I agree with this. This is definitely a good idea even though it might still promote an indirect distrust between the communities and police...

This is such an interesting concept to think about. We have created our civilization by tearing down large plots of vegetation that once kept our air quality top-grade housed thousands of animals, but now we pride ourselves on greenery and making sure the places we go to have some of that. It's kind of ironic. We want beauty, but there was more beauty when we weren't in the picture to ruin it.

My high school had a lot of technology which opened the windows to a lot of different opportunities. We has excess to everything at a touch of a button.

Author Response

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

Reviewer #1 (Public Review):

In this manuscript, the authors explore the effects of DNA methylation on the strength of regulatory activity using massively parallel reporter assays in cell lines on a genome-wide level. This is a follow-up of their first paper from 2018 that describes this method for the first time. In addition to adding more indepth information on sequences that are explored by many researchers using two main methods, reduced bisulfite sequencing and sites represented on the Illumina EPIC array, they now show also that DNA methylation can influence changes in regulatory activity following a specific stimulation, even in absence of baseline effects of DNA methylation on activity. In this manuscript, the authors explore the effects of DNA methylation on the response to Interferon alpha (INFA) and a glucocorticoid receptor agonist (dexamethasone). The authors validate their baseline findings using additional datasets, including RNAseq data, and show convergences across two cell lines. The authors then map the methylation x environmental challenge (IFNA and dex) sequences identified in vitro to explore whether their methylation status is also predictive of regulatory activity in vivo. This is very convincingly shown for INFA response sequences, where baseline methylation is predictive of the transcriptional response to flu infection in human macrophages, an infection that triggers the INF pathways.

Thank you for your strong assessment of our work!

The extension of the functional validity of the dex-response altering sequences is less convincing.

We agree. We note that genes close to dex-specific mSTARR-seq enhancers tend to be more strongly upregulated after dex stimulation than those near shared enhancers, which parallels our results for IFNA (lines 341-344). However, there is unfortunately no comparable data set to the human flu data set (i.e., with population-based whole genome-bisulfite sequencing data before and after dex challenge), so we could not perform a parallel in vivo validation step. We have added this caveat to the revised manuscript (lines 555-557).

Sequences altering the response to glucocorticoids, however, were not enriched in DNA methylation sites associated with exposure to early adversity. The authors interpret that "they are not links on the causal pathway between early life disadvantage and later life health outcomes, but rather passive biomarkers". However, this approach does not seem an optimal model to explore this relationship in vivo. This is because exposure to early adversity and its consequences is not directly correlated with glucocorticoid release and changes in DNA methylation levels following early adversity could be related to many physiological mechanisms, and overall, large datasets and meta-analyses do not show robust associations of exposure to early adversity and DNA methylation changes. Here, other datasets, such as from Cushing patients may be of more interest.

Thank you for making these important points. We have expanded the set of caveats regarding the lack of enrichment of early adversity-reported sites in the mSTARR-data set (lines 527-533). Specifically, we note that the relationship between early adversity and glucocorticoid physiology is complex (e.g., Eisenberger and Cole, 2012; Koss and Gunnar, 2018) and that dex challenge models one aspect of glucocorticoid signaling but not others (e.g., glucocorticoid resistance). Nevertheless, we also see little evidence for enrichment of early adversity-associated sites in the mSTARR data set at baseline, independently of the dex challenge experiment (lines 483-485; Figure 4).

We also agree that large data sets (e.g., Houtepen et al., 2018; Marzi et al., 2018) and reviews (e.g., Cecil et al., 2020) of early adversity and DNA methylation in humans show limited evidence of associations between early adversity and DNA methylation levels. However, the idea that early adversity impacts downstream outcomes remains pervasive in the literature and popular science (see Dubois et al., 2019), which we believe makes tests like ours important to pursue. We also hope that our data set (and others generated through these methods) will be useful in interpreting other settings in which differential methylation is of interest as well—in line with your comment below. We have clarified both of these points in the revised manuscript (lines 520-522; 536-539).

Overall, the authors provide a great resource of DNA methylation-sensitive enhancers that can now be used for functional interpretation of large-scale datasets (that are widely generated in the research community), given the focus on sites included in RBSS and the Illumina EPIC array. In addition, their data lends support that differences in DNA methylation can alter responses to environmental stimuli and thus of the possibility that environmental exposures that alter DNS methylation can also alter the subsequent response to this exposure, in line with the theory of epigenetic embedding of prior stimuli/experiences. The conclusions related to the early adversity data should be reconsidered in light of the comments above.

Thank you! And yes, we have revised our discussion of early life adversity effects as discussed above.

Reviewer #1 (Recommendations For The Authors):

While the paper has a lot of strengths and provides new insight into the epigenomic regulation of enhancers as well as being a great resource, there are some aspects that would benefit from clarification.

a. It would be great to have a clearer description of how many sequences are actually passing QC in the different datasets and what the respective overlaps are in bps or 600bp windows. Now often only % are given. Maybe a table/Venn diagram for overview of the experiments and assessed sequences would help here. This concern the different experiments in the K652, A549, and Hep2G cell lines, including stimulations.

We now provide a supplementary figure and supplementary table providing, for each dataset, the number of 600 bp windows passing each filter (Figure 2-figure supplement 1; Supplementary File 9), as well as a supplementary figure providing an upset plot to show the number of assessed sequences shared across the experiments (Figure 2-figure supplement 2).

b. It would also be helpful to have a brief description of the main differences in assessed sequences and their coverage of the old (2018) and new libraries in the main text to be able better interpret the validation experiments.

We now provide information on the following characteristics for the 2018 data set versus the data set presented for the first time here: mean (± SD) number of CpGs per fragment; mean (± SD) DNA sequencing depth; and mean (± SD) RNA sequencing depth (lines 169-170 provide values for the new data set; in line 194, we reference Supplementary File 5, which provides the same values for the old data set). Notably, the coverage characteristics of analyzed windows in both data sets are quite high (mean DNA-seq read coverage = 94x and mean RNA-seq read coverage = 165x in the new data set at baseline; mean DNA-seq read coverage = 22x and mean RNA-seq read coverage = 54x in Lea et al. 2018).

c. Statements of genome-wide analyses in the abstract and discussion should be a bit tempered, as quite a number of tested sites do not pass QC and do not enter the analysis. From the results it seems like from over 4.5 million sequences, only 200,000 are entering the analysis.

The reason why many of the windows are not taken forward into our formal modeling analysis is that they fail our filter for RNA reads because they are never (or almost never) transcribed—not because there was no opportunity for transcription (i.e., the region was indeed assessed in our DNA library, and did not show output transcription, as now shown in Figure 2-figure supplement 1). We have added a rarefaction analysis (lines 715-722 in Materials and Methods) of the DNA fragment reads to the revised manuscript which supports this point. Specifically, it shows that we are saturated for representation of unique genomic windows (i.e., we are above the stage in the curve where the proportion of active windows would increase with more sequencing: Figure 1figure supplement 4). Similarly, a parallel rarefaction curve for the mSTARR-seq RNA-seq data (Figure 1-figure supplement 4) shows that we would gain minimal additional evidence for regulatory activity with more sequencing depth. We now reference these analyses in revised lines 179-184 and point to the supporting figure in line 182.

In other words, our analysis is truly genome-wide, based on the input sequences we tested. Most of the genome just doesn’t have regulatory activity in this assay, despite the potential for it to be detected given that the relevant sequences were successfully transfected into the cells.

d. Could the authors comment on the validity of the analysis if only one copy is present (cut-off for QC)?

We think this question reflects a misunderstanding of our filtering criteria due to lack of clarity on our part, which we have modified in the revision. We now specify that the mean DNA-seq sequencing depth per sample for the windows we subjected to formal modeling was quite high:

93.91 ± 10.09 SD (range = 74.5 – 113.5x) (see revised lines 169-170). In other words, we never analyze windows in which there is scant evidence that plasmids containing the relevant sequence were successfully transfected (lines 170-172).

Our minimal RNA-seq criteria require non-zero counts in at least 3 replicate samples within either the methylated condition or the unmethylated condition, or both (lines 166-168). Because we know that multiple plasmids containing the corresponding sequence are present for all of these windows—even those that just cross the minimal RNA-seq filtering threshold—we believe our results provide valid evidence that all analyzed windows present the opportunity to detect enhancer activity, but many do not act as enhancers (i.e., do not result in transcribed RNA). Notably, we observe a negligible correlation between DNA sequencing depth for a fragment, among analyzed windows, and mSTARR-seq enhancer activity (R2 = 0.029; now reported in lines 183-184). We also now report reproducibility between replicates, in which all replicate pairs have r > 0.89, on par with previously published STARR-seq datasets (e.g., Klein et al., 2020; Figure 1-figure supplement 6, pointed to in line 193).

e. While the authors state that almost all of the control sequences contain CpGs sites, could the authors also give information on the total number of CpG sites in the different subsets? Was the number of CpGs in a 600 bp window related to the effects of DNA methylation on enhancer activity?

We now provide the number of CpG sites per window in the different subsets in lines 282-284. As expected, they are higher for EPIC array sites and for RRBS sites because the EPIC array is biased towards CpG-rich promoter regions, and the enzyme typically used in the starting step of RRBS digests DNA at CpG motifs (but control sequences still contain an average of ~13 CpG sites per fragment). We also now model the magnitude of the effects of DNA methylation on regulatory activity as a function of number of CpG sites within the 600 bp windows. Consistent with our previous work in Lea et al., 2018, we find that mSTARR-seq enhancers with more CpGs tend to be repressed by DNA methylation (now reported in lines 216-219 and Figure 1figure supplement 11).

f. In the discussion, a statement on the underrepresented regions, likely regulatory elements with lower CG content, that nonetheless can be highly relevant for gene regulation would be important to put the data in perspective.

Thanks for this suggestion. We agree that regulatory regions, independent of CpG methylation, can be highly relevant, and now clarify in the main text that the “unmethylated” condition of mSTARR-seq is essentially akin to a conventional STARR-seq experiment, in that it assesses regulatory activity regardless of CpG content or methylation status (lines 128-130).

Consequently, our study is well-designed to detect enhancer-like activity, even in windows with low GC content. We now show with additional analyses that we generated adequate DNA-seq coverage on the transfected plasmids to analyze 90.2% of the human genome, including target regions with no or low CpG content (lines 148-149; 153-156; Supplementary file 2). As noted above, we also now clarify that regions dropped out of our formal analysis because we had little to no evidence that any transcription was occurring at those loci, not because sequences for those regions were not successfully transfected into cells (see responses above and new Figure 1-figure supplement 4 and Figure 2-figure supplement 1).

g. To control for differences in methylation of the two libraries, the authors sequence a single CpGs in the vector. Could the authors look at DNA methylation of the 600 bp windows at the end of the experiment, could DNA methylation of these windows be differently affected according to sequence? 48 hours could be enough for de-methylation or re-methylation.

We agree that variation in demethylation or remethylation depending on fragment sequence is possible. We now state this caveat in the main text (lines 158-159), and specify that genomic coverage of our bisulfite sequencing data across replicates are (unfortunately) too variable to perform reliable site-by-site analysis of DNA methylation levels before and after the 48 hour experiment (lines 1182-1185). Instead, we focus on a CpG site contained in the adapter sequence (and thus included in all plasmids) to generate a global estimate of per replicate methylation levels. We also now note that any de-methylation or re-methylation would reduce our power to detect methylation-dependent activity, rather than leading to false positives (lines 163-165).

h. The section on the method for correction for multiple testing should be more detailed as it is very difficult to follow. Why were only 100 permutations used, the empirical p-value could then only be <0.01? The description of a subsample of the N windows with positive Betas is unclear, should the permutation not include the actual values and thus all windows - or were the no negative Betas? Was FDR accounting for all elements and pairs?

We have now expanded the text in the Materials and Methods section to clarify the FDR calculation (lines 691, 695-699, 702, 706). We clarify that the 100 permutations were used to generate a null distribution of p-values for the data set (e.g., 100 x 17,461 p-values for the baseline data set), which we used to derive a false discovery rate. Because we base our evidence on FDRs, we therefore compare the distribution of observed p-values to the distribution of pvalues obtained via permutation; we do not calculate individual p-values by comparing an observed test statistic against the test statistics for permuted data for that individual window.

We compare the data to permutations with only positive betas because in the observed data, we observe many negative betas. These correspond to windows which have no regulatory activity (i.e., they have many more input DNA reads than RNA-seq reads) and thus have very small pvalues in a model testing for DNA-RNA abundance differences. However, we are interested in controlling the false discovery rate of windows that do have regulatory activity (positive betas). In the permuted data, by contrast and because of the randomization we impose, test statistics are centered around 0 and essentially symmetrical (approximately equally likely to be positive or negative). Retaining all p-values to construct the null therefore leads to highly miscalibrated false discovery rates because the distribution of observed values is skewed towards smaller values— because of windows with “significantly” no regulatory activity—compared to the permuted data. We address that problem by using only positive betas from the permutations.

i. The interpretation of the overlap of Dex-response windows with CpGs sites associated with early adversity should be revisited according to the points also mentioned in the public review and the authors may want to consider exploring additional datasets with other challenges.

Thank you, see our responses to the public review above and our revisions in lines (lines 555559). We agree that comparisons with more data sets and generation of more mSTARR-seq data in other challenge conditions would be of interest. While beyond the scope of this manuscript, we hope the resource we have developed and our methods set the stage for just such analyses.

Reviewer #2 (Public Review):

This work presents a remarkably extensive set of experiments, assaying the interaction between methylation and expression across most CpG positions in the genome in two cell types. To this end, the authors use mSTARR-seq, a high-throughput method, which they have previously developed, where sequences are tested for their regulatory activity in two conditions (methylated and unmethylated) using a reporter gene. The authors use these data to study two aspects of DNA methylation:

1) Its effect on expression, and 2. Its interaction with the environment. Overall, they identify a small number of 600 bp windows that show regulatory potential, and a relatively large fraction of these show an effect of methylation on expression. In addition, the authors find regions exhibiting methylation-dependent responses to two environmental stimuli (interferon alpha and glucocorticoid dexamethasone).

The questions the authors address represent some of the most central in functional genomics, and the method utilized is currently the best method to do so. The scope of this study is very impressive and I am certain that these data will become an important resource for the community. The authors are also able to report several important findings, including that pre-existing DNA methylation patterns can influence the response to subsequent environmental exposures.

Thank you for this generous summary!

The main weaknesses of the study are: 1. The large number of regions tested seems to have come at the expense of the depth of coverage per region (1 DNA read per region per replicate). I have not been convinced that the study has sufficient statistical power to detect regulatory activity, and differential regulatory activity to the extent needed. This is likely reflected in the extremely low number of regions showing significant activity.

We apologize for our lack of clarity in the previous version of the manuscript. Nonzero coverage for half the plasmid-derived DNA-seq replicates is a minimum criterion, but for the baseline dataset, the mean depth of DNA coverage per replicate for windows passing the DNA filter is quite high: 12.723 ± 41.696 s.d. overall, and 93.907 ± 10.091 s.d. in the windows we subjected to full analysis (i.e., windows that also passed the RNA read filter). We now provide these summary statistics in lines 148-149 and 169-170 and Supplementary file 5 (see also our responses to Reviewer 1 above). We also now show, using a rarefaction analysis, that our data set saturates the ability to detect regulatory windows based on DNA and RNA sequencing depth (new Figure 1-figure supplement 4; lines 179-184; 715-722).

2) Due to the position of the tested sequence at the 3' end of the construct, the mSTARR-seq approach cannot detect the effect of methylation on promoter activity, which is perhaps the most central role of methylation in gene regulation, and where the link between methylation and expression is the strongest. This limitation is evident in Fig. 1C and Figure 1-figure supplement 5C, where even active promoters have activity lower than 1. Considering these two points, I suspect that most effects of methylation on expression have been missed.

Thank you for pointing this out. We agree that we have not exhaustively detected methylationdependent activity in all promoter regions, given that not all promoter regions are active in STARR-seq. However, there is good evidence that some promoter regions can function like enhancers and thus be detected in STARR-seq-type assays (Klein et al., 2020). This important point is now noted in lines 187-189; an example promoter showing methylation-dependent regulatory activity in our dataset is shown in Figure 3E.

We also now clarify that Figure 1C shows significant enrichment of regulatory activity in windows that overlap promoter sequence (line 239). The y-axis is not a measure of activity, but rather the log-transformed odds ratio, with positive values corresponding to overrepresentation of promoter sequences in regions of mSTARR-seq regulatory activity. Active promoters are 1.640 times more likely to be detected with regulatory activity than expected by chance (p = 1.560 x 10-18), which we now report in a table that presents enrichment statistics for all ENCODE elements shown in Figure 1C for clarity (Supplementary file 4). Moreover, 74.1% of active promoters that show regulatory activity have methylation-dependent activity, also now reported in Supplementary file 4.

Overall, the combination of an extensive resource addressing key questions in functional genomics, together with the findings regarding the relationship between methylation and environmental stimuli makes this a key study in the field of DNA methylation.

Thank you again for the positive assessment!

Reviewer #2 (Recommendations For The Authors):

I suggest the authors conduct several tests to estimate and/or increase the power of the study:

1) To estimate the potential contribution of additional sequencing depth, I suggest the authors conduct a downsampling analysis. If the results are not saturated (e.g., the number of active windows is not saturated or the number of differentially active windows is not saturated), then additional sequencing is called for.

We appreciate the suggestion. We have now performed a downsampling/rarefaction curve analysis in which we downsampled the number of DNA reads, and separately, the number of RNA reads. We show that for both DNA-seq depth and RNA-seq depth, we are within the range of sequencing depth in which additional sequencing would add minimal new analysis windows in the dataset (Figure 1-figure supplement 4; lines 179-184; 715-722).

2) Correlation between replicates should be reported and displayed in a figure because low correlations might also point to too few reads. The authors mention: "This difference likely stems from lower variance between replicates in the present study, which increases power", but I couldn't find the data.

We now report the correlations between RNA and DNA replicates within the current dataset and within the Lea et al., 2018 dataset (Figure 1-figure supplement 6). The between-replicate correlations in both our RNA libraries and DNA libraries are consistently high (r ≥ 0.89).

3) The correlation between the previous and current K562 datasets is surprisingly low. Given that these datasets were generated in the same cell type, in the same lab, and using the same protocol, I expected a higher correlation, as seen in other massively parallel reporter assays. The fact that the correlations are almost identical for a comparison of the same cell and a comparison of very different cell types is also suspicious.

Thanks for raising this point. We think it is in reference to our original Figure 1-Figure supplement 6, for which we now provide Pearson correlations in addition to R2 values (now Figure 1-Figure supplement 8). We note that this is not a correlation in raw data, but rather the correlation in estimated effect sizes from a statistical model for methylation-dependent activity. We now provide Pearson correlations for the raw data between replicates within each dataset (Figure 1-Figure supplement 6), which for the baseline dataset are all r > 0.89 for RNA replicates and r > 0.98 for DNA replicates, showing that replicate reproducibility in this study is on par with other published studies (e.g., Klein et al., 2020 report r > 0.89 for RNA replicates and r > 0.91 for DNA replicates).

We do not know of any comparable reports in other MPRAs for effect size correlations between two separately constructed libraries, so it’s unclear to us what the expectation should be. However, we note that all effect sizes are estimated with uncertainty, so it would be surprising to us to observe a very high correlation for effect sizes in two experiments, with two independently constructed libraries (i.e., with different DNA fragments), run several years apart—especially given the importance of winner’s curse effects and other phenomena that affect point estimates of effect sizes. Nevertheless, we find that regions we identify as regulatory elements in this study are 74-fold more likely to have been identified as regulatory elements in Lea et al., 2018 (p < 1 x10-300).

4) The authors cite Johnson et al. 2018 to support their finding that merely 0.073% of the human genome shows activity (1.7% of 4.3%), but:

a. the percent cited is incorrect: this study found that 27,498 out of 560 million regions (0.005%) were active, and not 0.165% as the authors report.

We have modified the text to clarify the numerator and denominator used for the 0.165% estimate from Johnson et al 2018 (lines 175-176). The numerator is their union set of all basepairs showing regulatory activity in unstimulated cells, which is 5,547,090 basepairs. The denominator is the total length of the hg38 human genome, which is 3,298,912,062 basepairs.

Notably, the denominator (the total human genome) is not 560 million—while Johnson et al (2018) tested 560 million unique ~400 basepair fragments, these fragments were overlapping, such that the 560 million fragments covered the human genome 59 times (i.e., 59x coverage).

b. other studies that used massively parallel reporter assays report substantially higher percentages, suggesting that the current study is possibly underpowered. Indeed, the previous mSTARR-seq found a substantially larger percentage of regions showing regulatory activity (8%). The current study should be compared against other studies (preferably those that did not filter for putatively active sequences, or at least to the random genomic sequences used in these studies).

We appreciate this point and have double checked comparisons to Johnson et al., 2018 and Lea et al., 2018. Our numbers are not unusual relative to Johnson et al., 2018 (0.165%), which surveyed the whole genome. Also, in comparing to the data from Lea et al., 2018, when processed in an identical manner (our criteria are more stringent here), our values of the percent of the tested genome showing significant regulatory activity are also similar: 0.108% in the Lea et al., 2018 dataset versus 0.082% in the baseline dataset. Finally, our rarefaction analyses (see our responses above) indicate that we are not underpowered based on sequencing depth for RNA or DNA samples. We also note that there are several differences in our analysis pipeline from other studies: we use more technical replicates than is typical (compare to 2-5 replicates in Arnold et al., 2013; Johnson et al., 2018; Muerdter et al., 2018), we measure DNA library composition based on DNA extracted from each replicate post-transfection (as opposed to basing it on the pre-transfection library: [Johnson et al., 2018], and we use linear mixed models to identify regulatory activity as opposed to binomial tests [Johnson et al., 2018; Arnold et al., 2013; Muerdter et al., 2018].

I find it confusing that the four sets of CpG positions used: EPIC, RRBS, NR3C1, and random control loci, add up together to 27.3M CpG positions. Do the 600 bp windows around each of these positions sufficient to result in whole-genome coverage? If so, a clear explanation of how this is achieved should be added.

Thanks for this comment. Although our sequencing data are enriched for reads that cover these targeted sites, the original capture to create the input library included some off target reads (as is typical of most capture experiments, which are rarely 100% efficient). We then sequenced at such high depth that we ultimately obtained sequencing coverage that encompassed nearly the whole genome. We now clarify in the main text that our protocol assesses 27.3 million CpG sites by assessing 600 bp windows encompassing 93.5% of all genomic CpG sites (line 89), which includes off-target sites (line 149).

scatter plot showing the RNA to DNA ratios of the methylated (x-axis) vs unmethylated (y-axis) library would be informative. I expect to see a shift up from the x=y diagonal in the unmethylated values.

We have added a supplementary figure showing this information, which shows the expected shift upwards (Figure 1-figure supplement 9).

Another important figure missing is a histogram showing the ratios between the unmethylated and methylated libraries for all active windows, with the significantly differentially active windows marked.

We have added a supplementary figure showing this information (Figure 1-Supplementary Figure 10).

Perhaps I missed it, but what is the distribution of effect sizes (differential activity) following the various stimuli?

This information is provided in table form in Supplementary Files 3, 10, and 11, which we now reference in the Figure 2 legend (lines 365-366).

Minor changes

It is unclear what the lines connecting the two groups in Fig.3C represent, as these are two separate groups of regions.

We now clarify in the figure legend that values connected by a line are the same regions, not two different sets of regions. They show the correlation between DNA methylation and gene expression at mSTARR-seq-identified enhancers in individuals before and after IAV stimulation, separately for enhancers that are shared between conditions (left) versus those that are IFNAspecific (right). The two plots therefore do show two different sets of regions, which we have depicted to visualize the contrast in the effect of stimulation on the correlation on IFNA-specific enhancers versus shared enhancers. We have revised the figure legend to clarify these points (line 458-460).

L235-242 are unclear. Specifically - isn't the same filter mentioned in L241-242 applied to all regions?

Yes, the same filter for minimal RNA transcription was applied to all regions. We have modified the text (lines 264-265, 271, 275-277) to clarify that the enrichment analyses were performed twice, to test whether the target types were: 1) enriched in the dataset passing the RNA filter (i.e., the dataset showing plasmid-derived RNA reads in at least half the sham or methylated replicates; n = 216,091 windows) and 2) enriched in the set of windows showing significant regulatory activity (at FDR < 1%; n = 3,721 windows).

To improve cohesiveness, the section about most CpG sites associated with early life adversity not showing regulatory activity in K562s can be moved to the supplementary in my opinion.

Thank you for this suggestion. Because ELA and the biological embedding hypothesis (via DNA methylation) were major motivations for our analysis (see Introduction lines 42-48; 75-79), and we also discuss these results in the Discussion (lines 518-520), we have respectfully elected to retain this section in the main manuscript. We have added text in the Discussion explaining why we think experimental tests of methylation effects on regulation are relevant to the literature on early life adversity (lines 520-522), and have added discussion on limits to these analyses (lines 527-533).

References:

Arnold CD, Gerlach D, Stelzer C, Boryń ŁM, Rath M, Stark A (2013) Genome-wide quantitative enhancer activity maps identified by STARR-seq. Science, 339, 1074-1077.

Cecil CA, Zhang Y, Nolte T (2020) Childhood maltreatment and DNA methylation: A systematic review. Neuroscience & Biobehavioral Reviews, 112, 392-409.

Dubois M, Louvel S, Le Goff A, Guaspare C, Allard P (2019) Epigenetics in the public sphere: interdisciplinary perspectives. Environmental Epigenetics, 5, dvz019.

Eisenberger NI, Cole SW (2012) Social neuroscience and health: neurophysiological mechanisms linking social ties with physical health. Nature neuroscience, 15, 669-674.

Houtepen L, Hardy R, Maddock J, Kuh D, Anderson E, Relton C, Suderman M, Howe L (2018) Childhood adversity and DNA methylation in two population-based cohorts. Translational Psychiatry, 8, 1-12.

Johnson GD, Barrera A, McDowell IC, D’Ippolito AM, Majoros WH, Vockley CM, Wang X, Allen AS, Reddy TE (2018) Human genome-wide measurement of drug-responsive regulatory activity. Nature communications, 9, 1-9.

Klein JC, Agarwal V, Inoue F, Keith A, Martin B, Kircher M, Ahituv N, Shendure J (2020) A systematic evaluation of the design and context dependencies of massively parallel reporter assays. Nature Methods, 17, 1083-1091.

Koss KJ, Gunnar MR (2018) Annual research review: Early adversity, the hypothalamic–pituitary– adrenocortical axis, and child psychopathology. Journal of Child Psychology and Psychiatry, 59, 327-346.

Marzi SJ, Sugden K, Arseneault L, Belsky DW, Burrage J, Corcoran DL, Danese A, Fisher HL, Hannon E, Moffitt TE (2018) Analysis of DNA methylation in young people: limited evidence for an association between victimization stress and epigenetic variation in blood. American journal of psychiatry, 175, 517-529.

Muerdter F, Boryń ŁM, Woodfin AR, Neumayr C, Rath M, Zabidi MA, Pagani M, Haberle V, Kazmar T, Catarino RR (2018) Resolving systematic errors in widely used enhancer activity assays in human cells. Nature methods, 15, 141-149.

Reviewer #2 (Public Review):

This work presents a remarkably extensive set of experiments, assaying the interaction between methylation and expression across most CpG positions in the genome in two cell types. To this end, the authors use mSTARR-seq, a high-throughput method, which they have previously developed, where sequences are tested for their regulatory activity in two conditions (methylated and unmethylated) using a reporter gene. The authors use these data to study two aspects of DNA methylation: 1. Its effect on expression, and 2. Its interaction with the environment. Overall, they identify a small number of 600 bp windows that show regulatory potential, and a relatively large fraction of these show an effect of methylation on expression. In addition, the authors find regions exhibiting methylation-dependent response to two environmental stimuli (interferon alpha and glucocorticoid dexamethasone).

The questions the authors address represent some of the most central in functional genomics, and the method utilized is currently the best method to do so. The scope of this study is very impressive and I am certain that these data will become an important resource for the community. The authors are also able to report several important findings, including that pre-existing DNA methylation patterns can influence the response to subsequent environmental exposures.

вместо встречи

关于 RunScanner 的综合回答 RunScanner 是一个免费的便携系统工具，用于扫描计算机上所有运行的进程、程序、服务、驱动程序和自启动位置[1]。主要特点包括：

系统扫描功能：RunScanner 可以分析系统运行的进程，提供快速或全面的扫描选项[4]。 自启动位置扫描：该工具能检测系统中的自启动程序、位置[2]。 兼容性：RunScanner 适用于 Microsoft Windows 操作系统[3]。 安全性：其目标是警告用户有关系统中运行的危险进程[10]。 此外，RunScanner 还有相关的 GitHub 存储库，用于监控文件系统中包含测序仪器（如 Illumina）输出的目录[5]。

总体而言，RunScanner 是一款用于系统分析和安全性检查的免费工具，适用于 Windows 操作系统。

参考资料：

RunScanner - Free download and software reviews [1] Download RunScanner [2] RunScanner - Wikipedia [3] RunScanner - Download [4] Run Scanner on GitHub [5] RunScanner for Windows - Download it from Uptodown [10]

Windows NT和Windows 9x/NT 3.5操作系统使用注册表文件（REGF）来存储系统设置和配置信息。注册表文件有时被称为"hives"。在Windows NT中，它们存储在C:\Windows\System32\config目录中，包括SYSTEM、SOFTWARE、SAM、SECURITY等文件[1][6]。而在Windows 9x/NT 3.5中，引入了C:\WINDOWS\目录下的SYSTEM.DAT文件，用于存储系统设置、所有用户的受保护存储区以及已安装程序的设置[5]。

Win 9x系统使用CREG文件格式，而Windows NT系统使用REGF文件格式[1][9]。Regedit是Win 9x注册表编辑器，而Regedit32.exe则是Windows NT中的注册表编辑器[3]。编辑注册表涉及到特定的工具和操作，如使用Regini.exe配置注册表[4]。

注册表是32位版本的Microsoft Windows中存储设置和选项的数据库[8]。了解注册表对于系统配置和故障排除至关重要。在Windows NT中，可以通过NT Explorer编辑注册表，包括HKEY_CURRENT_USER包含关于系统的信息[7]。

参考： 1. Windows NT Registry File (REGF) format 2. The Windows 9x Registry REGEDIT.EXE 3. 4. Windows 9x/NT 3.5 4. Windows 9x Registry File (CREG) format 5. Editing the Windows NT Registry

Windows 9x/NT 注册表文件是Windows操作系统中的关键组成部分，用于存储系统和应用程序的配置信息。注册表包含各种参数，直接影响Windows的功能和行为[1]。对于Windows NT，注册表数据存储在\windows\system32\config目录下的文件中，包括Default、Sam、Security、Software、System、Userdief等[6]。而对于Windows 9x/Me，注册表用于管理文件系统，指明文件类型、调用程序信息等[4]。

注册表在Windows系统中起到统一配置信息的作用，涉及应用程序和文件关联、硬件设备说明、状态属性等[3]。通过注册表，用户可以控制系统特性、限制功能，对硬件、软件和用户环境进行管理[5]。

总体而言，注册表在Windows系统中扮演重要角色，管理员和用户可以通过操作注册表来调整系统行为和配置信息[1][4]。

要在文件中生成备注，你可以根据所用的软件或工具采取不同的方法：

1. WPS文档： 打开文档，选择需要添加批注的文字，然后依次点击“插入” -> “批注”[3].

2. PDF文件： 在需要添加备注的地方双击，点击标识上方的图形，然后在弹出的备注框内添加备注[2].

3. 文件夹（Windows）： 在文件夹上右键，选择属性，然后点击“详细信息”选项卡，你可以在那里添加备注[4].

4. 任意文件： 对特定文件右击，选择属性，点击“详细信息”选项卡，找到“备注”栏，添加你的备注信息[6].

注意：确保保存修改以保存你的备注。

参考： 1. WPS学堂 - 如何在文档中插入批注 2. PDF文件如何快速添加备注? 3. WPS word单元格怎么插入批注 4. Win10/Win11如何给文件夹打上注释/备注? 5. win11给文件夹添加备注 6. 为任意文件添加备注信息

生成 Windows 下的 Keytab 文件

在 Windows 系统上生成 Keytab 文件，你可以使用 ktpass 工具。以下是简要步骤：

1. 打开命令提示符。

2. 运行以下命令，生成 BloxOne DDI 用户账户的 Keytab 文件： bash ktpass -princ username@REALM [3]

3. 使用 Microsoft Windows Server 的 ktpass 实用程序为在Active Directory中创建的每个用户账户生成 Keytab 文件： bash ktpass utility to generate a keytab file for each user account you created in Active Directory [8]

4. 如果需要为服务主体名称（SPN）生成 Keytab 文件，可以使用 Windows Server 工具包中的 ktpass 工具： bash Generate the Kerberos keytab file (krb5.keytab) for an SPN using the ktpass tool from the Windows Server toolkit [10]

请注意，Keytab 文件的创建通常在域控制器服务器或作为域的一部分的 Windows Server 计算机上完成，使用域管理员账户[7]。

要在Windows上生成Keytab文件，您可以使用ktpass工具。以下是简要步骤：

1. 打开命令提示符。

2. 运行以下命令以生成Keytab文件： plaintext ktpass -princ username@REALM -pass password -mapuser username -crypto DES-CBC-MD5 -ptype KRB5_NT_PRINCIPAL -out keytab_file.keytab 替换username@REALM和password为相应的用户名和域信息。

3. 这将生成Keytab文件，其中包含与指定用户关联的凭据。

请注意，此过程可能需要在域控制器服务器或运行Windows Server的计算机上进行[2][3][7][8][9][10][11]。

在Windows系统中，有多种方法可以生成文件信息。以下是一些建议：

1. 使用fsutil命令可以创建指定名称和大小的文件，内容为零。这可以通过以下命令完成：fsutil file createnew 文件名 大小[3][11]。

2. 使用CreateFile函数可以创建新文件或打开现有文件。必须指定文件名和创建说明。这是通过编程实现的，适用于Win32应用程序[4][6]。

3. 若要创建系统信息文件，可以在任务栏上的Cortana或搜索图标中键入“System”，然后点击System Information图标[5]。

4. 通过命令提示符使用cmd.exe可以获取文件的所有信息和属性[7]。

这些方法提供了在Windows系统中生成文件信息的多样选择。

cognitive effort, switching mode of working, sometimes tool (booth helper), sometimes not.

Like the way he softens up the audience with intro!

10 Points - Meaning over Money "Democratize" - Mantra 3 words No Mission Statement - Jump to next curve define yourself by what value you provide not what you do - Roll the dice Like the Windows reference. Wrestle it to the ground! - Don't be perfect Elements of crappiness to your revolution - 100 Flowers Let it bloom Don't be proud - Polarize - Iterate ignore naysayers - Niche thyself - Pitch custom to the audience - Customize - 10 slides on 20 min 30 point font - Don't let the Bozos grind you down My stuff ends in i New fun word! Bozocity We all say stupid things Guy!

He keeps it fun upbeat and sticks to his 20 minutes.

Connects with people.. a skill a lot of tech entrepreneurs don't have (Zuckerberg)

Reviewer #2 (Public Review):

Summary:<br /> This paper aimed to examine the spatial frequency selectivity of macaque inferotemporal (IT) neurons and its relation to category selectivity. The authors suggest in the present study that some IT neurons show a sensitivity for the spatial frequency of scrambled images. Their report suggests a shift in preferred spatial frequency during the response, from low to high spatial frequencies. This agrees with a coarse-to-fine processing strategy, which is in line with multiple studies in the early visual cortex. In addition, they report that the selectivity for faces and objects, relative to scrambled stimuli, depends on the spatial frequency tuning of the neurons.

Strengths:<br /> Previous studies using human fMRI and psychophysics studied the contribution of different spatial frequency bands to object recognition, but as pointed out by the authors little is known about the spatial frequency selectivity of single IT neurons. This study addresses this gap and they show that at least some IT neurons show a sensitivity for spatial frequency and interestingly show a tendency for coarse-to-fine processing.

Weaknesses and requested clarifications:<br /> 1. It is unclear whether the effects described in this paper reflect a sensitivity to spatial frequency, i.e. in cycles/ deg (depends on the distance from the observer and changes when rescaling the image), or is a sensitivity to cycles /image, largely independent of image scale. How is it related to the well-documented size tolerance of IT neuron selectivity?

2. The authors' band-pass filtered phase scrambled images of faces and objects. The original images likely differed in their spatial frequency amplitude spectrum and thus it is unclear whether the differing bands contained the same power for the different scrambled images. If not, this could have contributed to the frequency sensitivity of the neurons.

3. How strong were the responses to the phase-scrambled images? Phase-scrambled images are expected to be rather ineffective stimuli for IT neurons. How can one extrapolate the effect of the spatial frequency band observed for ineffective stimuli to that for more effective stimuli, like objects or (for some neurons) faces? A distribution should be provided, of the net responses (in spikes/s) to the scrambled stimuli, and this for the early and late windows.

4. The strength of the spatial frequency selectivity is unclear from the presented data. The authors provide the result of a classification analysis, but this is in normalized units so that the reader does not know the classification score in percent correct. Unnormalized data should be provided. Also, it would be informative to provide a summary plot of the spatial frequency selectivity in spikes/s, e.g. by ranking the spatial frequency bands for each neuron based on half of the trials and then plotting the average responses for the obtained ranks for the other half of the trials. Thus, the reader can appreciate the strength of the spatial frequency selectivity, considering trial-to-trial variability. Also, a plot should be provided of the mean response to the stimuli for the two analysis windows of Figure 2c and 2d in spikes/s so one can appreciate the mean response strengths and effect size (see above).

5. It is unclear why such brief stimulus durations were employed. Will the results be similar, in particular the preference for low spatial frequencies, for longer stimulus durations that are more similar to those encountered during natural vision?

6. The authors report that the spatial frequency band classification accuracy for the population of neurons is not much higher than that of the best neuron (line 151). How does this relate to the SNC analysis, which appears to suggest that many neurons contribute to the spatial frequency selectivity of the population in a non-redundant fashion? Also, the outcome of the analyses should be provided (such as SNC and decoding (e.g. Figure 1D)) in the original units instead of undefined arbitrary units.

7. To me, the results of the analyses of Figure 3c,d, and Figure 4 appear to disagree. The latter figure shows no correlation between category and spatial frequency classification accuracies while Figure 3c,d shows the opposite.

8. If I understand correctly, the "main" test included scrambled versions of each of the "responsive" images selected based on the preceding test. Each stimulus was presented 15 times (once in each of the 15 blocks). The LDA classifier was trained to predict the 5 spatial frequency band labels and they used 70% of the trials to train the classifier. Were the trained and tested trials stratified with respect to the different scrambled images? Also, LDA assumes a normal distribution. Was this the case, especially because of the mixture of repetitions of the same scrambled stimulus and different scrambled stimuli?

9. The LDA classifiers for spatial frequency band (5 labels) and category (2 labels) have different chance and performance levels. Was this taken into account when comparing the SNC between these two classifiers? Details and SNC values should be provided in the original (percent difference) instead of arbitrary units in Figure 5a. Without such details, the results are impossible to evaluate.

10. Recording locations should be described in IT, since the latter is a large region. Did their recordings include the STS? A/P and M/L coordinate ranges of recorded neurons?

11. The authors should show in Supplementary Figures the main data for each of the two animals, to ensure the reader that both monkeys showed similar trends.

12. The authors found that the deep nets encoded better the spatial frequency bands than the IT units. However, IT units have trial-to-trial response variability and CNN units do not. Did they consider this when comparing IT and CNN classification performance? Also, the number of features differs between IT and CNN units. To me, comparing IT and CNN classification performances is like comparing apples and oranges.

13. The authors should define the separability index in their paper. Since it is the main index to show a relationship between category and spatial frequency tuning, it should be described in detail. Also, results should be provided in the original units instead of undefined arbitrary units. The tuning profiles in Figure 3A should be in spikes/s. Also, it was unclear to me whether the classification of the neurons into the different tuning profiles was based on an ANOVA assessing per neuron whether the effect of the spatial frequency band was significant (as should be done).

14. As mentioned above, the separability analysis is the main one suggesting an association between category and spatial frequency tuning. However, they compute the separability of each category with respect to the scrambled images. Since faces are a rather homogeneous category I expect that IT neurons have on average a higher separability index for faces than for the more heterogeneous category of objects, at least for neurons responsive to faces and/or objects. The higher separability for faces of the two low- and high-pass spatial frequency neurons could reflect stronger overall responses for these two classes of neurons. Was this the case? This is a critical analysis since it is essential to assess whether it is category versus responsiveness that is associated with the spatial frequency tuning. Also, I do not believe that one can make a strong claim about category selectivity when only 6 faces and 3 objects (and 6 other, variable stimuli; 15 stimuli in total) are employed to assess the responses for these categories (see next main comment). This and the above control analysis can affect the main conclusion and title of the paper.

15. For the category decoding, the authors employed intact, unscrambled stimuli. Were these from the main test? If yes, then I am concerned that this represents a too small number of stimuli to assess category selectivity. Only 9 fixed + 6 variable stimuli = 15 were in the main test. How many faces/ objects on average? Was the number of stimuli per category equated for the classification? When possible use the data of the preceding selectivity test which has many more stimuli to compute the category selectivity.

What are the advantages and disadvantage of such a broad view? This reminds me of ML argument of crayons as technologies. [pencil; post-it; rock]

MI

SE of how streaming services (and technology) allow individuals to listen to music more than ever

SE of the accessibility of streaming apps, being able to use them on the go at all times leads to more time listening to music

C&C listening to music on streaming services vs owning music - more people now use a platform than own music

EX of one streaming platform typically targeted to smaller artists (Soundcloud)

C&C how followers enjoy music then and now - now listeners also are interested in their favorite artists life outside of music (largely due to social media)

EX of one of the downsides of social media - how easy it is for one post to ruin an individual's reputation, causing them to easily lose fans - the cost of a closer community

SE of how social media strengthens bond between musician and listener

EX of how widespread technology enabled music to become

C/E accessibility of music causes people to listen more

SE of how even though sales of physical media is down, it still is important for making that connection between listeners and artists

MI

SE of how piracy negatively affects artists/labels

C/E technology caused some individuals to illegally download music, damaging industry

C/E introduction of social media causes more connection between fans and artist and makes artists more human to followers

Mobile devices, such as phones and tablets, are evolving and becoming more accessible by the day. Most people in the world, as well as in Georgia, use phones for daily activities, such as shopping, paying bills, or studying. Whichever field your business belongs to, its website must be mobile and tablet friendly. Our team offers compatibility with all types of mobile devices and operating systems (iOS, Android, Windows, etc.) without sacrificing the site's functionality or aesthetics.

Papercut: Windows-based SMTP server/interceptor

ChangemakerStudios/Papercut-SMTP: Papercut SMTP -- The Simple Desktop Email Server

Tall dark arched windows, framed with stone bricks, punctuated the front wall, equidistant from each other.

這裡，punctuate 是句子的及物動詞，動詞的主詞是拱窗，不是石磚。

是拱窗點綴前牆，不是石磚延伸到牆。 是拱窗彼此間等距，不是石磚等距排列。

如要描述石磚，punctuated 會寫成 punctuating。

Author Response

Reviewer #1 (Public Review):

This article proposes a new statistical approach to identify which of several experimenter-defined strategies best describes a biological agent's decisions when such strategies are not fully observable by choices made in a given trial. The statistical approach is described as Bayesian but can be understood instead as computing a smoothed running average (with decay) of the strategies' success at matching choices, with a winner-take-all inference across the rules. The article tests the validity of this statistical approach by applying it to both simulated agents and real data sets in mice and humans. It focuses on dynamically changing environments, where the strategy best describing a biological agent may change rapidly.

The paper asks an important question, and the analysis is well conducted; the paper is well-written and easy to follow. However, there are several concerns that limit the strength of the contribution. Major concerns include the framing of the method, considerations around the strategy space, limitations in how useful the technique may be, and missing details in analyses.

Reviewer #2 (Public Review):

In this study, the goal is to leverage the power of Bayesian inference to estimate online the probability that any given arbitrarily chosen strategy is being used by the decision-maker. By computing the trial-by-trial MAP and variance of the posterior distribution for each candidate strategy, the authors can not only see which strategy is primarily being used at every given time during the task and when strategy changes occur but also detect when the target rule of a learning task becomes the front-running strategy, i.e., when successful learning occurs.

Strengths:

1) The proposed approach adds to recent methods for capturing the dynamics of decision-making at finer temporal resolution (trials) (Roy et al., 2021; Ashwood et al., 2022) but it is novel and differs from these in that it is suited especially well for analyzing when learning occurs, or when a rule switches and learning must recommence, and it does not necessitate large numbers of trials.

2) The manuscript starts with a validation of the approach using synthetic data and then is applied to datasets of trial-based two-alternative forced choice tasks ranging from rodent to non-human primate to human, providing solid evidence of its utility.

3) Compared to classic procedures for identifying when an animal has learned a contingency which typically needs to be conservative in favor of better accuracy, this method retrieves signs of learning happening earlier (~30 trials earlier on average). This is achieved by identifying the moment (trial) when the posterior probability of the correct "target" rule surpasses the probability of all other strategies. Having greater temporal precision in detecting when learning happens may have a very significant impact on studies of the neural mechanisms of learning.

4) This approach seems amenable to testing many different strategies depending on the purpose of the analysis. In the manuscript, the authors test target versus non-target strategies (correct versus incorrect) and also in another version of the analysis, they test what they call "exploratory" strategies.

5) One of the main appeals of this method is its apparent computational simplicity. It necessitates only updating on every trial the parameters of a beta distribution (prior distribution for a given strategy) with the evidence that the behavior on trial was either consistent or inconsistent with the strategy. Two scalars, the mode of the posterior (MAP) and the inverse of the variance, are all that are required for identifying the decision criterion (highest MAP and if tied lowest variance) and the learning criterion (first trial where MAP for target strategy is higher than chance).

Weaknesses:

1) It seems like a limitation of this approach is that the candidate strategies to arbitrate between must be known ex-ante. It is not clear how this approach could be applied to uncover latent strategies that are not mixtures of the strategies selected.

2) Different strategies may be indistinguishable from each other and thus it may not be possible to distinguish between them. Similarly, the fact that two strategies seem to be competing for the highest MAP doesn't necessarily mean that those are correct strategies and perhaps interchangeable as the manuscript seems to suggest.

3) The decay parameter is a necessary component to make the strategy selection non-stationary and accommodate data sets where the rules are changing throughout the task. However, the choice of the decay parameter value bounds does not seem very principled. Having this parameter as a free-parameter adds a flexibility that seems to have significant effects on when the strategy switch is detected and how stable the detected switch is.

4) This method is a useful approach for arbitrating between strategies and describing the behavior with a temporal precision that may prove important for studies attempting to tie these precise events to changes in neural activity. However, it seems limited in its explanatory power. In its current form, this method does not provide a prediction of the probability to transition from one strategy to another. And, because the MAP of different strategies may be close at any given moment, it is hard to imagine using this approach to tease out the different "mental states" that represent each strategy being at play.

The reviewers’ detailed comments, not shared here, helped us considerably to improve the paper, and we thank the reviewers for their time here. We are unsure of the merits of sharing public reviews of a paper that has now changed considerably from the version that these reviews address. Nonetheless we shall address some key points of potential misunderstanding here.

“The statistical approach is described as Bayesian but can be understood instead as computing a smoothed running average (with decay) of the strategies' success at matching choices, with a winner-take-all inference across the rules.“

This is inaccurate. The algorithm performs sequential Bayesian updates on the evidence for and against the use of each strategy considered; for a given strategy i, its output at each trial is a fully parameterised posterior distribution over the probability of that strategy being used by the subject.

We are careful in the paper to separate the algorithm’s output from our further use of that output. To plot and analyse the output we often make use of the maximum a posteriori (MAP) estimate from each posterior. Other choices are of course possible, and we discuss them in the text.<br /> In one set of simulations we quantify the results using a decision rule that chooses the strategy with the highest MAP - this is presumably the “winner-takes-all inference” in the quoted text. We do not use this anywhere else in the paper, including the analyses of the 4 datasets, and so do not consider it as part of our method, but one possible use of the output of the algorithm.

“Major concerns include the framing of the method, considerations around the strategy space, limitations in how useful the technique may be, and missing details in analyses”

Our goal for this paper was to develop a computationally lightweight, trial-resolution, Bayesian approach to tracking the probability of user-specified strategies, so that we can capture the observer’s evidence for learning or for the features driving exploratory choice (e.g. whether subjects are responding to losses or wins; are they responding to cues or choice etc). The above quote reflects their detailed review comments, where we felt this reviewer wanted a solution to a different problem, that of a parameterised latent model of strategy use: while a perfectly valid research goal, this was not what we addressed here.

“1) It seems like a limitation of this approach is that the candidate strategies to arbitrate between must be known ex-ante. It is not clear how this approach could be applied to uncover latent strategies that are not mixtures of the strategies selected.”

The problem of knowing which strategies to analyse in advance only applies when running our algorithm in real-time. The fact that it could be run in real-time on modest computing hardware is to us one of its strengths, so we consider this a good problem to have.

As noted above, rather than determine latent strategies, our goal was to build an observer model that allows users to specify whatever strategy they wanted in order to answer their scientific question(s) of their data. For example, to define when a particular rule has been learnt; or to look for changes in response to particular features of the environment, such as a cue, or to a drug treatment or other intervention.

2) Different strategies may be indistinguishable from each other and thus it may not be possible to distinguish between them. Similarly, the fact that two strategies seem to be competing for the highest MAP doesn't necessarily mean that those are correct strategies and perhaps interchangeable as the manuscript seems to suggest.

As noted above, this is an observer model, and it is thus necessarily true that there are strategies for which the observer does not have sufficient evidence to distinguish. For example, a subject who continually chooses the rewarded left-hand lever will be doing both a strategy of “go left” and of “win-stay” in response to their choice. The inability to distinguish strategies is a property of the data, not of the algorithm. Also as noted above, we do not here consider the competition between strategies.

3) The decay parameter is a necessary component to make the strategy selection non-stationary and accommodate data sets where the rules are changing throughout the task. However, the choice of the decay parameter value bounds does not seem very principled. Having this parameter as a free-parameter adds a flexibility that seems to have significant effects on when the strategy switch is detected and how stable the detected switch is.

The revised manuscript draws together the existing simulations and analysis of the method to directly address this point, showing that there is a principled range of the decay parameter in which the algorithm should operate. The Discussion also points out that this is no different to a free parameter than any frequentist approach to strategy analysis, which must choose some time windows over which to compute the frequentist probability.

4) This method is a useful approach for arbitrating between strategies and describing the behavior with a temporal precision that may prove important for studies attempting to tie these precise events to changes in neural activity. However, it seems limited in its explanatory power. In its current form, this method does not provide a prediction of the probability to transition from one strategy to another. And, because the MAP of different strategies may be close at any given moment, it is hard to imagine using this approach to tease out the different "mental states" that represent each strategy being at play.

As noted above, this is an observer model and does not intend to infer mental states. The goal is to make accurate statements about observable behaviour. We agree that an interesting extension to this approach would be to model the transitions between strategies, and had already outlined this in the Discussion.

**Reviewer 2. Luke Carroll **

The paper applies machine learning to publicly available proteomics data sets and assesses the ability to transfer learning algorithms between projects. The primary aim of these algorithms appears to be an attempt to increase consistency of retention time prediction for data-dependent acquisition data sets, however this is not explicitly stated within the text. The application of machine learning to derive insight from previous performed proteomics experienced is a worthwhile exercise.

1. The authors report ΔRT to determine fitting for their models. It would be interesting to see whether the models had other metrics used to assess fitting, or could be used to increase number of identifications within sample sets, and whether this was successful. ALternatively, was there any conclusions able to be drawn about peptide structure and RT determination from these models?

2. Project specific libraries are well known to improve results compared with publicly available databases, and the discussion on this point should be developed further through comparison of this work with other papers - particularly with advances in machine learning and neural networks in the data independent analysis field.

3. Comparison of Q-Exactiv models vs Orbitraps appears to be somewhat redundant, and possible a result of poor meta-data as Q-Exactiv instruments are orbitrap mass spectrometers. A more interesting comparison to make here would be between orbitrap and TOF instruments, though as the datasets have all been processed through MaxQuant, it is likely the vast majority were acquired on orbitrap instruments.

4. The paper uses ΔRT as the readout for all models tested, however the only chromatography variable considered in testing the models is gradient length. However, chromatography is also dependent on column chemistry, column dimensions, composition of buffer, use of traps, temperature etc. These are also likely to be contributing the variance observed between the PT datasets where these variables will be consistent and publicly available datasets. These factors are also likely to play a role in higher uncertainty for early and late eluting peptides where these factors are likely to vary most between sample sets. The metadata may not be available to use to compare within the data sets selected, so the authors should at minimum make discussion around these points.

5. Sample preparation is likely to have similar effects, and as the PT datasets are generated synthetically using ideal peptides, publicly available datasets will be generated from complex sample mixtures, and have increased variance due to inefficiencies of digestion, sample clean up and matrix effects. Previous studies on variance have also described sample preparation as the highest cause of variance. This needs further discussion

6. While the isolation windows of the m/z will lead to unobserved space, search engines setting will also apply here. From the text, it appears that the only spectra that were considered were those already identified in a search program (due to having Andromeda cut-off scores always apply). Typical setting for a database search will have a cut off of peptide sequences of at least 7 residues, making peptide masses appearing lower than 350 m/z unlikely. There is also significant amount of noise below 350 m/z and this also likely contributes to poorer fitting.

7. The authors identify differences in MSMS spectral features, however, most of these points are well known in the field. The authors should develop the discussion on the causes of the differences in fragmentation, as CID low mass drop off is expected, and the change in profile is expected with increasing activation energies. A more developed analysis could exclude precursor masses from these plots and focus solely on fragment ions generated.

8. The authors highlight that internal fragmentation of peptides could be used as a valuable resource to implement in machine learning. There has already been some success using these fragmentation patterns for sequence identification within both top-down and bottom up proteomic searches that the authors should consider discussing. However, these data do not appear to be incorporated into the machine learning models in this paper - or at least seem not to play a significant role in prediction, and this section appears to be a bit out of place.

Re-Review The changes and additions to the discussion for the paper address the key points, and have addressed some of the limitations imposed by the availability and ability to extract certain data elements particularly around sample preparation and LC settings. I think this strengthens their manuscript, and provides a more wholistic discussion of factor in the experimental setup.

Author Response

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

Note to all Reviewers

We appreciate the reviewers’ comments and suggestions for improving the manuscript. Below is a summary of new data added and a brief description of the major new results. A detailed pointby-point response follows.

New data:

• Figure 1f

• Figure 2b, f, g

• Figure 4b

• Figure S7 • Figure S8

• Figure S9

Summary of major new results/edits:

• At the request of Reviewer #1 we have updated the name of the degradation tag to be more specific and we now call it the “LOVdeg” tag.

• We have added new controls demonstrating that light stimulation does not cause photobleaching or toxicity issues (Fig. S7).

• We now show that LOVdeg can function at various points in the growth cycle, demonstrating robust degradation (Fig. 1f, Fig. S8).

• We have included relevant controls for the AcrB-LOVdeg efflux pump results (Fig. 2f-g).

• We have included important benchmarking controls, such as an EL222-only control and SsrA tag control to provide a clearer view of how LOVdeg performance compares to other systems (Fig. S9, Fig. 4b).

Additional note:

• While repeating experiments during the revision process we found that the results for the combined action of EL222 and the LOVdeg tag were not as dramatic as in our original measurements, though the overall findings are consistent with our original results. Specifically, we still find that the combination of EL222 and the LOVdeg tag produces a lower signal than either on their own. We have updated these data in the revised manuscript (Fig. 4b).

Reviewer #1:

Public Review:

Specifically controlling the level of proteins in bacteria is an important tool for many aspects of microbiology, from basic research to protein production. While there are several established methods for regulating transcription or translation of proteins with light, optogenetic protein degradation has so far not been established in bacteria. In this paper, the authors present a degradation sequence, which they name "LOVtag", based on iLID, a modified version of the blue-light-responsive LOV2 domain of Avena sativa phototropin I (AsLOV2). The authors reasoned that by removing the three C-terminal amino acids of iLID, the modified protein ends in "-E-A-A", similar to the "-L-A-A" C-terminus of the widely used SsrA degradation tag. The authors further speculated that, given the light-induced unfolding of the C-terminal domain of iLID and similar proteins, the "-E-A-A" C-terminus would become more accessible and, in turn, the protein would be more efficiently degraded in blue light than in the dark.

Indeed, several tested proteins tagged with the "LOVtag" show clearly lower cellular levels in blue light than in the dark. While the system works efficiently with mCherry (10-20x lower levels upon illumination), the effect is rather modest (2-3x lower levels) in most other cases. Accordingly, the authors propose to use their system in combination with other light-controlled expression systems and provide data validating this approach. Unfortunately, despite the claim that the "LOVtag" should work faster than optogenetic systems controlling transcription or translation of protein, the degradation kinetics are not consistently shown; in the one case where this is done, the response time and overall efficiency are similar or slightly worse than for EL222, an optogenetic expression system.

The manuscript and the figures are generally very well-composed and follow a clear structure. The schematics nicely explain the underlying principles. However, limitations of the method in its main proposed area of use, protein production, should be highlighted more clearly, e.g., (i) the need to attach a C-terminal tag of considerable size to the protein of interest, (ii) the limited efficiency (slightly less efficient and slower than EL222, a light-dependent transcriptional control mechanism), and (iii) the incompletely understood prerequisites for its application. In addition, several important controls and measurements of the characteristics of the systems, such as the degradation kinetics, would need to be shown to allow a comparison of the system with established approaches. The current version also contains several minor mistakes in the figures.

We thank reviewer #1 for the feedback and suggestions to strengthen the manuscript. We have addressed these comments in the points that follow and now include important controls and benchmarks for our molecular tool.

Major points

1. The quite generic name "LOVtag" may be misleading, as there are many LOV-based tags for different purposes.

We appreciate that it would be beneficial to have a more specific name. We have updated the name to “LOVdeg” tag, which captures both the inclusion of LOV and the degradation function of the tag.

Updated throughout the manuscript and figures

1. Throughout the manuscript, the authors use "expression levels". As protein degradation is a post-expression mechanism, "protein levels" should be used instead.

We have transitioned to using “protein levels” at many points in the manuscript.

Updated throughout the manuscript

1. Degradation dynamics (time course experiments) should be shown. The only time this is done in the current version (in Fig. 4), degradation appears to be in the same range (even a bit slower) than for EL222, which does not support the claim that the "LOVtag" acts faster than other optogenetic systems controlling protein levels.

In the revised manuscript, time course data are now shown at multiple points. These include new data in Fig. 1f and Fig. S8 that demonstrate degradation at various stages of growth. Fig. S4 also shows the dynamics of degradation when comparing to the addition of exogenously expressed ClpA. We have added text in the results section to point the reader to these data. In addition, we have made minor modifications to the text in the Introduction to avoid making claims about speed comparisons. Fig. 1f, Fig. S8, Fig. S4

Results: Design and characterization of the AsLOV2-based degradation tag, Introduction

1. "Frequency" is used incorrectly for Fig. 3. A series of 5 seconds on, 5 seconds off corresponds to a frequency of 0.1 Hz (1 illumination round / 10 s), not of 0.5 Hz. What the authors indicate as "frequency" is the fraction of illumination time. However, the (correct) frequency should be given, as this is likely the more important factor.

We have changed how we calculate frequency to use the proposed definition of one pulse per time period. We updated the values in the text and in the figure. Fig. 3c

Results: Tuning frequency response of the LOVdeg tag

1. To properly evaluate the system, several additional controls are needed:

a. To test for photobleaching of mCherry by blue light illumination, untagged controls should be shown for the mCherry-based experiments. Fluorescence always seems to be lower upon illumination, except for the AsLOV2*(546) data, where it cannot be excluded that fluorescence readings are saturated. Relatedly, the raw data for OD and fluorescence should be included. Showing a Western blot against mCherry in at least one case would allow to separate the effects of photobleaching and degradation.

We appreciate the suggestion and have conducted these important controls. We now include new data demonstrating that light induction does not change fluorescence levels using an untagged mCherry control, nor does it significantly affect endpoint OD levels. Based on these results, we did not perform a Western blot because there were no effects to separate. Fig. S7

b. In Fig. 2b, light + IPTG should be shown to estimate the activity of the system at higher expression levels.

We have added these to the figure. Light + IPTG modestly increases expression compared to IPTG only, likely due to the saturating level of IPTG added, which achieves near full induction. Fig. 2b

c. In Fig. 4, EL222 alone should be shown to allow a comparison with the LOVtag. From the data presented, it looks like EL222 is both slightly faster and more efficient than the LOVtag.

We have added the EL222-only case for comparison with LOVdeg only and EL222 + LOVdeg. We note that Reviewer #3 raised a similar concern. Fig. 4b

d. The effect of the used light on bacterial viability under exponential and stationary conditions should be shown.

In this revision, we have added new data on light exposure at various points during exponential and stationary phase (Fig. 1f, Fig. S8). These OD data show that growth curves are similar for all cultures, regardless of the time light is applied during the growth phase. Additionally, we also now include ODs for the photobleaching experiments. These data also show that growth is not significantly altered under continuous light exposure. Figure 1f, Fig. S7b

1. The claim that "Post-translational control of protein function typically requires extensive protein engineering for each use case" is not correct. The authors should discuss alternative options, e.g. based on dimerization, more extensively and in a less biased manner.

We have toned down the language in this location and at other points in the manuscript. However, we maintain that other types of post-translational control, such as dimerization or LOV2 domain insertion, require more protein engineering than inserting a degradation tag. For example, we and others have directly demonstrated this in previous work (e.g. DOI: 10.1021/acssynbio.9b00395, 10.1101/2023.05.26.542511, 10.1038/s41467-023-38993-6), where numerous split site or insertion variants need to be screened and fine-tuned for successful light control. In contrast, a degradation mechanism has the potential to require less fine tuning to achieve a light response. We have included the above sources to clarify this point. Introduction, Results: Modularity of the LOVdeg tag

Minor points

1. In Suppl. Fig. 1, amino acid numbers seem to be off. Also, the alterations in iLID (compared to AsLOV2) that are not used in "LOVtag" appear to be missing and the iLID sequence incorrect, as a consequence.

Thank you for catching this. The number indices in Fig. S1 have been corrected. We also realized we were reporting the iLID(C530M) variant in our amino acid sequence and have reverted the 530M back to C. Fig. S1

1. Why is AsLOV2(543) more efficiently degraded than AsLOV2(543) (blue column in Fig. 1d) when the dark state should be stabilized in AsLOV2(543)?

We are not sure of the exact reason for the increased degradation response in the AsLOV2*(543) variant. It may be that the dark-state stabilizing mutations introduced also have more favorable interactions with degradation machinery, although this is highly speculative.

1. Why does the addition of EL222 reduce protein levels so strongly in the dark for CpFatB1* (Fig. 5)?

We believe this effect stems from the EL222 responsive promoter (PEL222). With LOVdeg only, CpFatB1* is expressed from an IPTG inducible promoter (PlacUV5) whereas EL222 responsive constructs necessitate a promoter switch containing an EL222 binding site. We have clarified this point and expanded our discussion of these results.

Results: Optogenetic control of octanoic acid production

1. Fig. 2f / S10 are difficult to interpret. Why does illumination only lead to a significant effect at 2.5 and 5 µg/ml and not at lower concentrations, where the degradation system would be expected to be most efficient?

We have expanded our discussion on these results to explain that this likely stems from basal protein levels of AcrB-LOVdeg in the light that can provide resistance at low antibiotic concentrations. We have also added new controls to this figure to show the chloramphenicol sensitivity of a ΔacrB strain and a ΔacrB strain with an IPTG-inducible version of acrB with no induction, demonstrating the lowest achievable chloramphenicol resistance from a standard inducible system.

Results: Modularity of the LOVdeg tag, Fig. 2f-g

1. Fig. 2f / S10 do not measure the MIC (which is a clearly defined value), but the sensitivity to Chloramphenicol.

We have changed the text to use the term chloramphenicol sensitivity instead of MIC. Results: Modularity of the LOVdeg tag

1. "***" in Fig. S1 should be explained.

We have removed the ‘***’ to avoid confusion. Fig. S1

1. The fold-change differences between light and dark, indicated in some selected cases, should be listed for all figures.

We have added fold-change values where appropriate. Fig 1d, Fig. 2b

Reviewer #2:

Public Review:

In this manuscript the authors present and characterize LOVtag, a modified version of the bluelight sensitive AsLOV2 protein, which functions as a light-inducible degron in Escherichia coli. Light has been shown to be a powerful inducer in biological systems as it is often orthogonal and can be controlled in both space and time. Many optogenetic systems target regulation of transcription, however in this manuscript the authors target protein degradation to control protein levels in bacteria. This is an important advance in bacteria, as inducible protein degradation systems in bacteria have lagged behind eukaryotic systems due to protein targeting in bacteria being primarily dependent on primary amino acid sequence and thus more difficult to engineer. In this manuscript, the authors exploit the fact that the J-alpha helix of AsLOV2, which unwinds into a disordered domain in response to blue light, contains an E-A-A amino acid sequence which is very similar to the C-terminal L-A-A sequence in the SsrA tag which is targeted by the unfoldases ClpA and ClpX. They truncate AsLOV2 to create AsLOV2(543) and combine this truncation with a mutation that stabilizes the dark state to generate AsLOV2*(543) which, when fused to the C-terminus of mCherry, confers light-induced degradation. The authors do not verify the mechanism of degradation due to LOVtag, but evidence from deletion mutants contained in the supplemental material hints that there is a ClpA dominated mechanism. They demonstrate modularity of this LOVtag by using it to degrade the LacI repressor, CRISPRa activation through degradation of MCP-SoxS, and the AcrB protein which is part of the AcrAB-TolC multidrug efflux pump. In all cases, measurement of the effect of the LOVtag is indirect as the authors measure reduction in LacI repression, reduction in CRISPRa activation, and drug resistance rather than directly measuring protein levels. Nevertheless the evidence is convincing, although seemingly less effective than in the case of mCherry degradation, although it is hard to compare due to the different endpoints being measured. The authors further modify LOVtag to contain a known photocycle mutation that slows its reversion time in the dark, so that LOVtag is more sensitive to short pulses of light which could be useful in low light conditions or for very light sensitive organisms. They also demonstrate that combining LOVtag with a blue-light transcriptional repression system (EL222) can decrease protein levels an additional 269-fold (relative to 15-fold with LOVtag alone). Finally, the authors apply LOVtag to a metabolic engineering task, namely reducing expression of octanoic acid by regulating the enzyme CpFatB1, an acyl-ACP thioesterase. The authors show that tagging CpFatB1 with LOVtag allows light induced reduction in octanoic acid titer over a 24 hour fermentation. In particular, by comparing control of CpFatB1 with EL222 transcriptional repression alone, LOVtag, or both the authors show that light-induced protein degradation is more effective than light-induced transcriptional repression. The authors suggest that this is because transcriptional repression is not effective when cells are at stationary phase (and thus there is no protein dilution due to cell division), however it is not clear from the available data that the cells were in stationary phase during light exposure. Overall, the authors have generated a modular, light-activated degron tag for use in Escherichia coli that is likely to be a useful tool in the synthetic biology and metabolic engineering toolkit.

We thank Reviewer #2 for the constructive feedback. In the updated manuscript, we now include data demonstrating degradation at different growth stages and address other points brought up in the review to improve understanding of the degradation tag.

Overall, the authors present a well written manuscript that characterizes an interesting and likely very useful tool for bacterial synthetic biology and metabolic engineering. I have a few suggestions that could improve the presentation of the material.

Major Comments:

• Could the authors clarify, perhaps through OD measurements, that the cultures in the octanoic acid experiment are actually in stationary phase during the relevant light induction. It isn't clear from the methods.

We have updated the Methods to clarify that the cells are entering stationary phase (OD600 = 0.6) when light is either kept on or turned off for production experiments. Production is continued for the following 24 hours. Note that we now show OD measurements in a separate set of experiments (Fig. 1f, Fig. S8).

Methods: Octanoic acid production experiment. Fig. 1f, Fig. S8

• Can the authors clarify why there is an overall decrease in protein in the clpX deletion? And is it this initial reduction that is the source of the change in fold in 1C? Similarly, for hslU is it because overall protein levels are higher with the tag? In general, I feel that the interpretation of Supplemental Figures S6-S10 could be moved in more detail to the main text, or at least the main takeaway points. But this is a personal preference, and not necessary to the major flow of the story which is about the utility of the LOVtag tool.

As shown in Fig. S5, expression of mCherry without any degradation tag is decreased in a clpX knockout strain compared to wild type. This difference may be the result of reduced cell health, and we now note this in the text. The strains shown in Fig. 1c are in wild type cells with normal expression, so this is not the source of the fold change. As for hslU, we agree it is interesting that expression seems to increase. However, the increase is modest and could stem from gene network regulation differences in that strain compared to wild type and may not be related to LOVdeg tag degradation. Each endogenous protease is involved in a wide range of functions within the cell, and it is unknown how global gene expression is impacted. We acknowledge the suggestion of moving the protease results to the main text, but we have ultimately elected to keep these data in the Supplementary Information to maintain the flow in the manuscript. However, we have added additional text pointing the reader to the Supplemental Text and include a brief summary of the findings in the main text.

Results: Design and characterization of the AsLOV2-based degradation tag

• What is the source of the poor repression in Figure 2D?

Presumably, this stems from low levels of the CRISPRa MCP-SoxS activator, even in the presence of light. We have added this point to the text.

Results: Modularity of the LOVdeg tag

• In general, it would be nice to have light-only controls for many of the experiments to validate that light is not affecting the indicated proteins or their function.

We thank the reviewer for this suggestion and note that Reviewer #1 raised a similar concern. We have now included light-only data for a strain containing IPTG-inducible mCherry without the LOVdeg tag (Fig. S7). These data show that light itself, at the levels used in this study, does not affect mCherry expression or cell growth. This strain serves as a direct control for data presented in Fig. 1 and Fig. 2b, as the systems are identical except for the addition of the LOVdeg tag onto either mCherry or the LacI repressor. Additionally, the control translates to other experiments since mCherry is used as a reporter for other systems in this study. Fig. S7

• It would be nice to directly measure the function of the tool at different phases of E. coli growth to show directly that protein degradation works at stationary phase, rather than the more indirect measurements used in the octanoic acid experiment.

We thank the reviewer for this suggestion, which significantly strengthens our results. We have added an experiment that tests the LOVdeg tag at different phases of growth (Fig. 1f, Fig. S8). In this experiment, cultures are growth from early exponential to stationary phase, and light is introduced at various points. Exposure windows of 4 hours, ranging from early exponential to stationary phase, all show functional light inducible degradation. Fig. 1f, Fig. S8.

Results: Design and characterization of the AsLOV2-based degradation tag

Minor Comments:

• It would be nice to make clear that the data in S6d and S7 is repeated, but with the HslUV data in S7.

We clarified this point in the caption of Fig. S4 (the former Fig. S7 in the original manuscript). Fig. S4 caption

• Why was 5s picked for the frequency response in Figure 3

We picked 5s because 1) it is a substantially shorter timescale than overall degradation dynamics seen for the LOVdeg tag, and 2) we found that shorter pulses could not be reliably achieved with the light stimulation hardware and software we used (Light Plate Apparatus with Iris software). To ensure high fidelity pulses, we opted for 5 second pulses that we empirically determined to be stable throughout long experiments. We have added text clarifying this. Results: Tuning frequency response of the LOVdeg tag

Reviewer #3:

Public Review:

The authors present the mechanism, validation, and modular application of LOVtag, a light-responsive protein degradation tag that is processed by the native degradosome of Escherichia coli. Upon exposure to blue light, the c-terminal alpha helix unfolds, essentially marking the protein for degradation. The authors demonstrate the engineered tag is modular across multiple complex regulatory systems, which shows its potential widespread use throughout the synthetic biology field. The step-by-step rational design of identifying the protein that was most dark stabilized as well as most light-responsive for degradation, was useful in terms of understanding the key components of this system. The most compelling data shows that the engineered LOVTag can be fused to multiple proteins and achieve light-based degradation, without affecting the original function of the fused protein; however, results are not benchmarked against similar degradation tagging and optogenetic control constructs. Creating fusion proteins that do not alter either of the original functions, is often difficult to achieve, and the novelty of this should be expanded upon to drive further impact.

We appreciate the feedback from Reviewer #3 to improve the manuscript. We have included important controls and benchmarking experiments to address the reviewer’s concerns, which are detailed in the points below.

Benchmarking:

The similarity between the L-A-A sequence of SsrA and the E-A-A sequence of LOVtag is one of the pieces of evidence that led the authors to their current protein design. The differences in degradation efficiency between the SsrA degradation tag and LOVtag are not shown, and benchmarking against SsrA would be a valuable way to demonstrate the utility of this construct relative to an established protein tagging tool.

We thank the reviewer for suggesting an experiment to benchmark performance. We have added new experimental data where a full length SsrA tag is added to a fusion protein of nearly identical size (mCherry-iLID), allowing us to directly compare performance to mCherryLOVdeg (Fig. S9). These results show that light inducible control with LOVdeg tag decreases protein expression levels to near those achieved with the native SsrA tag. Fig. S9.

Results: Design and characterization of the AsLOV2-based degradation tag

Additionally, there is a lack of an EL222-only control presented in Figure 4b and in the results section beginning with "Integrating the LOVtag with EL222...". Without benchmarking against this control the claim that "EL222 and the LOVtag work coherently to decrease expression" is unsubstantiated. No assumptions of synergy can be made.

We appreciate this comment and note that Reviewer #1 raised a similar concern. We have added data to Fig. 4b with an EL222-only control for comparison. Fig. 4b

The dramatic change in dark octanoic acid titer between the EL222, LOVtag and combined conditions are surprising, especially in comparison to the lack of change in the dark mCherry expression shown in Figure 4b. This data is the only to suggest that LOVtag may perform better than EL222. However, the inconsistencies in dark state regulation presented in the two experiments, and between conditions in this experiment bring the latter claim to question. A recommendation is that the authors either repeat this experiment, or comment on the observed discrepancy in dark state octanoic acid titers in their discussion.

First, a key difference between the data presented in Fig. 4 and Fig. 5 is that the production experiment is conducted over a long time period (24 hours) and the EL222/LOVdeg reporter experiment is conducted over 5 hours. Likely, performance differences between EL222 and the LOVdeg tag become more pronounced as protein accumulation occurs. Second, the LOVdeg only construct is expressed from a non-EL222 promoter which is able to achieve higher expression (see response to Reviewer #1, Minor point #3). Lastly, a convoluting factor is that the relationship between expression of CpFatB1 and octanoic acid production is not completely linear, and there are likely thresholds or expressions windows that result in similar endpoint titers. We agree a more detailed examination of how CpFatB1 changes over the course of the production period would be very interesting. However, this is beyond the scope of the present study, whose goal is to introduce and showcase the utility of the LOVdeg tag as a tool. We have added new discussion on this in the Results section to clarify some of these points. We have also repeated all experiments in Fig. 4 and consistently see the LOVdeg tag performing as well as or better than EL222. As noted in the remarks to all reviewers, these data have been updated in the revised manuscript.

Results: Optogenetic control of octanoic acid production. Fig. 4d

Based on the methodology presented, no change in the duration in light exposure was tested, even though this may be an important part of the system response. The on/off, for example in Figure 4b, is either all light or all dark, but they claim that their system is beneficial especially at stationary phase. The authors should consider showing the effects of shifting from dark to light at set intervals. (i.e. 1 hr dark then light, 2hr dark until light, etc.) This data would also aid in supporting the utility of this tag for controlling expression during different growth phases, where light may be used after the cells have reached a certain phase.

We have added new data showing the effect of light stimulation at different times in the growth cycle (see response to Reviewer #2, bullet point #5). These data demonstrate that the LOVdeg tag performs well at various points in the growth cycle. Fig. 1f, Fig. S8.

Results: Design and characterization of the AsLOV2-based degradation tag

Minor Revisions Figures:

• Figure 1:

• More clarity is needed in the naming conventions for this figure and in the body of the text. For example, a different convention than 546 and 543 should be used to refer to the full and truncated lengths of the tag. It would greatly aid understanding for this to be made more clear. The authors could simply continue to use "full" and "truncated" to refer to them. In addition, the term "stabilizing mutations" in 1c could be changed to read "dark state stabilizing mutations" to aid in clarity.

When describing the design of the LOVdeg tag, we opted towards a more technically accurate description over clarity in order to make our engineering process easily comparable to other LOV2 systems. As such, we kept the number-based nomenclature (543 or 546) to represent the domain within the phototropin 1 protein from Avena sativa (AsLOV2). The domain used in this study, and many other studies, are only amino acids 404-546, i.e. not the full sequence, thus saying simply ‘full’ or ‘truncated’ is not technically accurate. We believe the detailed nomenclature, which is limited to one section, is important to provide clarity on exactly what we used for protein engineering. In the revised version we introduce the nickname “LOVdeg” tag earlier and use it throughout the rest of the manuscript.

Results: Design and characterization of the AsLOV2-based degradation tag

• 1b It is not clear that this is the dark state stabilized structure in the figure, but is referred to as such only in the body of the text.

We have added text in the manuscript to clarify this is AsLOV2, not iLID, and have labeled it in the figure caption as well.

Results: Design and characterization of the AsLOV2-based degradation tag

• 1d. Fold change is reported in Figure 2d, and may be relevant to include those values in 1d as well.

Done. Fig. 1d

• 1e. It is not clear which tag is being used in this bar plot. Please specify that this is the dark state stabilized, truncated tag.

We have added a title to the plot and language to the caption, both of which clarify this point. Fig. 1e

• In addition, the microscopy images provided in supplemental material should be included in the first figure as it adds a compelling observation of LOVtag activity.

We are pleased to hear that the microscopy results are beneficial, however we elected to leave them in Supplementary to preserve the flow of the manuscript in the text surrounding Fig. 1.

• Figure 2:

• 2d. It is unclear what the 2.5x fold change is relative to (the baseline or the dark)

We have added a line in the figure to clarify the comparison being made. Fig. 2d

• 2f. More discussion can be added to describe what concentration of chloramphenicol is biologically/bioreactor relevant.

Our previous studies on the relationship between AcrAB expression and mutation rate (cited in the text) were carried out at a concentration within the range in which the LOVdeg tag is effective (5 μg/ml), suggesting this range to be relevant to tolerance and resistance.

• Figure 3:

• We recommend that this data and discussion are better suited for supplementary figures. The results shown here essentially recapitulate the same findings of Zoltowski et al., 2009. In addition, the paper describing this mutation should be cited in this figure caption in addition to the body of the text

Although these results are in line with previous findings, we believe this dataset is important for several reasons. First, the agreement with known mutations validates the unfolding-based mechanism for degradation control. Second, degradation that is contingent on unfolding of LOV2 offers a direct actuating mechanism of photocycle properties. Other systems, like that in Zoltowski et al., examine properties of purified proteins but lack the mechanism to translate its effect in live cells. This figure demonstrates how degradation can do so and lays the groundwork for degradation-based frequency processing circuits. Last, there are discrepancies between photocycle kinetics in situ, as reported by Li et al. (DOI: 10.1038/s41467-020-18816-8), and in cell-free studies such as in Zoltowski et al. The studies use different methods of measuring photocycle kinetics (in situ vs cell-free). This dataset substantiates relaxation times from Li et al. and suggests cell-free relaxation time constants are over estimated relative to our live cell results.

• Figure 4:

• There is a lack of an EL222-only control presented in Figure 4b. Without this data present, the claim that "EL222 and the LOVtag work coherently to decrease expression" is unsubstantiated. No assumptions of synergy can be made.

We have added EL222-only data to the figure; we note that Reviewer #1 made a similar request. Figure 4b

Manuscript

Results

• Design and characterization...

• Due to the extensive discussion of ClpX at the beginning of this section, more of the results on evaluating the binding partners and mechanism of LOVtag degradation should be presented in the main body of the manuscript and not in supplementary materials.

To maintain flow of the manuscript and focus on how the LOVdeg tag works as a synthetic biology tool, we have opted to keep this section in the Supplement Information, but have several lines in the text related to Fig. 1 that point the reader to this material. Results: Design and characterization of the AsLOV2-based degradation tag

• In the second paragraph of this section, the authors theorize that the C-terminal truncated E-AA sequence will "remain caged as part of the folded helix". How did the authors determine this? Was there any evidence to suggest that the truncated state would be any more responsive than the full length sequence? More data or rationale may need to be introduced to support the overall hypothesis presented in this paragraph.

We determined this by examining the crystal structure which shows that the E-A-A sequence is part of the folded helix. As seen in Fig. 1b, addition of amino acids after the EAAKEL sequence would not be part of the folded helix which ends prior to the terminal leucine. We added text to clarify our logic.

Results: Design and characterization of the AsLOV2-based degradation tag

• The similarity between the L-A-A sequence of SsrA and the E-A-A sequence of LOVtag is one of the pieces of evidence that brought the authors to their current protein design. The differences in degradation efficiency between the SsrA degradation tag and LOVtag are not clear, and benchmarking against SsrA would be a valuable way to demonstrate the utility of this construct relative to an established protein tagging tool.

We added an SsrA comparison to benchmark the system. Fig. S9

Results: Design and characterization of the AsLOV2-based degradation tag

• Tuning frequency and response...

• Overall the results presented in this section essentially recapitulate the effects that mutation presented in Zoltowski et. al., 2009 have on AsLOV2 dark state recovery and although this is a useful observation of LOVtag performance, a recommendation is to move this into a supplementary section.

See above response to Fig. 3 comment.

• Integrating the LOVtag with EL222...

• The claim is made in this section that LOVtag and EL222 work synergistically, however the experiments presented do not test repression due to EL222 activity alone. Without benchmarking against this control, the claim of synergy is not supported and we recommend that the authors perform this experiment again with the EL222-only control.

We have added this important control. Fig. 4b

Discussion

• The statement "the LOVtag can easily be integrated with existing optogenetic systems to enhance their function" is not substantiated without benchmarking LOVtag against an EL222- only control. As mentioned above this condition should be included in the experiments discussed in Figure 4 and in the section "Integrating the LOVtag with EL222.."

We added EL222-only regulation to benchmark the LOVdeg tag and LOVdeg + EL222 experiments. Fig. 4b

Experiments

Applications:

The application of this tag to the metabolic control of octanoic acid production could be more impactful. For instance, using the LOVtag with two different enzymes to change the composition of long/short chain fatty acids with light induction., Or possibly integrating the tag into a switch to activate production. However, the authors address that "decreasing titers is not the overall goal in metabolic engineering" in their discussion, and therefore the pursuit of this additional experiment is up to the authors' discretion.

We appreciate the suggestions for further applications of the LOVdeg tag. We envision that follow up studies will focus on the application of the LOVdeg tag in metabolic engineering. However, this will require significant development of production systems. We believe this to be out of the scope of this work, where the goal is to present the design and function of the LOVdeg tag as a tool.

Windows Terminal<br /> wt.exe -w gns3 new-tab --title %d --suppressApplicationTitle plink -telnet -P %p %h

Windows Terminal Colour Tab wt.exe -w gns3 new-tab --title %d --suppressApplicationTitle --tabColor "#099ecb" plink -telnet -P %p %h

You have to know what are you collecting for. A good analogy is that you want to build a Lego Catherdral, and when you do that, you know exactly what you are looking for.

ipfs://bafybeidhfjjaqwwt2jlozgp7vumwigm6yxzzzldbzz5shtqsmgnwtqsw6y/

Reviewer #2 (Public Review):

In this study, Leighton et al performed remarkable experiments by combining in-vivo patch-clamp recording with two-photon dendritic Ca2+ imaging. The voltage-clamp mode is a major improvement over the pioneer versions of this combinatorial experiment that has led to major breakthroughs in the neuroscience field for visualizing and understanding synaptic input activities in single cells in-vivo (sharp electrodes: Svoboda et al, Nature 1997, Helmchen et al, Nature Neurosci 1999; whole-cell current-clamp: Jia et al, Nature 2010, Chen et al, Nature 2011. I suggest that these papers would be cited). This is because in voltage-clamp mode, despite the full control of membrane voltage in-vivo not being realistic, is nevertheless most effective in preventing back-propagation action potentials, which would severely confound the measurement of individual synaptically-induced Ca2+ influx events. Furthermore, clamping the cell body at a strongly depolarized potential (here the authors did -30mV) also facilitates the detection of synaptically-induced Ca2+ influx. As a result, the authors successfully recorded high-quality Ca2+ imaging data that can be used for precise analysis. To date, even in view of the rapid progress of voltage-sensitive indicators and relevant imaging technologies in recent years, this very old 'art' of combining single-cell electrophysiology and two-photon imaging (ordinary, raster-scanned, video-rate imaging) of Ca2+ signals still enables measurements of the best-level precision.

On the other hand, the interpretation of data in this study is a bit narrow-minded and lacks a comprehensive picture. Some suggestions to improve the manuscript are as follows:

1. The authors made a segregation of 'spine synapse' and 'shaft synapse' based solely on the two-photon images in-vivo. However, caution shall be taken here, because the optical resolution under in-vivo imaging conditions like this cannot reliably tell apart whether a bright spot within or partially overlapping a segment of the dendrite is a spine on top of (or below) it. Therefore, what the authors consider as a 'shaft synapse' (by detecting Ca2+ hotspots) has an unknown probability of being just a spine on top or below the dendrite. If there is other imaging data of higher axial resolution to validate or calibrate, the authors shall take some further considerations or analysis to check the consistency of their data, as the authors do need such a segregation between spine and shaft synapses to show how they evolve over the brain development stages.

2. The use of terminology 'bursts of spontaneous inputs' for describing voltage-clamp data seems improper. Conventionally, 'burst' refers to suprathreshold spike firing events, but here, the authors use 'burst' to refer to inward synaptic currents collected at the cell body. Not every excitatory synaptic input (or ensemble of inputs) activation will lead to spike firing under naturalistic conditions, therefore, these two concepts are not equivalent. It is recommended to use 'barrage of inputs' instead of 'burst of inputs'. Imagine a full picture of the entire dendritic tree, the fact that the authors could always capture spontaneous Ca2+ events here and there within a few pieces of dendrites within an arbitrary field-of-view suggests that, the whole dendritic tree must have many more such events going on as a barrage while the author's patch electrode picks up the summed current flow from the whole dendritic tree.

3. Following the above issue, an analysis of the temporal correlation between synaptic (not segregating 'spine' or 'shaft') Ca2+ events and EPSCs is absent. Again, the authors drew arbitrary time windows to clump the events for statistical analysis. However, the demonstrated example data already shows that the onset times of individual synaptic Ca2+ events do not necessarily align with the beginning of a 'barrage' inward current event.

4. The authors claim that "these observations indicate that the activity patterns investigated here are not or only slightly affected by low-level anesthesia". It would be nice to show some of the recordings in this work without any anesthesia to support this claim.

This should go in the previous section Daily Workflow.

I submitted a PR

relevant only for windows server 2016 and older

Reviewer #2 (Public Review):

Summary:<br /> In Ferrareti et al. they identify adaptively introgressed genes using VolcanoFinder and then identify pathways enriched for adaptively introgressed genes. They also use a signet to identify pathways that are enriched for Denisovan alleles. The authors find that angiogenesis and nitric oxide induction are enriched for archaic introgression.

Strengths:<br /> Most papers that have studied the genetic basis of high altitude (HA) adaptation in Tibet have highly emphasized the role of a few genes (e.g. EPAS1, EGLN1), and in this paper, the authors look for more subtle signals in other genes (e.g EP300, NOS 2) to investigate how archaic introgression may be enriched at the pathway level.

Looking into the biological functions enriched for Denisovan introgression in Tibetans is important for characterizing the impact of Denisovan introgression.

Weaknesses:<br /> The manuscript lacks details or justification about how/why some of the analyses were performed. Below are some examples where the authors could provide additional details.

The authors made specific choices in their window analysis. These choices are not justified or there is no comment as to how results might change if these choices were perturbed. For example, in the methods, the authors write "Then, the genome was divided into 200 kb windows with an overlap of 50 kb and for each of them we calculated the ratio between the number of significant SNVs and the total number of variants."

Additional information is needed for clarity. For example, "we considered only protein-protein interactions showing confidence scores {greater than or equal to} 0.7 and the obtained protein frameworks were integrated using information available in the literature regarding the functional role of the related genes and their possible involvement in high-altitude adaptation." What do the confidence scores mean? Why 0.7?

In the method section (Identifying gene networks enriched for Denisovan-like derived alleles), the authors write "To validate VolcanoFinder results by using an independent approach". Does this mean that for signet the authors do not use the regions identified as adaptively introgressed using volcanofinder? I thought in the original signet paper, the authors used a summary describing the amount of introgression of a given region.

Later, the authors write "To do so, we first compared the Tibetan and Denisovan genomes to assess which SNVs were present in both modern and archaic sequences. These loci were further compared with the ancestral reconstructed reference human genome sequence (1000 Genomes Project Consortium et al., 2015) to discard those presenting an ancestral state (i.e., that we have in common with several primate species)." It is not clear why the authors are citing the 1000 genomes project. Are they comparing with the reference human genome reference or with all populations in the 1000 genomes project? Also, are the authors allowing derived alleles that are shared with Africans? Typically, populations from Africa are used as controls since the Denisovan introgression occurred in Eurasia.

The methods section for Figures 4B, 4C, and 4D is a little hard to understand. What is the x-axis on these plots? Is it the number of pairwise differences to Denisovan? The caption is not clear here. The authors mention that "Conversely, for non-introgressed loci (e.g., EGLN1), we might expect a remarkably different pattern of haplotypes distribution, with almost all haplotype classes presenting a larger proportion of non-Tibetan haplotypes rather than Tibetan ones." There is clearly structure in EGLN1. There is a group of non-Tibetan haplotypes that are closer to Denisovan and a group of Tibetan haplotypes that are distant from Denisovan...How do the authors interpret this?

In the original signet paper (Guoy and Excoffier 2017), they apply signet to data from Tibetans. Zhang et al. PNAS (2021) also applied it to Tibetans. It would be helpful to highlight how the approach here is different.

Reducing the performance hit of copying data between processes:

Option #1: Just use threads

Processes have overhead, threads do not. And while it’s true that generic Python code won’t parallelize well when using multiple threads, that’s not necessarily true for your Python code. For example, NumPy releases the GIL for many of its operations, which means you can use multiple CPU cores even with threads.

``` # numpy_gil.py import numpy as np from time import time from multiprocessing.pool import ThreadPool

arr = np.ones((1024, 1024, 1024))

start = time() for i in range(10): arr.sum() print("Sequential:", time() - start)

expected = arr.sum()

start = time() with ThreadPool(4) as pool: result = pool.map(np.sum, [arr] * 10) assert result == [expected] * 10 print("4 threads:", time() - start) ```

When run, we see that NumPy uses multiple cores just fine when using threads, at least for this operation:

$ python numpy_gil.py Sequential: 4.253053188323975 4 threads: 1.3854241371154785

Pandas is built on NumPy, so many numeric operations will likely release the GIL as well. However, anything involving strings, or Python objects in general, will not. So another approach is to use a library like Polars which is designed from the ground-up for parallelism, to the point where you don’t have to think about it at all, it has an internal thread pool.

Option #2: Live with it

If you’re stuck with using processes, you might just decide to live with the overhead of pickling. In particular, if you minimize how much data gets passed and forth between processes, and the computation in each process is significant enough, the cost of copying and serializing data might not significantly impact your program’s runtime. Spending a few seconds on pickling doesn’t really matter if your subsequent computation takes 10 minutes.

Option #3: Write the data to disk

Instead of passing data directly, you can write the data to disk, and then pass the path to this file: * to the subprocess (as an argument) * to parent process (as the return value of the function running in the worker process).

The recipient process can then parse the file.

``` import pandas as pd import multiprocessing as mp from pathlib import Path from tempfile import mkdtemp from time import time

def noop(df: pd.DataFrame): # real code would process the dataframe here pass

def noop_from_path(path: Path): df = pd.read_parquet(path, engine="fastparquet") # real code would process the dataframe here pass

def main(): df = pd.DataFrame({"column": list(range(10_000_000))})

with mp.get_context("spawn").Pool(1) as pool:
    # Pass the DataFrame to the worker process
    # directly, via pickling:
    start = time()
    pool.apply(noop, (df,))
    print("Pickling-based:", time() - start)

    # Write the DataFrame to a file, pass the path to
    # the file to the worker process:
    start = time()
    path = Path(mkdtemp()) / "temp.parquet"
    df.to_parquet(
        path,
        engine="fastparquet",
        # Run faster by skipping compression:
        compression="uncompressed",
    )
    pool.apply(noop_from_path, (path,))
    print("Parquet-based:", time() - start)

if name == "main": main() `` **Option #4:multiprocessing.shared_memory`**

Because processes sometimes do want to share memory, operating systems typically provide facilities for explicitly creating shared memory between processes. Python wraps this facilities in the multiprocessing.shared_memory module.

However, unlike threads, where the same memory address space allows trivially sharing Python objects, in this case you’re mostly limited to sharing arrays. And as we’ve seen, NumPy releases the GIL for expensive operations, which means you can just use threads, which is much simpler. Still, in case you ever need it, it’s worth knowing this module exists.

Note: The module also includes ShareableList, which is a bit like a Python list but limited to int, float, bool, small str and bytes, and None. But this doesn’t help you cheaply share an arbitrary Python object.

A bad option for Linux: the "fork" context

You may have noticed we did multiprocessing.get_context("spawn").Pool() to create a process pool. This is because Python has multiple implementations of multiprocessing on some OSes. "spawn" is the only option on Windows, the only non-broken option on macOS, and available on Linux. When using "spawn", a completely new process is created, so you always have to copy data across.

On Linux, the default is "fork": the new child process has a complete copy of the memory of the parent process at the time of the child process’ creation. This means any objects in the parent (arrays, giant dicts, whatever) that were created before the child process was created, and were stored somewhere helpful like a module, are accessible to the child. Which means you don’t need to pickle/unpickle to access them.

Sounds useful, right? There’s only one problem: the "fork" context is super-broken, which is why it will stop being the default in Python 3.14.

Consider the following program:

``` import threading import sys from multiprocessing import Process

def thread1(): for i in range(1000): print("hello", file=sys.stderr)

threading.Thread(target=thread1).start()

def foo(): pass

Process(target=foo).start() ```

On my computer, this program consistently deadlocks: it freezes and never exits. Any time you have threads in the parent process, the "fork" context can cause in potential deadlocks, or even corrupted memory, in the child process.

You might think that you’re fine because you don’t start any threads. But many Python libraries start a thread pool on import, for example NumPy. If you’re using NumPy, Pandas, or any other library that depends on NumPy, you are running a threaded program, and therefore at risk of deadlocks, segfaults, or data corruption when using the "fork" multiprocessing context. For more details see this article on why multiprocessing’s default is broken on Linux.

You’re just shooting yourself in the foot if you take this approach.

Reviewer #2 (Public Review):

In this work, Urtecho et al. use genome-integrated massively parallel reporter assays (MPRAs) to catalog the locations of promoters throughout the E. coli genome. Their study uses four different MPRA libraries. First, they assayed a library containing 17,635 promoter regions having transcription start sites (TSSs) previously reported by three different sources. They found that 2,760 of these regions exhibited transcription above an experimentally determined threshold. Second, they assayed a library using sheared E. coli genome fragments. This library allowed the authors to systematically identify candidate promoter regions throughout the genome, some of which had not been identified before. Additionally, by performing experiments with this library under different growth conditions, the authors were able to identify promoters with condition-dependent activity. Third, to improve the resolution at which they were able to identify transcription start sites, the authors assayed a library that tiled all candidate promoter regions identified using the genomic fragments library. Data from the tiled library allowed the authors to identify minimal promoter regions. Fourth, the authors assayed a scanning mutagenesis library in which they systematically scrambled individual 10 bp windows within 2,057 previously identified active promoters at 5 bp intervals. After validation with known promoters, this approach allowed the authors to identify novel functional elements within regulatory regions. Finally, the authors fit multiple machine learning models to their data with the goal of predicting promoter activity from DNA sequences.

The work by Urtecho et al. provides an important resource for researchers studying bacterial transcriptional regulation. Despite decades of study, a comprehensive catalogue of E. coli promoters is still lacking. The results of Urtecho et al. provide a state-of-the-art atlas of promoters in the E. coli genome that is readily accessible through the website, http://ecolipromoterdb.com. The authors' work also provides an important demonstration of the power of genome-integrated MPRAs. Unlike many MPRA-based studies, the authors use the results of their initial MPRAs to design follow-up MPRAs, which they then carry out. Finally, the scanning mutagenesis MPRAs the authors perform provide valuable data that could lead to the discovery of novel transcription factor binding sites and other functional regulatory sequence elements.

Below I provide two major critiques and some minor critiques of the paper. The purpose of these critiques is simply to help the authors improve the quality of the manuscript.

Major points:<br /> 1. Ultimately, a comprehensive atlas of E. coli promoters should include nucleotide resolution TSS data, which is not present in the MPRA datasets reported by Urtecho et al.. The authors do use some methods to narrow down the positions of TSSs, but these methods do not provide the resolution one would ideally like to see in a TSS atlas. I understand that acquiring single-nucleotide-resolution data is beyond the scope of this manuscript, but it still might make sense for the authors to discuss this limitation in the Discussion section.

2. The authors should clarify which points in the Results section are novel conclusions or observations, and which points are simply statements that prior conclusions or observations were confirmed. This distinction can be unclear at times.

Minor points:<br /> 1. Line 200-203: "We conclude that inactive TSS-associated promoters lack -35 elements but may become active in growth conditions where additional transcription factors mobilize and facilitate RNAP positioning in the absence of a -35 motif." Making this type of mechanistic observations from the slight difference observed in the enrichment analysis seems too speculative to me. Also, I do not understand how the discrepancies can be explained in terms of transcription factor differences. If the previous studies from which the annotated TSS were extracted were also performed during the log phase in rich media, why would the transcription factors present be different?

2. Line 224-226: "Active TSSs not overlapping a candidate promoter region generally exhibited weak activity, which may indicate that greater sensitivity is achieved through testing of oligo-array synthesized regions (Figure S3)." The authors should clarify this statement. In particular, it is mechanistically unclear why one library would be more sensitive than another if they contain similar sequences.

3. Figure 2B. The authors should clarify that the heights of the arrows correspond to TSS activity as assayed by one library and that the pile-up plots represent promoter activity as assayed by a different library.

4. Line 255-257: "We also observed an enrichment for 150 bp minimal promoter regions, although these were generally weak indicating that our resolution is limited when tiling weaker promoters." The authors should clarify whether the peak at 150 bp is an artifact of using oligos containing 150 bp tiles to construct the library. Also, the authors should clarify why there are some minimal promoters with lengths > 150 bp when the length of the tiles was 150 bp.

5. Line 262 refers to "Supplementary Table 1", but I was not able to find this table in the supplement.

6. Line 324-325: "We used a σ70 PWM to identify the highest-scoring σ70 motifs within intragenic promoters and determined their relative coding frames". I find the term "relative coding frame" here to be unclear; the authors should clarify what they mean.

7. Figure 3 C , D: The authors should use the same terminology in the plots and the methods section describing them. They should also clarify how the values plotted in C and D were computed.

8. Line 329-332: "The observed depletion of -35 motifs positioned in the +2 reading frame and -10 motifs in the +1 reading frame is likely due to the fact that the canonical sequences for these motifs would create stop codons within the protein if placed at these positions." The definition of the reading frame here is unclear. Do the authors mean that the 0 frame is defined as occurring when the hexamer exactly overlaps 2 codons, the +1 frame is when the hexamer is shifted 1 nt downstream of that position, and the +2 frame is when the hexamer is shifted 2 nt downstream of that position?

9. Line 538-539: "We performed hyperparameter tuning for a three-layer CNN and achieved an AUPRC =0.44." The authors should explicitly describe the architecture used for the CNN, and perhaps include a diagram of this architecture. In addition, the authors should clarify the mathematical forms of the other methods tested.

10. Line 1204-1205: "We standardized all datasets as detailed above in 'Universal Promoter Expression Quantification and Activity Thresholding'". That title does not appear before in the text. I believe the appropriate subsection is called "Standardizing Promoter Expression Quantification and Activity Thresholding".

11. Line 1265-1266: "We include a k-mer if the absolute correlation with expression is greater than the 'random' k-mer frequency, resulting in 4800/5440 filtered k-mers." It is unclear to me which two correlations are being compared. Please clarify. For example, would this be accurate: "We include a k-mer if the absolute correlation of its frequency with expression is greater than the absolute correlation of its 'random' frequency with expression"?

• Quickly add a desktop by using the keyboard shortcut <kbd>Windows Key + Ctrl + D</kbd>.

• Quickly switch desktops by using the keyboard shortcuts <kbd>Windows Key + Ctrl + Left Arrow</kbd> or <kbd>Windows Key + Ctrl + Right Arrow</kbd>.

• To rename your desktops, open the Task View pane, right-click a desktop and click Rename.

• To change desktop backgrounds open the Task View pane, right-click a desktop and click Choose background.

• You can click and drag applications from one desktop to another through the Task View pane, or you can right-click an application, click Move to and then click which desktop you want to move the application to.

• To close a virtual desktop, open up the Task View pane and hover over the desktop you want to close until an X appears in the upper-right corner. Click the X to close the desktop. You can also open Task View by clicking <kbd>Windows Key + Tab</kbd>. Then, use your arrow keys to select a virtual desktop and clicking the <kbd>Delete</kbd> key on the virtual desktop you want to close.

SCALE????

this info should go into the introduction. This page should be devoted to image analysis and connection to the quotation.

Reviewer #1 (Public Review):

Summary:<br /> In this paper, Song, Shi, and Lin use an existing deep learning-based sequence model to derive a score for each haplotype within a genomic region, and then perform association tests between these scores and phenotypes of interest. The authors then perform some downstream analyses (fine-mapping, various enrichment analyses, and building polygenic scores) to ensure that these associations are meaningful. The authors find that their approach allows them to find additional associations, the associations have biologically interpretable enrichments in terms of tissues and pathways, and can slightly improve polygenic scores when combined with standard SNP-based PRS.

Strengths:<br /> - I found the central idea of the paper to be conceptually straightforward and an appealing way to use the power of sequence models in an association testing framework.<br /> - The findings are largely biologically interpretable, and it seems like this could be a promising approach to boost power for some downstream applications.

Weaknesses:<br /> - The methods used to generate polygenic scores were difficult to follow. In particular, a fully connected neural network with linear activations predicting a single output should be equivalent to linear regression (all intermediate layers of the network can be collapsed using matrix-multiplication, so the output is just the inner product of the input with some vector). Using the last hidden layer of such a network for downstream tasks should also be equivalent to projecting the input down to a lower dimensional space with some essentially randomly chosen projection. As such, I am surprised that the neural network approach performs so well, and it would be nice if the authors could compare it to other linear approaches (e.g., LASSO or ridge regression for prediction; PCA or an auto-encoder for converting the input to a lower dimensional representation).

- A very interesting point of the paper was the low R^2 between the HFS scores in adjacent windows, but the explanation of this was unclear to me. Since the HFS scores are just deterministic functions of the SNPs, it feels like if the SNPs are in LD then the HFS scores should be and vice versa. It would be nice to compare the LD between adjacent windows to the average LD of pairs of SNPs from the two windows to see if this is driven by the fact that SNPs are being separated into windows, or if sei is somehow upweighting the importance of SNPs that are less linked to other SNPs (e.g., rare variants).

- There were also a number of robustness checks that would have been good to include in the paper. For instance, do the findings change if the windows are shifted? Do the findings change if the sequence is reverse-complemented?

- It was also difficult to contextualize the present work in terms of recent results showing that sequence models tend to not do very well at predicting cross-individual expression changes (and such results presumably hold for predicting cross-individual chromatin changes). In particular, it would be good for the authors to contrast their findings with the work of Alex Sasse and colleagues (https://www.biorxiv.org/content/10.1101/2023.03.16.532969.abstract) and Connie Huang and colleagues (https://www.biorxiv.org/content/10.1101/2023.06.30.547100.abstract).

Reviewer #2 (Public Review):

Summary:<br /> In this work, Song et al. propose a locus-based framework for performing GWAS and related downstream analyses including finemapping and polygenic risk score (PRS) estimation. GWAS are not sufficiently powered to detect phenotype associations with low-frequency variants. To overcome this limitation, the manuscript proposes a method to aggregate variant impacts on chromatin and transcription across a 4096 base pair (bp) loci in the form of a haplotype function score (HFS). At each locus, an association is computed between the HFS and trait. Computing associations at the level of imputed functional genomic scores should enable the integration of information across variants spanning the allele frequency spectrum and bolster the power of GWAS.

The HFS for each locus is derived from a sequence-based predictive model. Sei. Sei predicts 21,907 chromatin and TF binding tracks, which can be projected onto 40 pre-defined sequence classes ( representing promoters, enhancers, etc.). For each 4096 bp haplotype in their UKB cohort, the proposed method uses the Sei sequence class scores to derive the haplotype function score (HFS). The authors apply their method to 14 polygenic traits, identifying ~16,500 HFS-trait associations. They finemap these trait-associated loci with SuSie, as well as perform target gene/pathway discovery and PRS estimation.

Strengths:<br /> Sequence-based deep learning predictors of chromatin status and TF binding have become increasingly accurate over the past few years. Imputing aggregated variant impact using Sei, and then performing an HFS-trait association is, therefore, an interesting approach to bolster power in GWAS discovery. The manuscript demonstrates that associations can be identified at the level of an aggregated functional score. The finemapping and pathway identification analyses suggest that HFS-based associations identify relevant causal pathways and genes from an association study. Identifying associations at the level of functional genomics increases the portability of PRSs across populations. Imputing functional genomic predictions using a sequence-based deep learning model does not suffer from the limitation of TWAS where gene expression is imputed from a limited-size reference panel such as GTEx.

However, there are several major limitations that need to be addressed.

Major concerns/weaknesses:<br /> 1. There is limited characterization of the locus-level associations to SNP-level associations. How does the set of HFS-based associations differ from SNP-level associations?

2. A clear advantage of performing HFS-trait associations is that the HFS score is imputed by considering variants across the allele frequency spectrum. However, no evidence is provided demonstrating that rare variants contribute to associations derived by the model. Similarly, do the authors find evidence that allelic heterogeneity is leveraged by the HFS-based association model? It would be useful to do simulations here to characterize the model behavior in the presence of trait-associated rare variants.

3. Sei predicts chromatin status / ChIP-seq peaks in the center of a 4kb region. It would therefore be more relevant to predict HFS using overlapping sequence windows that tile the genome as opposed to using non-overlapping windows for computing HFS scores. Specifically, in line 482, the authors state that "the HFS score represents overall activity of the entire sequence, not only the few bp at the center", but this would not hold given that Sei is predicting activity at the center for any sequence.

4. Is the HFS-based association going to miss coding variation and several regulatory variants such as splicing variants? There are also going to be cases where there's an association driven by a variant that is correlated with a Sei prediction in a neighboring window. These would represent false positives for the method, it would be useful to identify or characterize these cases.

Additional minor concerns:<br /> 1. It's not clear whether SuSie-based finemapping is appropriate at the locus level, when there is limited LD between neighboring HFS bins. How does the choice of the number of causal loci and the size of the segment being finemapped affect the results and is SuSie a good fit in this scenario?

2. It is not clear how a single score is chosen from the 117 values predicted by Sei for each locus. SuSie is run assuming a single causal signal per locus, an assumption which may not hold at ~4kb resolution (several classes could be associated with the trait of interest). It's not clear whether SuSie, run in this parameter setting, is a good choice for variable selection here.

3.. A single HFS score is being chosen from amongst multiple tracks at each locus independently. Does this require additional multiple-hypothesis correction?

4. The results show that a larger number of loci are identified with HFS-based finemapping & that causal loci are enriched for causal SNPs. However, it is not clear how the number of causal loci should relate to the number of SNPs. It would be really nice to see examples of cases where a previously unresolved association is resolved when using HFS-based GWAS + finemapping.

5. Sequence-based deep learning model predictions can be miscalibrated for insertions and deletions (INDELs) as compared to SNPs. Scaling INDEL predictions would likely improve the downstream modeling.

137 arrests is a high number, and a lot of paperwork, but overturning vehicles is something that doesn't always happen.

Author Response

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

eLife assessment

This study reports important findings regarding the systemic function of hemocytes controlling whole-body responses to oxidative stress. The evidence in support of the requirement for hemocytes in oxidative stress responses as well as the hemocyte single-nuclei analyses in the presence or absence of oxidative stress are convincing. In contrast, the genetic and physiological analyses that link the non-canonical DDR pathway to upd3/JNK expression and high susceptibility, and the inferences regarding the function of hemocytes in systemic metabolic control are incomplete and would benefit from more rigorous approaches. The work will be of interest to cell and developmental biologists working on animal metabolism, immunity, or stress responses.

We would like to thank the editorial team for these positive comments on our manuscript and the constructive suggestions to improve our manuscript. We are now happy to send you our revised manuscript, which we improved according to the suggestions and valuable comments of the referees.

Public Reviews:

Reviewer #1 (Public Review):

The study examines how hemocytes control whole-body responses to oxidative stress. Using single cell sequencing they identify several transcriptionally distinct populations of hemocytes, including one subset that show altered immune and stress gene expression. They also find that knockdown of DNA Damage Response (DDR) genes in hemocytes increases expression of the immune cytokine, upd3, and that both upd3 overexpression in hemocytes and hemocyte knockdown of DDR genes leads to increased lethality upon oxidative stress.

Strengths

1. The single cell analyses provide a clear description of how oxidative stress can cause distinct transcriptional changes in different populations of hemocytes. These results add to the emerging them in the field that there functionally different subpopulations of hemocytes that can control organismal responses to stress.

2. The discovery that DDR genes are required upon oxidative stress to limit cytokine production and lethality provides interesting new insight into the DDR may play non-canonical roles in controlling organismal responses to stress.

We are grateful to referee 1 to point out the importance and novelty of our snRNA-seq data and our findings on the role of DNA damage-modulated cytokine release by hemocytes during oxidative stress. We further extended these analyses in the revised manuscript by looking deeper into the transcriptomic alterations in fat body cells upon oxidative stress (Figure 4, Figure S4). We further provide additional data to support the connection of DNA damage signaling and regulation of upd3 release from hemocytes (Figure 6F). Here we show that upd3-deficiency can abrogate the increased susceptibility of flies with mei41 and tefu knockdown in hemocytes. In line with this finding, we also show that upd3null mutants show a reduced but not abolished susceptibility to oxidative stress overall (Figure 6F), underlining the role of upd3 as a mediator of oxidative stress response.

Weaknesses

1. In some ways the authors interpretation of the data - as indicated, for example, in the title, summary and model figure - don't quite match their data. From the title and model figure, it seems that the authors suggest that the DDR pathway induces JNK and Upd3 and that the upd3 leads to tissue wasting. However, the data suggest that the DDR actually limits upd3 production and susceptibility to death as suggested by several results:

According to the referee’s suggestion, we revised the manuscript and adjusted our title, abstract and graphical summary to be more precise that DNA damage signaling seem to have a modulatory or regulatory effect on upd3 release. Furthermore, we provide now additional data to support the connection between DNA damage signaling and upd3 release. For example, we added several genetic “rescue” experiments to strengthen the epistasis that modulation of DNA damage signaling and the higher susceptibility of the fly is connected to altered upd3 levels (Figure 6F). We now provide additional data showing that the loss of upd3 rescues the susceptibility to oxidative stress in flies, which are deficient for DDR components in hemocytes.

a. PQ normally doesn't induce upd3 but does lead to glycogen and TAG loss, suggesting that upd3 isn't connected to the PQ-induced wasting.

Even though in our systemic gene expression analysis of upd3 expression, we could not detect a significant induction of upd3 upon PQ feeding. However, we found upd3 expression within our snRNAseq data in a distinct cluster of immune-activated hemocytes (Figure 3B, Cluster 6). Upon knockdown of the DNA damage signaling in hemocytes, the levels then increase to a detectable level in the whole fly. This supports our assumption that upd3 is needed upon oxidative stress to induce energy mobilization from the fat body, but needs to be tightly controlled to balance tissue wasting for energy mobilization. Furthermore, we found evidence in our new analysis of the snRNA-seq data of the fat body cells, that indeed we can find Jak/STAT activation in one cell cluster here, which could speak for an interaction of Cluster 6 hemocytes with cluster 6 fat body cells. A hypothesis we aim to explore in future studies.

b. knockdown of DDR upregulates upd3 and leads to increased PQ-induced death. This would suggest that activation of DDR is normally required to limit, rather than serve as the trigger for upd3 production and death.

Our data support the hypothesis that DDR signaling in hemocytes “modulates” upd3 levels upon oxidative stress. We now carefully revised the text and the graphical summary of the manuscript to emphasize that oxidative stress causes DNA damage, which subsequently induces the DNA damage signaling machinery. If this machinery is not sufficiently induced, for example by knockdown of tefu and mei-41, non-canonical DNA damage signaling is altered which induces JNK signaling and induces release of pro-inflammatory cytokines, including upd3. Whereas DNA damage itself is only slightly increase in the used DDR deficient lines (Figure 5C) and hemocytes do not undergo apoptosis (unaltered cell number on PQ (Figure 5B)), we conclude that loss of tefu, mei-41, or nbs1 causes dysregulation of inflammatory signaling cascades via non-canonical DNA damage signaling. However, oxidative stress itself seems to also induce upd3 release and DNA damage signaling in the same cell cluster, as shown by our snRNA-seq data (Figure 3B). Hence, we think that DNA damage signaling is needed as a rate-limiting step for upd3 release.

c. hemocyte knockdown of either JNK activity or upd3 doesn't affect PQ-induced death, suggesting that they don't contribute to oxidative stress-induced death. It’s only when DDR is impaired (with DDR gene knockdown) that an increase in upd3 is seen (although no experiments addressed whether JNK was activated or involved in this induction of upd3), suggesting that DDR activation prevents upd3 induction upon oxidative stress.

Whereas the double knockdown of upd3 or bsk and DDR genes was resulting in insufficient knockdown efficiencies, we added a rescue experiment where we combined upd3null mutants with knockdown of tefu and mei-41 in hemocytes and found a reduced susceptibility of DDR-deficient flies to oxidative stress.

1. The connections between DDR, JNK and upd3 aren't fully developed. The experiments show that susceptibility to oxidative stress-induced death can be caused by a) knockdown of DDR genes, b) genetic overexpression of upd3, c) genetic activation of JNK. But whether these effects are all related and reflect a linear pathway requires a little more work. For example, one prediction of the proposed model is that the increased susceptibility to oxidative stress-induced death in the hemocyte DDR gene knockdowns would be suppressed (perhaps partially) by simultaneous knockdown of upd3 and/or JNK. These types of epistasis experiments would strengthen the model and the paper.

As mentioned before, we had some technical difficulties combining the knockdown of bsk or upd3 with DDR genes. However, we added a new experiment in which we show that upd3null mutation can rescue the higher susceptibility of hemocytes with tefu and mei41 knockdown.

1. The (potential) connections between DDR/JNK/UPD3 and the oxidative stress effects on depletion of nutrient (lipids and glycogen) stores was also not fully developed. However, it may be the case that, in this paper, the authors just want to speculate that the effects of hemocyte DDR/upd3 manipulation on viability upon oxidative stress involve changes in nutrient stores.

In the revised version of the manuscript, we now provide a more thorough snRNA-seq analysis in the fat body upon PQ treatment to give more insights on the changes in the fat body upon PQ treatment. We added additional histological images of the abdominal fat body on control food and PQ food, to demonstrate the elimination of triglycerides from fat body with Oil-Red-O staining (Figure S1). We also analyzed now hemocyte-deficient (crq-Gal80ts>reaper) flies for their levels of triglycerides and carbohydrates during oxidative stress, to support our hypothesis that hemocytes are key players in the regulation of energy mobilization during oxidative stress. Loss of hemocytes (and therefore also their regulatory input on energy mobilization from the fat body) results in increased triglyceride storage in the fat body during steady state with a decreased consumption of these triglycerides on PQ food compared to control flies (Figure 1J). In contrast, glycogen storage and mobilization, which is mostly done in muscle, is not altered in these flies during oxidative stress (Figure 1L). Interestingly, free glucose levels are drastically reduced in hemocyte-deficient flies, which could be due to insufficient energy mobilization from the fat body and subsequently results in a higher susceptibility of these flies on oxidative stress (Figure 1K). Additionally, we aim to point out here that “functional” hemocytes are needed for effective response to oxidative stress, but this response has to be tightly balanced (see also new graphical abstract).

Reviewer #2 (Public Review):

Hersperger et al. investigated the importance of Drosophila immune cells, called hemocytes, in the response to oxidative stress in adult flies. They found that hemocytes are essential in this response, and using state-of-the-art single-cell transcriptomics, they identified expression changes at the level of individual hemocytes. This allowed them to cluster hemocytes into subgroups with different responses, which certainly represents very valuable work. One of the clusters appears to respond directly to oxidative stress and shows a very specific expression response that could be related to the observed systemic metabolic changes and energy mobilization. However, the association of these transcriptional changes in hemocytes with metabolic changes is not well established in this work. Using hemocyte-specific genetic manipulation, the authors convincingly show that the DNA damage response in hemocytes regulates JNK activity and subsequent expression of the JAK/STAT ligand Upd3. Silencing of the DNA damage response or excessive activation of JNK and Upd3 leads to increased susceptibility to oxidative stress. This nicely demonstrates the importance of tight control of JNK-Upd3 signaling in hemocytes during oxidative stress. However, it would have been nice to show here a link to systemic metabolic changes, as the authors conclude that it is tissue wasting caused by excessive Upd3 activation that leads to increased susceptibility, but metabolic changes were not analyzed in the manipulated flies.

We thank the referee for the suggestion to better connect upd3 cytokine levels to energy mobilization from the fat body. We agree that this is an important point to support our hypothesis. First, we added now a detailed analysis of fat body cells in our snRNA-seq data to evaluate the changes induced in the fat body upon oxidative stress. We further added additional metabolic analyses of hemocyte-deficient flies (crq-Gal80ts>reaper) to support our hypothesis that hemocytes are key players in the regulation of energy mobilization during oxidative stress (see also answer to referee 1). Loss of the regulatory role of hemocytes in the energy mobilization and redistribution leads to a decreased consumption of these triglycerides on PQ food compared to control flies (Figure 1J). In contrast, glycogen storage and mobilization from muscle, is not affected in hemocyte-deficient flies during oxidative stress (Figure 1L). Interestingly, free glucose levels are drastically reduced in hemocyte-deficient flies compared to controls, which could be due to insufficient energy mobilization from the fat body resulting in a higher susceptibility to oxidative stress (Figure 1K). This data supports our assumption that “functional” hemocytes are needed for effective response to oxidative stress, but this response has to be tightly balanced (see also new graphical summary).

The overall conclusion of this work, as presented by the authors, is that Upd3 expression in hemocytes under oxidative stress leads to tissue wasting, whereas in fact it has been shown that excessive hemocyte-specific Upd3 activation leads to increased susceptibility to oxidative stress (whether due to increased tissue wasting remains a question). The DNA damage response ensures tight control of JNK-Upd3, which is important. However, what role naturally occurring Upd3 expression plays in a single hemocyte cluster during oxidative stress has not been tested. What if the energy mobilization induced by this naturally occurring Upd3 expression during oxidative stress is actually beneficial, as the authors themselves state in the abstract - for potential tissue repair? It would have been useful to clarify in the manuscript that the observed pathological effects are due to overactivation of Upd3 (an important finding), but this does not necessarily mean that the observed expression of Upd3 in one cluster of hemocytes causes the pathology.

We agree with the referee that the pathological effects and increased susceptibility to oxidative stress are mediated by over-activated hemocytes and enhanced cytokine release, including upd3 during oxidative stress. We edited the revised manuscript accordingly to imply a “regulatory” role of upd3, which we suspect and suggest as an important mediator for inter-organ communication between hemocytes and fat body. Whereas our used model for oxidative stress (15mM Paraquat feeding) is a severe insult from which most of the flies will not recover, we could not account and test how upd3 might influence tissue repair after injury, insults and infection. We believe that this is an important factor, we aim to explore in future studies.

Reviewer #3 (Public Review):

In this study, Kierdorf and colleagues investigated the function of hemocytes in oxidative stress response and found that non-canonical DNA damage response (DDR) is critical for controlling JNK activity and the expression of cytokine unpaired3. Hemocyte-mediated expression of upd3 and JNK determines the susceptibility to oxidative stress and systemic energy metabolism required for animal survival, suggesting a new role for hemocytes in the direct mediation of stress response and animal survival.

Strength of the study:

1. This study demonstrates the role of hemocytes in oxidative stress response in adults and provides novel insights into hemocytes in systemic stress response and animal homeostasis.

2. The single-cell transcriptome profiling of adult hemocytes during Paraquat treatment, compared to controls, would be of broad interest to scientists in the field.

We are grateful to these positive comments on our data and are excited that the referee pointed out the importance of our provided snRNA-seq analysis of hemocytes and other cell types during oxidative stress. In the revised, version we now extended this analysis and looked not only into hemocytes but also highlighted induced changes in the fat body (Figure 4).

Weakness of the study:

1. The authors claim that the non-canonical DNA damage response mechanism in hemocytes controls the susceptibility of animals through JNK and upd3 expression. However, the link between DDR-JNK/upd3 in oxidative stress response is incomplete and some of the descriptions do not match their data.

In the revised manuscript, we aimed to strengthen the weaknesses pointed out by the referee. We now included additional genetic crosses to validate the connection of DDR signaling in hemocytes with upd3 release. For example, we added now survival studies where we show that upd3null mutation can rescue the higher susceptibility of flies with tefu and mei41 knockdown in hemocytes during oxidative stress. Furthermore, we added additional data to highlight the importance of hemocytes themselves as essential regulators of susceptibility to oxidative stress. We analyzed the hemocyte-deficient flies (crq-Gal80ts>reaper) for their triglyceride content and carbohydrate levels during oxidative stress (Figure 1 I-L). As outlined above, loss of hemocytes leads to a decreased consumption of these triglycerides on PQ food compared to control flies (Figure 1J). In contrast, glycogen storage and mobilization from muscle, is not affected in hemocyte-deficient flies during oxidative stress (Figure 1L). Interestingly, free glucose levels are drastically reduced in hemocyte-deficient flies, which could be due to insufficient energy mobilization from the fat body resulting in a higher susceptibility to oxidative stress (Figure 1K).

1. The schematic diagram does not accurately represent the authors' findings and requires further modifications.

We carefully revised the text throughout the manuscript describing our results and edited the graphical abstract to display that upd3 levels and hemocytes are essential to balance and modulate response to oxidative stress.

Reviewer #1 (Recommendations For The Authors):

The summary doesn't say too much about what the specific discoveries and results of the study are. The description is limited to just one sentence saying, "Here we describe the responses of hemocytes in adult Drosophila to oxidative stress and the essential role of non-canonical DNA damage repair activity in direct "responder" hemocytes to control JNK-mediated stress signaling, systemic levels of the cytokine upd3 and subsequently susceptibility to oxidative stress" which doesn't provide sufficient explanation of what the results were.

In the revised version of our manuscript, we now provide further information for the reader to outline the findings of our study in a concise way in the summary.

Reviewer #2 (Recommendations For The Authors):

1. To strengthen the conclusion that the DDR response suppresses JNK, and thus Upd3, rescue of DDR by upd3 null mutation would help (knockdown by Hml>upd3IR might not work, RNAi seems problematic).

We would like to thank the referee for this suggestion and included now a genetic experiment where we combined upd3null mutants with hemocyte-specific knockdown of mei-41 and tefu to test their susceptibility to oxidative stress. Our data indeed provide evidence that loss of upd3 rescues the higher susceptibility of flies with hemocyte-specific knockdown for tefu and mei-41 (Figure 6F). Furthermore, we see that upd3null mutants show a diminished susceptibility to oxidative stress compared to control flies (Figure 6F).

1. To link the observed effects to systemic metabolic changes, it would be useful to measure glycogen and triglycerides in these flies as well:

2. crq-Gal80ts>reaper to see what role hemocytes play in the observed metabolic changes.

3. Hml-Upd3 overexpression and Upd3 null mutant (Upd3 RNAi seems to be problematic, we have similar experiences) to see if Upd3 overexpression leads to even more profound changes as suggested, and if Upd3 mutation at least partially suppresses the observed changes.

We agree with the referee that analyzing the connection of hemocyte activation to metabolic changes should be demonstrated in our manuscript to support our claim that hemocytes are important regulators of energy mobilization during oxidative stress. Hence, we analyzed triglycerides and carbohydrate levels in hemocyte-deficient flies (crq-Gal80ts>reaper) during oxidative stress. Indeed, we found substantial differences in energy mobilization in these flies supporting the assumption that the higher susceptibility of hemocyte-deficient flies could be caused by substantial decrease in free glucose and inefficient lysis of triglycerides from the fat body (Figure 1I-K).

1. To test whether the cause of the increased susceptibility to oxidative stress is due to Upd3 overactivation induced by DDR silencing, the authors should attempt to rescue DDR silencing with an Upd3 null mutation.

The suggestion of the reviewer was included in the revised manuscript and as outlined above we now added this data set to our manuscript (Figure 6F). Indeed, we can now provide evidence that upd3null mutation rescues the higher susceptibility of flies with DDR knockdown in hemocytes.

1. Lethality after PQ treatment varies widely (sometimes from 10 to 90%! as in Figure 5D) - is this normal? In some experiments the variability was much lower. In particular, Figure 5D is very problematic and for example the result with upd3 null mutant compared to control is not very convincing. This could be an important result to test whether Upd3, with normal expression likely coming from cluster 6, actually plays a beneficial role, whereas overexpression with Hml leads to pathology.

We agree with the referee that it would be more convincing if the variation cross of survival experiments would be less. However, we included a lot of flies and vials in many individual experiments to test our hypothesis and variation in these survivals was always the case. These effects can be caused by many factors for example the amount of food intake by the flies, genetic background or inserted transgenes. The n-number is quite high across our survivals; so that we are convinced, the seen effects are valid. This reflects also the power of using Drosophila melanogaster as a model organism for such survivals. The high n-number in our data falls into a normal Gauss distribution with a distinct mean susceptibility between the genotypes analyzed.

1. I like the conclusion at the end of the results: line 413: "We show that this oxidative stressmediated immune activation seems to be controlled by non-canonical DNA damage signaling resulting in JNK activation and subsequent upd3 expression, which can render the adult fly more susceptible to oxidative stress when it is over-activated." This is actually a more appropriate conclusion, but in the summary, introduction and discussion along with the overall schematic illustration, this is not actually stated as such, but rather as Upd3 released from cluster 6 causes the pathology. For example: line 435 "Hence, we postulate that hemocyte-derived upd3, most likely released by the activated plasmatocyte cluster C6 during oxidative stress in vivo and subsequently controlling energy mobilization and subsequent tissue wasting upon oxidative stress."

We thank the referee for this suggestion and edited our manuscript and conclusions accordingly.

Reviewer #3 (Recommendations For The Authors):

1. In Figure 2, the authors claim showed that PQ treatment changes the hemocyte clusters in a way that suppresses the conventional Hml+ or Pxn+ hemocytes (cluster1) while expanding hemocyte clusters enriched with metabolic genes such as Lpin, bmm etc. It is not clear whether these cells are comparable to the fat body and if these clusters express any of previously known hemocyte marker genes to claim that these are bona fide hemocytes.

We now included a new analysis of our snRNA-seq data in Figure S4, where we clearly show that all identified hemocyte clusters do not have a fat body signature and are hemocytes, which seem to undergo metabolic adaptations (Figure S4A). Furthermore, we show that the identified fat body cells have a clear fat body signature (Figure S4B) and do not express specific hemocyte markers (Figure S4C).

1. In Figure 4C, the authors showed that comet assays of isolated hemocytes result in a statistically significant increase in DNA damage in DDR-deficient flies before and after PQ treatment. However, the authors conclude that, in lines 324-328, the higher susceptibility of DDR-deficient flies is not due to an increase in DNA damage. To explicitly conclude that "non-canonical" DNA damage response, without any DNA damage, is specifically upregulated during PQ treatment, the authors require further support to exclude the potential activation of canonical DDR.

The referee is correct that we do not provide direct evidence for non-canonical DNA damage signaling. Therefore, we also decided to tune down our statement here a bit and removed that claim from the title. Increase in DNA damage can of course also increase the non-canonical DNA damage signaling pathway, loss of DNA damage signaling genes such as tefu and mei-41 seem to only have minor impacts on the overall amount of DNA damage acquired in hemocytes by oxidative stress. We therefore concluded that the induction in immune activation is most unlikely only caused by increased DNA damage but might be connected to dysregulation in non-canonical DNA damage signaling. Canonical DNA damage signaling leads essentially to DDR, which could be slow in adult hemocytes because they post-mitotic, or to apoptosis, which we could not observe in the analyzed time window in our experiments. Hemocyte number remained stable over the 24h PQ treatment without reduction in cell number (Figure 1H).

1. From Figure 4D-F, the authors showed that loss of DDR in hemocytes induces the expression of unpaired 2 and 3, Socs36E, which represent the JAK/STAT pathway, and thor, InR, Pepck in the InR pathway, and a JNK readout, puc. These results indicate that the DDR pathway normally inhibits the upd-mediated JAK/STAT activation upon PQ treatment, compared to wild-type animals during PQ treatment in Figure 1B-C, which in turn protects the animal during oxidative stress responses. However, the authors claim that "enhanced DNA damage boosts immune activation and therefore susceptibility to oxidative stress (lines 365-366); we show that this oxidative stress-mediated immune activation seems to be controlled by non-canonical DNA damage signaling resulting in JNK activation and subsequent upd3 expression (line 413-416)". These conclusions are not compatible with the authors' data and may require additional data to support or can be modified.

In the revised manuscript, we carefully revised now the text and our statements that it seems that DNA damage signaling in hemocytes has regulatory or modulatory effect on the immune response during oxidative stress. Accordingly, we also adjusted our graphical summary. We agree with the referee and used the term “non-canonical” DNA damage signaling more carefully throughout the manuscript. The slight increase in DNA damage seen after PQ treatment can contribute to immune activation but seems to be not correlative to the induced cytokine levels or the susceptibility of the flies to oxidative stress.

1. In Fig 1I, the authors showed that genetic ablation of hemocytes using UAS-repear induces susceptibility to PQ treatment. It is possible that inducing cell death in hemocytes itself causes the expression of cytokine upd3 or activates the JNK pathway to enhance the basal level of upd3/JNK even without PQ treatment. If this phenotype is solely mediated by the loss of hemocytes, the results should be repeated by reducing the number of hemocytes with alternative genetic backgrounds.

In the different genotypes analyzed across our manuscript we did not detect cell death of hemocytes or a dramatic reduction in hemocytes number (see Figure 1H, Figure 5B, Figure 6C). The higher susceptibility if hemocyte-deficient flies during oxidative stress is most likely caused by the loss of their regulatory role during energy mobilization. We tested triglyceride levels in hemocyte-deficient flies and found a decreased triglyceride consumption (lipolysis), with reduced levels of circulating glucose levels. This findings support our hypothesis that hemocytes are needed to balance the response to oxidative stress. In contrast, the flies with DDR-deficient hemocytes show higher systemic cytokine levels, which most likely enhance energy mobilization from the fat body and therefore result in a higher susceptibility of the fly to oxidative stress. Hence, we claim that hemocytes and their regulation of systemic cytokine levels are important to balance the response to oxidative stress and guarantee the survival of the organism.

1. Lethality of control animals in PQ treatment is variable and it is hard to estimate the effect of animal susceptibility during 15mM PQ feeding. For example, Fig1A shows that control animals exhibit ~10% death during 15mM PQ which is further enhanced by crq-Gal80>reaper expression to 40% (Fig 1I). However, in Fig 5D-E, the basal lethality of wild-type controls already reaches 40~50%, which makes them hard to compare with other genetic manipulations. Related to this, the authors demonstrated that the expression of upd3 in hemocytes is sufficient to aggravate animal survival upon PQ treatment; however, upd3 null mutants do not rescue the lethality, which indicates that upd3 is not required for hampering animal mortality. These data need to be revisited and analyzed.

As outlined above, we find the variability of susceptibility to oxidative stress across all of our experiments. This could be due to different effects such as food intake but also transgene insertion and genetic background. Crq-gal80ts>reaper flies are healthy, but show a shortened life span on normal food (Kierdorf et al., 2020) due to enhanced loss of proteostasis in muscles. We show in the revised manuscript that these flies have a higher susceptibility to oxidative stress and that this effect could be mediated by defects in energy mobilization and redistribution as shown by less triglyceride lysis from the fat body and decreasing levels in free glucose. This would explain the high mortality rate of these flies at 7 days after eclosion. Paraquat treatment (15mM) is a severe inducer of oxidative stress, which results in death of most flies when they are maintained for longer time windows on PQ food. Hence, it is a model, which is not suitable to examine and monitor recovery from this detrimental insult. upd3null mutants were extensively reexamined in this manuscript, and even though we could not see a full protection of these flies from oxidative stress induced death, we found a reduced susceptibility compared to control flies (Figure 6F). Furthermore, when we combined upd3null mutants with flies deficient for tefu and mei-41 in hemocytes, the increased susceptibility to oxidative stress was rescued.

Author Response

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

We appreciate the reviewers’ detailed corrections and insightful comments. We have revised our manuscript per reviewers’ recommendations by including new data and clarifications/expansion of the discussion on our findings. Please see below for details.

Reviewer #1 (Recommendations For The Authors):

1. The introduction notes that CD1d KO mice show reduced levels of Va3.2 T cells (Ruscher et al.), which is interesting because innate memory T cell development in the thymus often requires IL-4 production by NKT cells. Have the authors explored QFL T cells in CD1d KO and/or IL-4 KO mice? Since their QFL TCR Tg mice still develop QFL T cells (and these animals likely have very few thymic NKT cells), NKT cells may not be required for the intrathymic development of QFL T cells?

Answer: We agree that investigation on the role of NKT cells or IL-4 in QFL T cell development will greatly further our understanding of these cells.

We validated the finding that expression of the QFL TCR transgene largely repressed the expression of endogenous TCRα, as indicated by the low levels of endogenous Vα2 on mature CD8SP T cells in both thymus and spleen. However, the frequencies of Vα2 usage in CD4 SP thymocytes and splenocytes from QFL transgenic mice were similar to non-transgenic mice, confirming that they underwent positive selection using endogenous TCR rather than the QFL TCR. We thus do not exclude the possible presence of NKT cells in QFLTg mouse and their potential involvement in the QFL T cells development. Our manuscript here is mainly focused on investigating the peripheral phenotype of QFL T cells and their association with the gut microbiota environment. Investigations into the role of CD1d/IL-4 will be best addressed in our future studies.

1. The finding that Qa-1 expression is not required for the development of QFL T cells raises questions about other MHC products that may be involved. In this context, it is interesting that TAP-deficient mice develop few QFL T cells, for reasons that are unclear, but the authors may speculate a bit. In this context, it may be helpful for the authors to note whether TAP is required for QFL presentation to QFL T cells. Since Qa-1 is not required, and CD1d is still expressed in TAP KO mice, what then could be responsible for their defect in QFL T cell development?

Answer: This is a great point. Figure 2 (from (Valerio et al., 2023) on the development of QFL T cells) tested whether QFL TCR cross-react with other MHC I molecules.

We assessed the activation of pre-selection QFLTg thymocytes in response to various MHC I deficient DC2.4 cell lines. While the QFL thymocytes showed partially reduced activation when stimulated with Qa-1b deficient APCs, triple knock-out (KO) of Qa-1b, Kb, and Db in DC2.4 cells reduced activation close to background levels. However, double knock-out of Qa-1b with either Kb, or Db led to stimulation that was intermediate between the triple KO and Qa-1b-KO cell lines. These data suggest that Kb and Db may contribute to the positive selection of QFL T cells in Qa-1b-KO mice.

TAP is required for FL9 peptide presentation and is very likely needed for presentation of the yet unidentified MHC Ia presented peptide(s) that are essential to QFL T positive selection. While CD1d/NKT cells/IL-4 may be involved in supporting the maturation of QFL T cells, we think in the TAP-KO mice the absence of TAP led to deletion/altered selection of the QFL T population at early developmental stage. We have added clarification on this point in the revised manuscript (line 412~418).

1. It may be worthwhile for the authors to note that Qa-1 was also dispensable for the intrathymic selection of another Qa-1-restricted TCR (Doorduijn et al. 2018. Frontiers Immunol.), although this is presumably not the case for others (Sullivan et al. 2002. Immunity 17, 95).

Answer: We appreciate this recommendation. We have noted this point in the resubmitted manuscript (line 412~418).

1. Lines 122-124: The sentence "Interesting ..." seemed confusing to me; are the numbers (60 and 30%) correct?

Answer: The numbers 60% and 30% were referring to the largest number we have detected for percentages of Va3.2 QFL T cells and Va3.2 CD8 T cell respectively. Here in the revised version, we replaced these numbers with average percentages (20.1% and <10%) to avoid confusion (line 134).

1. Qa-1/peptide complexes may also be recognized by CD94/NKG2 receptors, which may complicate the interpretation of the data (e.g., staining of the dextramers). From their previous work, it appears that Qa-1/QFL does not bind CD94/NKG2, which would be helpful to note in the text.

Answer: We have noted this point in the revised manuscript (line 117~121).

1. It would be helpful to add a few comments about the potential relevance to HLA-E.

Answer: We have included discussion on this point (line 391~401).

1. Figure legends: Most legends note the total number of replicates, which is usually quite high. It would also be helpful to indicate the total number of independent experiments performed and, when relevant, that the data are pooled from multiple independent experiments.

Answer: Thank you for raising the concern. We have clarified the experimental repeats in figure legends.

Reviewer #2 (Recommendations For The Authors):

1. The work of Nilabh Shastri was the foundation of the present study. Unfortunately, he passed away in 2021. Since he can no longer assume the responsibilities of a senior author, I wonder if it would be more appropriate to dedicate this paper to him than to list him as a co-author.

Answer: We have removed Dr. Shastri’s name as a co-senior author and have dedicated this work to his memory.

1. The official symbol for ERAAP is Erap1.

Answer: We have replaced ERAAP with ERAP1.

1. Please refrain from editorializing. For example, "strikingly" appears eight times and "interestingly" 9 times in the manuscript. Most readers believe they do not need to be said when something is striking or interesting.

Answer: We appreciate the Reviewer’s suggestion and have removed ‘strikingly’ and ‘interestingly’ from the manuscript.

1. In WT mice, are there some cell types that express Qa-1b but not Erap1 and could therefore present the FL9 peptide?

Answer: This is a great question. Using our highly sensitive QFL T cell hybridoma line BEko8Z (sensitivity shown in Fig. 6b), we have so far not been able to detect steady-state FL9 presentation by cells isolated from the spleen, lymph nodes, various gut associated lymphoid tissues or intestinal epithelial cells (Supplementary Fig. 8 a left panel). However, we do not exclude the possibility of FL9 peptide being transiently presented under certain conditions (i.e. ER stress/transformed cells) at particular locations or within certain time windows, which is of great importance for understanding the function of these cells but is beyond the scope of this study.

1. Since you have not tested substitutions at other positions, could you explain your reasoning that P4 and P6 are the critical residues (lines 271-272)?

Answer: Thank you for raising the concern. We have expanded on explanation of our strategy for determining peptide homology (line 272~313) in the revised manuscript. We have also included data on the structure the QFL TCR: FL9-Qa-1b complex predicted by Alphafold2, conformation alignment of FL9 and Qdm (Figure 6. a, b) and the NetMHCpan prediction of Qa1b binding of Qdm, FL9 and various FL9 mutant peptides (Supplementary Fig. 8 c) to help readers visualize the reasoning behind our strategy.

1. Readers might appreciate having a Figure summarizing the differences between spleen and gut QFL T cells.

Answer: This is a great suggestion. We have added a table summarizing the characteristic features of the splenic and IEL QFL T cells (Table 1).

1. In the discussion, readers would like to know what plan you might have to elucidate the function of QFL T cells.

Answer: We appreciate the recommendation. We have elaborated on our opinions and future directions in the resubmitted manuscript (line 393~401, 446~455).  

Reviewer #3 (Public Review):

1. For most of the report, the authors use a set of phenotypic traits to highlight the unique features of QFL-specific CD8+ T cells - specifically, CD44high, CD8aa+ve, CD8ab-ve. In Supp. Fig. 4, however, completely distinct phenotypic characteristics are presented, indicating that IEL QFL-specific T cells are CD5low, Thy-1low. No explanation is provided in the text about whether this is a previously reported phenotype, whether any elements of this phenotype are shared with splenic QFL T cells, what significance the authors ascribe to this phenotype (and to the fact that Qa1-deficiency leads to a more conventional Thy-1+ve, CD5+ve phenotype), and whether this altered phenotype is also seen in ERAAP-deficient mice. At least some explanation for this abrupt shift in focus and integration with prior published work is needed. On a related note, CD5 expression is measured in splenic QFL-specific CD8+ T cells from GF vs SPF mice (Supp. Fig. 9), to indicate that there is no phenotypic impact in the GF mice - but from Supp. Fig. 4, it would seem more appropriate to report CD5 expression in QFL-specific cells from the IEL, not the spleen.

Answer: Expression of CD8αα and lack of CD4, CD8αβ, CD5 and CD90 expression was indeed reported as the characteristic phenotype of natIELs. We have clarified this point in the resubmitted manuscript (line 80). The CD8αα+ IEL QFL T cells have consistently showed CD5CD90- phenotype. While CD8αα expression was sufficient to describe their natIEL phenotype, we showed the CD5-CD90- data in Supplementary figures only to provide additional evidence.

The CD5 molecule by itself reflects the TCR signaling strength and high CD5 level is associated with self-reactivity of T cells (Azzam et al., 2001; Fulton et al., 2015). The implication of CD5 expression on QFLTg cells is discussed in our other manuscript where we investigate the development of these cells (Valerio et al., 2023). In Supplementary Fig. 9, because the donor splenic QFLTg cell have consistently showed comparable CD5 level between the GF and SPF group, we reasoned that it would not interfere with our interpretation of the CD44 expression.

1. The authors suggest the finding that QFL-specific cells from ERAAP-deficient mice have a more "conventional" phenotype indicates some form of negative selection of high-affinity clones (this result being somewhat unexpected since ERAAP loss was previously shown to increase the presentation of Qa-1b loaded with FL9, confirmed in this report). It is not clear how this argument aligns with the data presented, however, since the authors convincingly show no significant reduction in the number of QFL-specific cells in ERAAP-knockout mice (Fig. 3a), and their own data (e.g. Fig. 2a) do not suggest that CD44 expression correlates with QFL-multimer staining (as a surrogate for TCR affinity/avidity). Is there some experimental basis for suggesting that ERAAP-deficient lacks a subset of high affinity QFL-specific cells?

Answer: We think the presence of QFL T cells in ERAAP-KO mice is a result of the unconventional developmental mechanism of these cells which is better addressed in our complementary manuscript on the development of QFL T cells(Valerio et al., 2023). Valerio et al. found that the most predominant QFL T clone which expresses Vα3.2Jα21, Vβ1Dβ1Jβ2-7 received relatively strong TCR signaling and underwent agonist selection during thymic development, indicating that the QFL ligand is involved in selection of the innate-like QFL T population.

We agree that there is so far no direct evidence showing the QFL T cells that were absent in the ERAAP-KO mice were high-affinity clones. We have removed ‘high-affinity’ from the manuscript (line 180). While CD44 expression has been associated the antigen-experiences phenotype of T cells, it is yet unclear whether expression level of this molecule directly reflects TCR affinity/avidity. identification of clones of different affinities/avidities require high precision technologies that are not currently available to the research community. While we do have zMovi, a newly developed (developing) technology, in the lab claimed to measure relative avidity/affinity of different cell types for ligands, during the past two years working with this instrument has taught us that the technology is not yet advanced enough; it can only produce reliable data on extreme differences of single clones, i.e., high numbers of homogeneous cell types expressing very high affinity receptors.

1. The rationale for designing FL9 mutants, and for using these data to screen the proteomes of various commensal bacteria needs further explanation. The authors propose P4 and P6 of FL9 are likely to be "critical" but do not explain whether they predict these to be TCR or Qa-1b contact sites. Published data (e.g., PMID: 10974028) suggest that multiple residues contribute to Qa-1b binding, so while the authors find that P4A completely lost the ability to stimulate a QFL-specific hybridoma, it is unclear whether this is due to the loss of a TCR- or a Qa-1-contact site (or, possibly, both). This could easily be tested - e.g., by determining whether P4A can act as a competitive inhibitor for FL9-induced stimulation of BEko8Z (and, ideally, other Qa-1b-restricted cells, specific for distinct peptides). Without such information, it is unclear exactly what is being selected in the authors' screening strategy of commensal bacterial proteomes. This, of course, does not lessen the importance of finding the peptide from P. pentosaceus that can (albeit weakly) stimulate QFL-specific cells, and the finding that association with this microbe can sustain IEL QFL cells.

Answer: Thank you for raising the concern. We have expanded on explanation of our strategy for determining peptide homology (line 272~313) in the revised manuscript. We have also included data on the structure the QFL TCR: FL9-Qa-1b complex predicted by Alphafold2, conformation alignment of FL9 and Qdm (Figure 6. a, b) and the NetMHCpan prediction of Qa1b binding of Qdm, FL9 and various FL9 mutant peptides (Supplementary Fig. 8 c) to help readers visualize the reasoning behind our strategy.

References

Azzam, H.S., DeJarnette, J.B., Huang, K., Emmons, R., Park, C.S., Sommers, C.L., El-Khoury, D., Shores, E.W., and Love, P.E. (2001). Fine tuning of TCR signaling by CD5. J Immunol 166, 5464- 5472.10.4049/jimmunol.166.9.5464, PMID:11313384

Fulton, R.B., Hamilton, S.E., Xing, Y., Best, J.A., Goldrath, A.W., Hogquist, K.A., and Jameson, S.C. (2015). The TCR's sensitivity to self peptide-MHC dictates the ability of naive CD8(+) T cells to respond to foreign antigens. Nat Immunol 16, 107-117.10.1038/ni.3043, PMID:25419629

Valerio, M.M., Arana, K., Guan, J., Chan, S.W., Yang, X., Kurd, N., Lee, A., Shastri, N., Coscoy, L., and Robey, E.A. (2023). The promiscuous development of an unconventional Qa1b-restricted T cell population. bioRxiv, 2022.2009.2026.509583.10.1101/2022.09.26.509583,

Efectivamente. Esto es clave, particularmente para modalidades de educación bimodal que requieren participación remota. Y si bien se crearon versiones de Grafoscopio que facilitaban el proceso de instalación y disminuían los tiempos, se requieren unos mínimos de hardware y, desafortunadamente, Windows utiliza muy mal los recursos de cómputo de las máquinas donde está instalado.

Reviewer #1 (Public Review):

In this study, the authors offer a fresh perspective on how visual working memory operates. They delve into the link between anticipating future events and retaining previous visual information in memory. To achieve this, the authors build upon their recent series of experiments that investigated the interplay between gaze biases and visual working memory. In this study, they introduce an innovative twist to their fundamental task. Specifically, they disentangle the location where information is initially stored from the location where it will be tested in the future. Participants are tasked with learning a novel rule that dictates how the initial storage location relates to the eventual test location. The authors leverage participants' gaze patterns as an indicator of memory selection. Intriguingly, they observe that microsaccades are directed toward both the past encoding location and the anticipated future test location. This observation is noteworthy for several reasons. Firstly, participants' gaze is biased towards the past encoding location, even though that location lacks relevance to the memory test. Secondly, there's a simultaneous occurrence of an increased gaze bias towards both the past and future locations. To explore this temporal aspect further, the authors conduct a compelling analysis that reveals the joint consideration of past and future locations during memory maintenance. Notably, microsaccades biased towards the future test location also exhibit a bias towards the past encoding location. In summary, the authors present an innovative perspective on the adaptable nature of visual working memory. They illustrate how information relevant to the future is integrated with past information to guide behavior.

This short manuscript presents one experiment with straightforward analyses, clear visualizations, and a convincing interpretation. For their analysis, the authors focus on a single time window in the experimental trial (i.e., 0-1000 ms after retro cue onset). While this time window is most straightforward for the purpose of their study, other time windows are similarly interesting for characterizing the joint consideration of past and future information in memory. First, assessing the gaze biases in the delay period following the cue offset would allow the authors to determine whether the gaze bias towards the future location is sustained throughout the entire interval before the memory test onset. Presumably, the gaze bias towards the past location may not resurface during this delay period, but it is unclear how the bias towards the future location develops in that time window. Also, the disappearance of the retro cue constitutes a visual transient that may leave traces on the gaze biases which speaks again for assessing gaze biases also in the delay period following the cue offset.

Moreover, assessing the gaze bias before retro-cue onset allows the authors to further characterize the observed gaze biases in their study. More specifically, the authors could determine whether the future location is considered already during memory encoding and the subsequent delay period (i.e., before the onset of the retro cue). In a trial, participants encode two oriented gratings presented at opposite locations. The future rule indicates the test locations relative to the encoding locations. In their example (Figure 1a), the test locations are shifted clockwise relative to the encoding location. Thus, there are two pairs of relevant locations (each pair consists of one stimulus location and one potential test location) facing each other at opposite locations and therefore forming an axis (in the illustration the axis would go from bottom left to top right). As the future rule is already known to the participants before trial onset it is possible that participants use that information already during encoding. This could be tested by assessing whether more microsaccades are directed along the relevant axis as compared to the orthogonal axis. The authors should assess whether such a gaze bias exists already before retro cue onset and discuss the theoretical consequences for their main conclusions (e.g., is the future location only jointly used if the test location is implicitly revealed by the retro cue).

Para el repositorio de chiselapp se debe ingresar al comando, (Windterm o Powershell), con fossil y agregar el título de las sesiones.

fossil.exe para windows; fossil para Mac

Reviewer #2 (Public Review):

In this study the authors sought to understand the extent of similarity among species in intraspecific adaptation to environmental heterogeneity at the phenotypic and genetic levels. A particular focus was to evaluate if regions that were associated with adaptation within putative inversions in one species were also candidates for adaptation in another species that lacked those inversions. This study is timely for the field of evolutionary genomics, due to recent interest surrounding how inversions arise and become established in adaptation.

Major strengths

Their study system was well suited to addressing the aims, given that the different species of sunflower all had GWAS data on the same phenotypes from common garden experiments as well as landscape genomic data, and orthologous SNPs could be identified. Organizing a dataset of this magnitude is no small feat. The authors integrate many state-of-the-art statistical methods that they have developed in previous research into a framework for correlating genomic Windows of Repeated Association (WRA, also amalgamated into Clusters of Repeated Association based on LD among windows) with Similarity In Phenotype-Environment Correlation (SIPEC). The WRA/CRA methods are very useful and the authors do an excellent job at outlining the rationale for these methods.

Major weaknesses

The study results rely heavily on the SIPEC measure, but I found the values reported difficult to interpret biologically. For example, in Figure 4 there is a range of SIPEC from 0 to 0.03 for most species pairs, with some pairs only as high as ~0.01. This does not appear to be a high degree of similarity in phenotype-environment correlation. For example, given the equation on line 517 for a single phenotype, if one species has a phenotype-environment correlation of 1.0 and the other has a correlation of 0.02, I would postulate that these two species do not have similar evolutionary responses, but the equation would give a value of (1+0.02)*1*0.02/1 = 0.02 which is pretty typical "higher" value in Figure 4. I also question the logic behind using absolute values of the correlations for the SIPEC, because if a trait increases with an environment in one species but decreases with the environment in another species, I would not predict that the genetic basis of adaptation would be similar (as a side note, I would not question the logic behind using absolute correlations for associations with alleles, due to the arbitrary nature of signing alleles). I might be missing something here, so I look forward to reading the author's responses on these thoughts.

An additional potential problem with the analysis is that from the way the analysis is presented, it appears that the 33 environmental variables were essentially treated as independent data points (e.g. in Figure 4, Figure 5). It's not appropriate to treat the environmental variables independently because many of them are highly correlated. For example in Figure 4, many of the high similarity/CRA values tend to be categorized as temperature variables, which are likely to be highly correlated with each other. This seems like a type of pseudo replication and is a major weakness of the framework.

Below I highlight the main claims from the study and evaluate how well the results support the conclusions.

* "We find evidence of significant genome-wide repeatability in signatures of association to phenotypes and environments" (abstract)<br /> * Given the questions above about SIPEC, I did not find this conclusion well supported with the way the data are presented in the manuscript.

* "We find evidence of significant genome-wide repeatability in signatures of association to phenotypes and environments, which are particularly enriched within regions of the genome harbouring an  inversion in one species. " (Abstract) And "increased repeatability found in regions of the genome that harbour inversions" (Discussion)<br /> * These claims are supported by the data shown in Figure 4, which shows that haploblocks are enriched for WRAs. I want to clarify a point about the wording here, as my understanding of the analysis is that the authors test if *haploblocks* are enriched with *WRAs*, not whether *WRAs* are enriched for *haploblocks*. The wording of the abstract is claiming the latter, but I think what they tested was the former. Let me know if I'm missing something here.<br /> * Notwithstanding the concerns about highly correlated environments potentially inflating some of the patterns in the manuscript, to my knowledge this is the first attempt in the literature to try this kind of comparison, and the results does generally suggest that inversions are more likely capturing, rather than accumulating adaptive variation. However, I don't think the authors can claim that repeated signatures are enriched with haploblock regions, and the authors should take care to refrain from stating the relative importance of different regions of the genome to adaptation without an analysis.


* "While a large number of genomic regions show evidence of repeated adaptation, most of the strongest signatures of association still tend to be species-specific, indicating substantial genotypic redundancy for local adaptation in these species." (Abstract)<br /> * Figure 3B certainly makes it look like there is very little similarity among species in the genetic basis of adaptation, which leaves the question as to how important the repeated signatures really are for adaptation if there are very few of them. (Is 3B for the whole genome or only that region?). This result seems to be at odds with the large number of CRAs and the claims about the importance of haploblock regions to adaptation, which extend from my previous point.


* "we have shown evidence of significant repeatability in the basis of local adaptation (Figure 4, 5), but also an abundance of species-specific, non-repeated signatures (Figure 3)"<br /> * While the claim is a solid one, I am left wondering how much of these genomes show repeated vs. non-repeated signatures, how much of these genomes have haploblocks, and how much overlap there really is. Finding a way to intuitively represent these unknowns would greatly strengthen the manuscript.

Overall, I think the main claims from the study, the statistical framework, and the results could be revised to better support each other.

Although the current version of the manuscript has some potential shortcomings with regards to the statistical approaches, and the impact of this paper in its present form could be stifled because the biology tended to get lost in the statistics, these shortcomings may be addressed by the authors.

With some revisions, the framework and data could have a high impact and be of high utility to the community.

Author Response

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

Reviewer #1 (Public Review):

Soudi, Jahani et al. provide a valuable comparative study of local adaptation in four species of sunflowers and investigate the repeatability of observed genomic signals of adaptation and their link to haploblocks, known to be numerous and important in this system. The study builds on previous work in sunflowers that have investigated haploblocks in those species and on methodologies developed to look at repeated signals of local adaptations. The authors provide solid evidence of both genotype-environment associations (GEA) and genome-wide association study (GWAS), as well as phenotypic correlations with the environment, to show that part of the local adaptation signal is repeatable and significantly co-occur in regions harboring haploblocks. Results also show that part of the signal is species specific and points to high genetic redundancy. The authors rightfully point out the complexities of the adaptation process and that the truth must lie somewhere between two extreme models of evolutionary genetics, i.e. a population genetics view of large effect loci and a quantitative genetics model. The authors take great care in acknowledging and investigating the multiple biases inherent to the used methods (GEA and GWAS) and use a conservative approach to draw their conclusions. The multiplicity of analyses and their interdependence make them slightly hard to understand and the manuscript would benefit from more careful explanations of concepts and logical links throughout. This work will be of interest to evolutionary biologists and population geneticists in particular, and constitutes an additional applied example to the comparative local adaptation literature.

Some thoughts on the last paragraph of the discussion (L481-497): I think it would be fine to have some more thoughts here on the processes that could contribute to the presence/absence of inversions, maybe in an "Ideas and Speculation" subsection. To me, your results point to the fact that though inversions are often presented as important for local adaptation, they seem to be highly contingent on the context of adaptation in each species. First, repeatability results are only at the window/gene level in your results, the specific mutations are not under scrutiny. Is it possible that inversions are only necessary when sets of small effect mutations are used, opposite to a large effect mutation in other species? Additionally, in a model with epistasis, fitness effects of mutations are dependent on the genomic background and it is possible that inversions were necessary in only certain contexts, even for the same mutations, i.e. some adaptive path contingency. Finally, do you have specific demographic history knowledge in this system that maps to the observations of the presence of inversions or not? For example, have the species "using" inversions been subject to more gene flow compared to others?

Thank you for the great suggestions and helpful comments. Regarding the question of demography, each of the species actually harbours quite a large number of haploblocks (13 in H. annuus spanning 326Mb, 6 in H. argophyllus spanning 114 Mb, and 18 in H. petiolaris spanning 467 Mb; see Todesco et al. 2020 for more details) so there does not seem to be any clear association with demography. We agree about the complexities that might underly the evolution of inversions that you outline above, and have refined some of the text where we discuss their evolution in the Discussion.

Reviewer #2 (Public Review):

In this study the authors sought to understand the extent of similarity among species in intraspecific adaptation to environmental heterogeneity at the phenotypic and genetic levels. A particular focus was to evaluate if regions that were associated with adaptation within putative inversions in one species were also candidates for adaptation in another species that lacked those inversions. This study is timely for the field of evolutionary genomics, due to recent interest surrounding how inversions arise and become established in adaptation.

Major strengths

Their study system was well suited to addressing the aims, given that the different species of sunflower all had GWAS data on the same phenotypes from common garden experiments as well as landscape genomic data, and orthologous SNPs could be identified. Organizing a dataset of this magnitude is no small feat. The authors integrate many state-of-the-art statistical methods that they have developed in previous research into a framework for correlating genomic Windows of Repeated Association (WRA, also amalgamated into Clusters of Repeated Association based on LD among windows) with Similarity In Phenotype-Environment Correlation (SIPEC). The WRA/CRA methods are very useful and the authors do an excellent job at outlining the rationale for these methods.

Thank you!

Major weaknesses

The study results rely heavily on the SIPEC measure, but I found the values reported difficult to interpret biologically. For example, in Figure 4 there is a range of SIPEC from 0 to 0.03 for most species pairs, with some pairs only as high as ~0.01. This does not appear to be a high degree of similarity in phenotype-environment correlation. For example, given the equation on line 517 for a single phenotype, if one species has a phenotype-environment correlation of 1.0 and the other has a correlation of 0.02, I would postulate that these two species do not have similar evolutionary responses, but the equation would give a value of (1+0.02)10.02/1 = 0.02 which is pretty typical "higher" value in Figure 4. I also question the logic behind using absolute values of the correlations for the SIPEC, because if a trait increases with an environment in one species but decreases with the environment in another species, I would not predict that the genetic basis of adaptation would be similar (as a side note, I would not question the logic behind using absolute correlations for associations with alleles, due to the arbitrary nature of signing alleles). I might be missing something here, so I look forward to reading the author's responses on these thoughts.

The reviewer makes a very good point about the range of SIPEC, and we have changed our analysis to reflect this, now reporting the maximum value of SIPEC for each environment (across the axes of the PCA on phenotypes that cumulatively explain 95% of the variance), in Figure 4 and Supplementary Figures S2 and S13. For consistency among manuscript versions and to illustrate the effect of this change, we retain the mean SIPEC value in one figure in the supplementary materials (S12), which shows the small effect of this change on the qualitative patterns. Figure 4 now shows that the maximum SIPEC value is regularly quite strong, which should address the reviewer’s concern that this is not being driven by anomalous and small values. We appreciate this point and think this change now more closely reflects how we are trying to estimate the biological feature of interest – that some axis of phenotypic space is strongly (or not) responding to selection from the environmental variable.

With respect to the logic behind using absolute value, we still feel this is justified for traits, because if a trait evolves to be bigger or smaller, it may still use the same genes. For example, flowering time may change to be later or earlier, which would result in opposite correlations with a given environment, but might use the same gene (e.g. FT) for this. As such, we think keeping absolute value is more representative as otherwise species with strong but opposite patterns of adaptation would look like they were very different. We have added a statement on line 584 in the methods section to further clarify the reason for this choice.

An additional potential problem with the analysis is that from the way the analysis is presented, it appears that the 33 environmental variables were essentially treated as independent data points (e.g. in Figure 4, Figure 5). It's not appropriate to treat the environmental variables independently because many of them are highly correlated. For example in Figure 4, many of the high similarity/CRA values tend to be categorized as temperature variables, which are likely to be highly correlated with each other. This seems like a type of pseudo replication and is a major weakness of the framework.

This is a good point and we fully agree. It is for this reason that we didn’t present any p-values or statistical tests of the overall patterns that are shown in these figures (i.e. the linear relationship between SIPEC and number of CRAs in figure 4 and the tendency for most points to fall above the 1:1 line in figure 5). But to make sure this is even more clear, we have added statements to the captions of these figures to remind readers that points are non-independent. We still feel that in the absence of a formal test, the overall patterns are strongly consistent with this interpretation. A smaller number of non-pseudo-replicated points in Figure 4 would still likely show linear patterns. Similarly, there are almost no significant points falling below the 1:1 line in Figure 5, and it seems unlikely that pseudoreplication would generate this pattern.

Below I highlight the main claims from the study and evaluate how well the results support the conclusions.

"We find evidence of significant genome-wide repeatability in signatures of association to phenotypes and environments" (abstract)<br /> Given the questions above about SIPEC, I did not find this conclusion well supported with the way the data are presented in the manuscript.

We have changed the reporting of the SIPEC metric so that it more clearly reflects whichever axis of phenotypic space is most strongly correlated with environment in both species (using max instead of mean). This shows similar qualitative patterns but illustrates that this happens across much higher values of SIPEC, showing that it is in fact driven by high correlations in each species (or non-similar correlations resulting in low values of SIPEC). While we agree about the pseudo-replication problem preventing formal statistical test of this hypothesis, the visual pattern is striking and seems unlikely to be an artefact, so we think this does still support this conclusion.

"We find evidence of significant genome-wide repeatability in signatures of association to phenotypes and environments, which are particularly enriched within regions of the genome harbouring an inversion in one species. " (Abstract) And "increased repeatability found in regions of the genome that harbour inversions" (Discussion)<br /> These claims are supported by the data shown in Figure 4, which shows that haploblocks are enriched for WRAs. I want to clarify a point about the wording here, as my understanding of the analysis is that the authors test if haploblocks are enriched with WRAs, not whether WRAs are enriched for haploblocks. The wording of the abstract is claiming the latter, but I think what they tested was the former. Let me know if I'm missing something here.

We are actually not interested in whether WRAs are enriched for haploblocks; we want to know if WRAs tend to occur more commonly within haploblocks than outside of them. We have tried to clarify that this is our aim in various places in the manuscript. Our analysis for Figure 5 is the one supporting these claims, and it uses the Chi-square test statistic to assess the number of WRAs and non-WRAs that fall within vs. outside of inversions, and a permutation test to assess the significance of this observation, for each environmental variable and phenotype. We don’t think that this test has any direction to it – it’s simply testing if there is non-random association between the levels of the two factors. Thus, we think the wording we have used is consistent with the test result and our aims. Perhaps the confusion arose from the two methods that we present in the Methods (one is used for Figure 5, the other for Figure S6C & D), so we have added clarifications there.

Notwithstanding the concerns about highly correlated environments potentially inflating some of the patterns in the manuscript, to my knowledge this is the first attempt in the literature to try this kind of comparison, and the results does generally suggest that inversions are more likely capturing, rather than accumulating adaptive variation. However, I don't think the authors can claim that repeated signatures are enriched with haploblock regions, and the authors should take care to refrain from stating the relative importance of different regions of the genome to adaptation without an analysis.

Actually, we don’t have a strong feeling about whether inversions are capturing vs. accumulating adaptive variation, as these results could be consistent with either. As described above, we do not understand why we can’t claim that repeated signatures are enriched within haploblocks. We thought the reviewer is perhaps referring to the fact that the points are pseudo-replicated in the figures due to environment? We note that a very large number of points are significantly different from random in terms of the distribution of WRAs within vs. outside of haploblocks (light- vs. dark-shaded symbols), and that almost all of them fall above the 1:1 line. While there may be pseudo-replication preventing a test of the bigger multi-environment/multi-species hypothesis across all phenotypes and environments, there is almost a complete lack of significant results in the other direction. This seems like quite strong evidence about enrichment of WRAs within haploblocks, across many environments/species contrasts. We have added some text to the description of patterns in figure 5 to try to clarify this.

"While a large number of genomic regions show evidence of repeated adaptation, most of the strongest signatures of association still tend to be species-specific, indicating substantial genotypic redundancy for local adaptation in these species." (Abstract)<br /> Figure 3B certainly makes it look like there is very little similarity among species in the genetic basis of adaptation, which leaves the question as to how important the repeated signatures really are for adaptation if there are very few of them. (Is 3B for the whole genome or only that region?). This result seems to be at odds with the large number of CRAs and the claims about the importance of haploblock regions to adaptation, which extend from my previous point.

Figure 3B is for the whole genome, we have added text to the figure caption to clarify this. We think that both interpretations are possible: that most of the regions of the genome that are driving adaptation are non-repeated, but that a small but significant proportion of regions driving adaptation are repeated above what would be expected at random. Thus, it seems that there is high redundancy, coupled with adaptation via some genes that seem particularly functionally important and non-redundant, and therefore repeated. We added clarifying text on lines 541-548.

"we have shown evidence of significant repeatability in the basis of local adaptation (Figure 4, 5), but also an abundance of species-specific, non-repeated signatures (Figure 3)"<br /> While the claim is a solid one, I am left wondering how much of these genomes show repeated vs. non-repeated signatures, how much of these genomes have haploblocks, and how much overlap there really is. Finding a way to intuitively represent these unknowns would greatly strengthen the manuscript.

We agree, and really struggled to find the best way to communicate both the repeated patterns and the large amount of non-repeated signatures. Unfortunately, we have more confidence in the validity of repeated patterns because for the non-repeated patterns, a strong signature of association to environment in only one species could just be the product of structureenvironment correlation, as we didn’t control for population structure. Thus, trying to quantify the proportion of non-repeated signatures is difficult to do with any accuracy and we preferred to avoid putting too much emphasis on the simple calculation of the proportion of top candidate windows that were also WRAs.

Overall, I think the main claims from the study, the statistical framework, and the results could be revised to better support each other.

Although the current version of the manuscript has some potential shortcomings with regards to the statistical approaches, and the impact of this paper in its present form could be stifled because the biology tended to get lost in the statistics, these shortcomings may be addressed by the authors.

With some revisions, the framework and data could have a high impact and be of high utility to the community.

Thank you for your very helpful comments and suggestions on our paper, we really appreciate it.

Recommendations for the authors: please note that you control which revisions to undertake from the public reviews and recommendations for the authors

Editor's comments:

The reviewers make a series of reasonable suggestions that I echo. I found the paper quite hard to follow, and got fairly lost in the various layers of analyses done. Partially, this represents the complexity of empirical genomic data, which rarely deliver simple stories of convergence at a few genes. However, the properties of the various statistics used to detail local adaptation and convergence are not particularly clear and the figures presented were not intuitive representations of the data. This leaves the reader with an incomplete view of how much weight to put in the various lines of evidence marshaled. I would suggest simplifying the presentation of the results considerably. I add a few additional comments below.

Great suggestion, we’ve added a schematic overview of the methods and main research questions to Figure S1 in the supplementary materials.

A figure would help showing some of the signals of SNPs with putative signals of convergent environmental correlations across species, e.g. frequencies plotted against climate variables. This would help readers get a sense of how strong these signals were. These could be accompanied by the statistics calculated for these SNPs, that would allow the reader to start to get some intuitive sense of what the numbers mean.

Great suggestion, we have added a schematic overview of the methods to Figure S1 that shows some of the values and illustrates how the methods work using visual examples from our data.

In general, the introduction and some of the discussion of the inversion results feel oddly framed:<br /> Abstract line 36: "This shows that while inversions may facilitate local adaptation, at least some of the loci involved can still make substantial contributions without the benefit of recombination suppression."

We have changed “some of the loci involved can still make substantial contributions without the benefit of recombination suppression” here to “some of the loci involved can still harbour mutations that make substantial contributions without the benefit of recombination suppression in species lacking a segregating inversion” as it hopefully clarifies that we’re not talking about individual alleles that are present in both species.

Models of the role of local adaptation in the establishment of inversions (Kirkpatrick & Barton) assume that there are multiple locally adapted alleles already present. It is the load created by these alleles being constantly maintained in the face of migration and subsequent recombination that allow an inversion to be selected for because it keeps together locally adapted alleles. Thus these models predict that there could well be standing local adaptation at these loci in the absence of the inversion in other species, and that these locally adapted alleles while not fixed may be at high frequency. (After establishment, inversions housing locally adapted alleles, can shield more weakly, locally beneficial alleles from migration allow other alleles to build up.) Empirically it's interesting to find signals of local adaptation in other species that don't contain putative inversions. But the logic of the different predictions is not particularly clear from the introduction, and only becomes somewhat clearer in the discussion.

Thank you for pointing out this murkiness, we have re-written portions of both the Introduction and Discussion to clarify this aspect.

From the introduction: Inversions have been implicated in local adaptation in many species (Wellenreuther and Bernatchez 2018), likely due to their effect to suppress recombination among inverted and noninverted haplotypes, and thereby maintain LD among beneficial combinations of locally adapted alleles (Rieseberg 2001; Noor et al. 2001; Kirkpatrick and Barton 2006). This has been approached by models studying the establishment of inversions that capture combinations of locally adapted alleles present as standing variation (e.g., Kirkpatrick and Barton 2006), as well as models examining the accumulation of locally adapted mutations within inversions (e.g., Schaal et al. 2022). If there is variation in the density of loci that can potentially contribute to local adaptation, inversions would be expected to preferentially establish and be retained in regions harbouring a high density of such loci (and this expectation would hold for both the capture and accumulation models). We would also expect to see stronger signatures of repeated local adaptation in such high density regions. Despite mounting evidence of their importance in adaptation, it is unclear how inversions may covary with repeatability of adaptation among species. A fundamental parameter of importance in these models is the relationship between migration rate and strength of selection on individual alleles, which may not make persistent contributions to local adaptation without the suppressing effects of recombination if selection is too weak (Yeaman and Whitlock 2011; Bürger and Akerman 2011). If most alleles have small effects relative to migration rate and can only contribute to local adaptation via the benefit of the recombination-suppressing effect of an inversion, then we would expect little repeatability at the site of an inversion – other species lacking the inversion would not tend to use that same region for adaptation because selection would be too weak for alleles to persist. On the other hand, if some loci are particularly important for local adaptation and regularly yield mutations of large effect, with these patterns being conserved among species, repeatability within regions harbouring inversions may be substantial. Thus, studying whether adaptation at the same genomic region harbouring an inversion is observed in other species lacking the inversion can give insights about the underlying architecture of adaptation, and the evolution and maintenance of inversions.

From the Discussion: The observed repeatability associated with inversions further supports the local adaptation model as an explanation for the long-term persistence of segregating inversions (at least in sunflowers, rather than mechanisms based on dominance or meiotic drive (Rieseberg 2001). If there is variation across the genome in the density of loci with the potential to be involved in local adaptation, then the establishment and maintenance of inversions would be biased towards regions harbouring a high density such loci under this model. If the genomic basis for local adaptation is conserved amongst species, then these same regions are more likely to have high repeatability. Thus, our observation of genomic regions harbouring inversions also being enriched for WRAs is consistent with this general model for inversion evolution. Unfortunately, our observations do not provide much insight into whether inversions evolve through the capture (e.g. Kirkpatrick and Barton 2006) or accumulation (e.g. Schaal et al. 2022) type of model, as either model would be consistent with our results. Most of the sunflower inversions are >1 My old, and therefore predate any current local adaptation patterns, but likely do not predate the genes underlying local adaptation (which appear to be shared among the species we studied). As for the alleles underlying local adaptation, they may be younger than the inversions, but as our work suggests, these regions are prone to harbouring locally adaptive alleles so it is possible that they also harboured other ancestral locally adaptive alleles.

As a minor comment, there's a fair number of places where a more nuanced view of the field is needed, e.g.:<br /> "Models in evolutionary genetics tend to focus on extremes: population genetic approaches explore cases where strong selection deterministically drives a change in allele frequency" --This seems like a strange strawman. Population genetic models span a huge parameter range. The empirical approaches of looking for sweeps by detecting genome-wide statistical outliers is predicated on strong selection, but there are numerous papers that have looked for signals of weak selection genome-wide.

Good point, we have changed our wording here.

Reviewer #1 (Recommendations For The Authors):

Comments

My main comment on the manuscript is that the different levels and diversity of analyses are slightly hard to follow on the first, and even second, read. As there are several layers of correlations and comparisons, as well as some independent analyses, I wonder if it might be helpful to have a summary schematic figure of how all analyses fit together.

Great idea, we have added Figure S1 that summarizes the main flow of the methods and research questions.

• L169-171: Would it be more accurate to say that SIPEC is maximized when both species have strong correlations for an environmental variable across the same phenotypes? But maybe I misunderstood the index.

Good point, we have now simplified SIPEC, reporting the max instead of the mean, which we think better reflects when similar patterns are happening in both species for some phenotype.

• L191: Given the discussion in the introduction and elsewhere about the correction for population structure, which version is used here? Same for Figure 3.

We have added clarification there.

• L348: One [environmental] variable?

Added

• L353: Maybe add a percentage indication for 387 so that it is comparable to the following 23.3%.

Good point, added

-> L388 and paragraph: You mention "significant repeatability" but it is hard from the results at this point to have a broad idea of the amount of signal that is repeatable. Would it be possible to add here some quantitative measure of the proportion of signal repeatable or not, even if approximated?

I wish we could, but I think the precision implied by such an approximation would involve a huge amount of uncertainty and likely inaccuracy. Because it is so hard to conclusively identify how many loci are significant but non-repeated, we really don’t have a good handle on the denominator here. We are pretty confident that the repeated loci are strongly enriched for true positives, but the non-repeated loci are also almost certainly strongly enriched for false positives. While we really want to be able to quantify this explicitly, we don’t think it’s possible given our data.

-L415-418: "If there is variation [...] involved in local adaptation", I do not follow this argument, could you rephrase?

Changed

-L447-450: As you say in the supplementary methods, your analyses exclude 3/4 of the genome. Do you think this choice has a large impact on the number of outliers observed here as the genome-wide baseline would change?

This is a very good question, but one that is quite complex and without a clear answer – we chose not to delve into it in the paper to keep the discussion streamlined. My (SY) feeling is that it is unlikely that regions harbouring transposable elements would contribute much to adaptation, but I think we really don’t know if that is true. Even excluding ¾ of the genome harbouring TEs, ¼ of the genome still constitutes a huge amount of sequence and a very large number of genes and it seems plausible that most genes and genic regions would not contribute to adaptation for a given trait, so I don’t think this would change the results too much in a qualitative way – but would almost certainly change the number of windows that are significant, etc.

• L455-457: "As we are unable [...] potentially important drivers" Could you provide the logical link here between loci of small effect and them being important drivers. I presume you mean that the large effect loci found here only account for a small proportion of the heritability?

Yes that’s what we meant here, so we’ve added some clarification.

• L482: "enriched within inversions" should that be 'in genomic regions where there exist inversions in at least one species'? Thanks for catching that, yes. Changed.

• Methods/SIPEC L512: Compared to the Results section it is unclear here what is referred to as an "environment" Is it a variable or a set of environment variables?

This is done per environmental variable.

I find the presence of the PCA for environment variables in Figure 2 misleading as my first interpretation was that PCs for environment were also used.

Good point, we have clarified this on line 190-193.

Maybe one potential addition to the formula would be to add an environment variable $j$ notation such that it reads "$SIPEC_j = \sum_i (|r_{ij,1}| + ...) ...$ where ... between environment variable $j$". I had initial difficulties to understand how this SIPEC was computed relating to environmental variables and this might help.

Given the other changes we made to SIPEC, we felt it was simpler to just present it as a single calculation on a given combination of phenotype and environment for a pair of species, and then discuss taking the mean and maximum of this later.

Finally, PCA axes explaining 95% of the variance are used, I would find it interesting to see how many PCs are used in comparison to the number of traits being measured.

We have added the following sentence to the methods describing this:

"For comparisons including H. argophyllus, 95% of the variance was typically explained by 8-10 PC axes (out of 28 or 29 phenotypes), whereas for comparisons among other taxa this included 21 or 22 PC axes (out of 65 or 66 phenotypes."

Typos

L52: --

Changed

L254: portions [of] their

Changed

L399: additional closing parenthesis

Changed

L458: signatures [of] repeated association

Changed

L554: performed [on]

Changed

L578: 5 ~~kp~~/kb windows

Changed

L601: ~~casual~~/causal SNPs

Changed

L615: ~~widow~~/window

Changed

L732: ~~Banding~~/Banting Postdoctoral Fellowship

Changed

L1002 & L960: [Supplementary] Figure

Changed

Supplementary: Some figure titles are in bold and others are not.

Changed

Reviewer #2 (Recommendations For The Authors):

Overall I found the writing to be very clear and easy to follow. Despite my comments, it was clear that a lot of thought went into how to conduct the tests and visualize the results. I recommend ending the Discussion on a positive note, rather than an impossible test.

Thanks for the positive suggestion, we have done this.

In Figure 5, is the temperature variable missing in the legend and in the plot?

No, for this plot we just combined the temperature/precipitation variables into one variable called “climate”.

Reviewer #2 (Public Review):

In this study the authors sought to understand the extent of similarity among species in intraspecific adaptation to environmental heterogeneity at the phenotypic and genetic levels. A particular focus was to evaluate if regions that were associated with adaptation within putative inversions in one species were also candidates for adaptation in another species that lacked those inversions. This study is timely for the field of evolutionary genomics, due to recent interest surrounding how inversions arise and become established in adaptation.

Major strengths-

Their study system was well suited to addressing the aims, given that the different species of sunflower all had GWAS data on the same phenotypes from common garden experiments as well as landscape genomic data, and orthologous SNPs could be identified. Organizing a dataset of this magnitude is no small feat. The authors integrate many state-of-the-art statistical methods that they have developed in previous research into a framework for correlating genomic Windows of Repeated Association (WRA, also amalgamated into Clusters of Repeated Association based on LD among windows) with Similarity In Phenotype-Environment Correlation (SIPEC). The WRA/CRA methods are very useful and the authors do an excellent job at outlining the rationale for these methods.

Weaknesses-

The authors did an excellent job responding to the first set of reviews and overall I found the manuscript more streamlined and easier to read. The main weakness in the manuscript is that correlations among environmental variables were not controlled for in their results, and is a source of potential pseudoreplication. The authors are clear about the results that are affected by pseudoreplication.

The manuscript shows how to integrate many recent methods to study the repeatability of adaptation, and the methods and data are likely to be used in similar studies.

Reviewer #2 (Public Review):

Summary:

Reward and punishment learning have long been seen as emerging from separate networks of frontal and subcortical areas, often studied separately. Nevertheless, both systems are complimentary and distributed representations of rewards and punishments have been repeatedly observed within multiple areas. This raised the unsolved question of the possible mechanisms by which both systems might interact, which this manuscript went after. The authors skillfully leveraged intracranial recordings in epileptic patients performing a probabilistic learning task combined with model-based information theoretical analyses of gamma activities to reveal that information about reward and punishment was not only distributed across multiple prefrontal and insular regions, but that each system showed specific redundant interactions. The reward subsystem was characterized by redundant interactions between orbitofrontal and ventromedial prefrontal cortex, while the punishment subsystem relied on insular and dorsolateral redundant interactions. Finally, the authors revealed a way by which the two systems might interact, through synergistic interaction between ventromedial and dorsolateral prefrontal cortex.

Strengths:

Here, the authors performed an excellent reanalysis of a unique dataset using innovative approaches, pushing our understanding on the interaction at play between prefrontal and insular cortex regions during learning. Importantly, the description of the methods and results is truly made accessible, making it an excellent resource to the community.

This manuscript goes beyond what is classically performed using intracranial EEG dataset, by not only reporting where a given information, like reward and punishment prediction errors, is represented but also by characterizing the functional interactions that might underlie such representations. The authors highlight the distributed nature of frontal cortex representations and propose new ways by which the information specifically flows between nodes. This work is well placed to unify our understanding of the complementarity and specificity of the reward and punishment learning systems.

Weaknesses:

The conclusions of this paper are mostly supported by the data, but whether the findings are entirely generalizable would require further information/analyses.

First, the authors found that prediction errors very quickly converge toward 0 (less than 10 trials) while subjects performed the task for sets of 96 trials. Considering all trials, and therefore having a non-uniform distribution of prediction errors, could potentially bias the various estimates the authors are extracting. Separating trials between learning (at the start of a set) and exploiting periods could prove that the observed functional interactions are specific to the learning stages, which would strengthen the results.

Importantly, it is unclear whether the results described are a common feature observed across subjects or the results of a minority of them. The authors should report and assess the reliability of each result across subjects. For example, the authors found RPE-specific interactions between vmPFC and lOFC, even though less than 10% of sites represent RPE or both RPE/PPE in lOFC. It is questionable whether such a low proportion of sites might come from different subjects, and therefore whether the interactions observed are truly observed in multiple subjects. The nature of the dataset obviously precludes from requiring all subjects to show all effects (given the known limits inherent to intracerebral recording in patients), but it should be proven that the effects were reproducibly seen across multiple subjects.

Finally, the timings of the observed interactions between areas preclude one of the authors' main conclusions. Specifically, the authors repeatedly concluded that the encoding of RPE/PPE signals are "emerging" from redundancy-dominated prefrontal-insular interactions. However, the between-region information and transfer entropy between vmPFC and lOFC for example is observed almost 500ms after the encoding of RPE/PPE in these regions, questioning how it could possibly lead to the encoding of RPE/PPE. It is also noteworthy that the two information measures, interaction information and transfer entropy, between these areas happened at non overlapping time windows, questioning the underlying mechanism of the communication at play (see Figures 3/4). As an aside, when assessing the direction of information flow, the authors also found delays between pairs of signals peaking at 176ms, far beyond what would be expected for direct communication between nodes. Discussing this aspect might also be of importance as it raises the possibility of third-party involvement.

Reviewer #2 (Public Review):

The authors have studied in detail the embryogenesis of the ametabolan insect Thermobia domestica. They have also measured the levels of the two most important hormones in insect development: juvenile hormone (JH) and ecdysteroids. The work then focuses on JH, whose occurrence concentrates in the final part (between 70 and 100%) of embryo development. Then, the authors used a precocene compound (7-ethoxyprecocene, or 7EP) to destroy the JH producing tissues in the embryo of the firebrat T. domestica, which allowed to unveil that this hormone is critically involved in the last steps of embryogenesis. The 7EP-treated embryos failed to resorb the extraembryonic fluid and did not hatch. More detailed observations showed that processes like the maturational growth of the eye, the lengthening of the foregut and posterior displacement of the midgut, and the detachment of the E2 cuticle, were impaired after the 7EP treatment. Importantly, a treatment with a JH mimic subsequent to the 7EP treatment restored the correct maturation of both the eye and the gut. It is worth noting that the timing of JH mimic application was essential for correcting the defects triggered by the treatment with 7EP.

This is a relevant result in itself since the role of JH in insect embryogenesis is a controversial topic. It seems to have an important role in hemimetabolan embryogenesis, but not so much in holometabolans. Intriguingly, it appears important for hatching, an observation made in hemimetabolan and in holometabolan embryos. Knowing that this role was already present in ametabolans is relevant from an evolutionary point of view, and knowing exactly why embryos do not hatch in the absence of JH, is relevant from the point of view of developmental biology.

Then, the authors describe a series of experiments applying the JH mimic in early embryogenesis, before the natural peak of JH occurs, and its effects on embryo development. Observations were made under different doses of JHm, and under different temporal windows of treatment. Higher doses triggered more severe effects, as expected, and different windows of application produced different effects. The most used combination was 1 ng JHm applied 1.5 days AEL, checking the effects 3 days later. Of note, 1.5 days AEL is about 15% embryonic development, whereas the natural peak of JH occurs around 85% embryonic development. In general, the ectopic application of JHm triggered a diversity of effects, generally leading to an arrest of development. Intriguingly, however, a number of embryos treated with 1 ng of JHm at 1.5 days AEL showed a precocious formation of myofibrils in the longitudinal muscles. Also, a number of embryos treated in the same way showed enhanced chitin deposition in the E1 procuticle and showed an advancement of at least a day in the deposition of the E2 cuticle.

While the experiments and observations are done with great care and are very exhaustive, I am not sure that the results reveal genuine JH functions. The effects triggered by a significant pulse of ectopic JHm when the embryo is 15% of the development will depend on the context: the transcriptome existing at that time, especially the cocktail of transcription factors. This explains why different application times produce different effects. This also explains why the timing of JHm application was essential for correcting the effects of 7EP treatment. In this reasoning, we must consider that the context at 85% development, when the JH peaks in natural conditions and plays its genuine functions, must be very different from the context at 15% development, when the JHm was applied in most of the experiments. In summary, I believe that the observations after the application of JHm reveal effects of the ectopic JHm, but not necessarily functions of the JH. If so, then the subsequent inferences made from the premise that these ectopic treatments with JHm revealed JH functions are uncertain and should be interpreted with caution.

Those inferences affect not only the "JH and the progressive nature of embryonic molts" section, but also, the "Modifications in JH function during the evolution of hemimetabolous and holometabolous life histories" section, and the entire "Discussion". In addition to inferences built on uncertain functions, the sections mentioned, especially the Discussion, I think suffer from too many poorly justified speculations. I love speculation in science, it is necessary and fruitful. But it must be practiced within limits of reasonableness, especially when expressed in a formal journal.

Finally, In the section "Modifications in JH function during the evolution of hemimetabolous and holometabolous life", it is not clear the bridge that connects the observations on the embryo of Thermobia and the evolution of modified life cycles, hemimetabolan and holometabolan.

https://www.gaijin.at/en/infos/windows-version-numbers

a variation of this worked for me too

This work has been peer reviewed in GigaScience (see https://doi.org/10.1093/gigascience/giad091 ), which carries out open, named peer-review. These review is published under a CC-BY 4.0 license:

Reviewer name: Tazro Ohta

The manuscript describes Cellsnake, a user-friendly tool for single-cell RNA sequencing analysis that targets non-expert users in the field of bioinformatics. Cellsnake operates as a command-line application, providing offline analysis capabilities for sensitive data. The integration of popular single-cell RNA-seq analysis software within Cellsnake, as described in Table 1, enhanced its utility as a comprehensive workflow. Cellsnake has different execution options (minimal, standard, and advanced) with varying outputs and execution times. The authors have provided well-structured online documentation, including helpful quick-start examples that facilitated easy understanding and usage of Cellsnake.

The tool was tested using the Docker appliance and the provided fetal brain dataset and performed as expected. The manuscript explains the functions well, with the results reproduced from existing research using publicly available datasets. The following issues need to be addressed by the authors.

1. The authors should include the citation for the Snakemake paper to acknowledge its contribution. https://doi.org/10.1093/bioinformatics/bts480

2. To support the claim of unique features in Cellsnake, a comparison with other similar methods, such as that on Galaxy (https://doi.org/10.1093/gigascience/giaa102), should be included.

3. It is recommended to host the Docker container image on both the GitHub Container Registry and the Docker Hub for better availability and redundancy. The authors should publish the Dockerfile to enable users to build a container image, if needed.

4. Online documentation is missing a link to the fetal-liver example dataset (https://cellsnake.readthedocs.io/en/latest/fetalliver.html), which needs to be addressed. The fetalbrain dataset shared via Dropbox should also be deposited in the Zenodo repository to improve accessibility and long-term preservation.

5. To assist users who want to use Cellsnake as a Snakemake workflow, the tool documentation should provide clear instructions on how to run Cellsnake as a single snakemake pipeline. This would be useful for users who utilize existing workflow platforms to accept snakemake requests.

6. The benchmarking of Cellsnake must provide more precise specifications than simply referring to "a standard laptop" for computing requirements. My trial of "cellsnake integrated standard" with the fetalbrain dataset took more than 17 h via Docker execution on my M1 Max MacBook Pro. This may be because the provided Docker image is AMD-based, which let my MacBook run the container on a VM, but the recommended computational specifications will help users. The GitHub issue of the Cellsnake repository also mentioned that the software is not tested on Windows Conda, which should be mentioned at least in the online documentation.

7. In the Data Availability section, please ensure that the correct formatting and consistent identifiers are used for public data, such as replacing SRP129388 with PRJNA429950 and E-MTAB-7407 with PRJEB34784, specifying that these IDs are from the Bioproject database. It is important to mention that EGA files are under controlled access, requiring user permission for retrieval.

8. The references in the manuscript need to be properly formatted to ensure the inclusion of publication years and DOIs where available.

9. The help message from the Cellsnake command indicates that its default values are set for human samples. The authors should mention in the manuscript that the pipeline is configured for human samples and requires further configuration for use with samples from other organisms. A step-by-step guide to configuring the setting for the other species, including the reference data download, would be helpful in obtaining more audiences.

One method used to solve the context window issue is recursive summarization, but this method is lossy, and iteratively lossy at that.

Context window treated as RAM while the external context acts as hard drives, basically making the LLM its own OS that they call 'LLM OS'

Semiotic, structural: Everyone follows the same routine

search : - windows command turn screen blank powershell

%systemroot%\system32\scrnsave.scr /s

This will not turn off you screen but make it completely black

• multistep pages
• quickly checking a new page rather than a focussed read
• overloading the browser tab bar

Mir schwant Böses... oder sagen wir "Lästiges". Soll ich mich während ich z. B. einen Text bearbeite, ständig damit auseinandersetzen, was mir ein schlauer Copilot vorschlägt?

Lass mich doch einfach in Ruhe arbeiten.

rhymes for rhythm, imagery, and emphasis

Author Response

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

Thank you for reviewing and assessing our paper. Reviewer2 had only posive comments. Reviewer 1 also had posive comments but included a list of suggesons. The revised version includes text edits to address the suggesons.

Reviewer 1:

… First, it is unclear whether the experiments and analyses were set up to be able to rule out more specific candidate funcons of the ZI.

The list of possible funcons performed by the ZI is broad. Nevertheless, our study considers a rather long list of neural processes related to the behaviors listed below.

Second, many important details of the experiments and their results are hard to decipher given the current descripons and presentaons of the data.

The procedures used in the present study have all been used and described in our previous studies (cited). We used the same descripons and presentaons as in the prior studies. We have gone over the Methods and figures to ensure that all details required to understand the experiments are provided, but we also added further details following the suggesons noted below.

The paper could be significantly strengthened by including more details from each experiment, stronger jusficaons for the limited behaviors and experimental analyses performed, and, finally, a broader analysis of how the recorded acvity in the ZI relates to behavioral parameters.

The paper studied several behaviors including: 1) spontaneous movement of head-fixed mice on a spherical treadmill, 2) tacle (whisker, and body parts) and auditory (tones and white noise) smuli applied to head fixed mice, 3) spontaneous movement iniaon, change, and turns in freely moving mice, 4) auditory tone (frequency and SPL) mapping in freely behaving mice, 5) auditory-evoked orienng head movements (responses) in the context of several behavioral tasks, 6) signaled acve avoidance responses and escapes (AA1), 7) unsignaled/signaled passive avoidance responses (AA2ITI/AA3-CS2), 8) sensory discriminaon (AA3), 9) CS-US interval ming discriminaon (AA4), and 10) USevoked unsignaled escape responses.

In freely moving experiments, the behavior is connuously tracked and decomposed into translaonal and rotaonal movement components. Discrete responses are also evaluated (e.g., acve avoids, escapes, passive avoids, errors, intertrial crossings, latencies, etc.). These behavioral procedures evaluate many neural processes, including decision making (Go/NoGo in AA1-3), response control/inhibion (unsignaled and signaled passive avoidance in AA2/3), and smulus discriminaon (AA3). The applied smuli, discrete responses, and tracked movement are always related to the recorded ZI acvity using a variety of techniques (e.g., cross-correlaons, PSTHs, event-triggered me extracons, etc.), which relate the discrete and me-series parameters to the neural acvity. We do not think all this qualifies as, “limited behaviors”.

(1) Anatomical specificaon: The ZI contains many disnct subdivisions--each with its own topographically organized inputs/outputs and putave funcons. The current manuscript doesn't reference these known divisions or their behavioral disncons, and one cannot tell exactly which poron(s) of the ZI was included in the current study. Moreover, the elongated structure of the ZI makes it very difficult to specifically or completely infect virally. The data could be beter interpreted if the paper included basic informaon on the locaons of recordings, the extent of the AAV spread in the ZI in each viral experiment, and what fracon of infected neurons were inside versus outside ZI.

Our experiments employed Vgat-Cre mice to target ZI neurons. In this line, GABAergic neurons from the enre ZI express Cre, including the dorsal and ventral subdivisions (see (Vong et al., 2011; Hormigo et al., 2020)). Consequently, AAV injecons in Vgat-Cre mice produce restricted expression in the ZI that can fully delineate the nucleus as shown in the papers referenced above (including ours). There is nil expression in structures above or below ZI because they do not express Cre in these mice (e.g., thalamus and subthalamic nucleus), which allows for selecve targeng of ZI. Our optogenec manipulaons and photometry recordings were not aimed at specific ZI subdivisions. We targeted the area of ZI indicated by the stereotaxic coordinates (see Methods), which are aimed at the center of the structure to maximize success in recording/manipulang neurons within ZI. While all the animals included in the study expressed opsins and GCaMP within ZI that in many animals fully delineated the nucleus, there was normal variability in the locaon of opcal fibers, but we did not detect any differences in the results related to these variaons.

Fiber photometry and optogenecs experiments are performed with rather large diameter opcal probes, which record/manipulate relavely large areas of the targeted structure. This is useful because our goal was to idenfy funconal roles of the enre ZI, which could then be parsed. In the present study, we did not perform experiments to target specific ZI populaons (e.g., retrograde Cre expression from target areas), which may have revealed differences atributed to their projecon sites. However, in the last experiment, we selecvely excited ZI fibers targeng three different areas (midbrain tegmentum, superior colliculus, and posterior thalamus), which revealed clear differences on movement. Thus, future experiments should explore these different populaons (e.g., using retrograde/anterograde expression systems), which may be in different subdivisions.

We have enhanced the Methods secon to clarify these points, including the addion of these references.

(2) Electrophysiological recording on the treadmill: The authors are commended for this technically very difficult experiment. The authors do not specify, however, how they knew when they were recording in ZI rather than surrounding structures, parcularly given that recording site lesions were only performed during the last recording session. A map of the locaons of the different classes of units would be valuable data to relate to the literature.

We have added details about this procedure in the Methods secon. These recordings are performed based on coordinates, and categorizing neurons as belonging to ZI is obviously an esmate based on the final histological verificaon. Nevertheless, the marking lesions revealed that the electrodes were on target, which likely resulted from the care taken during the surgical procedure to define reference points used later during the recording sessions (see Methods). Regarding a map of the unit locaons, we performed several analyses that did not reveal clear differences based on site. For example, we compared depth vs cell class, “There was no difference in recording depth between the four classes of neurons (ANOVA F(3,337)= 1.06 p=0.3676)”. Future experiments that employ addional methods (labelling, opto-tagging, etc.) would be more appropriate to address mapping quesons. Finally, as we state in the paper, “However, these recordings do not target GABAergic neurons and may sample some neurons in the tissue surrounding the zona incerta. Therefore, we used calcium imaging fiber photometry to target GABAergic neurons in the zona incerta”.

(3) The raonale of the analysis of acvity with respect to “movement peak”: It is unclear why the authors did not assess how ZI acvity correlates with a broad set of movement parameters, but rather grouped heterogeneous behavioral epochs to analyze firing with respect to “movement peaks”.

The reviewer is referring to movement peaks on the spherical treadmill. On the treadmill, we used the forward locomotor movement of the animal because this is the main acvity of the mice on the treadmill. We considered “all peaks” (or movements) and “>4 sec peaks”, which select for movement onsets. Compared to the treadmill, in freely movement condions during various behavioral tasks, there is a richer behavioral repertoire, which was analyzed in more detail (i.e., translaonal, and rotaonal components during spontaneous ongoing movement and movement onsets, movement related to various behaviors such as orienng, acve and passive avoidance, escape, sensory smulaon, discriminaon, etc.). Thus, we focused on a broader set of movement parameters in the Cre-defined ZI cells of freely behaving mice.

(4) The display of mean categorical data in various figures is interesng, however, the reader cannot gather a very detailed view of ZI firing responses or potenal heterogeneity with so litle informaon about their distribuons.

The PCA performs the heterogeneity classificaon in an unbiased manner, which we feel is a thoughul approach. The firing rates and correlaons with movement for each category of neurons are detailed in the results. Furthermore, the sensory responses for these neurons are also detailed. Together, we think this provides a detailed view of the units we recorded in awake/head-fixed mice. As already stated, further study would benefit from an addional level of cell site verificaon.

(5) Somatosensory firing responses in ZI: It is unclear why the authors chose the specific smuli used in the study. How oen did they evoke reflexive motor responses? What was the latency of sensory-evoked responses in ZI acvity and the latency of the reflexive movement?

These are broad quesons, and we assume that the reviewer is asking about somatosensory evoked responses on the spherical treadmill. We used air-puffs applied to the whiskers and on the back (le vs right) because the whiskers represent an important sensory representaon for mice, and the back is a part of the body (trunk), which we oen use to movate the animals to move forward on the treadmill. Regarding the latency of the somatosensory evoked responses, in this case, we did not correct them based on the me it takes the air-puff to travel to the whiskers or body part, and therefore we did not provide latencies. Moreover, air-puffs are not a very good method to quanfy whisker-evoked latencies, which are beter measured using other methods (whisker deflecons of single/mulple whiskers using piezo-devices or other mechanical devices, as we and others have done in many studies). We are not sure what the reviewer means by “reflexive behavior”; we did not measure any reflexive behavior under these condions. We have gone over the Methods and Results to ensure that sufficient details are provided about these experiments.

(6) It would be valuable to see example traces in Figure 3 to get a beter sense of the me course and contexts under which Ca signals in ZI tracks movement. What is the typical latency? What is the typical range of magnitudes of responses? Does the Ca signal track both fast and slow movements? How are the authors sure that there are no movement arfacts contribung to the calcium imaging? It seems there is more informaon in the dataset that could be valuable.

As is well known, fiber photometry calcium imaging is a slow populaon signal. We do not think it would be valuable to get into ming issues beyond what is already detailed in the study (i.e., magnitudes measured as areas or peaks, and ming as me-to-peaks). Regarding “movement arfacts”, these signals are absent (flat) in animals that do not express GCAMP. We agree that there must be addional valuable informaon in our datasets (as in most me-series). However, the current paper is already rather extensive. We will connue to peruse our datasets and report addional findings in new papers.

(7) Figure 4: The raonale for quanfying the F/Fo responses over a 6-second window, rather than with respect to discrete movement parameters, is not well explained. What types of movement are binned in this approach and might this broad binning hinder the ability to detect more specific relaonships between acvity and movement?

Figure 4 is focused on characterizing the relaonship between turns (ipsiversive and contraversive) during movement and ZI acvity. We tested different binning windows to find differences, including the 6 sec window in figure 4 for populaon measures (-3 to 3 sec around the turns). This binning approach is effecve at revealing differences where they exist (e.g., superior colliculus) as shown in our previous studies (e.g. (Zhou et al., 2023)). Moreover, the turns in the different direcons can be considered discrete responses at their peak, and the ming of the related acvaons (e.g., me to peaks), which we evaluated, are rather sensive and would have revealed differences, but we did not find them.

(8) Separaon of sensory and motor responses in Figure 5: The current data do not adequately differenate whether the responses are sensory or motor given the high correlaon of the sensory inputs driving motor responses. Because isoflurane can diminish auditory responses early in the auditory pathway, this reviewer is not convinced the isoflurane experiments are interpretable.

The reviewer is referring to Fig. 5C,D. Indeed, the point of this experiment was to show that it is difficult to differenate whether neural responses are sensory or motor in awake and freely moving condions. As we stated in the Results secon, “Although arousal and movement were not dissected in the present experiment (this would likely require paralyzing and ventilating the animal), the results indicate that activation of zona incerta neurons by sensory stimulation is primarily associated with states when sensory-evoked movement is also present”. This is followed in the Discussion by, “…as already noted, the suppression of sensory responses may be due to changes in arousal (Castro-Alamancos, 2004; Lee and Dan, 2012) and not caused by the abolishment of the movements per se”.

(9) Given the broad duraon of the mean avoidance response (Fig. 6 C, botom), it would be useful to know to what extent this plot reflects a prolonged behavior or is the result of averaging different animals/trials with different latencies. Given that the shapes of the F/Fo responses in ZI appear similar across avoids and escapes (Fig. 6D), despite their apparent different speeds and movement duraons (Fig 6C), it would be valuable to know how the ming of the F/Fo relates to movement on a trial-by-trial basis.

The duraon of the avoidance response cannot be ascertained from CS onset (panel 6C botom) and avoids are not wide but rather sharp. We have now made this clearer when Fig. 6C is first menoned (“note that since avoids occur at different latencies after CS onset they are best measured from their occurrence as in Fig. 6D”). Like other related condioned and uncondioned responses, avoids and escapes are similar, varying in the noted parameters. Regarding ming, as already menoned above, we think that the characteriscs of the populaon calcium signal make it unsuitable for further ming consideraons than what we included, parcularly for movements occurring at the fast speeds of avoids and escapes.

(10) Lesion quanficaon: One cannot tell what rostral-caudal extent of ZI was lesioned and quanfied in this experiment. It would be easier to interpret if also ploted for each animal, so the reader can tell how reliable the method is. The mean ablaon would be beter shown as a normalized fracon of cells. Although the authors claim the lesions have litle impact on behavior, it appears the incompleteness of the lesions could warrant a more conservave interpretaon.

The lesion experiment was a complement to the optogenecs inacvaon experiments we performed in our preceding ZI paper and in the present paper. Thus, the finding that the lesions had litle impact on behavior is supporve of the optogenecs findings. Regarding cell counts, we did not select any parts of the ZI to quanfy the number of neurons in either control or lesion mice. We considered the full rostrocaudal extent in our measurements. We are not sure what “fracon” the reviewer is suggesng, considering that these counts are from two different groups of mice (control vs lesion). Note that the red-marked neurons, as shown in Fig. 8A, reveal healthy non-Vgat-Cre neurons outside ZI that mark the extent of the AAV diffusion, which as shown spanned the full extent of the ZI in the coronal plane (and in other planes as the AAV spreads in all direcons).

(11) Optogenecs: the locaon of infected neurons is poorly described, including the rostral-caudal extent and the fracon of neurons inside and outside of ZI. Moreover, it is unclear how strongly the optogenec manipulaons in this study are expected to affect neuronal acvity in ZI.

We discussed the first point in (1) above. Regarding, how optogenec manipulaons are expected to affect neuronal acvity in ZI and its targets, we have conducted extensive electrophysiological recordings in slices and in vivo to detail the effects of our manipulaons on GABAergic neurons (e.g. (Hormigo et al., 2016; Hormigo et al., 2019; Hormigo et al., 2021a; Hormigo et al., 2021b), including ZI neurons (Hormigo et al., 2020). In fact, we never use an opsin we have not validated ourselves using electrophysiology. Moreover, our experiments employ a spectrum of optogenec light paterns (including trains/cont at different powers) that trate the optogenec effects within each session/animal. As shown in fig. 11 and 12, these paterns produce different behavioral effects related to the different levels of neural firing they induce. For ChR2-expressing neurons in ZI, firing is frequency dependent and maximal during Cont blue light (at the same power). For Arch-expressing neurons only Cont is used, and inhibion is a funcon of the green light power. When blue light is applied in ZI fibers targeng different areas, this relaonship changes. Blue light trains (1-ms pulses) at 40-66 Hz become the most effecve means of inducing sustained postsynapc inhibion compared to Cont or low frequencies.

References

Castro-Alamancos MA (2004) Dynamics of sensory thalamocorcal synapc networks during informaon processing states. Progress in Neurobiology 74:213-247.

Hormigo S, Vega-Flores G, Castro-Alamancos MA (2016) Basal Ganglia Output Controls Acve Avoidance Behavior. J Neurosci 36:10274-10284.

Hormigo S, Zhou J, Castro-Alamancos MA (2020) Zona Incerta GABAergic Output Controls a Signaled Locomotor Acon in the Midbrain Tegmentum. eNeuro 7.

Hormigo S, Zhou J, Castro-Alamancos MA (2021a) Bidireconal control of orienng behavior by the substana nigra pars reculata: disnct significance of head and whisker movements. eNeuro. Hormigo S, Vega-Flores G, Rovira V, Castro-Alamancos MA (2019) Circuits That Mediate Expression of Signaled Acve Avoidance Converge in the Pedunculoponne Tegmentum. J Neurosci 39:45764594.

Hormigo S, Zhou J, Chabbert D, Shanmugasundaram B, Castro-Alamancos MA (2021b) Basal Ganglia Output Has a Permissive Non-Driving Role in a Signaled Locomotor Acon Mediated by the Midbrain. J Neurosci 41:1529-1552.

Lee SH, Dan Y (2012) Neuromodulaon of brain states. Neuron 76:209-222.

Vong L, Ye C, Yang Z, Choi B, Chua S, Jr., Lowell BB (2011) Lepn acon on GABAergic neurons prevents obesity and reduces inhibitory tone to POMC neurons. Neuron 71:142-154.

Zhou J, Hormigo S, Busel N, Castro-Alamancos MA (2023) The Orienng Reflex Reveals Behavioral States Set by Demanding Contexts: Role of the Superior Colliculus. J Neurosci 43:1778-1796.

Joint Public Review:

In the current paper, Jones et al. describe a new framework, named "coccinella", for real-time high-throughput behavioral analysis aimed at reducing the cost of analyzing behavior. In the setup used here each fly is confined to a small circular arena and able to walk around on an agar bed spiked with nutrients or pharmacological agents. The new framework, built on the researchers' previously developed platform Ethoscope, relies on relatively low-cost Raspberry Pi video cameras to acquire images at ~0.5 Hz and pull out, in real time, the maximal velocity (parameter extraction) during 10 second windows from each video. Thus, the program produces a text file, and not voluminous videos requiring storage facilities for large amounts of video data, a prohibitive step in many behavioral analyses. The maximal velocity time-series is then fed to an algorithm called Highly Comparative Time-Series Classification (HCTSA)(which itself is based on a large number of feature extraction algorithms) developed by other researchers. HCTSA identifies statistically salient features in the time-series which are then passed on to a type of linear classifier algorithm called support vector machines (SVM). In cases where such analyses are sufficient for characterizing the behaviors of interest this system performs as well as other state-of-the-art systems used in behavioral analysis (e.g., DeepLabCut)

In a pharmacobehavior paradigm testing different chemicals, the authors show that coccinella can identify specific compounds as effectively as other more time-consuming and resource-consuming systems.

The new paradigm should be of interest to researchers involved in drug screens, and more generally, in high-throughput analysis focused on gross locomotor defects in fruit flies such as identification of sleep phenotypes. By extracting/saving only the maximal velocity from video clips, the method is fast. However, the rapidity of the platform comes at a cost--loss of information on subtle but important behavioral alterations. When seeking subtle modifications in animal behavior, solutions like DeepLabCut, which are admittedly slower but far superior in terms of the level of details they yield, would be more appropriate.

The manuscript reads well, and it is scientifically solid. The comments listed below were directed to the original submission and were satisfactorily addressed in the revised version.

1- The fact that Coccinella runs on Ethoscopes, an open source hardware platform described by the same group, is very useful because the relevant publication describes Ethoscope in detail. However, the current version of the paper does not offer details or alternatives for users that would like to test the framework, but do not have an Ethoscope. Would it be possible to overcome this barrier and have coccinella run with any video data (and, thus, potentially be used to analyze data obtained from other animal models)?

2- Readers who want background on the analytical approaches that the platform relies on following maximal velocity extraction, will have to consult the original publications. In particular, the current manuscript does not provide much explanation on Highly Comparative Time-Series Classification (HCTSA) or SVM; this may be reasonable because the methods were developed earlier by others. While some readers may find that the lack of details increases the manuscript's readability, others may be left wanting to see more discussion on these not-so-trivial approaches. In addition, it is worth noting that the same authors that published the HCTSA method, also described a shorter version named catch22, that runs faster with a similar output. Thus, explaining in more detail how HCTSA operates, considering is a relatively new method, will make the method more convincing.

I did not know this and I own a windows computer as well but I have not noticed any big A.I features yet.

Is there any evidence to substantiate or reject these claims outside of Starbucks's statement?

Reviewer #3 (Public Review):

Summary:<br /> Blaeser et al. set out to explore the link between CSD and headache pain. How does an electrochemical wave in the brain parenchyma, which lacks nociceptors, result in pain and allodynia in the V1-3 distribution? Prior work had established that CSD increased the firing rate of trigeminal neurons, measured electrophysiologically at the level of the peripheral ganglion. Here, Blaeser et al. focus on the fine afferent processes of the trigeminal neurons, resolving Ca2+ activity of individual fibers within the meninges. To accomplish these experiments, the authors injected AAV encoding the Ca2+ sensitive fluorophore GCamp6s into the trigeminal ganglion, and 8 weeks later imaged fluorescence signals from the afferent terminals within the meninges through a closed cranial window. They captured activity patterns at rest, with locomotion, and in response to CSD. They found that mechanical forces due to meningeal deformations during locomotion (shearing, scaling, and Z-shifts) drove non-spreading Ca2+ signals throughout the imaging field, whereas CSD caused propagating Ca2+ signals in the trigeminal afferent fibers, moving at the expected speed of CSD (3.8 mm/min). Following CSD, there were variable changes in basal GCamp6s signals: these signals decreased in the majority of fibers, signals increased (after a 25 min delay) in other fibers, and signals remained unchanged in the remainder of fibers. Bouts of locomotion were less frequent following CSD, but when they did occur, they elicited more robust GCamp6s signals than pre-CSD. These findings advance the field, suggesting that headache pain following CSD can be explained on the basis of peripheral cranial nerve activity, without invoking central sensitization at the brain stem/thalamic level. This insight could open new pathways for targeting the parenchymal-meningeal interface to develop novel abortive or preventive migraine treatments.

Strengths:<br /> The manuscript is well-written. The studies are broadly relevant to neuroscientists and physiologists, as well as neurologists, pain clinicians, and patients with migraine with aura and acephalgic migraine. The studies are well-conceived and appear to be technically well-executed.

Weaknesses:<br /> 1) Lack of anatomic confirmation that the dura were intact in these studies: it is notoriously challenging to create a cranial window in mouse skull without disrupting or even removing the dura. It was unclear which meningeal layers were captured in the imaging plane. Did the visualized trigeminal afferents terminate in the dura, subarachnoid space, or pia (as suggested by Supplemental Fig 1, capturing a pial artery in the imaging plane)? Were z-stacks obtained, to maintain the imaging plane, or to follow visualized afferents when they migrated out of the imaging plane during meningeal deformations?<br /> 2) Findings here, from mice with chronic closed cranial windows, failed to fully replicate prior findings from rats with acute open cranial windows. While the species, differing levels of inflammation and intracranial pressure in these two preparations may contribute, as the authors suggested, the modality of measuring neuronal activity could also contribute to the discrepancy. In the present study, conclusions are based entirely on fluorescence signals from GCamp6s, whereas prior rat studies relied upon multiunit recordings/local field potentials from tungsten electrodes inserted in the trigeminal ganglion. As a family, GCamp6 fluorophores are strongly pH dependent, with decreased signal at acidic pH values (at matched Ca2+ concentration). CSD induces an impressive acidosis transient, at least in the brain parenchyma, so one wonders whether the suppression of activity reported in the wake of CSD (Figure 2) in fact reflects decreased sensitivity of the GCamp6 reporter, rather than decreased activity in the fibers. If intracellular pH in trigeminal afferent fibers acidifies in the wake of CSD, GCamp6s fluorescence may underestimate the actual neuronal activity.

critical info

It was a cold and rainy night. The birds had taken their journey far away and the evening gust set a somber mood over all the homes. Nothing moved in the night other than the sound of the wind rustling against the windows.

notice the for USER objects

typo - "which contain..."

[!NOTE] 如何加快克隆包含大量大文件的仓库？

flashcard

git lfs clone git lfs clone等待切换完成，然后把必需的Git LFS文件打包下载，这利用并行下载的优势，明显减少了HTTP请求的数量和进程的陡增（这尤其在Windows中作用明显）

Docker Desktop can run Wasm applications (binaries) instead of OS (Linux/Windows)

Explanation on how Wasm runs on servers

Not much new has happened in desktop computing since then.

x

run with bash

测试

style: Who is talking? People capote interviewed?

got this when running as root on ubuntu WSL while running code as /mnt/c/Users/Testr/AppData/Local/Programs/Microsoft\ VS\ Code/ ..

once su-d to user npx create-apprun-app my-app

Has this entire poem been the conversation of the speaker receiving a taro card reading?

Yes, energy efficiency is important, but there will always be heating needs, and Ewall isn't getting to the point.

x

Windows Terminal Preview

It sounds like there has been a lot of thought put into making passkeys reliable and easier to deal with than normal passwords. The fact that they can work across multiple servers and have two different keys made when you create one, makes one feel secure and confident about the success and reliability of passkeys.

Reviewer #3 (Public Review):

This study by Guan and co-workers focuses on a model neuronal lineage in the developing Drosophila nervous system, revealing interesting aspects about: a) the generation of supernumerary cells, later destined for apoptosis; and, b) new insights into the mechanisms that regulate this process. The two RNA-binding proteins, Imp and Syp, are shown to be expressed in temporally largely complementary patterns, their expression defining early vs later born neurons in this lineage, and thus also regulating the apoptotic elimination. Moreover, neuronal 'fate' transcription factors that are downstream of Imp and signatures of early-born neurons, can also be sufficient to convert later born cells to an earlier 'fate', including survival.

The authors provide solid evidence for most of their statements, including the temporal windows during which the early and the later-born motoneurons are generated by this model lineage, how this relates to patterns of cell death by apoptosis and that mis-expression of early-born transcription factors in later-born cells can be sufficient to block apoptosis (part of, and perhaps indicative of the late-born identity).

Other studies have previously outlined analogous, mutually antagonistic roles for Imp and Syp during nervous system development in Drosophila, in different parts and at different stages, with which the working model of this study aligns.

Overall, this study adds to and extends current working models and evidence on the developmental mechanisms that underlie temporal cell fate decisions.

Le maréchal Vendri a réclamé une modeste salle de réunion au bout du couloir de la salle du conseil endommagée par l'attaque du seigneur Soth. La pièce ne contient guère plus qu'une longue table couverte de rapports, mais de hautes fenêtres offrent une vue imprenable sur la ville et le port.

Le maréchal se tient près d'une fenêtre, le regard tourné vers l'horizon. Assis à la table, le seigneur Bakaris vous jette un coup d'œil, puis constate amèrement : "Les voilà maintenant, maréchal. Peut-être que si nous n'avions pas fait confiance aux épées, mon fils serait à mes côtés et le gouverneur serait encore en vie."

study involving an intermediate of indoor classroom setting with windows open. How much air movement and removal of airborne pathogens would really occur?

There was a whole page for every venue and all the speciality about the venue and have some well designed pictures of them.

All the information with photos make this page a perfect page for venue.

I wonder if the creeping women she sees from the window are simply John's sister. As the narrator's condition worsens, it seems she has also been tasked with keeping an eye on her as she later offers to sleep with her in her room. If this is true though, why does the narrator claim she can only see her from one window? Is she misleading as a false narrator or is there something else going on?

He mentions only mentions all things that keep her trapped and statonary.

It's a small detail only mentioned twice, but the barred windows make this room feel like a prison, which, for our protagonist, it effectively is.

Lorsque que l'on utilise ubuntu sur windows, les commandes citées indiquent seulement les dossier et fichier présent dans le lecteur linux et non le lecteur C: du pc. Comment faire?

Microsoft collects Windows diagnostic data to solve problems and to keep Windows up to date, secure, and operating properly. It also helps us improve Windows and related Microsoft products and services and, for customers who have turned on the Tailored experiences setting, to provide more relevant tips and recommendations to enhance Microsoft and third-party products and services for each customer’s needs.

Icacls is a Windows command-line utility that IT admins can use to change access control lists on files and folders. To find out what they are click here https://learn.microsoft.com/en-us/windows-server/administration/windows-commands/icacls . F is full control, M is modify, OI is Object Inherit, CI is Container Inherit

eLife assessment

This is an important study showing how dendritic plateau potentials can enable neurons to perform reliable 'binary' computations in the face of realistic spike time jitter in cortical networks. The authors make a surprising and novel claim that dendritic plateau potentials perform equally well in short integration windows of only 10 ms and detail a biophysical mechanism for how this effect may occur. While the authors make many good arguments, and the general concept underlying the paper is sound, the evidence as of now is incomplete, with some unsupported statements that should be more thoroughly defended in the manuscript.

make sure you haven't selected powershell ise x86 version!

Dickinson stated that she heard a fly buzz, then it appears to be a brief pause and she stated “when I died”. This is important because it sets the tone for what she wants the reader to know about her setting. It places emphasis on the fact that dying is an experience as well as living. The fact that she stated “when I died” allows us to ponder on whether there is life after death since she seems to be speaking from the other side. Most often death is sullen but in this poem we don’t get that sentiment because she is focusing on something that is typically so trivial, but in this context so significant. The fly buzzing interrupted “the stillness in the room” that everyone was experiencing as they simultaneously were waiting to experience the author passing away as she mentioned that they had basically passed the point of crying and were waiting for the last stage of their experience which was indicated as she mentioned that there was a “stumbling buzz between the light and me”. She was trying to focus on crossing over to the other side when the buzzing fly ‘interrupted’. The windows she spoke of I assume were her eyes and/or her ability to see the present moment as she ultimately died and the buzzing fly was at that moment no more of importance as she closes her final moment in her living body.

I think in this part of the poem she was trying to get the fly out of the house and when the fly was almost out the window closed it self.the light was so bright in her eye she couldn't see the fly anymore.

Here, I believe Dickinson illustrates the soul's and body's indifference after death. "With Blue - uncertain - stumbling Buzz" "Between the light - and me" depicts the narrator's soul/humanity struggling to maintain its will to survive as it is drawn towards the light and away from the body. On the other hand, the line "And then the windows failed- and then i could not see-" means that the body is succumbing to death as the "windows failed" indicates the body not having he strength to keep it's eyes open thus "I could not see see to see" meaning I can see physically and spiritually.

Hardie panels are as expensive as brick?

Contrast to how now they can only see their reflection out the window and are forced to look at a broken street lamp

a collapsed state.

DOI: 10.1111/bph.16245

Resource: XPP-Aut: X-Windows Phase Plane plus Auto (RRID:SCR_001996)

Curator: @scibot

SciCrunch record: RRID:SCR_001996

What is this?

Reviewer #3 (Public Review):

The study examines how different cell types in various regions of the mouse dorsal cortex respond to visuomotor integration and how antipsychotic drugs impacts these responses. Specifically, in contrast to most cell types, the authors found that activity in Layer 5 intratelencephalic neurons (Tlx3+) and Layer 6 neurons (Ntsr1+) differentiated between open loop and closed loop visuomotor conditions. Focussing on Layer 5 neurons, they found that the activity of these neurons also differentiated between negative and positive prediction errors during visuomotor integration. The authors further demonstrated that the antipsychotic drugs reduced the correlation of Layer 5 neuronal activity across regions of the cortex, and impaired the propagation of visuomotor mismatch responses (specifically, negative prediction errors) across Layer 5 neurons of the cortex, suggesting a decoupling of long-range cortical interactions.<br /> The data when taken as a whole demonstrate that visuomotor integration in deeper cortical layers is different than in superficial layers and is more susceptible to disruption by antipsychotics. Whilst it is already known that deep layers integrate information differently from superficial layers, this study provides more specific insight into these differences. Moreover, this study provides a first step into understanding the potential mechanism by which antipsychotics may exert their effect.<br /> Whilst the paper has several strengths, the robustness of its conclusions is limited by weaknesses in statistical analyses. A summary of the paper's strengths and weaknesses follow.

Strengths:

The authors perform an extensive investigation of how different cortical cell types (including Layer 2/3, 4 , 5, and 6 excitatory neurons, as well as PV, VIP, and SST inhibitory interneurons) in different cortical areas (including primary and secondary visual areas as well as motor and premotor areas), respond to visuomotor integration. This investigation provides strong support to the idea that deep layer neurons are indeed unique in their computational properties. This large data set will be of considerable interest to neuroscientists interested in cortical processing.<br /> The authors also provide several lines of evidence that visuomotor information is differentially integrated in deep vs. superficial layers. They show that this is true across experimental paradigms of visuomotor processing (open loop, closed loop, mismatch, drifting grating conditions) and experimental manipulations, with the demonstration that Layer 5 visuomotor integration is more sensitive to disruption by the antipsychotic drug clozapine, compared with cortex as a whole.

The study further uses multiple drugs (clozapine, aripiprazole and haloperidol) to bolster its conclusion that antipsychotic drugs disrupt correlated cortical activity in Layer 5 neurons, and further demonstrates that this disruption is specific to antipsychotics, as the psychostimulant amphetamine shows no such effect.

In widefield calcium imaging experiments, the authors effectively control for the impact of hemodynamic occlusions in their results, and try to minimize this impact using a crystal skull preparation, which performs better than traditional glass windows. Moreover, they examine key findings in widefield calcium imaging experiments with two-photon imaging.

Weaknesses:

A critical weakness of the paper is its statistical analysis and data representations. The study does not use mice as its independent unit for statistical comparisons but rather relies on other definitions (see authors' Tabe S1), without appropriate justification, which results in an inflation of sample sizes. For example, in Figure 2, the independent statistical unit is defined as sessions instead of mice, and in Figures 6 and 7 its pairs of cortical regions of interest. This greatly inflates N by at least 1-2 orders of magnitude compared to using N = number of mice. With such inflated sample sizes, it becomes more likely to find spurious differences between groups as significant.

It should be noted, however, that the authors have redone some analyses in their revision, specifically for Figure 1L, in which mice are used as independent units (shown in Figure S4) without any change in conclusion. However, this is not done for all other problematic figures in the manuscript.

Furthermore - and related to the previous comment - trace averages and SEMs across the figures of the manuscript come from hundreds to thousands of data points (e.g. locomotion onsets or cells) repeatedly measured from only a handful of mice. This can be visually misleading for the reader (even if statistics are not being formally performed on these traces) as it artificially reduces the size of the SEM masking the true variability (and size) of the effects portrayed in the paper. Again, this practice is only justified if the data (e.g. locomotion onsets) within a mouse is actually statistically independent, which the authors do not test for or justify.

It should be noted that the authors do show some trace averages and SEMs for a some of their data (Figure S2), in which N = individual mice, without any change in conclusion. However, this is not done for all other problematic figures in the manuscript.

The above statistical problems are apparent throughout the manuscript. The more disciplined approach would be to average the data within a mouse, and then use the mouse as an independent unit for statistical comparison and/or for the purposes of presenting means and SEMs for aggregate data. Alternatively, the authors should provide clear justification in the manuscript for opting for other definitions of N.

Finally, it is important to note that whilst the study demonstrates that antipsychotics may selectively impact visuomotor integration in L5 neurons, it does not show that this effect is necessary or sufficient for the action of antipsychotics; though this is likely beyond the scope of the study it is something for readers to keep in mind.