficient time for subunit exchange (if it occurs), the reactions were quenched, and then samples were analyzed. The result was that no evidence of dual-colored holoenzymes was present; if subunits had mixed between holoenzymes, dual-colored spots should have been evident after quenching the reactions. This was not the case. Further, experiments repeated with pools of differentially labeled kinase dead enzymes produced no colocalization, as predicted, if activation of the catalytic domains is necessary to establish IHP.

Finally, the authors employ mass photometry to investigate the potential for interholoenzyme interactions. At basal conditions, only a mass peak consistent with CaMKII dodecamers was evident. Upon activation, a small fraction of dimeric complexes was evident (with Ca2+/CaM bound) but the majority of the peak was a dodecamer with 12 associated CaM molecules, and importantly, a significant fraction of a mass population was found consistent with a pair of holoenzymes with associated CaM. As an aside, the holoenzyme population appeared to be modestly destabilized as evidence of a minor fraction of dimers appeared as the authors diluted the enzyme, but the pools of holoenzyme and pairs of holoenzymes (with CaM) remained the dominant species when activated under all three enzyme concentrations assessed. Supporting the importance of activation for interactions between holoenzymes, the catalytically dead kinase even under activating conditions, shows no evidence of dimers of holoenzymes.

Each of the approaches is well-controlled, the data is of uniformly high quality, and the authors' interpretations are generally well-supported.

We are very grateful for these supportive comments.

Reviewer #3 (Public Review):

CaMKII is a multimeric kinase of great biologic interest due to its crucial roles in long-term memory, cardiac pacemaking, and fertilization. CaMKII subunits organize into holoenzymes comprised of 1214 subunits, adopting a donut-like, double-ringed structure. In this manuscript, Lucic et al challenge two models in the CaMKII field, which are somewhat related. The first is a longstanding topic in the field about whether the autophosphorylation of a crucial residue, Thr286, can be phosphorylated between intact holoenzymes (inter-holoenzyme phosphorylation). The second is a more recent biochemical finding, which tested the long-running theory that CaMKII exchanges subunits between holoenzymes to create mixed oligomers. These two models are connected by the idea that subunit exchange could facilitate phosphorylation between subunits of different holoenzymes by allowing subunits to integrate into a different holoenzyme and driving transphosphorylation within the CaMKII ring. Here, the authors attempt to show that one intact holoenzyme phosphorylates another intact holoenzyme at Thr286. The authors also provide evidence suggesting that subunit exchange is not occurring under their conditions, and therefore not driving this phosphorylation event. The authors propose a model where instead of exchanging subunits, two holoenzymes interact via their kinase domains to enable transphosphorylation at Thr286 without integrating into the holoenzyme structure. In order for the authors to successfully convince readers of all three facets of this new model, they need to provide evidence that 1) transphosphorylation at Thr286 happens when subunit exchange is blocked, 2) subunit exchange does not occur under their conditions, and 3) there are interactions between kinases of different holoenzymes that lead to productive autophosphorylation at Thr286.

Strengths:

The authors have designed and performed a battery of cleverly designed and orthogonal experiments to test these models. Using mutagenesis, they mixed a kinase-dead mutant with an active kinase to ask whether transphosphorylation occurs. They observe phosphorylation of the kinase-dead variant in this experiment, which indicates that the active kinase must have phosphorylated it. A few key questions arise here: 1) whether this phosphorylation occurred within a single CaMKII holoenzyme ring (which is the canonical mechanism for Thr286 phosphorylation), 2) whether the phosphorylation occurred between two separate holoenzyme rings, and 3) why was this not observed in previous literature? To address questions 1 and 2, the authors implemented an innovative strategy introducing a geneticallyencoded photocrosslinker in the oligomerization domain, which when crosslinked using UV light, should lock the holoenzyme in place. The rate of phosphorylation was the same when comparing uncrosslinked and crosslinked CaMKII variants, indicating that phosphorylation is occurring between holoenzymes, rather than through a subunit exchange mechanism that would require some type of disassembly and reassembly (presumably blocked by crosslinking). The 3rd question remains as to why this has not been previously observed, as it has not been for lack of effort. The authors mention low temperature and low concentration as culprits, however, Bradshaw et al, JBC v. 277, 2002 carry out a series of careful experiments that indicated that autophosphorylation at T286 is not concentration-dependent (meaning that the majority of phosphorylation occurs via intra-holoenzyme), and this is done over a concentration and temperature range. It is possible that due to the mutants used in the current manuscript, it allows for the different behavior of the kinase-dead domains, which will have an empty nucleotide-binding pocket. Further studies will need to elucidate these details, and importantly, understand what physiological conditions facilitate this mechanism.

We thank the reviewer for their assessment of our work.

The paper cited by the reviewer (Bradshaw et al, JBC v. 277, 2002) is indeed a carefully designed biochemical investigation of CaMKII activity. As the reviewer pointed out, one of the conclusions of the paper is that the autophosphorylation of CaMKII is not concentration dependent, implying that it has to occur exclusively intra-holoenzyme. However, there are some limitations which colour the interpretation of this classic paper. Bradshaw and colleagues used only CaMKII wild-type protein, so the autophosphorylation which is taking place in their reactions is possible both within holoenzymes and between holoenzymes, but this is impossible to distinguish. The authors of the cited paper then used “Autonomous activity assay” (not any measurement of pT286 on CaMKII itself) in which they first stopped the initial autophosphorylation reaction at T286 by adding a quench solution which contained a mixture of EDTA and EGTA, and then measured phosphorylation of the peptide-substrate of CaMKII (autocamtide-2), in the absence of Calmodulin binding (autonomous activity). They also diluted the autophosphorylation reaction to 10 nM CaMKII before adding it to the “Autonomous activity assay”.

As a side point, each reaction was quenched and diluted to the same final CaMKII concentration of 10 nM. They measured the activity of this dilution with phosphorylation of a peptide-substrate (autocamptide-2), in the absence of CaM binding. The authors contend that autonomous activity reported in this way reflects the amount of pT286, which is not impossible, but it is not a direct measure of pT286.

All this adds up to allowing the autophosphorylation of wild-type CaMKII at various concentrations ranging from 0.1 to 4.6 µM in the presence of 10 µM Ca/CaM and 500 µM Mg/ATP. This is a very fast reaction, concentrations of enzyme (CaMKII wild-type), activator (Ca/CaM) and ATP/Mg are all high at the beginning of the autophosphorylation reaction and would expect to allow for maximal autophosphorylation in very short times (seconds). Most importantly, this experiment does not exclude a inter-holoenzyme reaction slower than the intra-holoenzyme one. It certainly could not detect it.

In any case, to relate these concepts to our experiments and current understanding of CaMKII, we performed a new set of experiments modelled on the Bradshaw paper. Critically, we used CaMKII wild-type as the enzyme, and CaMKII kinase-dead, as the substrate. Intraholoenzyme phosphorylation cannot occur in this reaction, which was designed to detect a concentration-dependent phosphorylation reaction. We used a fixed concentration of the substrate kinase (4 µM), and 4 different concentrations of CaMKIIWT ranging from 0.5 -100 nM. In our assay, the level of phosphorylation on substrate CaMKII(CaMKIIKD) was dependent on concentration of enzyme CaMKII (CaMKIIWT) (Figure 1-figure supplement 3), adding more evidence to the hypothesis that CaMKII autophosphorylation can occur inter-holoenzyme.

The possibility that empty nucleotide binding pocket is influencing the phosphorylation status of T286 in the regulatory domain of kinase-dead CaMKII is highly unlikely. One could maybe envision that empty nucleotide binding pocket might expose the regulatory domain in kinase-dead CaMKII for phosphorylation, which would be prevented in CaMKIIWT, but in all available structures of CaMKII (Chao et al, 2011; Myers et al., 2017, Buonarati et al., 2021), the regulatory domain is docked to the kinase domain of CaMKII, although the nucleotide binding pocket is empty (either by mutation of residue K42 and/or simply by not adding the ATP/Mg to reduce chemical dispersity of the sample). The only time the regulatory domain was not docked on the kinase domain is when CaMKII was in complex with Calmodulin (Rellos et al., 2010). Finally, in our crosslinking mass spectrometry experiments, we used both heavy and light forms of CaMKII wild-type, and there we can clearly see interactions between kinase/regulatory domains of two different species of CaMKIIWT, which are dependent on activation.

The most convincing data that subunit exchange does not occur is from the crosslinking mass spectrometry experiment. The authors created mixtures of 'light' and 'heavy' CaMKII holoenzymes, either activated or not and then used a Lys-Lys crosslinker (DSS) to trap the enzyme in its final state. The results of this experiment indicate that subunit exchange is not occurring under their conditions. A caveat here is that there are not many lysines at hub-hub interfaces, which is the crux of this experiment. If there is no subunit exchange under their conditions, how does transphosphorylation occur between holoenzymes? The authors show very nice mass photometry data indicating that there are populations of 24-mers, which corresponds to a double-holoenzyme. Paired with the data from their crosslinking mass spectrometry which shows crosslinks between kinase domains of different holoenzymes, this indicates that perhaps kinases between holoenzymes do interact, and they do so in a competent manner to allow transphosphorylation to occur.

It is true that there are “only” 6 Lysines in the hub domain of CaMKII. However, it is clear from our crosslinking mass spectrometry data that we can detect hub:hub peptides coming from the same holoenzymes (homocrosslinks, either 14N: 14N or 15N: 15N species), but never between holoenzymes (14N with 15N). The fact that peptides can be detected in the homocrosslinks speaks to the validity of using Lysine crosslinkers in this experiment.

Weaknesses:

The authors should be commended for performing three orthogonal experiments to test whether CaMKII holoenzymes exchange subunits to form heterooligomers. However, there are technical issues that dampen the strength of the results shown here. For simplicity, let's consider that CaMKII holoenzymes are comprised of two stacked hexameric rings. It has been proposed that the stable unit of CaMKII assembly and perhaps also disassembly and subunit exchange is a vertical dimer unit (comprised of one subunit from each hexameric ring). In the UV crosslinking data shown in this paper, the authors have a significant number of monomers, some crosslinked dimers (of which there are two populations), and fewer higher-order oligomers. To effectively block subunit exchange, robust crosslinking into hexamers is necessary, which the authors have not done. Incomplete crosslinking results in smaller species that can still exchange (and/or dissociate), confounding the results of this experiment. In addition, Figure 3 shows a trapping experiment, where if the exchange was occurring, there would be an oligomeric band in Lane 8, which is visible and highlighted with a blue arrow by the authors. This result is explained by nonspecific UV effects, however by eye it is not clear if there is an equivalent band in lane 10. The overall issue here is inefficient crosslinking.

We agree with the reviewer that the robustness of the UV-induced crosslinking is not extremely high. However we do observe higher order oligomers on the gel (Figure 2 and Figure 3B, pT286 blot), which states that at least a portion of the holoenzymes is crosslinked. On the other hand, the UVinduced crosslinking is not slowing down the trans-phosphorylation reaction, which would be expected if the subunit exchange would be the prevailing mechanism for spread of kinase activity between holoenzymes.

In figure 3, lanes 8 and 10 show a small portion of dimers (less than 5% by densitometry), and at the absolute limit of detection. This dimer band is most likely due to unspecific UV-induced disulfide bridging (we already lessened it by adding 50 mM TCEP prior to UV treatment (Figure 3-figure supplement 1B and C). Previous reviewers of this manuscript criticized the small dimer band in lane 8, and we wanted to address this transparently in the submission to eLife.

Unfortunately, if we absolutely crank up the contrast to see this band in lane 10, we start to see other features in the noise as well. We have now edited the image in Figure 3B to highlight these minor bands more clearly, but this is also not ideal.

With regard to the trapping experiment, the overall problem is not inefficient crosslinking, because we see that P-T286 signal is quite nicely represented in higher order bands from F394BzF protein, but kinase dead protein (Avi-tagged signal in Figure 3) is almost entirely absent. Any crosslinking of Avitagged protein (possibly corresponding to subunit exchange) is a minor process at the limit of detection on WB.

Unfortunately we did not yet find any better crosslinking sites than the two we report (we have tried about 10). But the results we did obtain encouraged us to employ other techniques to probe subunit exchange (for example, the MS X-linking).

The authors also employ a single-molecule TIRF experiment to further interrogate subunit exchange. Upon inspection of the TIRF images, it is not clear that the authors are achieving single molecule resolution (there are evident overlapping and distorted particles). The analysis employed here is Pearson's correlation coefficient, which is not sufficient for single molecule analysis and would not account for particle overlap, particles that are too bright, and/or particles that are too dim. For example, an alternative explanation for the authors' results is that activation results in aggregation (high correlation), and subsequent EGTA treatment leads to dissociation at these low concentrations (low correlation). However, further experimentation and analysis are necessary.

In the manuscript we present raw images, not processed. As we wrote in the material and methods, we thresholded the images for further processing. All colocalization methods have drawbacks, but we found that our thresholding combined with the Pearson coefficient was highly reproducible. We did also look at Manders coefficients, but these are less straightforward to understand, whilst still giving in our hands the same answer. We agree, there are more experiments that can be done, with particular predictions based on our new mechanism. And we are doing them and will report them when they are ready.

At the risk of repeating ourselves, the reversible loss of overlap of the two labelled populations is the key result and cannot be explained by spurious dim or bright particles, or by a few overlapping profiles.

Taken together, the authors have provided important food for thought regarding inter-holoenzyme phosphorylation and subunit exchange. However, given the shortcomings discussed here, it remains unclear exactly what mechanisms are at play within and between CaMKII holoenzymes once activated.

We thank the reviewer for their critical assessment of our manuscript. We will continue to investigate the relevant points and refine the overall picture of CaMKII, to better clarify the mechanisms.

Author Response

Reviewer #1 (Public Review):

Sučević and Schapiro investigated a neurobiologically inspired model of human hippocampal structure and computation in category learning. In three separate simulations, the model (CHORSE) is presented with learning tasks defined by various category structures from prior work and evaluated for its ability to learn the category structure, generalize categorization to novel stimuli, and accurately recognize previously encountered stimuli. Although originally conceived of as a computational model of associative memory, C-HORSE is demonstrated to quite naturally account for human-like learning of the three category tasks. Notably, the authors characterize the mechanisms underlying the model's learning by way of additional simulations in which "lesions" to the model's monosynaptic pathway (MSP; direct connections between ERC and CA1) are contrasted with lesions to its trisynaptic pathway (TSP; pathway connecting ERCDG-CA3-CA1). These in silico lesions offer key insight into the computational principles underlying theorized hippocampal functions in category learning: whereas MSP provides incremental learning of shared features diagnostic to category membership that are important for category generalization, TSP learns item-specific information that drives recognition behaviour. The authors propose that C-HORSE's successful account of a broad set of category learning datasets provides clear support for the role of complementary hippocampal functions mediated by MSP and TSP in category learning. This work adds compelling computational evidence to a growing literature linking hippocampus to a broader role in cognition that extends beyond declarative memory.

The model simulations are clear and properly conducted. The three datasets examined offer a relatively broad set of findings from the category learning literature; that the models provide reasonable accounts of human performance in all three speaks to the model's generalizability. Overall, I find this work exciting and an important step in linking longstanding well-established formal learning theories of psychology with neurobiological mechanism. Several weaknesses dampen this excitement, each of which are detailed below:

1) C-HORSE is presented as a new entry into a rich field of formal computational models of category learning. As noted above, the datasets examined span a broad range of learning contexts and structures and the model's ability to account for learning behaviour is compelling. However, no other models are leveraged to perform a direct evaluation. In other words, CHORSE's predictions are compelling, but is it better than other competing models in the literature? To be clear, C-HORSE offers a novel alternative with its fundamental mechanisms originating from anatomical structure and connectivity. As such, a proof-of-concept showing that such a neurobiologically inspired framework can account for category learning behaviour is a worthwhile contribution in its own right and a clear strength of this paper. However, how to consider this model relative to existing theoretical frameworks is not well described in the manuscript.

We very much appreciate this point — see response to Editor summary point #3 above.

2) Relatedly, C-HORSE is evaluated in terms of qualitative fit to behaviour measures from prior studies and in all three simulations restricted to measure of end of learning performance. Again, an appeal to the proof-of-concept nature of the current work may provide an appropriate context for this paper. But, a hallmark of well-established category learning models (e.g., SUSTAIN, DIVA, EBRW, SEA, etc.) is their ability to account for both end of learning generalization (and in some cases, recognition) and behaviour throughout the learning process. C-HORSE does provide predictions of how learning unfolds over time, but how well this compares to human measures is not considered in the current manuscript. Such comparisons would strengthen the support for C-HORSE as a viable model of category learning and help position it in the busy field of related formal models.

We completely agree about the value of this, and we have added empirical timecourse data for comparison with all simulations, as described in response to Editor summary point #7, above.

3) A consistent finding across all three simulations is that the TSP provides item-specific encoding. Evidence for this can be inferred by contrasting categorization and recognition performance across the TSP- and MSP-only model variants. In the discussion, the authors draw a parallel between exemplar theories of category learning and the TSP, which is a compelling theoretical position. However, as noted by the authors, unlike exemplar theories, the TSP-only model was notably impaired at categorization. The author's suggestions for extensions to CHORSE that would enable better TSP-based categorization are interesting. But, I think it would be helpful to understand something about the nature of the representations being formed in the TSP-only model. For example, are they truly item-specific, are the shared category features simply lost to heightened encoding of item-unique features, are category members organized similarly to the intact model just with more variability, and so on. Characterizing the nature of these representations to understand the limitations of the TSP-only model seems important to understanding the representational dynamics of C-HORSE, but are not included in the current manuscript.

The RSA results, now included for Simulations 2 and 3 in addition to Simulation 1, provide the information needed to characterize the nature of the TSP representations. Generally speaking, they are truly item specific, meaning that each item is represented by its own distinct set of units. This is a demonstration of the classic pattern separation function of this pathway, taking similar inputs and projecting them to orthogonal populations of neurons. Simulation 1 is the clearest example of this, where there is virtually no similarity and very low variability in the item similarity structure in DG and CA3. The new Simulation 3 RSA shows us where the limit is to this pattern separation ability of the TSP, with highly typical items being represented by somewhat overlapping populations of neurons in DG and CA3. To the extent that the TSP can succeed in generalization, it seems to involve this pattern separation failure.

We have made these points more explicit in new discussion of the RSA results:

• Simulation 1: “In the initial response, there was no sensitivity at all to category structure in DG and CA3 — items were represented with distinct sets of units. This is a demonstration of the classic pattern separation function of the TSP, applied to this domain of category learning, where it is able to take overlapping inputs and project them to separate populations of units in DG and CA3.” • Simulation 3: “As in the prior simulations, DG and CA3 represented the items more distinctly than CA1, and settled activity after big-loop recurrence increased similarity, especially in CA1. This simulation was unique, however, in that DG and CA3 showed clear similarity structure for the prototype and highly prototypical items. There is a limit to the pattern separation abilities of the TSP, and these highly similar items exceeded that limit. This explains why, at high typicality levels, the TSP could be quite successful on its own in generalization (Figure 5e), and why it struggled with atypical feature recognition for these items (Figure 5f).”

4) In general, a detailed description that links model mechanisms and analyses to the learning constructs of interest for the different simulations is lacking. For example, RSA results for simulation 1 are contrasted for initial and settled representations, but what is meaningful about these two timepoints is not directly stated (moreover, what initial and settled response mean in terms of the current model is not explained). The authors do briefly suggest that differences between initial and settled representations may reflect encoding dynamics before and after bigloop recurrence, but this is not established as a key metric for evaluating the nature of the model representations. In general, more motivation is needed to understand what the chosen analyses reveal about the nature of the model's learning process and representations.

We have added more description of the motivation for our analyses. See response to Editor summary point #6 above.

5) I appreciate the comparison in the discussion to extant models of categorization. Certainly, the exemplar and prototype models are fixtures of the category learning literature and they somewhat align with the type of learning that TSP and MSP, respectively, provide. REMERGE and SUSTAIN are also briefly mentioned, but their discussion is limited which is unfortunate as they are actually more functionally equivalent to C-HORSE. I think, however, that the authors are missing an opportunity to discuss how C-HORSE offers a means for bridging levels of analysis to connect neurobiological mechanisms with these notably successful psychological models of category learning. Rather than framing C-HORSE as a competitor to existing models, it should be viewed as an account existing on a different level of analysis. In this sense, it complements existing approaches and potentially extends a theoretical olive branch between the psychology and neuroscience of category learning.

We love this point about bridging levels of analysis and have added it to our discussion of the model’s relationship to other models, see Editor summary point #3 above.

6) The discussion takes a broad perspective on covering evidence concerning hippocampal contributions to category learning. Although comprehensive, some sections are not well connected back to the main thrust of the paper. For example, a section on neuropsychological accounts of the hippocampus and category learning summarizes central aspects of this literature but is never reflected on through the lens of the current findings. I do think this prior work is relevant, especially since it a central theme of the hippocampus not being necessary for category/concept learning, but its connection back to the current study is not well argued. Similarly, the section on consolidation and sleep is relevant, but in its current form does not seem to fit with the rest of the paper.

We have implemented these suggestions through very significant revisions to the Discussion. We now better connect the sections to the main argument of the paper and made cuts throughout, including removing the section on consolidation and sleep.

Reviewer #2 (Public Review):

The authors present a model of the hippocampal region that incorporates both the (indirect) trisynaptic and (direct) mono-synaptic pathways from entorhinal cortex (EC) to CA1 - the former incorporating projections from EC to dentate gyrus (DG), DG to CA3, and CA3 to CA1, and exhibiting a higher learning rate. They demonstrate that exposing this network to stimuli consistent with standard empirical tests of category learning (e.g. where within-category exemplars share a set of common features) allows the network to reliably assign both novel and previously encountered stimuli to the correct category (e.g. the network can learn to classify stimuli and generalise this knowledge to new examples). They show that the tri-synaptic pathway (TSP) preferentially supports the encoding of individual exemplars (e.g. analogous to episodic memory) while the mono-synaptic pathway (MSP) preferentially supports category learning.

The manuscript is well written, the simulation details appear sound, and the results are clearly and accurately presented. This model builds on a long tradition of computational modelling of hippocampal contributions to human memory function, strongly grounded in anatomical and electrophysiology data from both rodents and humans, and is therefore able to link phenomena at the level of individual cells and circuits to emergent behaviour - a major strength of this, and similar, work. However, I have two major concerns relating to the relationship between these findings and previously published work by the same and other authors.

First, it is not clear to me - from the manuscript - whether these results represent a significant novel advance on previous publications from the same senior author. Figures 1 and 3D are almost identical to figures published in Schapiro et al. (2017) Phil Trans B, and the take-home message (that the MSP might support statistical learning) is the same. In brief, it seems that the authors have subjected an identical network to some new (but related) tasks and reached the same set of conclusions. I see no distinction between learning to extract 'statistical regularities' (in previous work) and learning 'the structure of new categories' (described here). As an aside, demonstrating that an autoencoder network can learn stimulus categories and generalise to new exemplars is also well established.

We appreciate the opportunity to better articulate the novelty and importance of applying the model to the domain of category learning. There are crucial differences between statistical learning and category learning that make these simulations nontrivial (it did not have to be the case that the results would replicate for these category learning paradigms), and, importantly, many of the insights in the current work are category-learning specific (e.g., the effects of atypical features, trade-offs between generalization and recognition of exemplar-specific features). On the other hand, we of course agree that there are principles in common between statistical learning and category learning that are leading to the consistent findings. We added new material to the Introduction to explain the importance of these new simulations in the domain of category learning, and the value we see in demonstrating convergence across domains. See response to Editor point #1 above.

Second, I have some concerns with the relationship between the properties of this hippocampal network model and well described properties of single cells in the rodent and human hippocampus. In particular, the CA1 units in this model (and to some extent, also the CA3 units) come to respond strongly to all exemplars from within each category (e.g. as shown in Figure 3D, bottom right panel). This appears to be at odds with the known properties of place and concept cells from the rodent and human hippocampus, respectively, which show little generalisation across related concepts (i.e. the Jennifer Aniston neuron does not fire in response to other actors from Friends, for example). If the emergent properties of this model are not consistent with existing data, then it is not a valid model.

We appreciate the opportunity to discuss connections to the physiology literature. See response to Editor summary point #2 above.

More generally, the authors are clear that this model is "a microcosm of [the] hippocampusneocortex relationship" and that the properties of the MSP "mirror those of neocortex". Why not assume that category learning is supported by an interaction between hippocampus and neocortex, then, as in the complementary learning systems (CLS) model? Aside from some correlational fMRI data and partial deficits in hippocampal amnesics - either of which could have a myriad of different explanations - what empirical data is better accounted for by this model than CLS? Put differently, what grounds are there for rejecting the CLS model? To some extent, this model appears to account for less empirical data than CLS, with the exception of a few recent neuroimaging studies (which are hard to interpret at the level of single cells)

This is an important point for us to clarify, so we very much appreciate this comment. The crucial issue with CLS that motivated the microcosm theory is that the neocortex in the CLS framework learns far too slowly to support the kind of category learning studied in these paradigms, which unfolds over the course of minutes or hours. The neocortex in CLS was proposed to learn novel structure across days, months, and years.

We have added the following to the Introduction:

• “Despite its analogous properties, the MSP is not redundant with neocortex in this framework: the MSP allows rapid structure learning, on the timescale of minutes to hours, whereas the neocortex learns more slowly, across days, months, and years. The learning rate in the MSP is intermediate between the TSP (which operates as rapidly as one shot) and neocortex. The proposal is thus that the MSP is crucial to the extent that structure must be learned rapidly.”

We also have this description in the Discussion:

• “The MSP in our model has properties similar to the neocortex in that framework, with relatively more overlapping representations and a relatively slower learning rate, allowing it to behave as a miniature semantic memory system. The TSP and MSP in our model are thus a microcosm of the broader Complementary Learning Systems dynamic, with the MSP playing the role of a rapid learner of novel semantics, relative to the slower learning of neocortex.”

Reviewer #3 (Public Review):

The current work aimed to determine how the hippocampus may be able to detect regularities across experiences and how such a mechanism may serve to support category learning and generalization. Rapid learning in the hippocampus is critical for episodic memory and encoding of individual episodes. However, the rapid binding of arbitrary associations and one-shot learning was long thought suboptimal for finding regularities across experiences to support generalization, which were instead ascribed to other, slower-learning memory systems. More recent work has started to highlight hippocampal role in generalization, renewing the question of how generalization can be accomplished alongside memory for episodic details within a single memory structure. The current paper offers a reconciliation, presenting a biologically-inspired model of the hippocampus that is able to learn categories alongside stimulus-specific information comparably to human performance. The results convincingly demonstrate how distinct pathways within the hippocampus may differentially serve these complementary memory functions, enabling the single structure to support both episodic memory and categorization.

Major strengths and contributions

The paper includes simulation of three distinct categorization tasks, with a clear explanation of the unique aspects of each task. The key results are consistent across tasks, lending further support to the main conclusions of the role of distinct hippocampal pathways in learning specific details vs. regularities. Together with prior work on how the same architecture can support statistical learning in other types of tasks, this work provides important evidence of the broad role of the hippocampus in rapid integration of related information to serve many forms of cognition.

Throughout the paper, the authors nicely explain in conceptual terms how the same underlying computations may serve all three categorization tasks as well as statistical learning and episodic inference tasks. Thus, the paper will be of broad interest, beyond researchers focused on modeling and/or categorization.

On a conceptual level, this work provides a fruitful framework for understanding hippocampal functions, representations and computations. It provides a highly plausible mechanistic explanation of how category learning and generalization can be accomplished in the hippocampus and how distinct types of representations may emerge in distinct hippocampal subfields. The framework can be used to derive new testable predictions, some of which the authors themselves introduced. It also provides new insights into how the outputs of different pathways influence each other, providing a more nuanced view of the division of labor and interactions between hippocampal subfields. For example, the big loop recurrence would eventually lead to category influences even on the initially sparse, pattern separated representations in the CA3, which is an idea consistent with empirical observations.

The presented computational model of the hippocampus is currently the most detailed and biologically plausible hippocampal model easily applicable in the area of cognitive neuroscience and behavioral simulations. The commonalities and differences with other related models (conceptual and computational) are well explained. Both the conceptual and technical descriptions of the model are exceptionally clear and detailed. The model is also publicly available for download for any researcher to use with their own task and data. All these aspects make it likely that other researchers may adopt the model in a wider range of tasks, stimulating new discoveries.

The autoencoder nature of the model and the use of categorization tasks meant that some measures of interest, like recognition of exemplar-specific information, could not be evaluated by direct reading of the output layer to compare with some label (like old/new). The authors however came up with clever ways how to evaluate recognition performance in each task that was sensible and highlighted the multiple ways how one may think about information contained in neural representations in each layer. This approach can also be utilized by others for evaluating item-specific and category information in activation patterns, for example in analyses of fMRI.

Finally, I thought the current paper and provided model may also serve as an excellent introduction to computational modeling for those new to this approach. The exceptional clarity of the conceptual and technical description of this model and the clear logic of how one may model a cognitive task and interpret results made this paper fairly accessible. Furthermore, the paper offered new insights and predictions based on analyzing the model's hidden layers, lesion performance, and/or noting some patterns of behavior unique to specific tasks. This was also instructive for highlighting the distinctive contributions that the computational modeling approach can have for furthering our understanding of cognition and the brain.

We are extremely appreciative of the value the Reviewer sees in this work.

Weaknesses

The paper's strengths far outnumbered the weaknesses, that are minor. For one, the selected categorization tasks nicely complemented each other, but only covered stimuli with discretevalue dimensions (features like color, shape, symbol, etc). The degree to which the results generalize (or not) to continuous-value stimuli and different category structures (for instance information-integration or rule-based in COVIS framework) is not clear. How the model could be adjusted for continuous-value stimuli was not specified.

We agree that the simulation of only discrete valued dimensions is a limitation. We chose to do this simply because it is easier to use discrete values in the model as currently implemented, but future work will certainly need to test whether the model can simulate the various paradigms that make use of continuous-valued dimensions. We have added an explicit acknowledgement of this issue in the Methods:

• “The inhibition simulates the action of inhibitory interneurons and is implemented using a set-point inhibitory current with k-winner-take-all dynamics (O’Reilly, Munakata, Frank, Hazy, & Contributors, 2014). All simulations involved tasks with discrete-valued dimensions, as these are more easily amenable to implementation across input/output units whose activity tends to become binarized as a result of these inhibition dynamics. It will be important for future work to extend to implementations of category learning tasks with continuous-valued dimensions.”

There is compelling evidence for the dissociation between different hippocampal pathways and subfields (CA1 vs. CA3) that the model is based on. As the authors noted, there is also compelling evidence for functional dissociations along the long hippocampal axis, with anterior portions more geared towards coarse, generalized representations while posterior towards more detailed, specific representations. The authors nicely pointed out that these proposals of withinhippocampus division of labor are less orthogonal than they may first appear, as there is greater proportion of CA1 in the anterior hippocampus. However, it is premature to imply that this resolves the CA1/CA3 vs. anterior/posterior question; the idea that existing anterior findings may be simply CA1 findings is currently only speculation. Furthermore, first studies indicating that anterior/posterior representational gradients may exist within each subfield are beginning to emerge.

We completely agree that this is speculative at this point, which needed acknowledgment. See response to Editor summary point #2 above.

Under Our Team can Maternity Leave be replaced with just "on leave" -- i don't think it's necessary to share why they are on leave. May as well add Jazz Cook as well. Also, should we not include our pronouns here?

Trades are academic programs. In Hairstyling alone we learn history - how has the trade evolved? Which cultures developed certain styles and why? Science - Formulating colours is chemistry, Mixing disinfectants is math and science! Technical Terminology is language. My trade may not identify as one single category but it includes several dimensions of learning. It offers students the opportunity to indulge in a variety of aspects and perhaps thats why it has become increasingly interesting to students who possess multiple intelligences.

Author Response:

Reviewer #1 (Public Review):

Summary:<br /> In this study, the authors generate a Drosophila model to assess disease-linked allelic variants in the UBA5 gene. In humans, variants in UBA5 have been associated with DEE44, characterized by developmental delay, seizures, and encephalopathy. Here, the authors set out to characterize the relationship between 12 disease-linked variants in UBA5 using a variety of assays in their Drosophila Uba5 model. They first show that human UBA5 can substitute all essential functions of the Drosophila Uba5 ortholog, and then assess phenotypes in flies expressing the various disease variants. Using these assays, the authors classify the alleles into mild, intermediate, and severe loss-of-function alleles. Further, the authors establish several important in vitro assays to determine the impacts of the disease alleles on Uba5 stability and function. Together, they find a relatively close correlation between in vivo and in vitro relationships between Uba5 alleles and establish a new Drosophila model to probe the etiology of Uba5-related disorders.

Strengths:<br /> Overall, this is a convincing and well-executed study. There is clearly a need to assess disease-associated allelic variants to better understand human disorders, particularly for rare diseases, and this humanized fly model of Uba5 is a powerful system to rapidly evaluate variants and relationships to various phenotypes. The manuscript is well written, and the experiments are appropriately controlled.

Reviewer #2 (Public Review):

Relative simplicity and genetic accessibility of the fly brain make it a premier model system for studying the function of genes linked to various diseases in humans. Here, Pan et al. show that human UBA5, whose mutations cause developmental and epileptic encephalopathy, can functionally replace the fly homolog Uba5. The authors then systematically express in flies the different versions of the gene carrying clinically relevant SNPs and perform extensive phenotypic characterization such as survival rate, developmental timing, lifespan, locomotor and seizure activity, as well as in vitro biochemical characterization (stability, ATP binding, UFM-1 activation) of the corresponding recombinant proteins. The biochemical effects are well predicted by (or at least consistent with) the location of affected amino acids in the previously described Uba5 protein structure. Most strikingly, the severity of biochemical defects appears to closely track the severity of phenotypic defects observed in vivo in flies. While the paper does not provide many novel insights into the function of Uba5, it convincingly establishes the fly nervous system as a powerful model for future mechanistic studies.

One potential limitation is the design of the expression system in this work. Even though the authors state that "human cDNA is expressed under the control of the endogenous Uba5 enhancer and promoter", it is in fact the Gal4 gene that is expressed from the endogenous locus, meaning that the cDNA expression level would inevitably be amplified in comparison. The fact that different effects were observed when some experiments were performed at different temperatures (18 vs. 25) is also consistent with this. While I do not think this caveat weakens the conclusions of this paper, it may impact the interpretation of future experiments that use these tools, and thus should be clearly discussed in the paper. Especially considering the authors argue that most disease variants of UBA5 are partial loss-of-functions, the amplification effect could potentially mask the phenotypes of milder hypomorphic alleles. If the authors could also show that the T2A-Gal4 expression pattern in the brain matches well with that of endogenous RNA or protein (e.g. using HCR-FISH or antibody), it would help to alleviate this concern.

We thank the reviewer for pointing out this limitation.

Regarding the humanization strategy we used in the study, we agree that this is a binary system which may lead to overexpression of the target protein. However, as the

reviewer also points out, this temperature-sensitive system also enables us to flexibly adjust the expression level of the target protein, which is especially useful to study

partial LoF variants such as the UBA5 variants in this study. In our study we have successfully compared the relevant allelic strength of most of the variants, which

supports the use of our system in future studies. However, we do agree that the gene dosage effect could vary widely, so it is difficult to directly predict the effects of one variant in humans based upon results obtained in a model organism.

We agree with the reviewer that a masking effect may exist in our system due to its gene overexpression nature. However, we cannot conclude that this masking effect

really affects the interpretation of Group IA variants in our tests. The three variants are mild LoF, which is also supported by the biochemical assays. Hence, the variants may not cause any phenotype even when they are expressed at a physiological level.

Regarding the temporal and spatial expression pattern of the T2A-GAL4, the Bellen lab has generated T2A-GAL4 lines for more than 3,000 genes. The expression pattern of the vast majority of these GAL4 lines faithfully reflects the expression pattern of the endogenous genes, which has been documented in our previous publications (PMIDs 25824290, 29565247, 31674908, 35723254).

Reviewer #3 (Public Review):

Summary:<br /> Variants in the UBA5 gene are associated with rare developmental and epileptic encephalopathy, DEE44. This research developed a system to assess in vivo and in vitro genotype-phenotype relationships between UBA5 allele series by humanized UBA5 fly models and biochemical activity assays. This study provides a basis for evaluating current and future individuals afflicted with this rare disease.

Strengths:<br /> The authors developed a method to measure the enzymatic reaction activity of UBA5 mutants over time by applying the UbiReal method, which can monitor each reaction step of ubiquitination in real time using fluorescence polarization. They also classified fruit fly carrying humanized UBA5 variants into groups based on phenotype. They found a correlation between biochemical UBA5 activity and phenotype severity.

Weaknesses:<br /> In the case of human DEE44, compound heterozygotes with both loss-of-function and hypomorphic forms (e.g., p.Ala371Thr, p.Asp389Gly, p.Asp389Tyr) may cause disease states. The presented models have failed to evaluate such cases.

We agree with the reviewer that our model did not reflect the situation of the individuals who are compound heterozygous for a Group IA variant (p.Ala371Thr, p.Asp389Gly, or p.Asp389Tyr) and a strong LoF variant. However, we argue that our results do show that the Group IA variants alone do not cause disease. As discussed in the manuscript, individuals homozygous for the p.Ala371Thr variant are healthy and do not present with obvious phenotype. This is consistent with our findings in flies, and shows that the p.Ala371Thr variant is a mild LoF variant.

Author Response

The following is the authors’ response to the original reviews.

We greatly appreciate the thoughtful suggestions made by the Reviewers. We have addressed all of their comments below, with our responses bulleted and in italics. We believe these changes have helped clarify the manuscript and strengthen it overall.

Reviewer 1

1) Figures 1B and Supp. Figure 1A: It would be worth mentioning that the wave-form in the 129 strain in response to QLA starts out like AJ and B6, but transitions to looking like the wild-derived strain. So, although not quite as drastic as the NZO and NOD strains, it is not quite like the other classical inbred strains.

• We thank the reviewer for pointing this out. We have added further language to clarify the point:

“Additionally, even with the clear separation between the clusters, inter-strain variation was still observed within the clusters (e.g. more 129 islets had plateau responses to 8G/QLA than the B6 or AJ).”

2) The figures are generally excellent and really help to clarify the work in the paper. For Figure 2A, it would help even further if you could number the six different Ca++ parameters that are measured. They're all there, but it takes a bit of time to find them on the figure and numbering will make it easier on your reader.

• We appreciate this suggestion and have implemented it in our revised Figure 2A. The Ca2+ parameters are now numbered, and the description of this figure has been adjusted accordingly in the results section.

We added the revised text in the results section:

“To elucidate strain differences in Ca2+ dynamics, we focused on six parameters of the Ca2+ waveform (Figure 2A): 1) peak Ca2+ (the top of each oscillation); 2) period (the length of time between two peaks); 3) active duration (the length of time for each Ca2+ oscillation measured at half of the peak height, also known the oxidative “secretory” phase, or “MitoOx” (8); 4) pulse duration (active duration plus extra time for Ca2+ extrusion); 5) silent duration (the electrically-silent “triggering” phase, also known as “MitoCat” (8), which culminates in KATP closure and membrane depolarization); and 6) plateau fraction (the active duration divided by the period, or the fraction of time spent in the active “secretory” phase).”

3) Figure 4A, B: I was expecting to see Ca++ vs insulin parameters in the different strains/sexes. In addition to the heat maps, it would be useful to see the regression plots, showing where each strain and sex falls for the insulin and Ca++ parameters.

• This is an excellent suggestion, and we have added a new Supplemental Figure 5 to provide examples of various strain/sex patterns that drive the correlations used for the heatmap and histogram in Figure 4A and B.

We added text in the results section referring to this point:

“Clustering the Ca2+ responses into distinct groups based on our observations of the waveforms (Figure 1B, Figure 4C-E, and Supplemental Figures 1 and 2) also occurs when correlating individual Ca2+ parameters to ex vivo secretion and clinical data (Supplemental Figure 5). For example, the anticorrelation between the 1st frequency component in 8G and percent insulin secreted in 8.3G/QLA (Supplemental Figure 5A) separates the classic inbred, wild-derived, and diabetes-susceptible strains into distinct groups despite the variability in the trait. Correlation between the silent duration in 8G/QLA to insulin secretion in 8.3G/QLA, likewise groups by strain (Supplemental Figure 5B). Finally, some correlations, such as that between 8G/QLA/GIP silent duration and plasma insulin at sacrifice (Supplemental Figure 5C), can be strongly influenced by outlier strains; e.g., NZO. Collectively, these data demonstrate that genetics has a profound influence on key parameters of islet Ca2+ oscillations.”

4) Please include methods for the insulin measurements collected in Fig. 4.

• Thank you for pointing out this missing information. We have clarified that prior insulin measurements (plasma insulin and ex vivo static insulin secretion that were used in Figure 4 for correlation analysis) were completed in another previously published cohort of mice (reference 17: Mitok KA, Freiberger EC, Schueler KL, Rabaglia ME, Stapleton DS, Kwiecien NW, et al. Islet proteomics reveals genetic variation in dopamine production resulting in altered insulin secretion. The Journal of biological chemistry. 2018;293(16):5860-77).

We added this new text (highlighted) to the results section to help clarify this point:

“Fasting blood glucose and insulin levels were measured in mice at 19 weeks of age, except for the NZO males which were measured at 12 weeks of age. Glucose was analyzed by the glucose oxidase method using a commercially available kit (TR15221, Thermo Fisher Scientific), and insulin was measured by radioimmunoassay (RIA; SRI13K, Millipore). This is the same assay that was used to measure plasma insulin for the previously published cohort used for the correlation analysis in Figure 4 (17).”

5) In the methods, please include details on the four conditions used for Ca++ imaging of the islets, and the timing for each condition.

• We appreciate this guidance in clarifying our manuscript, and we have now included the conditions and timing for each condition in the methods section.

We added the following text to the results section to help clarify this:

“The solutions included 8 mM glucose (8G), 8 mM glucose + 2 mM glutamine, 0.5 mM leucine, and 1.5 mM alanine (8G/QLA), 8G/QLA + 10 nM glucose-dependent insulinotropic polypeptide (8G/QLA/GIP), and 2 mM glucose (2G), each of which were kept in a 37°C water bath.”

Reviewer 2

One major critique is that the authors studied "the human orthologues of the correlated mouse proteins that are proximal to the glycemia-associated SNPs in human GWAS". This implies two assumptions - (1) human and mouse proteins do not differ in terms of islet physiology and calcium signaling; (2) the proteins proximal to the SNPs are the causal factors for functional differences, though the SNPs could affect protein/gene function distant from the SNPs.

• Thank you very much for highlighting this limitation in our study. We think this is very important to address which we have done in our discussion section.

We have added the following text to discuss this important issue:

“Our approach to merge human GWAS with our findings in mouse assumes that the glycemic-related SNPs we nominated alter the abundance or function of the human orthologues. Most SNPs that are strongly associated with phenotypes in human GWAS are noncoding, residing within introns, promoters, 3’UTRs, or intergenic regions (e.g. Figure 6). Therefore, a limitation of our approach is the assumption that SNPs regulate the gene they are proximal to, which is not always accurate (76-78). To infer a more direct link between SNPs and potential target genes, we incorporated human islet chromatin data (37). Physical contact between a region containing SNPs and a distal gene supports a regulatory role, as for ACP1 (Figure 6B). Additionally, SNPs within regions of open chromatin (ATAC-seq) and actively transcribed regions (histone markers) suggest a higher likelihood of regulating transcription factor access. While this approach does not conclusively show a link between the SNPs and expression of the orthologue for our candidate proteins, these chromatin data more strongly suggest that the orthologue expression may be regulated by the candidates’ SNPs.”

Reviewer #3 (Public Review):

The authors report a study in which they use intracranial recordings to dissociate subjectively aware and subjectively unaware stimuli, focusing mainly on prefrontal cortex. Although this paper reports some interesting findings (the videos are very nice and informative!) the interpretation of the data is unfortunately problematic for several reasons. I will detail my main comments below. If the authors address these comments well, I believe the paper may provide an interesting contribution to further specifying the neural mechanisms important for conscious access (in line with Gaillard et al., Plos Biology 2009).

The main problem with the interpretation of the data is that the authors have NOT used a so-called "no-report paradigm". The idea of no report paradigms is that subjects passively view a certain stimulus without the instruction to "do something with it", e.g., detect the stimulus, immediately or later in time. Because of the confusion of this term, specifically being related to the "act of reporting", some have argued we should use the term no-cognition paradigm instead (Block, TiCS, 2019, see also Pitts et al., Phil Trans B 2018). The crucial aspect is that, in these types of paradigms, the critical stimulus should be task-irrelevant and thus not be associated with any task (immediately or later). Because in this experiment subjects were instructed to detect the gratings when cued 600 ms later in time, the stimuli are task relevant, they have to be reported about later and therefore trigger all kinds of (known and potentially unknown) cognitive processes at the moment the stimuli are detected in real-time (so stimulus-locked). You could argue that the setup of this delayed response task excludes some very specific report related processes (e.g., the preparation of an eye-movement), which is good, however this is usually not considered the main issue. For example when comparing masked versus unmasked stimuli (Gaillard et al., 2009 Plos Biology), these conditions usually also both contain responses but these response related processes are "averaged out" in the specific contrasts (unmasked > masked). In this paper, RT differences between conditions (that are present in this dataset) are taken care of by using this delayed response in this paper, which is a nice feature for that and is not the case for the above example set-up.

Given the task instructions, and this being merely a delayed-response task, it is to be expected that prefrontal cortex shows stronger activity for subjectively aware versus subjectively unaware stimuli. Unfortunately, given the nature of this task, the novelty of the findings is severely reduced. The authors cannot claim that prefrontal cortex is associated with "visual awareness", or what people have called phenomenal consciousness (this is the goal of using no-cognition paradigms). The only conclusion that can be drawn is that prefrontal cortex activity is associated with accessing sensory input: and hence conscious access. This less novel observation has been shown many times before and there is also little disagreement about this issue between different theories of consciousness (e.g., global workspace theory and local recurrency theories both agree on this).

The best solution at this point seems to rewrite the paper entirely in light of this. My advice would be to state in the introduction that the authors investigate conscious access using iEEG and then not refer too much to no-cognition paradigm or maybe highlight some different strategies about using task-irrelevant stimuli (see Canales-Johnson et al., Plos Biology 2023; Hesse et al., eLife 2020; Hatamimajoumerd et al Curr Bio 2022; Alilovic et al., Plos Biology 2023; Pitts et al., Frontiers 2014; Dwarakanth et al., Neuron 2023 and more). Obviously, the authors should then also not claim that their results solve debates about theories regarding visual awareness (in the "no-cognition" sense, or phenomenal consciousness), for example in relation to the debate about the "front or the back of the brain", because the data do not inform that discussion. Basically, the authors can just discuss their results in detail (related to timing, frequency, synchronization etc) and relate the different signatures that they have observed to conscious access.

I think the authors have to discuss the Gaillard et al PLOS Biology 2009 paper in much more detail. Gaillard et al also report a study related to conscious access contrasting unmasked and masked stimuli using iEEG. In this paper they also report ERP, time frequency and phase synchronization results (and even Granger causality). Because of the similarities in approach, I think it would be important to directly compare the results presented in that paper with results presented here and highlight the commonalities and discrepancies in the Discussion.

In the Gaillard paper they report a figure plotting the percentage of significant frontal electrodes across time (figure 4A) in which it can be seen that significant electrodes emerge after approximately 250 ms in PFC as well. It would be great if the authors could make a similar figure to compare results. In the current paper there are much more frontal electrode contacts than in the Gaillard paper, so that is interesting in itself.

In my opinion, some of the most interesting results are not highlighted: the findings that subjectively unaware stimuli show increased activations in the prefrontal cortex as compared to stimulus absent trials (e.g., Figure 4D). Previous work has shown PFC activations to masked stimuli (e.g., van Gaal et al., J Neuroscience 2008, 2010; Lau and Passigngham J Neurosci 2007) as well as PFC activations to subjectively unaware stimuli (e.g., King, Pescetelli, and Dehaene, Neuron 2016) and this is a very nice illustration of that with methods having more detailed spatial precision. Although potentially interesting, I wonder about the objective detection performance of the stimuli in this task. So please report objective detection performance for the patients and the healthy subjects, using signal detection theoretic d'. This gives the reader an idea of how good subjects were in detecting the presence/absence of the gratings. Likely, this reveals far above chance detection performance and in that case I would interpret these findings as "PFC activation to stimuli indicated as subjectively unaware" and not unconscious stimuli. See Stein et al., Plos Biology 2021 for a direct comparison of subjectively and objectively unaware stimuli.

In Figure 7 of the paper the authors want to make the case that the contrast does not differ between subjectively aware stimuli and subjectively unaware stimuli. However so far they've done the majority of their analyses across subjects, and for this analysis the authors only performed within-subject tests, which is not a fair comparison imo. Because several P values are very close to significance I anticipate that a test across subjects will clearly show that the contrast level of the subjectively aware stimuli is higher than of the subjectively unaware stimuli, at the group level. A solution to this would be to subselect trials from one condition (NA) to match the contrast of the other condition (NU), and thereby create two conditions that are matched in contrast levels of the stimuli included. Then do all the analyses on the matched conditions.

Related, Figure 7B is confusing and the results are puzzling. Why is there such a strong below chance decoding on the diagonal? (also even before stimulus onset) Please clarify the goal and approach of this analysis and also discuss/explain better what they mean.

I was somewhat surprised by several statements in the paper and it felt that the authors may not be aware of several intricacies in the field of consciousness. For example a statement like the following "Consciousness, as a high-level cognitive function of the brain, should have some similar effects as other cognitive functions on behavior (for example, saccadic reaction time). With this question in mind, we carefully searched the literature about the relationship between consciousness and behavior; surprisingly, we failed to find any relevant literature." This is rather problematic for at least two reasons. First, not everyone would agree that consciousness is a high-level cognitive function and second there are many papers arguing for a certain relationship between consciousness and behavior (Dehaene and Naccache, 2001 Cognition; van Gaal et al., 2012, Frontiers in Neuroscience; Block 1995, BBS; Lamme, Frontiers in Psychology, 2020; Seth, 2008 and many more). Further, the explanation for the reaction time differences in this specific case is likely related to the fact that subjects' confidence in that decision is much higher in the aware trials than in the unaware trials, hence the speeded response for the first. This is a phenomenon that is often observed if one explores the "confidence literature". Although the authors have not measured confidence I would not make too much out of this RT difference.

I would be interested in a lateralized analysis, in which the authors compare the PFC responses and connectivity profiles using PLV as a factor of stimulus location (thus comparing electrodes contralateral to the presented stimulus and electrodes ipsilateral to the presented stimulus). If possible this may give interesting insights in the mechanism of global ignition (global broadcasting), supposing that for contralateral electrodes information does not have to cross from one hemisphere to another, whereas for ipsilateral electrodes that is the case (which may take time). Gaillard et al refer to this issue as well in their paper, and this issue is sometimes discussed regarding to Global workspace theory. This would add novelty to the findings of the paper in my opinion.

You would imagine that by now you would be able to share some of your highlights without having to add some weird annotation to it especially when you are trying to share it on a private group.

it is also quite worrisome that the last time there was a comment about this it was in 2019. it is almost like there's no work or effort put into this one.

so as much as there's a comment about "thinking of effective ways", there's no clear indication that there's some were going into it.

I think there is no mention of experts already having problems with the way NASA are dealing with this. If there is ignore this comment. But I feel that a mention of this should be high above and then saying to see down here for more info on this.

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

Learn more at Review Commons

---

Reply to the reviewers

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

The manuscript describes that simultaneous inhibition of LOXL2 and BRD4 reduces proliferation of TNBC in vitro and reduces growth in vivo.

This observation is followed by extensive mechanistic studies that suggest physical interaction between LOXL2 and short isoform of BRD4-MED1. Inferences from Chip-seq analyses suggest that this interaction is involved in regulation of multiple transcriptional programs. Authors focus on differential activation of DREAM complex, to claim that this interaction "is fundamental for proliferation of TNBC". The manuscript is very well written and mechanistic inferences are based on a set of sophisticated epigenetic analyses and bioinformatical inferences. The phenotypic effects from LoxL2 inhibition by itself, or in combination with BRD4 inhibition are relatively modest. These modest effects, as well as many of the reported changes in gene expression are clearly inconsistent with the frequently used adjectives as "dramatic", "fundamental", "deeply affected", "drastically hampered" etc. Given the modest phenotypic effects, many of the key claims and conclusions are not supported by the data.

We thank the reviewer for appreciating our work, defining the manuscript as well-written, and saying that it comprises extensive mechanistic studies as well as sophisticated epigenetic analysis.

We apologize if some of our statements seemed exaggerated. In this revised version, we revisited some of our conclusion to moderate them.

Moreover, we took the reviewer's criticism as an opportunity to strengthen our findings. In the revised version of the manuscript, we included an additional TNBC PDX (PDX-127), and results from this experiment clearly reinforce our claims (Fig. 6D and Fig. EV9E-F). In this new in vivo experiment, we selected a PDX model in which the expression of BRD4L is not detectable, while BRD4S is clearly expressed. Therefore, the treatment with JQ1 would specifically affect the activity of BRD4S, making the treatment selective. Additionally, we reduced by half the dose of JQ1 administrated to limit the effect of BRD4S inhibition alone on tumor growth. The combinatorial treatment (JQ1+PXS) induced a clear superior effect in this setting as compared with single-agent treatments. In addition to this, we discarded that the observed growth reduction is not the result of the sole inhibition of LOXL2, which could affect FAK/Src activity or extracellular Collagen crosslinking. In conclusion, our data show that the combinatorial inhibition of LOXL2 and BRD4S is effective in reducing tumor proliferation in TNBC in vivo models, independently of the inhibition of BRD4S and of other pathways known to be regulated by LOXL2.

Specifically:

1) It is unclear why authors generalize their conclusions to TNBC. Figure 1B demonstrates synergy for 1/3 cell lines, which is chosen for the follow up study. Even for MDA231, the synergy is confined to low concentrations of BRD4i (S1c). While MDA231 cell line is frequently used in experimental studies of TNBC, it is quite dissimilar to majority of clinical TNBC, and contains mutant RAS, which is rare in this disease.

The synergistic effect is observed in MDA-MB-231 cells because only this cell line expresses both BRD4S and LOXL2. Indeed, in Fig. 1C we show that MDA-MB-468 cells do not express LOXL2, while BT549 only express minimal BRD4 levels.

To corroborate this hypothesis, in the revised version of the manuscript we added:

1. A new cell line (Cal51) expressing the same LOXL2 and BRD4 levels (Fig. EV8C) but showing greater resistance to JQ1 than MDA-MB-231 (Fig. EV8D). Also, in this cell line, we could show that the combinatorial treatment had a superior effect on cell viability than the single agents’ treatment (Fig. EV8E).
2. A western blot panel of different TNBC PDXs shows that the majority of them express medium to high levels of both BRD4S and LOXL2 proteins, as is the case of MDA-MB-231 (Fig. EV9E) and Cal51 (Fig. EV8C). This result suggests that the combinatorial treatment could be used in the majority of TNBC patients as they are expected to express both BRD4S and LOXL2.
3. Finally, as explained above, we performed another in vivo choosing a PDX that expresses BRD4S (but not BRD4L) and LOXL2 (PDX-127) (Fig. 6D and Fig. EV9E-F). Also, in this new model, we could observe that the combinatorial inhibition had a superior effect than single treatments.

   2) In vivo, the effect appears to be modest even in the MDA231 model, selected for evidence of synergy in vitro. In vivo, the combination appears to have an additive effect. Tumor growth rates are reduced, but no shrinkage is occurring. In the PDX model, LOXL2i does not have an effect as a monotherapy, while modestly enhancing the impact of BRD4i. These results are at odds with the claim of the interaction being fundamental for proliferation.

We agree with the reviewer that the combinatorial inhibition appears to have an additive effect in vivo using the MDA-MB-231 model.

1. For that reason, we have now performed the in vivo PDX experiment mentioned above (PDX-127; Fig. 6D and Fig. EV9E-F) in which we decreased the dose of JQ1 by half to avoid strong tumor growth effect due to BRD4 inhibition alone. In this new experiment, the synergistic effect is evident. While single-agent treatment showed a very moderate effect (0% or 20% tumor growth reduction for LOXL2 and JQ1, respectively), the combinatorial treatment showed a 50% reduction in tumor volume, further supporting our conclusions.
2. We also performed either BRD4 or MED1 pull-down experiments in the presence of PXS and JQ1. We show that upon PXS treatment, the interaction between LOXL2 and BRD4S is maintained while the interaction with MED1 is reduced (Fig. 5A-C). However, in the presence of JQ1, the interaction between LOXL2 and MED1 is maintained while BRD4S-LOXL2 and BRD4S-MED1 interactions are impaired (Fig. 5D-F). These new results explain why monotherapy does not have a sufficient effect in vivo and set the rationale for the use of the combinatorial treatment. We believe that these new results corroborate our initial findings and we hope to have been able to satisfy the reviewer comments.

3) No analysis of cell proliferation was shown in vivo. Authors should have performed BrdU or KI67 staining to support the claim. For in vitro analyses, authors also used indirect assays for proliferation. PI staining by itself does not have sufficient resolution to clearly capture modest effects that authors demonstrate. BrdU-PI double staining would have been much more useful.

We appreciate the reviewer’s comment. In the revised manuscript we have added Ki67 and H3S10p staining in the tumor samples for the new in vivo PDX experiment (Fig. 6E and Fig. EV10A-C). We show that the combinatorial treatment significantly induces a reduction of both proliferation markers, which is in agreement with a reduced tumor volume. Regarding the in vitro analysis, we did not only use PI staining to show a reduced proliferation state but also H3S10p staining (Fig. 4B) and an SLBP1 fluorescent reporter MDA-MB-231 cell line (Fig. 4D, Fig. EV6B, E, and Movie EV). In the revised version of the manuscript, we included a new FACS-PI analysis (Fig. 4A, C) to better represent the effects we see on the cell cycle.

Minor points:

Dose dependent decrease in phosphorylated H3 is not at all obvious from eyeballing the data in S1A; the only effect that I see is a modest reduction at the highest concentration of the inhibitor. Authors need to quantify the results to support the claim.

We agree with the reviewer and we apologize for the misinterpretation. We have changed the revised manuscript as follows: “The selective LOXL2 inhibitor PXS-538224 (hereafter, PXS) efficiently reduced the levels of oxidized histone H3 (H3K4ox) in MDA-MB-231 cells at 40 μM (Fig. EV6C), indicating an efficient inhibition of LOXL2 catalytic activity in the nucleus.”

Most of breast cancer cell lines are derived from metastatic disease, including pleural effusion, thus the point that because MDA231 cell line is derived from pleural effusion, it is metastatic does not have sufficient logical foundation.

Many publications have shown the high metastatic capacity of MDA-MB-231 (e.g. https://doi.org/10.1016/j.bbabio.2011.04.015, doi: 10.1038/s41467-017-01829-1), which are therefore used as TNBC metastatic model. The scope of the analysis reported in Fig. 6C was just to show whether any of the used treatments could reduce the metastatic capacity of this cell line. We believe we do not overstate the results but just report them as they are.

How is loss of cell-cell junction in vitro consistent with LOXL2 role in modulating ECM? There is no evidence of ECM production in MDA231 in vitro. On the other hand, this loss is associated with EMT.

We thank the reviewer for identifying this mistake. In the revised manuscript we changed the text as follows: “Gene set enrichment analysis (GSEA) revealed that LOXL2 KD induced upregulation of processes involved in cell morphology, secretion, membrane trafficking, and cell differentiation, with cell-cell junction being one of the most significantly affected pathways (Fig. EV5E). These results agree with the role of LOXL2 in regulating epithelial-to-mesenchymal transition, corroborating the high quality of our dataset.”

Reviewer #1 (Significance (Required)):

Discovery and characterization of LOXL2-BRD4 interaction is advancing the ever-deepening understanding of molecular mechanisms of regulation of gene expression. The studies and analyses appear to be sufficiently rigorous and reported with clarity, and the claimed discovery of the biological interaction between LOXL2 and BRD4 is well supported. However, given the magnitude of the reported (rather than claimed) effects of this interaction, and concerns about generalizability of authors conclusions, it is not clear how these results are promising for the development of new therapies in TNBC. Moreover, in contrast to luminal BC, there is no clear evidence for utility of cytostatic drugs in constraining TNBC. Therefore, biological and clinical significance of the authors discovery is unclear and claims in this regard appear to be overblown

We thank the reviewer for stating that our analysis is rigorous and reported with clarity. We really took the criticisms as an opportunity to strengthen our findings, as explained above.

For the newly presented in vivo PDX model, we performed immunohistochemistry of Ki67, H3S10p and Cleaved Caspase 3 to check whether the reduction of tumor volume observed in the combinatorial treatment was a result of a cytotoxic and/or a cytostatic effect (Fig. 6E and Fig. EV10A-C). As shown in the figure, the combination of the two inhibitors induced a superior decrease of Ki67, H3S10p, and a clear increase of Cleaved Caspase 3. Therefore, these new data indicate that the combinatorial treatment does not only have a cytostatic effect but also cytotoxic, suggesting a clinical exploitability for the treatment of TNBC patients.

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

In their study, Pascual-Reguant et al. show that combined inhibition of BRD4 and LOXL2 can synergize to restrict triple-negative breast cancer (TNBC) proliferation. BRD4 and LOXL2 are transcription regulators that can read and write epigenetic information, respectively. The authors employ three distinct breast cancer cell lines and mouse models with cell line-derived xenografts, and they show that combined inhibition of BRD4 and LOXL2 can be superior to single BRD4/LOXL2 inhibition in these model systems. In an attempt to identify a connection between BRD4 and LOXL2, the authors find that the two proteins can bind to each other. The authors performed most of the experiments in the breast cancer cell line MDA-MB-231. To assess the impact of LOXL2-inhibition on transcription, the authors assessed changes of the transcriptome in MDA-MB-231 cells following LOXL2 knockdown. They found that genes related to cell differentiation and morphology were upregulated, while genes related to the cell cycle were downregulated. ChIP-seq data of BRD4 showed that BRD4 can bind to cell cycle gene promoters and that this binding was enhanced upon loss of LOXL2. The authors found that LOXL2 and BRD4 interacted with the transcriptional cell cycle regulators B-MYB, FOXM1, and LIN9, which are components of the MYB-MuvB-FOXM1 (MMB-FOXM1) complex that is known to promote the expression of late cell cycle genes with important functions during mitosis. The authors conclude that LOXL2/BRD4 interact with each other and with the MMB-FOXM1 complex to drive the expression of cell cycle genes and cell proliferations. Vice versa, they conclude that inhibition of LOXL2/BRD4 reduces cell proliferation through inhibiting the expression of cell cycle genes.

Major:

• The data and methods are presented well. The experiments are adequately replicated and analyzed. However, except for the first section, all experiments were performed using only one cell line. It is important to validate key findings in at least a second cell line.

We thank the reviewer for valuing our work.

To address the reviewer’s comment, in the revised manuscript we added an additional cell line (Cal-51), that expresses similar levels of LOXL2 and BRD4 as compared to MDA-MB-231 (Fig. EV8C). Even though this cell line is clearly more resistant to JQ1 than the MDA-MB-231 cell line (Fig. EV8D), the combinatorial treatment is significantly more effective as compared with single agents’ treatment (Fig. EV8E).

Moreover, we have also performed an additional in vivo experiment using another TNBC PDX (PDX-127) that expresses LOXL2 and BRD4S, but not BRD4L. Given that JQ1 can inhibit both BRD4 isoforms, this in vivo system allowed us to demonstrate that the tumor antiproliferative capacity of the combinatorial treatment is due to the simultaneous inhibition of LOXL2 and BRD4S (rather than BRD4S and L) (Fig. 6D and Fig. EV9E-F).

• There appears to be a misunderstanding of the concept of cell cycle-dependent gene regulation by the DREAM complex and its related factors. Early (G1/S) cell cycle genes contain E2F promoter motifs, while late (G2/M) cell cycle genes contain CHR promoter motifs. The DREAM complex can bind both, while RB-E2F and MuvB recognize only E2F and CHR motifs, respectively. B-MYB and FOXM1 bind to MuvB and regulate late cell cycle genes, but they do not bind to early cell cycle genes. Given this concept, the authors' rationale to connect BRD4/LOXL2 through MuvB/B-MYB/FOXM1 with E2F promoter sequences and early cell cycle genes and the subsequent conclusions must be corrected.

We thank the reviewer for their expert explanation. We corrected our conclusion in the revised version of the manuscript following the reviewer’s comment.

• I felt that the suggested functional connection between LOXL2/BRD4 and DREAM is not strongly supported by the authors' data. Figure S6E: A similarity score of Fig. EV6E: We agree with the reviewer that a similarity score of Fig. 4E: We thank the reviewer for this comment. The performed pulldown showed that BRD4S, LOXL2, and MED1 interact with Lin9 and B-Myb, but not with FOXM1, thus FOXM1 itself is an internal negative control of the pulldown. Additionally, BRD4L does not show the same interaction pattern as BRD4S, LOXL2, and MED1, again acting as an internal negative control. We, therefore, believe that the pulldown is properly controlled and that the observed interaction is trustful. We furthermore agree with the reviewer that it would be interesting to characterize the interactions between the DREAM complex and BRD4S, LOXL2, and MED1. However, we believe that the dissection of these interactions at the mechanistic levels would require a deeper study, which can be a project in itself that we aim to explore in the future. For example, it would be interesting to investigate whether either the inhibition or the downregulation of LOXL2 and/or BRD4S specifically impairs the formation of the DREAM complex or the recruitment of specific DREAM complex subunits, as well as how these effects impair the DREAM complex chromatin binding. We are afraid that the suggested pulldowns would not be sufficient to answer these questions, which would require extensive cross-interaction studies in either BRD4/LOXL2 and BRD4+LOXL2 inhibition or downregulation followed by ChIP-seq and transcriptomics for all the conditions. We believe that the provided data, together with the functional characterization (both, in vitro and in vivo), of the phenotypes triggered by BRD4S and LOXL2 inhibition make a strong case for our manuscript and leave out of scope the suggested experiments. We hope the reviewer will understand our explanation and will appreciate that we are planning to pursue this further in the future.

Fig. 3: We thank the reviewer for this important comment. The ChIP-seq technique very often does not provide exhaustive results due to sequencing depth limits and antibody performance. We believe that the fraction of DREAM target genes found in our dataset as bound by BRD4S is not exhaustive and that the analysis proposed by the reviewer would not lead to clear conclusive results. However, we understand the importance of verifying that DREAM target genes whose promoter is bound by BRD4 are indeed downregulated when LOXL2 is inhibited. To give an answer to this question, in the revised manuscript we added gene expression analysis of selected DREAM target genes upon treatment with JQ1, PXS their combination. We could successfully show that both JQ1 and PXS treatment impairs the transcription of the selected DREAM target genes, however, the combinatorial treatment almost shut down their expression, in agreement with our hypothesis (Fig. 5J).

• The authors state that it is surprising to find that LOXL2 can promote target gene transcription because it is rather known as a transcriptional repressor. To this point, the authors should perform standard analyses using their RNA-seq and ChIP-seq data. Compare differential expression of genes that are bound by BRD4S/L/S+L and genes not bound by BRD4. Perform motif search and enrichment analyses for transcription factor and co-factor binding data (public ChIP-seq repositories). Such analyses may suggest what gene sets are up- and downregulated by LOXL2 through BRD4S/L and what other factors could be involved in LOXL2-dependent up- and downregulation of gene transcription.

We thank the reviewer for this valuable comment that certainly provides the rationale for a follow-up project. However, we believe that the proposed study goes beyond the scope of our work at this moment.

Minor:

• I felt that background information on the BRD4 isoforms was missing. The short and long isoforms of BRD4 should be introduced briefly.

We agree with the reviewer. In the revised manuscript, we addressed this by presenting BRD4 isoforms in the introduction part of the manuscript.

• Given that BRD4 inhibition is known to activate p53 (e.g., PMID 23317504 and 33431824) and p21 (PMID 31265875), the authors should discuss the p53 status of their cell lines (largely mutant). In general, I felt that the authors could better cite and discuss the current literature on BRD4 and LOXL2.

We appreciate the comment of the reviewer regarding p53. Given the fact that p53 is mutant in MDA-MB-231, we believe that the proliferation defect observed with the combinatorial treatment may be due to the activation of alternative cytostatic or cytotoxic signaling cascades, independently of P53 activation. We have now briefly mentioned this point in the manuscript discussion.

• It was unclear to me why the authors did not actually test experimentally whether their predicted interaction models 2 or 4 are likely true (Figure 2E+G).

We understand the reviewer’s comment. The fact that JQ1 treatment almost abrogates the interaction between LOXL2 and BRD4S strongly suggests that models 1 and 3 are likely wrong, therefore pointing towards models 2 and 4 as the correct ones. To test whether models 2 and 4 are indeed the correct models we are now performing extensive mutagenesis studies, which are producing preliminary results suggesting indeed that models 2 and 4 are correct. The reason why we did not include this study in the current manuscript, is that we started a parallel line of investigation aimed at identifying residues fundamental for the interaction that can be exploited in compound screening campaigns to identify molecules able to block the described interaction and thus cancer proliferation. Publishing these preliminary results at this stage could jeopardize the drug discovery campaign and we hope that the reviewer will understand our constraints.

• The transcription of cell cycle genes depends on the cell cycle (i.e., reduced cell cycle entry correlates with reduced cell cycle gene expression). Given that the authors showed LOXL2 inhibition reduce MDA-MB-231 cell proliferation, they should note that reduced expression of cell cycle-related genes is expected upon LOXL2 knockdown.

We understand the reviewer’s comment. We believe that we provide sufficient data supporting our hypothesis that LOXL2 controls the expression of cell cycle genes at the transcriptional level together with BRD4S. In addition, the sole inhibition of LOXL2 has practically no effect on tumor proliferation in vivo but largely enhances the antiproliferative effect of low-dose JQ1 (Fig. 6D). We hope these clarifications would satisfy the reviewer.

• The authors specify in their discussion that their data show a function of LOXL2/BRD4 in the cell cycle interphase, while there were no experiments that support that specific conclusion. At least it is unclear to me why the authors rule out a function in mitosis?

We thank the reviewer for this comment. We referred to interphase genes because these are the early cell cycle genes, while mitotic genes are the late ones. We do not discard a possible function for BRD4S and LOX2 regulating mitotic progression, however, we believe this would be a consequence of dysregulated G1-S-G2 gene expression, rather than a direct transcriptional effect. This conclusion derives from the fact that while we observe interactions between LOXL2, BRD4S, and MED1 with Lin9 and B-Myb, these are not fully conserved with FOXM1, which is typically required for the transcription of mitotic genes. To avoid confusion, we have now anyway removed the word “interphase” from the text.

• I felt that the first part of the manuscript (combination of BRD4 and LOXL2 inhibitors in TNBC) was a bit uncoupled from the functional studies on LOXL2 and its connection to BRD4. The transition between these parts and the final discussion on why the joint control of cell cycle genes by LOXL2/BRD4 may be important for the synergistic effect of LOXL2/BRD4 inhibitors. To this point, the authors' model was not clear to me.

We really appreciate the reviewer’s comment. To better connect the functional studies with the clinical significance of the proposed combinatorial treatment, we restructured the manuscript. In the revised version, the use of the combinatorial treatment is shown in Figure 6. Moreover, to better explain why we focused all the studies on BRD4 and LOXL2, we also included data from the Cancer Cell Line Encyclopedia (CCLE)-associated chemotherapeutics sensitivity (Fig. 1A and Fig. EV1) showing that LOXL2 expression levels can predict the response to BRD4 inhibition, suggesting a functional interaction between BRD4 and LOXL2 and the possibility to exploit it for therapeutical purposes. We believe that these data set the rationale to further explore the connection between LOXL2 and BRD4, both at the mechanistic and functional levels.

Reviewer #2 (Significance (Required)):

The study by Pascual-Reguant et al. shows that inhibitors of BRD4 and LOXL2 can be combined to achieve better efficacy in reducing proliferation of breast cancer cell lines and breast tumor growth in xenograft models. They provide strong evidence for a functional interaction between LOXL2 and BRD4 and investigate their common transcriptional targets. Intriguingly, some evidence points towards a direct regulation of the DREAM complex and its cell cycle gene targets.

The findings are novel and can be the basis for further research on TNBC combination therapy using BRD4 and LOXL2 inhibitors. The link to the DREAM complex is preliminary.

The study is of interest for a basic research audience with some translational aspects.

I reviewed this manuscript as a researcher in gene regulatory mechanisms, with cell cycle genes as one focus area. I have no expertise in the computational modeling of protein-protein interactions and I am no expert for breast cancer.

We thank the reviewer for the positive comments. We also would really like to thank the reviewer for their criticism, which, we believe, contributed to a new and improved manuscript version.

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

Summary:

In this manuscript, Laura Pascual-Reguant et al. identified a novel role of the LOXL2 oxidase in sustaining cell cycle progression through a so far uncharacterized gene-activating function is mediated by the BRD4S epigenetic reader and exerted on key DREAM-target genes in TNBC. Moreover, the authors showed that combinatorial treatment of TNBC with LOXL2- and BRD4-specific inhibitors result in a tremendous anti-tumorigenic effect. For all findings, they leveraged in vitro and in vivo settings as well as high-throughput sequencing approaches. However, the following points should be addressed and explained.

Major points:

-The authors on their working hypothesis propose that dual inhibition of BRD4 and LOXL2 is a novel strategy for curing TNBC. For my taste, just because both targets are quite promising for TNBC, the jump to this combinatorial treatment is kind of abrupt. Knowing the difficulty and time-/financial- investment, authors could optionally perform a mass spectrometry analysis on nuclei lysates with LOXL2 pull down to identify physical interactors. Due to the augmented resources and analysis of raw data, authors may necessitate a generous revision period (approx. 4 months for starters). By that, this can provide a more unbiased approached to look at nucleus-specific gene-regulatory functions and particularly at epigenetic readers. It would be also interesting to see if LOXL2 interacts with other members of the BRD family. Selecting BRD4 and no other members of the bromodomain family cannot be the only choice given that other BRD members can also interact with several of these mediator subunits.

We thank the reviewer for the suggestion and we agree with the fact that the rationale for combining BRD4 and LOXL2 inhibitors was not sufficiently argued in the first version of the manuscript. For that reason, in the revised manuscript, we added new data to explain why we explored this topic. In particular, to better explain why we focused all the studies on BRD4 and LOXL2, we included data from the Cancer Cell Line Encyclopedia (CCLE)-associated chemotherapeutics sensitivity (Fig. 1A and Fig. EV1) showing that LOXL2 expression levels can predict the response to BRD4 inhibition (but not to other approved chemotherapeutic drug), suggesting a functional interaction between BRD4 and LOXL2 and the possibility to exploit it for therapeutical purposes. Moreover, we restructured the manuscript to make the story more linear, explaining first the functionality of BRD4S-LOXL2 interaction at the molecular and cellular levels, and then presenting the in vivo systems in the last part of the manuscript.

We agree with the reviewer that it may be interesting to explore whether LOXL2 interacts with other BRD family members. However, given the prominent role of BRD4 in promoting cancer proliferation, we believe that understanding the relevance of BRD4S-LOXL2 interaction in TNBC is, per se, of great interest and provide a novel mechanistic understanding of how TNBC proliferation is controlled at the transcription level. In the specific case of TNBC, it has been shown that BRD4S has an oncogenic effect, while BRD4L is an oncosuppressor. In the manuscript, we now showed that LOXL2 downregulation sensitizes cells to JQ1 treatment (Fig. 1D). Additionally, while the downregulation of BRD4L does not have any additional effect on cell treated with PXS, the downregulation of BRD4S sensitize them to LOXL2 inhibition (Fig. EV8B). These results, once again, indicate the relevance of studying the functional interaction between BRD4S and LOXL2.

-LOXL enzymes have been shown to promote collagen and fibronectin assembly, thereby sustaining the pro-survival effect of the ITG5A/FN1/FAK/SRC signaling cascade and shielding TNBC cells against chemotherapy treatment (32415208). Did authors observe if LOXL2 loss or inhibition decreased the active status of FAK and SRC, which are well known to promote G1-S transition (25381661)?

Probably the cell cycle defects upon LOXL2 loss may also partially arise from the impairment of this cascade.

We really appreciate the reviewer’s suggestions. In the revised version of the manuscript, we checked FAK and Src activation status in tumor samples from one of our in vivo experiments (Fig. EV10D). We did not observe any difference in phospho-FAK or phospho-Src upon treatment either with PXS, JQ1, or their combinations, suggesting that alterations in the activity of these factors were not driving the observed proliferation defects.

-Authors exclusively use JQ1 as a BRD4 inhibitor. As JQ1 may have an unspecific effect on BRD2 as well, authors should consider reproducing key experiments with siControl- and siBRD4-treated cells and increasing doses of PSX as well as repeating the JQ1 dose response assay in Figure 1B using siRNA-mediated silencing of LOXL2. Given that both players are part of the same complex, silencing of one and inhibition of the other should sensitize cells compared to their control counterparts.

We agree with the reviewer and we addressed this comment in the revised manuscript. In particular, we have added two additional experiments:

• We transduced MDA-MB-231 cells with isoform-specific shBRD4s (shBRD4L and shBRD4S) (Fig. EV5H) and checked cell sensitivity to PXS treatment (Fig. EV8B). As explained also above, we observed that only when the short isoform of BRD4 was downregulated cells displayed higher sensitivity to PXS treatment. This result corroborates that BRD4S and LOXL2 are required for TNBC proliferation.

• We transduced MDA-MB-231 cells with shLOXL2 and assessed JQ1 sensitivity (Fig. 1D). We showed that upon LOXL2 downregulation, cells became more sensitive to JQ1 treatment, again corroborating the fact that TNBC proliferation requires BRD4S and LOXL2.

-Moreover, in Figures 1G and S3D the differential sensitivity of low and high LOXL2 cell lines is unclear. Do authors know if any of these growth kinetic lines represent one of the tested cell lines in Figure 1A-B? Authors should provide respective legends. In addition, authors should take advantage of their homemade data given that they have already selected a panel of TNBC cell lines with various LOXL2 expression at basal state (Figure 1A) for which dose response assays have been performed (Figure 1B). Therefore, I would perform an IC50 graph for JQ1 (without PSX treatment) using the existing data from Figure 1B.

We apologize if our representation was confusing. In the revised manuscript we have changed the sensitivity plots (Fig. 1A and Fig. EV1) to make them easier to grasp. Additionally, in Figure 1A we included the analysis of CCLE cell lines stratified based on their LOXL2 expression levels. This analysis showed that LOXL2 expression levels could overall predict the response to BETi treatment. As suggested by the reviewer, we also plotted the IC50 of the 3 cell lines tested. However, their JQ1 sensitivity curves did not show any difference that could be attributed to their different LOXL2 levels. Our speculation is that only 3 cell lines do not provide a sufficient size to reach a meaningful conclusion, which, in contrast, can be achieved by comparing the CCLE BETi sensitivity.

-In Figure 2D, the pull-down assay is inconclusive, as the molecular weight for each construct is not mentioned. I would probably add this information also in all performed western blots. Also, the overexpression of the BD1/BD2-mutated and especially the BD1/BD2-lacking construct is unclear if it still interacts with LOXL2, probably because of the lack of molecular weight reference of each band. Therefore, the authors should make this pull-down assay more descriptive regarding the size of the bands. Also, BD1 mutagenesis at N140 was shown to dislodge the binding of JQ1 to BRD4 (24497639), which implies that BD1 mutagenesis or overexpression of the BD1-deficient construct should abrogate the interaction of LOXL2 with BRD4, reminiscent to the abrogated interaction of BRD4/LOXL2 upon JQ1 that binds to both BDs (Figure 2F). And, what happens if a BD2-deficient construct is expressed?

We thank the reviewer for spotting this distraction. We apologize for this and in the revised version of the manuscript we included molecular weights for all western blots.

We acknowledge that BD1 mutagenesis displaces JQ1 binding, however, we respectfully disagree that because of this BD1-N140 mutant should not bind to LOXL2. Our docking analysis indeed showed that none of the poses is impaired either by BD1 or BD2 mutagenesis (Fig. EV4D). The fact that JQ1 disrupts the interaction between BRD4S and LOXL2 (Fig. 2F, G) is not due to the fact that they compete for the same binding residue, but rather for the space occupied by JQ1 inside the AcK binding pocket of either BD1 or BD2, which impedes proper binding to LOXL2. Our pulldown data indeed showed that mutant BD1 and BD2 retain the ability to bind to LOXL2 (Fig. 2C), as predicted by the docking.

We did not try to express constructs either lacking BD1 or BD2 and we cannot speculate what could happen to the BRD4S-LOXL2 interaction in this scenario. Even though this experiment could help dissect the interaction between LOXL2 and BRD4S, we decided to rather perform mutagenesis of specific residues that have been predicted to be important for the interaction. The reason why we did not include this study in the current manuscript, is that we started a parallel line of investigation aimed at identifying residues fundamental for the interaction that can be exploited in compound screening campaigns to identify molecules able to block the described interaction and thus cancer proliferation. Publishing these preliminary results at this stage could jeopardize the drug discovery campaign and we hope that the reviewer will understand our constraints.

-If authors support that BRD4S is the predominant isoform driving the expression of DREAM-targets, this means that DREAM-targets are mainly bound by BRD4S, relying on Figure 3E-F. However, based on the author's ChIPseq tracks in Figure 3H, DREAM targets such as EZH2 and HMGB2 are co-occupied by both BRD4 isoforms at the basal state on their promoter region. Also, especially for EZH2 and PLK4, authors should set to 'group auto-scale' both conditions in a smaller scale range for ChIPseq- and RNAseq tracks, although I do not these two genes as good candidates representing your analysis. Therefore, authors should initially show all genes (e.g in a table format) that enrich the 'DREAM-targets' signature and select for a greater panel of genes (like for AURKB and HMGB2) demonstrating a preferential occupancy of the BRD4S at their promoter region. Finally, authors are recommended to perform a ChIP-qPCR on these genomic regions at basal state (no LOXL2 silencing) to validate the predominant occupancy of BRD4S and the low/absent occupancy of BRD4L at these genomic sites.

We apologize for the confusion. To make the figure more understandable, we now scaled all the panels to the same scale and highlighted in grey the promoter region of each selected DREAM target gene. As the reviewer can appreciate, none of these genes is bound by BRD4L in basal conditions (Fig. 3F).

To better characterize the differential binding, following the reviewer’s suggestion, we performed ChIP-qPCR using Ab2 (which recognizes both BRD4 isoforms), in cells either downregulated for BRD4L or BRD4S with isoform-specific shRNAs (Fig. EV5H). Results showed that only the downregulation of BRD4S reduced the binding of Ab2 to the promoter of the selected DREAM target genes (Fig. 3D), corroborating our hypothesis and validating our ChIPseq strategy.

-Authors in Figure 3G should select an equal-sized population of randomly chosen non-DREAM-target genes, otherwise, the comparison of log2FC difference between these two gene cohorts is unreliable and difficult to make. Mann-Whitney test should also be performed.

We thank the reviewer for this suggestion, which was added to the revised version of the manuscript (Fig. 3E, lower panel).

-Authors should repeat the cell cycle analysis (Figure 4A) as the number of cells subjected to flow cytometry is quite discrepant between the conditions. Also, it is not clear if the experiment was performed in at least biological triplicates (although in the respective legend, it is stated so). If performed in biological triplicates, authors should make a new graph where each cell cycle phase cell population differs between the two conditions. Moreover, the difference in cell cycle defects in LOXL2-inhibited cells (Figure 4C) is indifferent compared to their control counterpart. Therefore, authors should address these inconsistencies.

We thank the reviewer for the suggestion. In the revised version of the manuscript, we represent the cell cycle also as a bar plot with statistical analysis (Fig. 4A, C). Even though the number of cells was the same across conditions, the sub-G1 population of the LOXL2 KD cells may have distorted the profile of the cell cycle. To avoid misinterpretations, we repeated the analysis in the revised version of the manuscript. Statistical analysis supports that LOXL2 inhibition or downregulation has a significant effect on cell cycle progression (Fig. 4A, C, right panel).

-Furthermore, authors should explain what was the rational selecting a mediator subunit and specifically MED1 as a possible interacting partner of LOXL2 and BRD4s since MED12 and MED24 were also highly essential (Figure 4F).

We selected MED1 as a Mediator Complex proxy. In our essentiality analysis MED 1, 9, 10, 12, 15, 16, 19, 23, 24, 25 score as significant, suggesting a functional interaction between LOXL2 and the Mediator Complex, rather than a specific subunit. MED1 has been previously described as a BRD4 partner and it is often used in immunofluorescence to visualize transcriptional foci, which made it the best candidate for follow-up study in our project.

-Moreover, do authors also observe this functional relationship of LOXL2 and BRD4S in cell cycle progression in other breast cancer subtypes presenting a high proliferation index e.g HER2+?

Presumably, the author's proposed mechanism applies to a wide panel of breast cancer entities, for which, only key experiments could be performed.

We thank the reviewer for the suggestion. We hypothesized that other cancer types expressing LOXL2 and BRD4S could also benefit from the combinatorial treatment. Indeed, the CCLE drug sensitivity panel in Fig. 1A comprises cancer cell lines of different origins, not just TNBC, and corroborates that the relationship between LOXL2 expression levels and BRD4 sensitivity exist also beyond TNBC. Even though it is important to experimentally verify this hypothesis, we decided to pursue it in the future to broaden the applicability of the proposed strategy in preclinical settings.

-Authors in Figure 5H represent LOXL2 and BRD4s as integral chromatin looping factors together with MED1 at promoter and enhancer regions. However, this illustration is an overrepresentation of their finding because authors did not address the differential occupancy of BRD4S upon LOXL2 loss in DREAM-target-specific enhancer regions. If they wish to do so, they may use the RANK ORDERING OF SUPER-ENHANCERS (ROSE) package to call for super-enhancer regions in the proximity of DREAM-targets and confirm similar results as for their TSS-proximal sites.

We thank the reviewer for the useful suggestion. In the new version of the manuscript, we have simplified the representation, which now does not show super-enhancers. However, following the reviewer’s suggestion, we performed super enhancer analysis using ROSE. Results showed that BRD4S binds to super-enhancers more than BRD4L, including DREAM target gene super-enhancers. Additionally, while LOXL2 KD did not alter the binding of LOXL2 to DREAM target gene super-enhancers, it decreased the binding of BRD4S to them (Fig. EV7D, E). Overall, these data are in agreement with our hypothesis that BRD4S together with LOXL2 controls the expression of DREAM target genes.

-In the current manuscript, authors did not address the translational relevance of their proposed mechanism in the context of conventional therapies. Knowing that several BRD-specific compounds currently undergo clinical trials, authors should address if LOXL2 low (MDAMB468) and high (BT549) cells demonstrate a differential sensitivity to increasing doses of chemotherapy, in the presence or absence of BRD4. By doing that, LOXL2 apart from being a therapeutic target could be also used as a prognostic marker to stratify patients and achieve better response to standard therapies.

We really appreciate the reviewer’s suggestion and we think this is a fundamental point. In the new version of the manuscript, we have performed further analysis using a greater panel of chemotherapeutic agents from the CCLE sensitivity database. We now show that LOXL2 low-expressing cells show significantly more sensitivity to BETi treatments, but not to conventional chemotherapeutic agents (e.g. doxorubicin, Olaparib, 5-fluorouracil, paclitaxel, etc.) (Fig. 1A and Fig. EV1), which set the rationale to further explore the functional relationship between BRD4 and LOXL2.

Minor points:

-In Figure 1D, the authors should convert the y-axis to a logarithmic scale to better represent the differences between JQ1, PXS, and combo. Also, One-way Anova should be performed between JQ1, PXS and combo.

We don’t understand the reviewer’s suggestion since Fig. 1D (Fig. 6B, right panel in the revised version) is a tumor picture for which the y-axis cannot be converted to a logarithmic scale.

-In Figure S6F, authors did not show the sensitivity of LOXL2 low and high cell lines for BRD4 KO. If LOXL2-proficient cells are less sensitive to JQ1, based on Figure 1B, authors should consider showing something similar from the gene essentiality database.

We agree with the reviewer and we apologize for this mistake. We have included the sensitivity of LOXL2 low and high cell lines for BRD4 KO and also for MYC KO (Fig. EV6G).

-Authors failed to discuss the work from Ozge Saatci et al (PMID: 32415208) regarding LOXL2 in TNBC and ECM reorganization as well as in other cancer entities (PMID: 35428659) in the context of ECM remodeling. Authors should realize that these published works and the current ones are not conflicting but complement each other.

We thank the reviewer for the suggestion. In the revised version of the manuscript, we discussed this work.

Reviewer #3 (Significance (Required)):

SIGNIFICANCE

The conception and findings are of enlightening significance for TNBC therapy, especially given the lack of targeted therapies in this particularly aggressive breast cancer subtype. Hence, I posit this work as highly relevant for the cancer epigenetics research community interested in characterizing unknown factors that facilitate the gene-activating function of epigenetic readers in health and disease.

My field of expertise is to uncover epigenetic vulnerabilities responsible for transcriptional plasticity driving drug tolerance in aggressive forms of breast cancer.

We would like to take the opportunity to thank the reviewer for the relevant suggestions. We strongly believe the revised version of the manuscript has been substantially improved by addressing the comments the reviewer made.

Author Response

Many thanks for the detailed and sometimes sharp, yet appropriate criticism of our study. It was an incentive for us to carry out additional analyses and to devote more effort to an elaboration of concepts. The outcome is that the results have changed slightly and that we now give more space to a discussion of concepts. We first address here the points raised by more than one reviewer before responding to comments contributed by individual reviewers.

The points raised can be divided into three thematic groups, 1) conceptual issues, 2) experimental and analytical questions, and 3) comments challenging the novelty of our results. On the first theme, we think it is essential to make a clear distinction between the conceptual and observational domains. As such, the criteria defining a “mirror neuron” and what is meant by the term "mirror mechanism" belong to the conceptual domain. This understanding of terms requires agreement among scientists, but is not experimentally testable. Unfortunately, there is no agreement on how to define a “mirror neuron” and what is meant by “mirror mechanism”. Thus, for the present work, the only option is to refer to specific definitions or to use our own, definitions which try to capture what others, and here most importantly Rizzolatti and colleagues, probably meant. We have adjusted the introduction in an attempt to convey our understanding and usage of the two terms in a hopefully comprehensible manner. Briefly, we use a definition for "mirror neuron" that we take from the first paragraph of the results section of Gallese et al. (Brain, 1996). We do not consider the "properties of mirror neurons" described in that paper as defining a mirror neuron (MN). Classifying neurons as MNs only on the basis of the presence of a modulation of discharge rate during an executed and an observed action compared with a baseline is a common practice also in other single neuron studies on MNs, consistent with this definition. Regarding "mirror mechanism", we refer to Rizzolatti and Sinigaglia (2016) and make a distinction between a broad and a strict definition. Given our finding that there are almost no F5 MNs whose activity during observation is a motor representation according to our strict definition of a mirror mechanism, and also given the problem that the term “mirror mechanism” itself is not uniformly understood, the question arises whether and how the term "mirror neuron" should be used in the future. The answer to this may vary and belongs to the conceptual domain. We briefly address this question at the end of the discussion of the revised manuscript.

From that understanding of terms, conceptual hypotheses are to be distinguished, which of course must allow experimental predictions, i.e., must be falsifiable. We now distinguish more clearly between a "representation hypothesis" and an "understanding hypothesis". Both hypotheses focus on F5 MNs and are based on the strictly defined mirror mechanism. We test the “representation hypothesis” in our study, and just because it is the basis for the “understanding hypothesis”, falsifying the “representation hypothesis” would allow us to conclude that the “understanding hypothesis” is not valid. In contrast, confirmation of the “representation hypothesis” would not, of course, allow us to conclude that the “understanding hypothesis” holds. That would really be circular reasoning (this conclusion was drawn by some and rightly criticized). However, support for the “representation hypothesis” would be the necessary prerequisite for the “understanding hypothesis” to be true. These two hypotheses take up the original argument that a certain understanding of observed actions could follow from an equality of action-specific F5 MN activity during execution and observation. Because we considered the data on equality of action- specific F5 MN activity to be insufficient, we designed this study. Since our result largely argues against the "representation hypothesis" and thus against the "understanding hypothesis," we now discuss alternative concepts for the function of F5 MNs in more detail. It should be noted here that our fourth concept ("goal-pursuit-by-actor") could well represent the observed action without contradiction to our broad definition of a mirror mechanism, which in principle could also serve a subjective experience (which could be conceived as a kind of understanding). The way we structure the concepts in the discussion of this revised manuscript is, in our opinion, a useful overview of the concepts. The third concept is new in this context. We would like to emphasize that we focus on F5 MNs and intentionally avoid a discussion of mirror neurons beyond F5 in this paper. With the data from this study, we cannot say anything about MNs outside of F5.

Regarding the key question of how the "understanding hypothesis" is testable, or whether it may not be testable at all, we agree, of course, that for the conclusion of whether F5 MNs contribute to perception, only a manipulation of F5 MNs can clarify it. We now say that explicitly in the introduction. We agree with reviewer #2 that "understanding" here is not limited to "action recognition" or "action categorization”, which in principle could be implemented by purely sensory processing. Therefore, we also do not believe that the approach proposed by reviewer #3, which builds on the distinction of actions, would allow for a critical examination of the "understanding hypothesis”. But we disagree that the "understanding hypothesis" is not testable at all. Operationalization is necessary. If we accept that we can measure certain visual or auditory perceptions of an animal by operationalization (e.g., the subjective visual vertical, see for example Khazali et al., PNAS, 2020), then we must also accept that we can, in principle, measure other subjective experiences by operationalization, such as pain or aiming at a goal or even the co- experience of pain. An example of how to approach this is the study by Carrillo et al. (Curr Biol, 2019), which reviewer #2 and colleagues discussed in a recent review article (Bonini et al., TCS, 2022).

With regard to the second theme, experimental and analytical questions, we noticed while reading the comments that in our first version we did not distinguish clearly enough between statements about single neurons and statements about populations of neurons. Therefore, we now clearly separate single neuron analysis and population code analysis in the structure of the article. In view of the fact that statements about mirror neurons in the literature mostly refer to single neurons, we added extensive single neuron analyses, so that only now statistically reliable statements about single neurons are possible. This has led to the realization that the number of neurons with exclusively shared code is so small that these neurons should be considered a rare exception. Given the small number of time periods with shared code, we additionally tested against a hypothesis already rightly proposed as an alternative explanation by G. Csibra in 2005 (Mirror neurons and action observation: Is simulation involved? In: What do mirror neurons mean? Interdisciplines Web Forum 2005). We were able to reject this hypothesis based on two of three methods for testing for a shared code. This is the second piece of evidence besides the clustering of time periods with shared code already described in the first version that time periods with shared code cannot be considered random.

We discuss in more detail the question of whether neurons that exhibit a shared code at least at times support the representation hypothesis. To this end, we additionally examined whether certain action segments are more frequently represented with a shared than with a non-shared code, whether neurons with shared code differ from those with non-shared code in anatomical location, and whether an accuracy can be achieved with a time bin-wise selection of neurons with shared code by population cross-task classifiers as with within-task classifiers in the whole population.

Another issue was how to test for shared code and how to decide if a code has enough sharing. To answer the question, the exact hypothesis we intended to test here is crucial. The representation hypothesis states that the representation of the observed actions in F5 MNs corresponds to the representation as it occurs during the execution of the same actions. Therefore, the relationship between discharge rate and actions that holds during execution should also hold during observation, which is measurable with a classifier trained on execution trials and tested on observation trials. Moreover, the actions should not be more distinguishable during observation with a classifier other than the execution-trained classifier, because if that were so, it would mean that the representation of observed actions is different from that of executed actions. The detection of a cluster of time bins for which both conditions are satisfied confirms that it is possible to discover in this way the shared codes postulated by the representation hypothesis.

With respect to concerns that the monkey may not have used the cue at all when the action was executed, we added a comparison with control trials with a non-informative cue and also compared the duration of the approach phase between the three actions. Regarding oculomotor behavior, we verified that the monkey had actually directed his gaze toward the action during action observation for all three actions.

On the third issue, concerning the novelty of our results, we have now explained in more detail in the introduction why we felt it necessary to conduct a study we considered fundamental. As a result of our study, it can be clearly stated now that representations of observed actions as predicted by the strictly defined mirror mechanism are rare in F5 MNs, but nevertheless cannot be dismissed as random. This dispels the objection rightly raised by Csibra in 2005 and contradicts the currently prevailing view that such a representation can only be found at a population level. Even if these representations are ultimately explained by a concept other than the strictly defined mirror mechanism, their existence must be accounted for by any theory of the function of F5 neurons. Moreover, it is also shown that the observed actions are well discriminated with a non- shared code, at times even optimally. This contradicts the notion – which has been widespread for a long time since the work of Gallese et al. (Brain, 1996) – that mapping to motor representations in terms of broad congruence is simply not perfect. The applied cross-task decoding approach seems promising to test also in the future for a shared action code. Finally, reconsideration of alternative concepts has led us to highlight the possibility of a representation of a goal pursuit by the observer.

Reviewer #1 (Public Review):

The authors set out to investigate the hypothesis that mirror neurons in ventral premotor area F5 code actions in a common motor representation framework. To achieve this, they trained a linear discriminant classifier on the neural discharge of three types of action trials and test whether the thus trained classifier could decode the same categories of actions when observed. They showed that codes were fully matched for a small subset of neurons during the action epoch, while a wider set of "mirror neurons" showed only poorly matched codes for different epochs.

This is one of the descriptions of our results, where we realized that in our first version we did not distinguish clearly enough between statements about single neurons and statements about populations of neurons. This prompted us to perform a detailed single neuron analysis.

The authors controlled for potential visual object confounds by having identical objects be manipulated in three different ways and by having the animal carry out the motor execution in the dark. The main strength of the study lies in the clever decoding approach testing the matched tuning to behavioural categories in a model-free way. The central result is in the identification of the small sub-group of mirror neurons that show true matching during the execution epoch, which can dissociate the three types of action almost perfectly. This aligns well with some previous work while offering a novel avenue to identify and investigate those neurons. The underlying neuronal mechanism and behavioural relevance of these neurons remain an open question. It would have been interesting to understand better whether the specific motor representations at a recording site, for instance identified through microstimulation prior to recording (see Methods), the reaction times on individual trials or the specific gaze targets (object/hand) had a bearing on the decoding performance for a neuron/trial.

We agree that these are interesting questions.

In this study, the focus is on testing for a shared code according to a strictly defined mirror mechanism. We have now compared the anatomical locations of neurons with only time bins in which observed actions were discriminated with a shared code (according to one of the methods) to the locations of neurons with only time bins with non-shared code (see last paragraph in Results). We did not find any relevant difference and this is why one cannot expect topographically specific effects of microstimulation.

We do not expect the reaction time (i.e., the time interval between LED onset and start button release, or the duration of the approach epoch) during execution or observation to have any effect on our results on shared coding as the analysis was based on relative time bins. The observed actions were predominantly distinguished late in the approach epoch, but especially in the manipulation epoch. At this time, reaction time is not expected to have a relevant influence.

The relationship between gaze/eye position and the activity of mirror neurons, during execution or observation, is an interesting topic in itself. However, for testing for a shared code according to a strictly defined mirror mechanism, it is only relevant that the observing monkey actually observes the action. We have ensured this in our experiment by a fixation window and have now also confirmed that the monkey actually looked into the area of the object during all three actions (see Results, lines 209-219 in the manuscript with tracked changes).

Ultimately, the uncovered matched mirror representations should in future experiments be tested with causal interventions and linked trial-by-trial to action selection performance.

The authors put the focus of their discussion on the wider, less well-matched neuronal pool to support an action selection framework, which is of course a valid view and well established in motor representations. From a sensory perspective, sparse coding, as suggested by the small group of "true" mirror neurons identified with the decoding approach, should also be considered as the basis for a possible neuronal mechanism. A particular strength of the paper is that it could give new data and impetus to the important discussion about how motor and sensory coding frameworks come together in cortical processing.

We have expanded the discussion considerably and also address the possibility of sparse coding.  

Reviewer #2 (Public Review):

The paper by Pomper and coworkers is an elegant neurophysiological study, generally sound from a methodological point of view, which presents extremely relevant data of considerable interest for a broad audience of neuroscientists. Indeed, they shed new light on the mirror mechanism in the primate brain, trying to approach its study with a novel paradigm that successfully controls for some important factors that are known to impact mirror neuron response, particularly the target object. In this work, a rotating device is used to present the very same object to the monkey or the experimenter, in different trials, and neurons are recorded while the monkey (motor response) or the experimenter (visual response) performed a different action (twist, shift, lift) cued by a colored LED.

The results show that there is a small set of neurons with congruent visual and motor selectivity for the observed actions, in line with classical mirror neuron studies, whereas many more cells showed temporally unstable matched or even completely non-matched tuning for the observed and executed actions. Importantly, the population codes allow to accurately decode both executed and observed actions and, to some extent, even to cross-decode observed actions based on the coding principles of the executed ones.

In my view, however, the original hypothesis that an observer understands the actions of others by the activation of his/her motor representations of the observed actions constitutes circular reasoning that cannot be challenged or falsified, as the author may want to claim. Indeed, 1) there is no causal evidence in the paper favoring or ruling out this hypothesis (and there couldn't be), 2) there is no independent definition (neither in this paper nor in the literature) of what "action understanding" should mean (or how it should be measured). Instead, the findings provide important and compelling evidence to the recently proposed hypothesis that observed actions are remapped onto (rather than matched with) motor substrates, and this recruitment may primarily serve, as coherently hypothesized by the authors, to select behavioral responses to others (at least in monkeys).

1) One of the main problems of this manuscript is, in my view, a theoretical one. The authors follow a misleading, though very influential, proposal, advanced since the discovery of mirror neurons: if there are (mirror) neurons in the brain of a subject with an action tuning that is matched between observation and execution contexts, then the subject "understands" the observed action. This is clearly circular reasoning because the "understanding" hypothesis uniquely derives from the neuron firing features, which are what the hypothesis should explain. In fact, there is no independent, operational definition of the term "understanding". Not surprisingly there is no causal evidence about the role of mirror neurons in the monkey, and the human studies that have claimed to provide causal evidence of "action understanding" ended up using, practically, operational definitions of "recognition", "match-to-sample", "categorization", etc. Thus, "action understanding" is a theoretical flaw, and there is no way "to challenge" a theoretical flaw with any methodologically sound experiment, especially when the flaw consists of circular reasoning. It cannot be falsified, by definition: it must simply be abandoned. On these bases, I strongly encourage the authors to rework the manuscript, from the title to the discussion, by removing any useless attempt to falsify or challenge a circular concept and, instead, constructively shed new light on how mirror neurons may work and which may be their functional role.

Please see the response to all.

2) An important point to be stressed, strictly related to the previous one, concerns the definition of "mirror neuron". I premise that I am perfectly fine with the definition used by the authors, which is in line with the very permissive one adopted in most studies of the last 20 years in this field. However, it does not at all fulfill the very restrictive original criteria of the study in which "action understanding" concept was proposed (see Gallese et al. 1996 Brain): no response to object, no response to pantomimed action or tool actions, activation during execution in the dark and during the observation of another's action.

We do not agree that the enumerated "very restrictive original criteria" emerge from the Gallese et al. (Brain, 1996) study. Except for the first paragraph in the results section, there is no clear statement on how mirror neurons should be defined.

If the idea (which I strongly disagree with) was to simply challenge a (very restrictive) definition of mirroring (a very out-of-date one, indeed, and different from the additional implication of "action understanding"), the original definition of this concept should be at least rigorously applied. In the absence of additional control conditions, only the example neuron in Figure 2A could be considered a mirror neuron according to Gallese et al. 1996.

We have the impression that the question does not distinguish clearly enough between the definition of "mirror neuron" and the definition of "mirror mechanism". In defining "mirror mechanism", we refer to the work of Rizzolatti and Sinigaglia (Nat Rev Neurosci, 2016). We do not think that this definition is out-of-date (see for example the 2018 article by Rizzolatti and Rozzi in Handbook of Clinical Neurology). If the term "mirror mechanism" is to be defined differently, then another term should be used for a new definition or an annotation should be added (such as "version 2"). This would be necessary to avoid unnecessary confusion resulting from unclear terms.

Permissive criteria implies that more "non-mirror" neurons are accepted as "mirror": simply because they are permissively named "mirror", does not imply they are mirroring anything as initially hypothesized

Even for a neuron that would be classified as a "mirror neuron" according to your previously stated "very restrictive original criteria”, it does not follow that it "mirrors” according to a mirror mechanism. And, of course, it is quite possible that more neurons do not "mirror” according to a mirror mechanism if one tests more neurons.

(Example neuron in Fig 2B, for example, could be related to mouth, rather than hand, movements, since it responds strongly and similarly around the reward delivery also during the observation task, when the monkey should be otherwise still).

We agree, it is not excluded that this neuron has a relation to mouth movements. However, since the neuron meets the conditions to be classified as a "mirror neuron", an additional relation to mouth movements would not be relevant. If mouth movements are to be an exclusion criterion, then this would have to be included and justified in the definition of a "mirror neuron".

Clearly, these concerns impact all the action preference analyses. To practically clarify what I mean, it should be sufficient to note that 74% (reported in this study) is the highest percentage ever reported so far in a study of neurons with "mirror" properties in F5 (see Kilner and Lemon 2013, Curr Biol) and it is similar to the 68% recently reported by these same authors (Pomper et al. 2020 J Neurophysiol) with very similar criteria. Clearly, there is a bias in the classification criteria relative to the original studies: again, no surprise if by rendering most of the recorded neurons "mirror by definition" then they don't "mirror" so much. I suggest keeping the authors' definition but removing the pervasive idea to challenge the (misleading) concept of understanding.

We think that it is very important to clearly separate "mirror neuron" from "mirror mechanism". And the question arises whether one should not include a mirroring criterion, which is derived from a definition of a mirror mechanism, in the definition of mirror neurons. We address this briefly in the discussion. Ultimately, the point of our study is to find out how many of the - if you want to put it that way - "permissively defined" mirror neurons actually “mirror”. And the answer depends on how one defines “mirror mechanism”. We provide an answer by resorting to a “strictly defined mirror mechanism”. We have now also given throughout the results section the percentages of neurons with certain properties with respect to all measured F5 neurons. This is a reference that allows comparisons among studies, provided that no neurons were directly discarded during recording, which we avoided in our study.

3) It would be useful to provide more information on the task. Panel B in Figure 1 is the unique information concerning the type of actions performed by the monkey and the experimenter. Although I am quite convinced of the generally low visuomotor congruence, there are no kinematics data nor any other evidence of the statement "the experimental monkey was asked to pay attention to the same actions carried out by a human actor". First, although the objects were the same, the same object cannot be grasped or manipulated in the same way by a human and a macaque, even just because of the considerable difference in the size of their hands; this certainly changes the way in which monkeys' and experimenter's hands interact with the same object, and this is a quantifiable (but not quantified) source of visuomotor difference between observed and executed actions and a potential source of reduced congruency.

We agree, of course, that there are kinematic differences in how a monkey and how a human manipulate the same object. We have not measured the kinematics and thus cannot make a systematic statement about this. We now report in the results section the rather incidental observation that already the reaching trajectories for the three actions differed and show corresponding differences in the timing of the approach epoch. However, for the question of this study, how many neurons are eligible to represent observed actions according to a strictly defined mirror mechanism, the kinematic repertoire of the observed actor is irrelevant. The reference is the F5 mirror neuron activity during the monkey's own action, i.e., how the monkey approaches the object with his hand, how he grasps it, and how he brings it to a certain target position and holds it there. The observed action, according to the strictly defined mirror mechanism, is to be mapped to this reference. Therefore, we did not collect kinematic data. But it is of course a possible explanation for a non-shared code if the strictly defined mirror mechanism does not apply.

Second, there is little information about monkey's oculomotor behavior in the two conditions, which is known to affect mirror neuron activity when exploratory eye movements are allowed (Maranesi et al. 2013 Eur J Neurosci), potentially influencing the present findings: a {plus minus}7 (vertical) and {plus minus}5 (horizontal) window at 49 cm implies that the monkey could explore a space larger than 10 cm horizontally and 14 cm vertically, which is fine, but certainly leaves considerable freedom to perform different exploratory eye movements, potentially different among observed actions and hence capable to account for different "attention" paid by the monkey to different conditions and hence a source of neural variability, in addition to action tuning.

We agree that the topic of the relationship between F5 MNs activity and eye movements is interesting. And we know from the work of Maranesi et al. (2013) that at least larger eye movements during action observation are related to the activity of F5 MNs. In our study, we ensured that the observing monkey was actually observing the action. For this purpose, we used a fixation window. We now additionally verified that the monkey really looked into the area of the object during all three actions (see Results, lines 209-219 in the manuscript with tracked changes). In our study, the fixation window was so small that the monkey could not see the face of the human actor, in contrast to the study of Maranesi et al. (2013). It was mainly the face that attracted the monkey's attention in that study (measured by gaze position). In our study, the risk that the gaze of observing monkey was out of the fixation window was high when he looked at the human actor's hand above the wrist. The execution of the action by the monkey took place in darkness. We did not use a fixation window because the monkey's own execution of the action can be assumed to direct his attention to the action.

We cannot rule out the possibility that smaller eye movements during observation, larger eye movements during execution in darkness, covert shifts of spatial attention, or more generally attentional fluctuations have an influence on F5 MNs that might have counteracted a shared action code in our study. However, if this were the case, then the investigated hypothesis that the activity of F5 MNs during action observation is a motor representation according to the strictly defined mirror mechanism would also have to be rejected.

4) Information about error trials and their relationship with action planning. The monkey cannot really "make errors" because, despite the cue, each object can be handled in a unique way. The monkey may not pay attention to the cue and adjust the movement based on what the object permits once grasped, depending on online object feedback. From the behavioral events and the times reported in Table 1, I initially thought that "shift" action was certainly planned in advance, whereas "lift" and "twist" could in principle be obtained by online adjustments based on object feedback; nonetheless, from the Methods section it appears that these times are not at all informative because they seem to depend on an explicit constraint imposed by the experimenters (in a totally unpredictable way). Indeed, it is stated that "to motivate the monkey even more to use the LED in the execution task, another timeout was active in 30% (rarely up to 100%) of trials for the time period between touch of object to start moving the object: 0.15 (rarely 0.1) for a twist and shift, 0.35 (rarely 0.3s) for a lift". This is totally confusing to me; I don't understand 1) why the monkey needed to be motivated, 2) how can the authors be sure/evaluate that the monkeys were actually "motivated" in this way, and 3) what kind of motor errors the monkey could actually do if any. If there is any doubt that the monkeys did actually select and plan the action in advance based on the cue, there is no way to study whether the activity during action execution truly reflects the planned action goal or a variety of other undetermined factors, that may potentially change during the trials. Please clarify.

It is true that the three actions could in principle be performed without using the LED as an informative cue. While this is unlikely under the assumption that a monkey prefers the easiest and fastest way to get reward, it remains a possibility. For this reason, we introduced time constraints in a part of the trials. The selection of time constraints and the proportion of trials in which they were applied, was a pragmatic compromise between a time limit, at which the LED must be used as an informative cue for action selection in order to comply with the task, and a time span that allows the task to be completed even when overall motivation is low. The latter takes into account the general experimental experience that a monkey's engagement or motivation in such experiments varies across trials, sessions, and days. To evaluate whether the LED color was, indeed, used as a cue for action planning in the execution task, we randomly interleaved trials with a different LED, non-informative regarding the type of object, as a control in 5% of the trials. We compared the behavioral responses in trials with informative cues and those with a non-informative cue. The behavioral analysis established that both monkeys indeed used the informative cues to guide their choices (see Fig. 1D).

Further evidence that the monkey used the cue for action selection and planning is the finding that the type of action was encoded before the release of the start button and then further during the approach phase, i.e., much earlier than somatosensory feedback about the manipulability of the object was available (see Fig. 3A and Fig. 6A).

Regarding the question, which "motor errors" were possible: The answer can be found in the description of the cases in which a trial was aborted (see Material and methods): releasing the start button too early (< 100 ms after turning on the LED), manipulating the object too slowly after touching it (the time constraints mentioned), not holding the object until the reward was given, or not performing the task at all (10 s timeout).

5) Classification analysis. There seems to be no statistical criterion to establish where and when the decoding is significantly higher than chance: the classifier performance should be formally analyzed statistically. I would expect that, in this way, both the exe-obs and the obs-exe decoding may be significant. Together with the considerations of the previous point 2 about the permissive inclusion criteria for mirror neurons, this is a remarkable (even quite unexpected) result, which would prove somehow contrary to what the authors claim in the title of the paper. The fact that in any classification the "within task" performance is significantly better than the "between task" performance does not appear in any way surprising, considering both the inclusive selection criteria for "mirror neurons" and the unavoidably huge different sources of input (e.g. proprioceptive, tactile, top-down, etc. afferences) between execution and observation. So, please add a statistical criterion to establish and show in the figures when and where the classifications are significantly above chance.

We have added - in addition to the statistics already performed in the first version (Fig. 3A in the previous version, now Fig. 6A) - a number of analyses including statistics. This mainly concerns the analyses regarding a shared code at the single neuron level, in which we additionally tested against the null hypothesis proposed by Csibra in 2005 using permutation tests. And we have now also calculated confidence intervals for the population classifications that allow the comparison with chance level. We re-performed the classification analyses using eight-fold cross-validation. We also added a statistical analysis to the finding of clustering of time periods with shared code (Fig. 4). In Figure 5, we additionally compared the frequency of action segments with shared and non-shared codes, which is a descriptive, exploratory analysis. For this reason, it does not make sense to perform inferential statistics. Overall, these analyses represent a significant expansion of the analyses in the first version. We have done this primarily to arrive at statistically sound conclusions at the single neuron level.

Regarding the comparison between within-task classification (o2o) and cross-task classification (e2o), it is important to keep in mind that the goal was to test the hypothesis that the activity of F5 MNs during action observation is a motor representation of the observed action according to the strictly defined mirror mechanism. This hypothesis requires both, 1) an above chance level accuracy of the e2o classifier and 2) no better accuracy of the o2o classifier as compared to the e2o classifier. If the o2o classifier were better, then the actions would not be represented as they are executed. And the reference in this hypothesis is the motor representation, that is, the code at execution. Thus, the direction e2o classification is the crucial one, not the reverse direction (o2e). One explanation for the fact that o2o shows better accuracy in the population may be the different sensory inputs mentioned above. In this case, the tested hypothesis has to be rejected and replaced by another one, which should then have a different name.

Nevertheless, we also show the result of the o2e cross-task classification in Fig. 6 (yellow curve), which was already included in Fig. 3 of the first version. However, we do not address it in more detail in the main text because it is not relevant for the hypothesis to be tested. It is only a reportable additional result.

6) "As the concept of a mirror mechanism posits that the observation performance can be led back to an activation of a motor representation, we restricted this analytical step to a comparison of the exe-obs and the obs-obs discrimination performance". I don't understand the rationale of this choice. The so-called "concept" of mirror mechanism in classical terms posits that mirror neurons have a motor nature and hence their functioning during observation should follow the same principle as during action execution. But this logical consideration has never been demonstrated directly (it is indeed costated by several papers), and when motor neurons are concerned (e.g. pyramidal tract neurons, see Kraskov et al. 2009) their behavior during action observation is by far more complex (e.g. suppression vs facilitation) than that hypothesized for classical "mirror neurons". Furthermore, when across-task decoding for execution and observation code has been used, both in neurophysiological (e.g. Livi et al. 2019, PNAS) and neuroimaging (Fiave et al. 2018 Neuroimage) data, the visual-to-motor direction typical produce better performance than the opposite one. Thus, I don't see any good reason not to show also (if not even just) the obs-exe results. Furthermore, I wonder whether it is considered the possible impact of a rescaling in the single neuron firing rate across contexts, as the observation response is typically less strong than the execution response in basically all brain areas hosting neurons with mirror properties, and this should not impact on the matching if the tuning for the three actions remains the same (e.g. see Lanzilotto et al. 2020 PNAS). The analysis shown in Figures 4 and 5 is, for the rest, elegant and very convincing - somehow surprising to me, as the total number of "congruent" neurons (7.5%) is even greater than in the original study by Gallese et al. (5.4%).

As to the rationale of our approach, please see our response to the previous point.

On the issue of rescaling: the hypothesis tested here requires that the F5 MNs activity on observation is a motor representation of the observed action. Hence, from the activity during observation the action should be just as readable as from the execution-related activity. If we had to use rescaling to find a shared code, then observed actions would not be represented in F5 MNs in the same way as on execution. Additional information on whether the action is being executed or observed would be needed. This would of course be possible in principle, but would contradict the hypothesis. And we then not only have the difficulty of which readout is the physiological one (here we make a parsimonious assumption with a linear readout), but we would have to make an additional assumption about rescaling. For this study, we have now chosen the solution of performing the action preference analysis on a single neuron level in a statistically clean way. This represents a very liberal form of rescaling, as it only tests whether the action with the highest or lowest discharge rate is the same when executed and observed. That is, if the result here is not fundamentally different, which is the case, then it can also be assumed that one does not get qualitatively different results for other forms of rescaling.

7) The discussion may need quite deep revision depending on the authors' responses and changes following the comments; for sure it should consider more extensively the numerous recent papers on mirror neurons that are relevant to frame this work and are not even mentioned.

The discussion has been thoroughly revised considering the comments raised and suggestions of this and the other two reviewers.

Reviewer #3 (Public Review):

Mirror neurons are a big deal in the neuroscience literature and have been for thirty years. I (and many others) remain skeptical of whether they serve the functions often attributed to them - specifically, whether they are motor planning neurons that contribute to understanding the actions of others. Testing their functions, therefore, is of great interest and importance. The present study, however, is not a cogent or convincing test. I do not think this study helps to answer the questions surrounding mirror neurons. It purports to provide a crucial test, that comes out mostly against the mirror neuron hypothesis, but the test has too many weaknesses to be convincing.

Thank you for the clear words. We take from it, first of all, that in the first version of the manuscript we failed to convey the relevance of our study for the discussion of mirror neuron function. The concerns of this reviewer are in line with those of the others and are addressed in our response to all three reviewers.

First, consider that the motor tuning and the visual tuning match "poorly." How poor or good must the match be before the mirror neuron hypothesis is rejected? I do not know, and the study does not help here. Even a "poor" match could contribute significantly to a social perception function.

The specific hypothesis tested here assumes that an action-specific activity of F5 MNs evoked by observed actions corresponds to an action-specific activity of these actions if executed. The approach taken here to compare cross-task classification accuracy (execution-trained, tested in observation) with within-task classification accuracy (observation-trained, tested in observation) tests this hypothesis. The fact that we found a cluster of time periods of single neurons in which both accuracies are almost equal supports this approach and also the hypothesis for these time periods. In principle, of course, the decision for the presence of a difference or equality is always only a statistical statement and contains assumptions. For example, the assumption that a linear readout has physiological relevance enters here. But this problem exists in all studies that ultimately try to understand biological neuronal networks in order to explain perceptions and behavior. However, it is such studies that attempt to elucidate what information is contained in which neurons that set the stage for experiments that, in the optimal case, manipulate certain neurons in a particular way in order to then measure the behavior of an animal that is just right for those neurons.

Second, the results remind me in some ways of other multi-modal responses in the brain. For example, in the visual area MST, neurons are tuned to optic flow fields that imply specific directions of self-motion. Many of the same neurons are tuned to vestibular signals that also imply specific directions of self-motion. But the optic flow tuning and the vestibular tuning are not perfectly matched. There is considerable slop and complexity in how the two tunings compare within individual neurons. That complexity is not evidenced against multi-modal tuning. Instead, it suggests a hidden-layer complexity that is simply not fully understood yet. Just so here, the fact that the apparent motor tuning and apparent visual tuning match "poorly" is not evidence against both a motor planning and a visual encoding function.

We hope that it is now clearer, in contrast to the first version, that we tested a specific hypothesis that is only a prerequisite for the hypothesis of a very specific form of understanding. Referring to the example, the hypothesis analogous to ours would be that the representation of self-motion direction due to optic flow ("observation") corresponds to the representation of self-motion direction due to vestibular stimulation ("execution"). If it were then found that the self-motion direction due to optic flow cannot be predicted from a classifier trained on vestibular stimulation, and that another classifier trained on optic flow performs better, then the hypothesis would have to be rejected. This is then a reason to realize that "everything is a bit more complex" and to search for better explanations.

Third, the animals are massively over-trained in three actions. They perform these actions and see them performed thousands of times toward the same object. Surely, if I were in the place of the monkey, every time I saw the object, I'd mentally imagine all three actions. As I saw a person act on the object, I'd mentally imagine the alternative two actions at the same time. Even if the mirror neuron hypothesis is strictly correct, this experiment might still find a confusion of signals, in which neurons that normally might respond mainly to one action begin to respond in a less predictable way during all three trial types.

In our study, we tested a specific hypothesis related to the time an action is observed. Here, you suggest an alternative hypothesis. The question is whether this alternative hypothesis better explains the result of our study. The alternative hypothesis can be formulated as follows: the F5 MNs activity elicited by an observed action in this experiment corresponds to a mixture of the activities that occur when the other two actions are executed. This hypothesis is to be rejected because it fails to explain why a shared code occurs in single neurons and why cross-task population classifiers show an accuracy above chance level. A modified alternative hypothesis, which states that what is represented in the experiment during observation is a mixture of all three actions, cannot explain why the three actions are very well represented in the population and are optimally represented exactly when the target position of the object is reached.

Fourth, the experiment relies on a colored LED that acts as an instructional cue, telling the monkey which action to perform. What is to stop the neurons from developing a cue-sensitive response, as in classic studies from Steve Wise and others in the premotor cortex? Perhaps the neuronal signal that the experimenters are trying to measure is partly obscured by other, complex responses influenced in some manner by the instructional cue?

In principle, there is the possibility that purely sensory information is also represented in area F5, at least in some neurons or at certain points in time. We take your suggestion and discuss this as one of the alternative concepts (we call it "sensory concept"). However, several findings argue against this concept. For example, neural responses to cues usually represent the subsequent action, but not sensory information of the cue such as the color of the cue. In our study, it is evident from Figure 3A, 6A and 6B that during action execution, actions are discriminated even before the start button is released. Since this discrimination of actions occurs with a time delay after the cue and then increases continuously, this is evidence that the action to be executed is represented, but not the cue itself.

Fifth, finally, and most importantly, the fundamental problem with this study is that it is correlational. Studies that purport to test the function of a set of neurons, and do so by use of correlational measurements, cannot provide strong answers. There are always half a dozen different interpretations and caveats, such as the ones I raised here. Both sides of a debate can always spin the results, and the arguments are never resolved. To test the mirror neuron hypothesis properly would require a causal study. For example, lesion area F5 and test if the monkey is less able to discriminate the actions of others. Or, electrically microstimulate in area F5 and test if the stimulation interferes (either constructively or destructively) with the task of discriminating the actions of others. Only in this way will it be possible to answer the question: do mirror neurons functionally participate in understanding the actions of others? The present study does not answer that question.

We would like to reiterate that studies aimed at elucidating what information is contained in which neurons or areas are necessary to understand neural network processes and are a prerequisite for conducting well-considered experiments that measure behavioral effects through specific manipulation of the neural network. Without the work of Gallese, Rizzolatti and colleagues, the idea of associating F5 neurons with action understanding would not have occurred in the first place. The current tricky question is whether at all, and if so, to what understanding, to what perception, to what behavior that uses information about mental states of another, F5 MNs might be able to contribute. And for this, it helps to have a clearer idea of what information is contained in F5 MNs during action observation.