eLife Assessment

This important study provides evidence that Clock and fat body proteasome subunits contribute to dietary restriction-mediated lifespan extension in Drosophila. The evidence supporting these conclusions is solid, yet whether dietary restriction-induced daily rhythmicity of proteasomal genes is CLK-dependent remains incompletely demonstrated, and alternative explanations involving feeding differences and non-circadian functions of CLK cannot be, with the current dataset, excluded. The work will definitely be of interest to researchers studying biological rhythms, nutrition, and aging.

Reviewer #1 (Public review):

The studies by Hwangbo et al. diligently attempt to account for many of the typically neglected dietary and non-dietary factors.

Strengths:

• Work addresses many potential artifacts of dietary (e.g., dehydration stress, macronutrient ratios, and protein source) and non-dietary (e.g., leaky expression of S106-GAL4) manipulations-important factors that are too often overlooked.

• Balanced and complementary behavioral, molecular, and bioinformatic experiments

• Show necessity of proteostatic subunits in the fat body for DR-mediated longevity. The findings in the current manuscript lay the ground for future studies that test sufficiency of fat body prosβ3 and rpn7, or necessity of other proteostatic genes in other tissues.

Comments on revised version:

The revised manuscript is substantially improved and addresses many of the prior concerns. I have only a few minor recommendations and remaining issues:

Clarify the interpretation of the Con‑Ex feeding data. The authors describe the ~70% higher intake on 1SY in Clk^Jrk as modest, and note a ~40% higher mean intake on 5SY that is not statistically significant. However, lack of significance can reflect limited power, and these differences are potentially biologically meaningful, given that relatively small changes in nutrient ingestion can substantially affect lifespan. If the average effects are real, the Clk^Jrk flies would be ingesting an effective diet closer to ~1.7SY and ~7SY relative to controls. A shift of the diet-lifespan response curve in Clk^Jrk therefore cannot be fully excluded, particularly given the absence of intermediate diets between 1SY and 5SY and the observation that Clk^Jrk is sometimes shorter‑ and sometimes longer‑lived than controls across different trials and diets.

Although the core finding is strengthened by using several diet formulations, most additional experiments continue to rely on whole‑food dilution, even as the field is moving toward more defined DR regimens (e.g., yeast‑only or yeast‑extract-based protocols). There remains considerable variability and, in some cases, a lack of clear DR‑mediated lifespan extension in control cohorts (for example, in some GeneSwitch experiments using whole‑food dilution). It would be helpful if the authors briefly commented on this variability and justified their continued use of whole‑food dilution in these experiments.

Please add a clear Methods description of the feeding assay (Con‑Ex), including fly age, assay duration, dye or tracer conditions, sample processing, quantification, and statistical analysis.

Please ensure that the survival data shown in Figure 5 and associated supplements are explicitly linked to Cox proportional hazards analyses in the text or figure legends, with clear indication of the models used (e.g., gene, diet, and gene×diet interaction terms). The Methods state that diet is used as a continuous variable; given the non‑linear (U‑shaped) lifespan-diet reaction norm (reduced survival at both 1SY and higher yeast), it would be important to clarify whether 1SY was excluded from these Cox models, or alternatively, to model diet categorically, restrict the continuous analysis to 5-20SY, or apply an appropriate non‑linear transformation (e.g., splines). As written, it is not clear how the Cox model accommodates the non‑linear diet response.

Reviewer #2 (Public review):

Summary:

Dietary restriction (DR) increases lifespan, an effect that has been consistently observed in several organisms, but we still lack a clear mechanism to explain this phenomenon. In this work, Hwangbo et al. revisited the role of the circadian clock in DR-mediated lifespan effects. They found that the increase in lifespan produced by DR is missing on a clock mutant, a clock dependency that is also observed at the level of nutrient-dependent egg laying. By conducting RNA-seq with an impressive temporal resolution, they showed that DR triggers an increment in the number of cycling genes expressed in the fat body, the fly functional analog of the mammalian liver. Interestingly, from these genes, a group of them are de novo daily expressed genes, meaning that their expression was not rhythmic under the control diet but appear rhythmically expressed under DR. Among those, genes encoding proteasome subunits are enriched. The authors finally showed that adult-specific knockdown of these genes in the fat body prevents the increase in lifespan under DR, further supporting a role of the proteasome in this process. Overall, the conclusions are mostly supported by the evidence presented, and the authors' discussion nicely frame their results with other research in the field.

Strengths:

- Many studies have limited their observations of DR on lifespan to a few dietary conditions which makes the reach of some previous conclusions somewhat limited. The dilution strategy that the authors used in this work provides a strong indication that the effect of DR on lifespan relies on clock expression regardless of the conditions used. Furthermore, the inclusion of the egg-laying assay is a good addition to support this hypothesis.

- Because the strength of the rhythmicity statistics relies heavily on the number of data points collected, the temporal resolution used for the RNA-seq experiments (every 2 hrs per 48hrs) is remarkable. This allows exquisite dissection of the phase of rhythmic genes in different conditions. The dataset produced in this work might be of use to other groups interested in weighting the role of other represented gene clusters in DR.

Weaknesses:

I see only minor flaws in this work, that if addressed, might strengthen the authors' conclusions, particularly:

- The results of the lifespan assays are quite variable and in some instances contradictory (Fig. S8) across trials, possibly because there are other unaccounted variables we still do not understand. The fecundity assay, in contrast, seems to be a better readout (Fig. 2). Confirming at least the two genes picked for the study (Fig. 5) would be good support for the claim that the proteasome mediates the effects of DR.

- According to the model, the acute effect of DR on gene expression is related to CLOCK protein function. However, I am not sure how this link was established. It is tempting to assume that CLOCK upstream is the reason for having an increase in rhythmic genes under DR, but the experiments did not test this. The tests conducted either assessed the role of clk or the effect of an impaired proteasome on DR-dependent extension of lifespan. Thus, it is difficult to assert the authors' claims on the link between CLK and the changes in cycling genes and to the proteasome upon DR.

Comments on revised version:

In this new version, Hwangbo and colleagues add new data to support a role of the clock in the effects of DR on lifespan. While adding this new data helps to alleviate some of the concerns previously raised, I think there are still some gaps. Below are my main concerns:

ClkJrk transcriptomic data: Adding this data supports a role of the clock in daily rhythmicity of genes in the fat body, likely due to a circadian role. However, it does not show that de novo rhythmicity of proteasomal genes is clock-related since there is no ClkJrk DR dataset. Thus, there is still a possibility this is a pleiotropic effect. While redoing an entire RNA-seq dataset might not be feasible, a possible way to support the circadian claim would be to use proxy genes observed in the Ctrl vs DR conditions and compare them by qPCR in ClkJrk Ctrl vs DR.

Feeding data: Are the flies reared in DR conditions, or do they just start the DR at the beginning of the experiment? If so, is it possible the flies will show a different feeding pattern after consecutive days of DR affecting overall (5-10 days) food consumption?

Clk expression is important for non-circadian roles in the ovaries (Wang et al., Cell Mol Life Sci, 2025). Therefore, it is possible that the fecundity effect is at the low level of the ovary/egg development instead of integration and processing of DR. This might be a confounding effect when interpreting data in Fig 2 in ClkJrk as solely the effect of DR.

Line 215: Considering the discussion above, I'd rather change "circadian-dependent change" to "daily"<br /> Considering that the ClkJrk RNA-seq transcriptomic was generated, presumably, at a different date/time than the original transcriptomic data from Control vs DR, comparative metrics (seq depth, mapping rate, etc.) between these are needed.

How is the feeding analysis conducted? I believe the method information was not updated.

Author response:

The following is the authors’ response to the original reviews

Correction: In the process of revising the preprint, we discovered that for the 15SY dataset that a single time point (ZT2) out of the 12 timepoint series was inadvertently combined with temporally adjacent time samples (ZT20, 22, 24). We corrected the accompanying GEO submission (Series GSE145509). With this update, we repeated the rhythm analysis with an updated RAIN algorithm as the original Boot-eJTK could not be run as it was outdated with dependent packages no longer maintained or supported, and some are no longer available through standard package managers. The analysis with the corrected ZT2 sample and did not find any significant changes in the major claims of the papers. One minor change is that we no longer observe significant DR-dependent increases in proteasome gene levels. Nonetheless, we still find DR-dependent cycling of proteasome genes and proteasome module network connectivity consistent with the DR-sensitivity of the proteasome pathway. Figure 4 has been updated to reflect this change. After correcting this issue, we revised the manuscript in response to the reviewer comments.

eLife Assessment

This study describes important findings on how a core component of the circadian clock impacts the effect of dietary restriction (DR) on longevity and fecundity in Drosophila, which lead the authors to postulate rhythmic control of proteostasis in the fat body as a critical aspect of DR effects. The evidence presented is still incomplete, not fully supporting the conclusions of the study, as alternative hypotheses/explanations have not yet been systematically explored. The work will nevertheless be of substantial interest to researchers working in circadian and cell biology, metabolism, and aging, with an interesting hypothesis to be explored further.

We sincerely thank eLife for considering our manuscript and express our gratitude to all three reviewers for their time and constructive comments. While acknowledging that there are alternative hypotheses for some of our findings, which make our evidence incomplete, we appreciate that our manuscript was recognized as important and of substantial interest. We have revised the manuscript to acknowledge that the major findings under light-dark conditions could be attributable to being driven by light rather than the circadian clock. We add new data demonstrating that the far majority of cycling genes in LD in control flies are disrupted in Clk^Jrk consistent with circadian regulation (see also below). Future investigations are needed to test the alternative hypotheses and explanations. Nevertheless, we believe that the manuscript still represents a meaningful advancement in understanding how molecular circadian clocks in peripheral tissues interact with diet to influence systemic lifespan and aging.

Public Reviews:

Reviewer #1 (Public Review):

The studies by Hwangbo et al. diligently attempt to account for many of the typically neglected dietary and non-dietary factors.

Strengths:

- Work addresses many potential artifacts of dietary (e.g., dehydration stress, macronutrient ratios, and protein source) and non-dietary (e.g., leaky expression of S106-GAL4) manipulations-important factors that are too often overlooked.

- Balanced and complementary behavioral, molecular, and bioinformatic experiments

- Show necessity of proteostatic subunits in the fat body for DR-mediated longevity. The findings in the current manuscript lay the ground for future studies that test sufficiency of fat body prosβ3 and rpn7, or necessity of other proteostatic genes in other tissues.

Weaknesses:

- Could the lack of DR response in clock mutants across dietary concentrations be simply because the clock mutants are better at compensatory feeding adjustments to dietary dilutions? If this were the case, there are two major implications to the authors' conclusions:

a) The Clk mutants are differently responding to dietary dilutions, not to dietary restriction, per se.

b) Nutritional intake was unaffected by the dietary manipulations. If the changes in fat body proteostasis and lifespan were due to nourishment, it would be expected that the physiology and lifespan do not change.

Accurate measurements of food consumption and the resulting protein intake could potentially clarify this critical question.

We thank the reviewer for their positive feedback and also appreciate their raising the important issue of whether Clk^Jrk mutants may be more effective at compensatory feeding. Xu et al. (2008) reported that overall food consumption in Clk^Jrk flies was indistinguishable from control flies.

We also directly assessed food intake in ~1 week old flies on three diets (1% SY, 5% SY, 15% SY) over two days (48 hours) using the Con-Ex method (Shell et al. 2018). We observed either no significant changes or relatively modest changes in food consumption between iso31 controls and Clk^Jrk flies that are limited compared to the large differences in caloric content between the diets. While we cannot rule out changes in feeding patterns throughout the lifespan, we believe that minimal differential compensatory feeding in the Clk^Jrk mutants are not sufficient to be the primary cause of the lifespan differences observed across diets. These data are added as new Figure 1-figure supplement 3.

Reviewer #1 (Recommendations For The Authors):

Hard to find information:

- Type of yeast used. Should be addressed in the Methods section at least, instead of having to dig through several paragraphs into the Results section.

Missing information:

- Agar type and concentration

- Mifepristone diets: pipetted on top or mixed into food?

The relevant information has been updated in the Materials and Methods

Wrong information:

- Line #159-160: "Fig. 1 and Fig. S3" should be "Fig. 1 and Fig. S2" and then "(Fig. S3)" added to the end of the sentence.

This information has been corrected in the revision

Presentation:

- Paragraphs are very long.

- Interactions should be denoted by ×, not * or x.

- Remove markers from mortality graphs. Having bulky markers reduces perceived differences between curves.

These changes have been updated in the revised version.

Reviewer #2 (Public Review):

Dietary restriction (DR) increases lifespan, an effect that has been consistently observed in several organisms, but we still lack a clear mechanism to explain this phenomenon. In this work, Hwangbo et al. revisited the role of the circadian clock in DR-mediated lifespan effects. They found that the increase in lifespan produced by DR is missing on a clock mutant, a clock dependency that is also observed at the level of nutrient-dependent egg laying. By conducting RNA-seq with an impressive temporal resolution, they showed that DR triggers an increment in the number of cycling genes expressed in the fat body, the fly functional analog of the mammalian liver. Interestingly, from these genes, a group of them are de novo daily expressed genes, meaning that their expression was not rhythmic under the control diet but appear rhythmically expressed under DR. Among those, genes encoding proteasome subunits are enriched. The authors finally showed that adult-specific knockdown of these genes in the fat body prevents the increase in lifespan under DR, further supporting a role of the proteasome in this process. Overall, the conclusions are mostly supported by the evidence presented, and the authors' discussion nicely frame their results with other research in the field.

Strengths:

- Many studies have limited their observations of DR on lifespan to a few dietary conditions which makes the reach of some previous conclusions somewhat limited. The dilution strategy that the authors used in this work provides a strong indication that the effect of DR on lifespan relies on clock expression regardless of the conditions used. Furthermore, the inclusion of the egg-laying assay is a good addition to support this hypothesis.

- Because the strength of the rhythmicity statistics relies heavily on the number of data points collected, the temporal resolution used for the RNA-seq experiments (every 2 hrs per 48hrs) is remarkable. This allows exquisite dissection of the phase of rhythmic genes in different conditions. The dataset produced in this work might be of use to other groups interested in weighting the role of other represented gene clusters in DR.

We are grateful for the reviewer’s positive feedback regarding the robust experimental design in the manuscript.

Weaknesses:

I see only minor flaws in this work, that if addressed, might strengthen the authors' conclusions, particularly:

- The results of the lifespan assays are quite variable and in some instances contradictory (Fig. S8) across trials, possibly because there are other unaccounted variables we still do not understand. The fecundity assay, in contrast, seems to be a better readout (Fig. 2). Confirming at least the two genes picked for the study (Fig. 5) would be good support for the claim that the proteasome mediates the effects of DR.

We appreciate the reviewer’s comment of seeing “only minor flaws”. We reiterate that we focused on those results which were replicated across trials, providing confidence in the overall conclusions. Nonetheless, we agree that exploring the proteasome role on the DR effect on fecundity would be intriguing and may complement the lifespan data. We now acknowledge this point in our discussion.

- According to the model, the acute effect of DR on gene expression is related to CLOCK protein function. However, I am not sure how this link was established. It is tempting to assume that CLOCK upstream is the reason for having an increase in rhythmic genes under DR, but the experiments did not test this. The tests conducted either assessed the role of clk or the effect of an impaired proteasome on DR-dependent extension of lifespan. Thus, it is difficult to assert the authors' claims on the link between CLK and the changes in cycling genes and to the proteasome upon DR.

We have updated our text and model figure suggesting a direct CLK role in the Discussion.

Reviewer #2 (Recommendations For The Authors):

As mentioned before, the experiments, in particular the RNA-seq datasets are excellent. Additionally, the discussion provides a good overview of other relevant papers on DR, and the conclusions are mostly supported by the data. Here I provide a couple of suggestions that I believe might improve this work:…

- Although the model is simple and understandable (Fig. 6), the inclusion of an overall summary or explanation in the figure legend would be appreciated, especially for readers that are not familiar with the terminology.

- It might be the formatting while parsing the files but some of the in-text citations are between curly brackets (e.g., lines 80, 92, 93).

- By definition, and unlike Canton-S or Oregon-R strains, w1118 flies are not wild-type but a genetic control. I believe that reference to this on the figures and text may need correction.

We updated and/or our corrected each of these in the revised version. For wild-type, we more explicitly define this as wild-type for the relevant genetic locus.

Reviewer #3 (Public Review):

In this study, Hwangbo and co-workers investigate the extent to which the well-established life extending effects of DR rely on the molecular circadian clock and how the landscape of clock-controlled gene expression changes in the face of DR within the fat body of the fly, a tissue that performs the functions associate with both the liver and adipose tissue of mammals. The authors evidence that DR extends lifespan in a manner that depends on only one of the two major limbs of the fly's molecular circadian clock, namely the positive limb, that DR produces major changes in the identities of cycling clock output genes, and that genes related to the proteosome represent a major component of DR-induced transcript cycling. Though interesting, these conclusions are not strongly supported by the data and there are two major reasons for this. First, the authors rely on only one loss of function genotype each for the loss of positive and negative limb clock gene function. Second, though they wish to address the "circadian transcriptome" under normal and DR conditions, the authors conduct all their work under strong Light/Dark cycles, making it impossible to address circadian phenomena. These shortcomings are problematic in the extreme, as they leave open obvious alternative explanations for the results and fail to directly determine if the rhythmic expression, they observe are clock controlled or merely driven by the light/dark cycles, which themselves produce major effects on activity, feeding, etc., that may be responsible for differentially driving rhythmic transcripts under normal and DR conditions in the fat bodies.

Major Weakness One: The use of only genotype each for the loss of positive (Clk^JRK) and negative (Per^01) limb of the circadian represents a major challenge for a central conclusion of the study. Phenotypes caused by the loss of a single clock gene may be due to the loss of circadian timekeeping, or they may represent a pleiotropic effect of the loss of function mutant being used. There are multiple precedents for pleiotropic (non-circadian) effects of clock gene mutants. It is, therefore, possible that the differences in the extent of DR mediated life extension between Clk^JRK and Per^01 may not represent a difference between breaking the positive and negative limbs of the clock but may simply reflect a pleiotropic effect of the dominant negative Clk^JRK. This possibility is acknowledged by the authors (lines 343-344). This could be addressed quite easily by extending the analysis to other loss of function mutants, for example, tim01 for the negative limb and cyc01 for the positive. Given the central focus here on the "circadian transcriptome," leaving open this alternative explanation for Clk's role in DR induced life extension represents a major weakness of the study. Furthermore, given the fact that Clk^JRK appears to be short lived on most of the media tested in the study, is it really surprising or informative that they would display lower life extension under DR?

We confirmed that the large majority of LD oscillating genes in wild-type controls are disrupted in ClkJrk consistent with circadian clock regulation (Figure 3-figure supplement 1). As noted, we formally acknowledged that the circadian clock mutant alleles used here, and in fact any circadian clock alleles, can have pleiotropic, i.e., non-circadian, clock effects. This would only be partially mitigated by adding more (but also potentially pleiotropic) clock mutant alleles. Very challenging circadian resonance experiments (see Xu et al, 2019) are the gold standard for resolving circadian clock v. non-clock effects which are beyond the scope of this study which we now add to our discussion.

We also note that foxo mutants are both short-lived and exhibit a robust lifespan extension to dietary restriction and thus the ClkJrk mutant is distinct in this regard. We have added this point to the Discussion.

Major Weakness Two: The authors have not established that any of cycling transcripts they have detected in the fat body under normal and DR conditions are driven by the circadian clock. This is because: 1.) they have conducted their transcriptomic analysis on cells taken from flies entrained to light dark cycles, which can themselves drive daily changes in expression levels and 2.) they have not shown that the cycling measured on normal diet or DR conditions depends on a functional circadian clock. The "significant reorganization of the circadian transcriptome" is presented as a major conclusion of this study, but the authors have not addressed circadian control of transcription at all here, either by an examination of transcription under free-running conditions and/or in loss of function clock mutants.

In addition, there is a logical gap in this study. The authors have shown that DR produces less life extension in Clk^JRK mutants than Per^01 or wild-type controls. They then show that DR produces changes in the rhythmic transcriptome when flies are place on DR. The central model presented in Fig. 6 shows/concludes that CLK drives increases in proteome-related transcript rhythms under DR. This conclusion could have been directly tested by asking if the changes in rhythmic gene expression induced by DR are gone the loss of function Clk mutants, or if the transcriptomic landscapes fail to differ between feeding conditions in these mutants.

In conclusion, the study falls far short of directly testing the ideas it puts forth, greatly limiting its impact and interest.

As noted above, we also examined the diurnal transcriptome in ClkJrk (at 4 hour resolution) and found that of the 290 genes that were detectably rhythmic in wild-type just 13 were rhythmic in ClkJrk consistent with the notion that oscillations depend on Clk (Figure 3-figure supplement 1). We now add this analysis to the manuscript. Nonetheless, we cannot exclude a role for light and thus have opted to use “diurnal” in place of “circadian” where appropriate for observed rhythms under LD conditions.

Reviewer #3 (Recommendations For The Authors):

Line 140 "showed an almost identical response" was a little hard to understand at first. Consider clarifying.

This has been rephrased for clarity

The authors claim that Clk mutants are "much longer lived" than wild-type controls on two of the relatively low calorie diets. Figure S3C certainly argues otherwise, and it's not clear how the data in 1C and S2C and warrant the use of "much" here.

This wording has been rephrased and corrected in the revised version. The low-calorie diet shown in Figure 1-figure supplement 3 contains a higher sucrose concentration (5%) than those used in Figure 1 and Figure 1-figure supplement 2 (1%). This observation suggests that sucrose may play an independent role in the survival of ClkJrk mutants under malnutrition conditions.

The authors should provide the rationale for the use of a dominant negative form of Clk for their experiments. Would the available amorphic allele be a better choice?

As ClkJrk is the first described Clk allele and it is probably the most well characterized. As a dominant negative version which is still capable of dimerizing and binding DNA it is less susceptible to compensation by redundant bHLH transcription factors as has been observed for between mouse Clock and NPAS2 (Debruyne et al, 2006).

It is not clear why the authors have chosen to examine transcriptomes so soon after transfer to DR. Why not wait longer. The authors provide context that changes are already taking place at the early time-point used, but would waiting a bit provide a more robust indication of how DR is changing the fat body?

We noted in the manuscript that the effects of DR on survival are evident relatively soon (~2d) after a diet shift. We were interested in identifying those changes in daily transcription that would be occurring during that early time span and potentially be a cause rather than an effect of survival changes.

eLife Assessment

This valuable study provides a detailed three-dimensional characterization of primary cilia organization in the postnatal mouse growth plate, revealing reproducible spatial patterns in ciliation, ciliary length, and orientation. The evidence supporting these descriptive findings is solid, based on high-quality quantitative imaging and complementary genetic, mechanical, and transcriptomic approaches. However, evidence for the broader mechanistic conclusions is incomplete, particularly regarding whether ciliary orientation is uncoupled from basal-body and cell orientation and whether its stability under altered mechanical loading reflects a cell-intrinsic program. The study provides a helpful foundation for understanding primary cilia organization and mechanobiology in the growing skeleton, while the mechanisms underlying these observations remain to be established.

Reviewer #1 (Public review):

Summary:

This manuscript provides fundamental insight into ciliary biology, specifically, how ciliary axoneme orientation is governed by microenvironmental bending or intrinsic cytoskeletal steering rather than strictly basal body docking coordinates.

Strengths:

There are three major strengths in this manuscript. First, combining high-resolution imaging with deep-tissue sectioning yields impressive lateral resolution, enabling robust separation of dual centrioles within the crowded chondrocyte extracellular matrix. Second, the authors established an automated pipeline that evaluates thousands of individual cells across multiple anatomical regions and differentiation zones, lending strong statistical weight to positional and volumetric measurements. Lastly, they demonstrate that ciliation peaks in the peripheral resting zone and ciliary length peaks in hypertrophic cells, providing a compelling cellular explanation for why Ift88 deletion impacts peripheral growth plate geometry and hypertrophic expansion.

Weaknesses:

There are three major weaknesses in this manuscript. First, the paper lacks explicit descriptions of data mentioned in the Abstract and Methods, including the RNA-seq differential expression, WGCNA modules, and immobilization/ciliary alignment data. Second, while the Methods section mentions correcting for "Z-blur" / point-spread function distortion in 3D spherical coordinate calculations, further detail is required on how orientations are disambiguated from optical sectioning depth artifacts. Lastly, the RNA-seq analysis demonstrates that ambulatory unloading alters hedgehog and primary cilia gene signatures, yet axoneme orientation itself remains static. The narrative requires a clearer mechanistic synthesis regarding how mechanical loading modulates ciliary signaling if physical alignment of the cilia is refractory to mechanical force in chondrocytes.

Reviewer #2 (Public review):

Summary:

The aim of this work was to characterise in detail the cellular organisation of the growth plate in the growing mouse limb, the effect of mechanical loading due to physical activity, and the role of primary cilia in mediating this effect as putative mechanosensors. Primary cilia are generally thought to function as mechanosensors in a range of different organs and tissues, e.g. in the kidney. Exposure to mechanical loading is a normal part of post-natal limb growth, and the central hypothesis underlying this work was that primary cilia will play an important role in transducing the effects of mechanical loading into cellular responses during this process. To investigate this, the authors used a mouse model encoding fluorescent markers for primary cilia, and also allowing conditional knockout of IFT88, a protein that plays a key role in primary cilium biogenesis. The authors compared the effects of mechanical loading by comparing tissue from mice with normal or surgically immobilised limbs. To investigate the role of primary cilia, the authors performed the same experiments with mice treated with tamoxifen to induce IFT88 knockout.

Strengths:

A major strength of this work is that it studied primary cilia in a fully in vivo system. The authors employed cutting-edge imaging approaches and image analysis pipelines to rigorously study cellular organisation, centriole positioning, cilium length, and orientation across thousands of cells in limbs from 18 animals. The imaging results presented are of an exceptionally high-quality. The analysis of imaging data is extremely quantitative and robust and utilised appropriate statistical analyses, which were clearly stated throughout. The imaging data were complemented with transcriptomic data and analyses, which provided an orthogonal dimension for understanding cellular responses. An interesting and unexpected outcome from this is that the primary cilia are oriented in the same direction, and inclined at a roughly 45{degree sign} angle to the mediolateral and proximal-distal axes of the limb. The authors speculate as to how this might arise and the role it may play in sensing.

Weaknesses:

Overall, the work reported in this study was of a very high quality, and I could not find any significant shortcomings. However, I did feel that the paper was not very well written in many places, which made it difficult to read.

Overall, I think that the authors did achieve their aims in this study. It will be interesting to unpick the cellular mechanisms that lead to alignment of the cilia, the role of this alignment in mechanosensing, and the molecular mechanisms by which the cilia sense mechanical strain. Thus, this work provides a fertile ground for future studies, which will have important consequences for the study of primary cilia in vivo.

Reviewer #3 (Public review):

Summary:

This study aims to characterize the three-dimensional organization of primary cilia in the postnatal mouse growth plate and to determine how ciliary prevalence, length, position, and orientation vary across anatomical regions and stages of chondrocyte differentiation. The authors combine volumetric imaging and automated image analysis with conditional disruption of IFT88, limb immobilization, and bulk transcriptomics. The work provides a valuable anatomical dataset and identifies several interesting spatial patterns, including increased ciliation in the lateral resting zone, longer cilia in hypertrophic chondrocytes, and, most notably, a non-random orientation of ciliary projections despite a much broader distribution of centriole positions.

The descriptive evidence is largely solid, and the imaging dataset should be useful to researchers studying primary cilia, skeletal development, and tissue mechanobiology. However, the evidence is incomplete for several of the central mechanistic conclusions. In particular, the current analyses do not establish that ciliary orientation is uncoupled from basal-body position, and they leave unresolved how ciliary orientation relates to cell orientation. The immobilization experiment also supports a narrower conclusion than the proposed cell-intrinsic orientation program, while the final model linking ciliary angle to multidirectional signal integration remains speculative. Overall, the authors succeed in identifying an interesting and reproducible anatomical pattern, but the mechanism underlying that pattern remains largely open.

Strengths:

A major strength is the scale and anatomical context of the imaging. Quantifying thousands of cilia and tens of thousands of centrioles in three dimensions while preserving information about growth-plate zone and position across the limb is technically demanding. This allows the authors to identify regional differences that would be lost in dissociated cells or bulk tissue measurements.

The study also provides several potentially useful observations. Ciliation is higher in the lateral periphery, particularly in the resting zone, cilia are longer in hypertrophic chondrocytes, and ciliary projections show a reproducible non-random orientation. The latter is the most interesting result of the study and provides a useful foundation for asking how organelle orientation is established within a developing tissue.

I also appreciated that the authors place these observations in several biological contexts rather than stopping at a descriptive atlas. IFT88 disruption alters ciliation and growth-plate cell organization, while immobilization changes growth-plate dimensions, cell morphology and orientation, and the transcriptome even though the measured ciliary properties remain comparatively stable. These perturbations give the anatomical observations useful biological context.

Weaknesses

The main concern is that the claim that ciliary orientation is uncoupled from basal-body position is not directly demonstrated. The manuscript shows that centriole positions are broadly distributed and, separately, that ciliary projections have a preferred tissue-level orientation. Different population-level distributions, however, do not establish independence within individual cells. For example, basal bodies could be broadly distributed while their position determines which of two opposite directions along a common tissue axis the cilium adopts. This possibility is particularly relevant because Centrin-2 labels both centrioles, whereas only one serves as the basal body. The current data therefore support a difference between the population distributions of position and orientation, but not yet the stronger claim that the two are uncoupled.

A related gap is the relationship between cell orientation and ciliary orientation. The manuscript measures both, and cell orientation changes with growth-plate region, IFT88 deletion, and immobilization, while ciliary orientation appears relatively stable. Yet the two measurements are never directly related within the same cells. It is therefore unclear whether cilia adopt a reproducible angle relative to the major axis of their own cell, whether this relationship changes between the center and periphery, or whether changes in cell organization can occur independently of local ciliary alignment. This seems important for interpreting the tissue-level orientation pattern.

The statistical treatment of orientation also deserves caution. These are circular or spherical measurements, yet much of the analysis relies on linear distributions and Kolmogorov-Smirnov tests. This is particularly problematic around the 0{degree sign}/360{degree sign} boundary, where values near 350{degree sign} and 30{degree sign} are geometrically close but appear separated in a linear representation. The manuscript also does not clearly distinguish between a preferred axis, where opposite directions are equivalent, and a preferred polarity, where one direction is favored. In addition, thousands of cilia are nested within a much smaller number of mice, so the apparent statistical power should not be driven primarily by pooled object counts. Given that regional differences are a central theme of the paper, it would also be useful to know more clearly whether the preferred axis or the strength of the orientation bias differs between the middle and lateral growth plate at the animal level. I also could not identify a formal comparison of the centriole or ciliary distributions with an appropriate uniform circular or spherical null. The reported tests mainly compare zones and regions, so the claims that centriole position is non-preferential and ciliary orientation is non-random are not yet statistically established in the form presented.

The IFT88 conditional knockout is not carried through to the principal orientation question. The authors examine cell size, cell-axis organization, ciliation, and cilium length, but do not report whether the remaining cilia retain the preferred orientation or whether centriole positioning changes. Given the central role of this genetic perturbation in the manuscript, this leaves the genetic and orientation arms of the study somewhat disconnected.

Finally, the immobilization experiment and the mechanistic interpretation should be separated more carefully. Only four animals were analyzed in the offloaded and contralateral conditions, and medial and lateral regions were averaged because of the small sample size. The experiment shows that the established ciliary orientation remains relatively stable over a two-week postnatal interval despite clear changes elsewhere in the tissue. It does not exclude a role for mechanical forces earlier in establishing the axis, nor does a nonsignificant difference with four animals demonstrate equivalence. Similarly, the proposed cell-intrinsic orientation program and the final "single-axis blindness" model are interesting hypotheses, but the study does not yet identify what establishes the axis or how the observed ciliary angle would alter sensitivity to a force or biochemical gradient. Those ideas are worth discussing, but they should remain clearly separated from the observations directly supported by the data.

Author response:

In response to the valuable reviewers’ comments and suggested changes, we are finalising changes to the manuscript, in order to resubmit a revised version, and a document with full author responses, that reflects all the review comments.

These changes include adding points of clarity, improving accuracy on wording of key messages, adding additional interpretation of the data, including additional data and analyses that reflect open questions raised, and more discussion concerning these unanswered questions, which are subjects of future work.

In response to specific points we were asked to provisionally address (actions in italics):

Reviewer #1. We are pleased the reviewer sees the insight these data bring. We indeed think it likely that cilia axoneme orientation is governed by the immediate microenvironment and/or changes to the cytoskeleton and are actively looking to explore this.

To address the 3 areas of concern:

(1) Our re-writing of the abstract and the results concerning transcriptomic data seeks to overcome weaknesses in descriptions of these data.

(2) Further detail is being added on how z-distortion is corrected for, so accuracy is the same in all axis and orientation measurements are robust.

(3) We will add to the discussion to add our thoughts as to why ciliary and cilia signalling genes are regulated by immobilisation, but that immobilisation does not apparently affect cilia structure.

Reviewer #2. Thank you for such broadly positive comments, we are pleased the scale and depth of the quantitative analyses comes across, but will make sure that revisions throughout improve the quality of the writing describing these. We are actively exploring means to test ideas for how cilia axoneme become orientated in this way and what the function is. These preliminary ideas will be reflected in the discussion.

Reviewer #3. Thank you for such a detailed and thoughtful review. They will ensure the data are presented to their very best and we will address the concerns raised. This descriptive study had a hypothesis, and made discoveries which surprised us, we have tried, as you say, to put this in some context of the role of cilia and the role of mechanical forces in GP biology. Most notably we are considering that uncoupled is not the correct term here. To address areas of concern;

(1) We agree ‘uncoupled’ is not the correct word here. We cannot find a pattern of correlation between centriole position and orientation. However, the two can’t be ‘uncoupled’ and there is no proven independence on a single-cell level (that cilia position and cilia orientation are not in any way mechanistically linked). We will make changes and add more details on what we have considered in this area.

(2) Similarly, we have not correlated in each cell, cellular orientation and ciliary orientation, though we have made attempts and not yet found a relationship. However, again, this is not the same as one being absent. We might have expected ciliary orientation to change as cell orientation does (through zones or with pertubations) as we have seen in vitro but this remains to be fully explored and is one subject of follow-up work. We are considering column populations and per animal considerations of the data.

(3) We did take a cautious approach to statistics and specialist advice, but advice was not to overcomplicate things when there are 2 main messages related to centriole position and cilia orientation. Firstly, centriolar position appears random or without preference thus distribution of position on cell, is homogenous. Second, ciliary orientation angle is not a homogenous distribution, as would be expected if random with this number of measurements. We do, and will add comments to this effect, have to mindful of large dataset, but do not think this means we are looking at false discoveries due to number of comparisons. We are considering how better to reflect this. To address these important points we will add a section to the methods and discussion and will endeavour to change results to this end as it is a central point of the manuscript.

(4) We will add new data concerning IFT88cKO and centriole position and orientation.

(5) We will add a critique of the immobilisation experiments to ensure the relatively diminished power is clear. We agree force may have set things up initially, we will ensure this is discussed and we will ensure our proposal for why cilia orientation is this way is framed as a hypothesis. We have preliminary data, but this is the subject of an entire new project thus not yet supported by robust experimental evidence so is speculative at this stage and we will ensure this is clear.

eLife Assessment

This Review Article addresses a significant and timely topic that concerns the role of cryo-EM in transforming RNA structural biology from static, individual conformations toward the reconstruction of dynamic conformational ensembles and energy landscapes. The authors describe eight case studies spanning different RNA classes and argue that conformational heterogeneity is a source of mechanistic information, rather than a limitation.

Reviewer #1 (Public review):

Summary:

The review addresses an important and timely topic that concerns the role of cryo-EM in transforming RNA structural biology from a "static" discipline to one increasingly concerned with conformational ensembles and molecular dynamics. The scope is well within the eLife standards, and the style and general architecture do fit eLife.

Strengths:

The review is extremely well written, well-conceived and clear. The main strengths are in the breadth of coverage, the clear theme, and the inclusion of practical examples that explain in detail the construct design, sample preparation, vitrification, and data analysis. The manuscript will certainly be impactful and valuable, especially for readers who are not specialists in cryo-EM, as it provides an accessible overview of recent advances across a wide range of RNA systems.

Weaknesses:

My only reservation is that, currently, the review reads too much like a list of examples. The authors should make an effort, and I am sure they are well up to it, to try to synthesise the message, provide more critical insights and amalgamate the text better, to really reach a wider audience.

If revised along the lines detailed below, I am sure that the review will become an authoritative and influential resource for the RNA structural biology community.

Reviewer #2 (Public review):

Summary:

In this review, the authors set out to synthesize how cryo-EM is reshaping RNA structural biology, moving the field from the determination of static, individual conformations toward the reconstruction of dynamic conformational ensembles and energy landscapes. Through eight case studies spanning ribozymes, riboswitches, viral RNAs, and synthetic RNA assemblies, they aim to show how cryo-EM has revealed mechanisms of RNA motion, including folding, ligand-dependent switching, and cooperative assembly, and to provide a practical account of the experimental and computational challenges (construct design, sample preparation, vitrification, data analysis) that are specific to dynamic RNA targets.

Strengths:

The manuscript succeeds in bringing together a wide and genuinely current range of case studies illustrating the field's shift toward dynamics-focused cryo-EM, several published within the last one to two years. The dedicated "Challenges" section, which walks through construct engineering, buffer and vitrification optimization, grid screening, and heterogeneity-resolving computational approaches, is a particularly useful and practical contribution; it goes beyond simply cataloguing structures and gives readers new to the area a genuine methodological roadmap. The figures are detailed and well matched to the quantitative claims made in the text (helix rotations, distances, RMSDs), which strengthens the paper's value as a reference resource.

Weaknesses:

The coverage of two areas in particular, the SL5 viral RNA element and RNA quaternary/multimeric assemblies, would benefit from incorporating additional recent primary literature that is directly relevant but currently omitted. This does not undermine the manuscript's core narrative, but it means the review is presently less complete than it could be as a field synthesis, particularly for readers using it to identify the full body of recent work on these specific RNA classes.

More concerning is that one specific structural claim, describing conformer heterogeneity in the cobalamin riboswitch (Case study 3, holo dimer 4), appears to invert the finding reported in its own source paper (Ding, Deme et al., 2023). As written, the manuscript states that P2 and the distal half of P6 are structured in dimer 4, whereas the source paper reports that these are the regions that could not be modeled. This is worth flagging prominently because it is a factual claim about a specific structure, not an interpretive point, and readers relying on this review as a secondary source could come away with an inverted understanding of that structure's flexibility.

The manuscript also contains several minor internal inconsistencies. None of these individually threatens the paper's core arguments, but together they suggest the manuscript would benefit from a careful proofreading and reference-list audit pass.

Overall, the authors largely achieve their stated aim. The case studies convincingly illustrate that cryo-EM can now resolve discrete and continuous conformational states of RNA at near-atomic resolution, and the Challenges section substantiates the claim that construct design, sample preparation, and computational innovations have been jointly responsible for this progress. The gaps in coverage of SL5 and multimeric RNA literature, and the inverted claim in Case study 3, are the main respects in which the manuscript falls short of being a fully comprehensive and accurate synthesis at this stage, but these are correctable issues rather than flaws in the overall argument or framework.

This review is likely to be a useful entry point and practical reference for researchers moving into RNA cryo-EM, particularly given the level of methodological detail in the Challenges section. Its impact would be strengthened by two additions. First, a short discussion of recent cryo-EM advances in tRNA would round out the manuscript's coverage of classical RNA structural targets alongside the ribozyme, riboswitch, and viral RNA case studies already included. Second, the raiA non-coding RNA, currently mentioned only briefly, has in the last two years become a genuine model system for cryo-EM-based ncRNA structure determination, progressing from a single novel-fold discovery to a comparative structural framework spanning multiple raiA subtypes and candidate protein partners. Expanding this into a full case study would let the manuscript showcase, in a single worked example, exactly the kind of field-level progression (from novel fold to scaffold-based strategy and further to comparative structural framework) that the review's own framing describes as the trajectory of the field as a whole.

Reviewer #3 (Public review):

This review describes how cryo-EM is moving RNA structural biology from the determination of static structures toward the characterization of conformational ensembles. Eight case studies spanning self-splicing introns, riboswitches, viral RNA elements and synthetic assemblies show how cryo-EM has captured folding intermediates, hinge-mediated domain motions and ligand-dependent switching, and the authors pair these examples with practical guidance on construct design, sample preparation, vitrification and heterogeneity analysis. The argument that conformational heterogeneity is a source of mechanistic information, rather than a limitation, is well supported, and Table 1 will be a useful reference for laboratories entering the field. The manuscript is timely and of broad interest, and I recommend publication after minor revision.

(1) Scaffold-based structure determination (page 14, lines 5-9). The section on chimeric RNAs presents the scaffold strategies as a single group and cites Haack et al. (2025) and Langeberg & Kieft (2023) in one parenthetical, so individual results are not attributed to their sources. The distinction between these studies is substantive. Earlier scaffolds based on the Tetrahymena group I intron (Langeberg & Kieft, 2023) or on RNA origami (Sampedro Vallina et al., Nucleic Acids Res. 51:4613-4624, 2023) resolved the appended RNAs at approximately 4.4-5 Å. Haack et al. (2025) reported the first RNA scaffold to yield a high-resolution structure of the target itself, resolving the ligand-binding pocket of the thiamine pyrophosphate (TPP) riboswitch at 2.5 Å and extending nucleotide-level cryo-EM analysis to small RNAs that had previously been intractable. I ask the authors to attribute each result to its source and to state explicitly that Haack et al. (2025) achieved the first high-resolution structure of a scaffolded target RNA. Because the same study captured the ligand-free TPP riboswitch in an open, Y-shaped conformation, a direct example of the ligand-dependent switching that is central to this review, the TPP riboswitch should also appear in the riboswitch section (page 8, lines 15-36), together with the fluoride riboswitch of Langeberg & Kieft (2023).

(2) Page 6, lines 8-14. Self-splicing introns are described collectively as evolutionary ancestors of the spliceosome that reside in pre-mRNA transcripts. The proposed ancestral relationship to spliceosomal introns and snRNAs applies to group II introns. Group I introns, including the Tetrahymena intron, which interrupts a pre-rRNA, initiate splicing with an exogenous guanosine and are not considered spliceosomal precursors.

eLife Assessment

This is an important study of critical period plasticity, focused on temperature manipulations, and how different parts of the Drosophila larval motor circuit adapt or maladapt. The work convincingly demonstrates that components of the motor network respond in distinct ways to heat shocks, and the combination of functional, structural, and electrophysiological approaches makes the study of significant interest. The work points to central interneurons as primary drivers of maladaptive changes, while motoneurons and neuromuscular junctions show compensatory or homeostatic adjustments. The authors have thoughtfully and substantially addressed the concerns raised in the original reviews through new temporal-window experiments, additional receptor and behavioral analyses, clearer discussion of the study's limitations, and an integrative model.

Reviewer #1 (Public review):

Summary:

The authors examine the impact of heat stress during an embryonic CP in Drosophila, focusing on the larval locomotor network. They show that elevated temperature increases neuronal activity and, when applied during the CP, results in long-term instability of the network which manifests in prolonged seizure recovery times. At the neuromuscular junction, substantial structural changes occur, including terminal overgrowth and altered receptor composition, yet synaptic transmission remains preserved due to homeostatic regulation. Motoneurons display reduced excitability but receive increased synaptic input from premotor interneurons. These findings suggest that maladaptive instability originates within the central circuitry rather than at the neuromuscular junction, where changes seem to be homeostatically compensated. The study concludes that different network components exhibit distinct and hierarchical responses to CP perturbations, with premotor interneurons setting the tone for downstream adjustments in motoneurons.

Strengths:

The work takes advantage of the unique accessibility of the Drosophila system. A major strength of the study is the integration of structural, physiological, and behavioral analyses, which allows the authors to draw a comprehensive picture of how CP perturbations shape the locomotor network. The choice of an ecologically relevant stimulus (heat stress) is particularly convincing, as it links experimental manipulations more closely to natural environmental conditions. The experiments are carefully designed, and the results are robust and consistent with previous findings in the field, while also extending them in new directions. Importantly the work clarifies how temperature perturbations within distinct developmental time windows affect different properties of motor circuit formation.

Weaknesses:

A small limitation of the study is that it remains difficult to integrate maladaptive (seizure recovery) and adaptive/homeostatic phenotypes within a single mechanistic framework, leaving some space for interpretation.

Comments on revised version.

I think the authors did a great job at revising the manuscript and they addressed all my comments.

Reviewer #2 (Public review):

Summary:

This manuscript presents a thoughtful and well-executed study of critical period plasticity in the Drosophila larval motor circuit. The authors examined how transient heat, 32C, during embryonic stage, altered network properties, showing that premotor interneurons A27h increase excitatory drive onto motoneurons, which respond with a reduction in excitability. At the NMJ, synaptic terminals expand and GluRIIA distribution shifts, yet synaptic transmission remains largely unaffected. Despite these local compensations, the treated larvae display slower crawling and prolonged recovery from seizures, indicating that the network is functionally compromised.

Strengths:

(1) One of the major strengths of this study is the elegant dissection of a defined circuit, tracking changes from premotor interneurons through motoneurons to the NMJ. The multimodal approach provides a comprehensive view of how connected elements respond to CP perturbations.

(2) An interesting finding is that NMJ morphology changes dramatically without corresponding deficits in synaptic transmission, challenging the common assumption that larger boutons necessarily indicate stronger synapses.

(3) Another intriguing result is that even with two layers of homeostatic compensation, locomotor behavior is still impaired, highlighting the limits of compensation and underscoring the critical role of CP timing.

(4) Beyond these scientific insights, the study benefits from a well-defined, tractable system and simple experimental manipulations, which together make the results highly interpretable and reproducible.

Comments on revised version.

The authors have carefully considered my comments and recommendations and have made substantial efforts to improve the clarity and validity of the study. Although additional electrophysiology experiments using shorter heat stress windows were not feasible, the authors performed additional analyses of postsynaptic GluRs and provided a clearer discussion of the study's limitations. Overall, this is a strong and well-written paper that establishes a valuable foundational framework for addressing interesting and important questions about adaptive responses in developing neural circuits.

Author response:

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

eLife Assessment:

This is an important study of critical period plasticity, focused on temperature manipulations, and how different parts of the Drosophila larval motor circuit adapt or maladapt. The work convincingly demonstrates that components of the motor network respond in distinct ways to the heat shock, and the combination of functional, structural, and electrophysiological approaches makes the study of significant interest. The work points to central interneurons as primary drivers of maladaptive changes, while motoneurons and neuromuscular junctions show compensatory or homeostatic adjustments. The study is methodologically rigorous, contributing important insights into critical period biology using a tractable invertebrate model.

We thank the reviewers for their thoughtful critique and suggestions. We agree with these and, where possible, we have attempted to address these, improving this study. As outlined below, we have revised the manuscript substantively and included additional data and figures.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors examine the impact of heat stress during an embryonic CP in Drosophila, focusing on the larval locomotor network. They show that elevated temperature increases neuronal activity and, when applied during the CP, results in long-term instability of the network, which manifests in prolonged seizure recovery times. At the neuromuscular junction, substantial structural changes occur, including terminal overgrowth and altered receptor composition, yet synaptic transmission remains preserved due to homeostatic regulation. Motoneurons display reduced excitability but receive increased synaptic input from premotor interneurons. These findings suggest that maladaptive instability originates within the central circuitry rather than at the neuromuscular junction, where changes seem to be homeostatically compensated. The study concludes that different network components exhibit distinct and hierarchical responses to CP perturbations, with premotor interneurons setting the tone for downstream adjustments in motoneurons.

Strengths:

The work takes advantage of the unique accessibility of the Drosophila system. A major strength of the study is the integration of structural, physiological, and behavioral analyses, which allows the authors to draw a comprehensive picture of how CP perturbations shape the locomotor network. The choice of an ecologically relevant stimulus (heat stress) is particularly convincing, as it links experimental manipulations more closely to natural environmental conditions. The experiments are carefully designed, and the results are robust and consistent with previous findings in the field, while also extending them in new directions.

Weaknesses:

The study leaves some uncertainty regarding the experimental design and interpretation. The change from short to prolonged heat shock manipulations raises the possibility that the effects observed may not be confined to the critical period alone - this could be experimentally addressed or simply rephrased in the text.

We agree that clarity about the experimental paradigm is important and have addressed this as suggested, within text, figures and figure legends: the duration of embryo exposure to 32˚C heat stress is now unambiguously stated and each figure has a graphical illustrations of the heat stress paradigm. For example, experiments represented in Figures 1, 3 (new data) and 8 (new data) used short, defined periods of a few hours of heat stress, aimed to identify specific windows of development that are sensitive to 32˚C heat stress. These also show that behavioural changes result from heat stress experienced during the specific 2-hour window that defines the critical period of the developing central locomotor circuitry, namely from 17-19 hours after egg laying - previously identified by Giachello & Baines (2015). Longer exposure to 32˚C heat stress during embryogenesis result in the same phenotypes when this 2-hour window is included, causing the same level of reduced larval crawling speed and lowered network stability, which manifests in increased seizure recovery times. This is as one might expect from a critical period of nervous system development.

Where the neuromuscular junction is concerned, where we identified embryonic heat stress causing phenotypes that are evident at late larval stages, a more complex model has emerged. Following suggestions from both reviewers to explore shorter heat stress exposures during embryogenesis, additional experiments (see Figure 3) we identified what might be a critical period for the body wall muscles. This is an earlier window of development, within 13-16 hours after egg laying, which is sensitive to heat stress in terms of the levels of the GluRIIA glutamate receptor subunit that will be expressed in the late larva. This developmental period is characterised by muscles acquiring their electrical properties (Broadie & Bate, 1993), i.e. comparable to the central locomotor network transitioning through its critical period at the time that it becomes active. The neuromuscular junction is composed of both presynaptic motoneurons and postsynaptic muscles, and therefore this composite structure is subject to multiple, sequential critical periods. The characterisation of changes to neuromuscular junction synaptic physiology was carried out using heat stress throughout most of embryogenesis (Figure 4). We think this appropriate from the perspectives of having included all relevant critical periods (muscle and CNS) to explore how this composite structure responds to environmental heat stress; also based on our observations that for each critical period phenotypes are defined by the experience during the critical period and not exacerbated by prolonged heat stress either side.

In addition, the maladaptive (seizure recovery) and adaptive/homeostatic phenotypes are not always clearly distinguished or highlighted, which makes it harder to appreciate how the different levels of the network plasticity fit together into a single mechanistic framework.

Following the suggestion, we have tried to clarify the mechanistic framework in a new figure that aims to summarise the model in Figure 9.

The question of whether phenotypes that result from an embryonic heat stress manipulation are adaptive or maladaptive is difficult to resolve. This is partly due to the nature of critical periods, since perturbations during these developmental windows can cause significant, long-lasting maladaptations that are challenging to reconcile from a perspective of adaptive plasticity. Secondly, in light of the nature of this animal, which has evolved a particularly rapid development and large brood sizes, any deviation from the evolved optimum developmental temperature of 25˚C could constitute a reduction in fitness. We interpret the phenotypes we see along those lines: network instability that results from critical-period perturbations is a manifestation of a sub-optimally tuned network, as is a reduction in larval crawling speed.

Reviewer #2 (Public review):

Summary:

This manuscript presents a thoughtful and well-executed study of critical period plasticity in the Drosophila larval motor circuit. The authors examined how transient heat, 32 {degree sign}C, during the embryonic stage, altered network properties, showing that premotor interneurons A27h increase excitatory drive onto motoneurons, which respond with a reduction in excitability. At the NMJ, synaptic terminals expand and GluRIIA distribution shifts, yet synaptic transmission remains largely unaffected. Despite these local compensations, the treated larvae display slower crawling and prolonged recovery from seizures, indicating that the network is functionally compromised.

Strengths:

(1) One of the major strengths of this study is the elegant dissection of a defined circuit, tracking changes from premotor interneurons through motoneurons to the NMJ. The multimodal approach provides a comprehensive view of how connected elements respond to CP perturbations.

(2) An interesting finding is that NMJ morphology changes dramatically without corresponding deficits in synaptic transmission, challenging the common assumption that larger boutons necessarily indicate stronger synapses.

(3) Another intriguing result is that even with two layers of homeostatic compensation, locomotor behavior is still impaired, highlighting the limits of compensation and underscoring the critical role of CP timing.

(4) Beyond these scientific insights, the study benefits from a well-defined, tractable system and simple experimental manipulations, which together make the results highly interpretable and reproducible.

Weaknesses:

There are a few areas where the manuscript could be strengthened.

(1) Although A27h premotor neurons are well characterized, the claim that they are the causal driver of downstream changes would be strengthened by additional experiments or a clearer discussion of the temporal hierarchy.

We have tried to clarify the model of the temporal hierarchy (new Figure 9). This is a model and as such will hopefully help us collectively to think about this system and underlying processes, while also inviting this perspective to be challenged. The model we propose is compatible with our observations, namely that the premotor circuitry might change in response to a critical period heat stress (e.g. increasing their synaptic drive onto motoneurons), followed by homeostatic adjustment by the postsynaptic motoneurons (e.g. by reduction of their excitability), thus serving to maintain overall normal motoneuron firing patterns (see Figure 6).

However, synaptic communication is commonly regulated in both antero- and retrograde directions. Therefore, while compatible with the observations we have made, bi-directional information flow could also be instructive during the CNS critical period.

(2) While 32 {degree sign}C heat stress is presented as ecologically relevant, it produces maladaptive behavioral outcomes, raising questions about the ecological and mechanistic interpretation of the model. In particular, most experiments, with the exception of Figure 1, used prolonged (24h) heat treatments, which could introduce developmental effects beyond the CP itself. Comparing shorter and longer heat exposures would help clarify the specificity of the CP response.

We agree. For a detailed response on this point, please response to Reviewer #1 above.

(3) While there are schematics for experimental procedures, a circuit diagram tracing information flow and indicating where structural and functional changes occur would help readers better understand the findings.

We have created Figure 9 as a working model.

(4) Finally, the main paradox of the study, that robust homeostatic compensations occur yet behavior remains impaired, could be explored in more depth in the Discussion.

We have tried to address this in the discussion.

Reviewer #3 (Public review):

Summary:

During development, neural circuits undergo brief windows of heightened neuronal plasticity (e.g., critical periods) that are thought to set the lifelong functional properties of underlying circuits. These authors, in addition to others within the Drosophila community, previously characterized a critical period in late fly embryonic development, during which alterations to neuronal activity impact late-stage larval crawling behavior. In the current study, the authors use an ethologically-relevant activation paradigm (increased temperature) to boost motor activity during embryogenesis, followed by a series of electrophysiology and imaging-based experiments to explore how 3 distinct levels of the circuit remodel in response to increases in embryonic motor activity. Specifically, they find that each level of the circuit responds differently, with increased excitatory drive from excitatory pre-motor neurons, reduced excitability in motor neurons, and no physiological changes at the NMJ despite dramatic morphological differences. Together, these data suggest that early life experience in the motor neuron drives compensatory changes at each level of the circuit to stabilize overall network output.

Strengths:

The study was well-written, and the data presented were clear and an important contribution to the field.

Weaknesses:

The sample sizes and what they referred to throughout the distinct studies were unclear. In the legends, the authors should clearly state for each experiment N=X, and if N refers to an NMJ, for example, instead of an individual animal, they should state N=X NMJs per N=X animals. This will help readers better understand the statistical impact of the study.

This is a good point. For the majority, each data point is derived from a unique specimen, unless explicitly stated otherwise, for NMJ size on muscle DA1 (Figure 3) and for larval crawling data, where each larva was measured up to three time, once per unique 5-minute crawling interval.

Recommendations for the authors:

Reviewing Editor Comments:

In addition to revising the text and making interpretive changes as suggested by the reviewers, we invite you to consider the following:

(1) Either rephrase the conclusions on the role of the 2h CP and discuss the effects of temperature during embryonic development. Alternatively, to validate the idea of a longer CP window, directly compare the results of a few key experiments using the 2h and 24h heat treatment.

We have addressed this within the text, as suggested. In the text and figure legends, we have made clear distinctions between exposure to 32˚C heat stress during most of embryogenesis vs a specific developmental window of a few hours. In figures, we have provided diagrams that graphically illustrate the period of heat stress exposure.

Our ability to experimentally test differences between precise vs broader heat stress periods during embryonic development have been constrained due to the departure of scientists, who were able to carry out electrophysiological recordings (as also explained below). As the next best alternative, we focus on imaging and behavioural analyses. These demonstrated that the developing body wall muscles are sensitive to heat stress during an earlier phase, from 13-16 hours after egg laying, when the body wall muscles become electrically active. It precedes the critical period of the central locomotor circuitry (17-19 hours after egg laying), when neurons in the CNS become electrically and synaptically active (16 hours after egg laying). We think this an exciting additional insight, demonstrating sequential critical periods as different parts of the locomotor network become active: first the body wall muscles, followed by the central circuitry.

(2) Clarifying the homeostatic responses and shedding light on how they engage with the maladaptive changes described would benefit the study. Furthermore, adding more information about the anatomical and structural changes and how they relate to the intrinsic and synaptic changes would also benefit the study.

We have tried to address this within the text and with a summary diagram (Figure 8), as suggested.

Reviewer #1 (Recommendations for the authors):

(1) It remains unclear whether the authors want to conclude that reduced network stability is not due to changes at the motoneuron level, but rather at the premotor level. Although this idea is mentioned in the results and discussion, it does not appear in the abstract or introduction, which leaves the different findings disconnected. Clarifying and highlighting this conclusion throughout the manuscript would strengthen the narrative.

We have added additional experiments and changed the manuscript to address this point. These showed that there are distinct phases of embryonic development during which heat stress causes changes to NMJ structure vs to larval behaviour (seizure recovery times/network stability and crawling speed) - additional data in suppl. Fig. 2 and Fig. 7). In the Drosophila embryo, the body wall muscles develop and acquire their electrical properties before central neurons do, and these phases correlate with sensitivity to heat stress.

(2) In the results section related to Figure 1, the logical link between the CP protocol and the functional assessment of the locomotor network at different temperatures is not sufficiently explained. It is not clear what this assessment is meant to test or demonstrate. A more explicit statement of the rationale and correction of what seems to be a typographical error in the final sentence of the paragraph would help to clarify the authors' intent.

We have tried to rectify this by changes in the manuscript and to Figure 1, to make the sequence of panels more intuitive.

(3) In the second results section, the experimental strategy shifts from using a short 2-hour heat shock to a 24-hour manipulation. The reasoning - that short manipulations in different windows yield no phenotype - is understandable, but a 24-hour perturbation may have broader consequences beyond the CP, simply by virtue of its longer duration. Moreover, 24h is roughly the duration of embryonic development at 25C. When at 32C, embryos should develop faster; therefore, is the 24h heat shock extending to L1?

Yes, the 24 hour heat stress extends into the first few hours of the L1 larval stage.

In order to validate the use of a longer window, the author should show how it affects the developmental time. Moreover, one should test that a few important observations remain the same with 2h and 24h heat perturbation. Alternatively, one cannot conclude that the phenotype is due to the rather narrow previously defined CP rather than to other effects associated with the overall embryonic developmental time and coordination. This would not make the results less interesting, but it would be important to assess whether the effects can be solely attributed to the 2h CP.

We have compared the impact of heat stress experience during the majority of embryogenesis, including the CP that had been defined for the central locomotor network (17-19 hours after egg laying) with shorter heat stress manipulations during consecutive phases of embryogenesis until larval hatching. As outlined above in response to point (1) by the Reviewing Editor, reduced stability of the central network and associated reduction in larval crawling occurs when heat stress is experienced during the CP of the central locomotor network (17-19 hours after egg laying). Prolonged heat stress experience for 24 hours leads to indistinguishable outcomes, as long as this 2-hour CP window is included (see Fig. 1 and Fig. 8).

However, this suggestion by Reviewer #1 led us to identify a second CP for the body wall muscles (see Fig. 3). NMJ overgrowth and changes to the postsynaptic glutamate receptor composition result from earlier heat stress experiences, and those are comparable to the effects caused by 24-hour heat stress exposure when this earlier muscle CP is included.

Therefore, NMJ development is affected by consecutive CPs, an earlier one linked to body wall muscle development, followed by a later one that impacts the presynaptic motoneurons and their upstream circuitry. Nevertheless, the larval NMJ and behavioural phenotypes that we have identified appear to result from sensitivity to heat stress during these respective CP windows, with no clear evidence of cumulative effects on these phenotypes resulting from longer heat stress exposure during embryogenesis.

(4) In session 3, the authors note that GluRIIA reductions were most pronounced in proximal regions of the NMJ. However, this is not explicitly quantified in the figures or methods. Including such quantification, or clarifying where it can be found, would make this observation more convincing.

We have analysed anti-GluRIIA signal intensities in proximal vs distal boutons, comparing different ROI selection processes (e.g. thresholding to a full NMJ/anti-HRP mask and to an anti-GluRIIB mask, which is more selective to postsynaptic sites). Analysis of multiple data sets did not show statistical significance, but instead confirmed that comparable reductions in anti-GluRIIA signal manifest in both proximal and distal boutons, following an embryonic 32C heat stress, relative to controls. We have therefore removed relevant speculative statements.

(5) In session 4, the authors conclude that motoneurons undergo a decrease in excitability to adjust to greater premotor drive. Is there anatomical evidence for this, such as an increase in input synapses?

We previously quantified change in excitatory presynaptic synaptic contact number onto aCC motoneuron dendrites in third instar larvae following an embryonic pharmacological activity manipulation: overexcitation of the developing network following introduction of PTX via feeding to gravid females*. No significant structural changes were seen. Although this is a different manipulation of the developing network, all our data to date suggest that heat stress manipulations during the embryonic critical period signal via the same pathways, at least in part due to temperature increases leading to activity increases. Because such a quantification is technically challenging and extremely time-consuming due to the low level of marking individual motoneurons, we did not think it informative or in scope for this project.

*See Figure 5 in this publication: Hunter I, Coulson B, Pettini T, Davies JJ, Parkin J, Landgraf M, Baines RA. Balance of activity during a critical period tunes a developing network. Elife. 2024 Jan 9;12:RP91599. doi: 10.7554/eLife.91599. PMID: 38193543; PMCID: PMC10945558.

The interpretation of the optogenetic experiments would also benefit from clarification. If motoneurons are less excitable yet receive more drive, one might expect no net change, rather than the differences observed. Alternatively, could the excitability of the premotor neuron itself have changed, either intrinsically or in relation to Chronos expression? Measuring premotor activity directly during optogenetic activation could help to resolve this ambiguity.

These are good suggestions. Yes, we think that motoneuron excitability has changed as a result of heat stress - see paper submitted in parallel and published since: Sobrido-Cameán et al., 2025, PLoS Biology. Unfortunately, the team members, who could have carried out this type of analysis had moved on by submission of the manuscript. Therefore, we were no able to experimentally pursue these questions further.

(6) In session 6, the authors report slower propagation of premotor activity waves after CP heat stress, but the logic of the experiment is not sufficiently explained. How does this finding relate to the enhanced premotor drive described earlier? Only timing is quantified; information about amplitude and wave dynamics would strengthen the interpretation. These results could also be discussed in relation to Figure 1B, where acute heat stress increased motoneuron activity. One interesting possibility could be that CP manipulations might adaptively prepare the larva to function at different temperatures. Experiments testing wave propagation at 32 {degree sign}C (Figure 6) or, conversely, motoneuron activity after CP manipulations (Figure 1B paradigm) would provide valuable evidence for such an adaptive role.

As per above, unfortunately, the team member who could have carried out this type of analysis had moved on by submission of the manuscript. However, we have tried to address the question of whether there is an adaptive element to the adjustments that result from embryonic heat stress experience. Specifically, we carried out behavioural tests on how larvae respond with changes in crawling speed to acute changes in ambient temperature (new Fig. 7).

We interpret our findings as follows: that heat stress during embryonic development leads to sub-optimal outcomes with regard to network stability as well as default and maximum crawling speed. The precise causes for this will be difficult to unpick. Behaviourally, when challenging larvae with an acute change in ambient temperature, we saw that slow crawling larvae do respond comparatively normally to a relative increase in ambient temperature by speeding up (in effect an escape response). This demonstrates that embryonic heat stress causes a change in the default crawling speed, while principally maintaining behavioural responses to changes in ambient temperature. It appears that animals that had experienced heat stress during embryonic development, by adjusting their default speed downward, maintain a dynamic response range into a higher temperature range than controls (35C vs 29C, respectively). Potentially, this could be an adaptive outcome to living at higher temperatures, though such an interpretation would require a body of work. 

Nevertheless, every aspect we have assayed suggests that heat stress experience during the CP leads to sub-optimal outcomes: of network instability, slower default and slower maximal crawling speeds.

Minor points:

(1) Abstract: "has suboptimal outcomes;" should be corrected to "has suboptimal outcomes,".

Corrected

(2) Abstract: "we find that transient embryonic..." would improve readability with a capitalized "We".

We are unsure about the sentence this refers to. If this sentence, then we suggest that this could remain as was.

"Within the central nervous system, we find transient embryonic CP perturbation leads to increased synaptic drive from premotor interneurons to motoneurons..."

The intention here is to differentiate between changes within the CNS vs at the NMJ.

(3) Abstract: The sentence "Present the larva ... as an experimental model system..." overstates novelty, as the system has already been established in prior work. Instead, this study could highlight temperature manipulation as an ecologically relevant way to probe CPs.

We have adopted this suggestion.

Reviewer #2 (Recommendations for the authors):

I would recommend:

(1) Perform additional experimental support and dissuasions for the causal role of the premotor neuron and the network disability.

Unfortunately, it has not been possible to carry out additional e-phys experimental work due to key people having moved on and now unable to carry out such experiments, and no replacements in sight to do so. Instead, we have focused on other work that we could do, namely to test the effect of different heat stress windows during embryonic development on GluRIIA vs GluRIIB expression at postsynaptic sites. This shows that indeed the effect seen following a 24-hour heat stress is replicated by a much shorter window of heat stress. For NMJs GluRII composition the critical period is different from the critical period of the CNS. This replicates the different developmental timings of maturation: the body wall muscles express ion channels and attain their electrical properties several hours before central neurons. We have provided these additional data as a new supplementary figure to Fig. 2. We have changed the main text to note the caveat of longer heat stress manipulations potentially leading to additional or more exacerbated phenotypes.

(2) I would suggest that the authors expand the discussion on why two layers of homeostatic adjustment fail to preserve behavior. Is this simply a limit of plasticity?

This is a difficult aspect to address well, beyond the purely speculative and potentially confusing. We would like to suggest that to do so requires a basic understanding on what pressures neurons/networks respond to (heat-caused over-activation and/or metabolic); and from that perspective to gauge how they adjust to those pressures, i.e. what the adjustments are trying to "achieve". This in turn should inform on whether such adjustments are homeostatic or anti-homeostatic in nature, and whether there are limits beyond which we consider a system "breaking".

(3) 32C heat was described as "ecologically relevant". However, it produces maladaptive outcomes. The author should consider reframing it as a stressor that reveals CP sensitivity, rather than an adaptive signal.

This is a good suggestion, and we have implemented changes accordingly.

(4) It is important to distinguish the effects of transient (2h) vs prolonged heat exposure to confirm the manipulation targeted CP specifically.

We have tried to make these distinctions clearer within text and figure legends. As per response to (1) above, we generated and analysed additional data to show that changes in GluRIIA are induced during a defined shorter developmental time window, not exacerbated by prolonged heat stress exposure during embryogenesis.

(5) I would definitely recommend adding a diagram tracking information flow and showing where structural and functional changes occur.

This is a helpful suggestion, which we have tried to implement with a new figure (Fig. 8).

Overall, this is a strong and well-written paper that produced some unexpected results, and added a solid model circuit to study CP plasticity at the circuit level.

Reviewer #3 (Recommendations for the authors):

I identified one typo: "activity manipulations during the embryonic CP are artificial, We asked to what extent" The "W" of "We" should be lower case.

Now corrected.

eLife Assessment

This paper represents a valuable contribution to our understanding of how local field potential (LFP) oscillations and beta band coordination between the hippocampus and prefrontal cortex of rats may relate to learning. Through a set of appropriate analyses, including oscillation-oscillation coupling, oscillation-spike modulation, and behavioral dependant measurements, the study presents convincing evidence for uncoupled beta activity between the two regions. This work will be of interest to researchers working on interaction of brain regions during spatial learning in rodents.

Reviewer #1 (Public review):

Wang, Zhou et al. investigated coordination between prefrontal cortex (PFC), and hippocampus (Hp), during reward delivery via analyzing beta oscillation. Beta oscillations are associated with various cognitive functions but their role in coordinating brain networks during learning is still not thoroughly studied. Authors focused on the changes in power, peak frequencies and coherence of beta oscillations in two regions when rats learn a spatial task thru days. Contradicting with authors hypothesis, beta oscillations in those two regions during reward delivery were not coupled in spectral or temporal aspects. They were, however, able to show reverse changes in beta oscillations in PFC and Hp as the animal's performance got better. Authors were also able to show a small subset of cell population in PFC that are modulated by both beta oscillations in PFC and sharp wave ripples in Hp. A similarly modulated cell population was not observed in Hp. These results are valuable in pointing out distinct periods during a spatial task when two regions modulate their activity independent from each other.

Authors made a detailed analysis of the data to support their conclusions. Few more points of discussion would clarify the results of the paper.

(1) One of the big conclusions of the paper is how the beta burst power is changing after learning the task (Figure 3). Authors have also showed in Figure 6-1, how the SWR power and rate are changing thru the training days. Did they observe a change in coordination of Beta bursts and SWR between the days, which would also reflect how experience changes the coordination?

(2) Authors have shown in detail the opposite relationship between Beta phase locking and SWR modulation in Hippocampus in Figure 7I. This might require a different analysis, but is it possible to make a discussion on predicting a cell firing in a SWR after it fires in a beta burst.

Other than these two points, authors have addressed previous comments and made a convincing analysis of their data.

Reviewer #2 (Public review):

Using electrophysiological recordings in freely moving rats during a spatial navigation task, this study investigated the role of beta oscillations in the hippocampal-prefrontal network. Through a set of appropriate analyses-including oscillation-oscillation coupling, oscillation-spike modulation, and behavioral dependant measurements-the study presents convincing evidence for uncoupled beta activity between the two regions. These findings offer important insights into the network mechanisms of spatial navigation and may have significant implications for related neurological disorders.

Comments on revised version.

The authors have carried out additional analyses and made corresponding revisions to the manuscript in response to the earlier review comments, which have made the conclusions more convincing. However, please note that the last question about coexistence of beta and SWR has not been fully answered. Please supplement your response to address this point completely.

Reviewer #3 (Public review):

This paper explored the role of beta rhythms in the context of spatial learning and mPFC-hippocampal dynamics. The authors characterized mPFC and hippocampal beta oscillations, examining how their coordination and their spectral profiles related to learning and prefrontal neuronal firing. Rats performed two tasks, a Y-maze and F-maze, with the F-maze task being more cognitively demanding. Across learning, prefrontal beta oscillation power increased while beta frequency decreased. In contrast, hippocampal beta power and beta frequency decreased. This was particularly for the well-performed and well-learned Y-maze paradigm. The authors identified the timing of beta oscillations, revealing an interesting shift in beta burst timing relative to reward entry as learning progressed. They also discovered an interesting population of prefrontal neurons that were tuned to both prefrontal beta and hippocampal sharp-wave ripple events, revealing a spectrum of SWR-excited and SWR-inhibited neurons that were differentially phase locked to prefrontal beta rhythms.

Author response:

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

Reviewer #1 (Public review):

Wang, Zhou et al. investigated coordination between the prefrontal cortex (PFC) and the hippocampus (Hp), during reward delivery, by analyzing beta oscillations. Beta oscillations are associated with various cognitive functions, but their role in coordinating brain networks during learning is still not thoroughly understood. The authors focused on the changes in power, peak frequencies, and coherence of beta oscillations in two regions when rats learn a spatial task over days. Inconsistent with the authors' hypothesis, beta oscillations in those two regions during reward delivery were not coupled in spectral or temporal aspects. They were, however, able to show reverse changes in beta oscillations in PFC and Hp as the animal's performance got better. The authors were also able to show a small subset of cell populations in PFC that are modulated by both beta oscillations in PFC and sharp wave ripples in Hp. A similarly modulated cell population was not observed in Hp. These results are valuable in pointing out distinct periods during a spatial task when two regions modulate their activity independently from each other.

The authors included a detailed analysis of the data to support their conclusions. However, some clarifications would help their presentation, as well as help readers to have a clear understanding.

(1) The crucial time point of the analysis is the goal entry. However, it needs a better explanation in the methods or in figures of what a goal entry in their behavioral task means.

We appreciate Reviewer 1 pointing out this shortcoming and will clarify the description in the revised manuscript. Each goal is located at the end of the arm, and is equipped with a reward delivery unit. The unit has an infrared sensor. The rat breaks the infrared beam when it enters the goal. Figures 1 and 2 have been updated to clearly indicate the time of goal entry. The main text and methods have been updated with the explanation.

(2) Regarding Figure 2, the authors have mentioned in the methods that PFC tetrodes have targeted both hemispheres. It might be trivial, but a supplementary graph or a paragraph about differences or similarities between contralateral and ipsilateral tetrodes to Hp might help readers.

We appreciate this suggestion, which has led to an interesting finding. The coherence and burst coordination were similar for ipsi- and contralateral PFC and hippocampus. Interestingly, we found PFC beta activity was more coherent within each PFC hemisphere compared with across hemispheres. This was observed for coherence and burst time. This suggests there is hemispheric localization of beta oscillations. These results are shown in Fig. 2-1.

(3) The authors have looked at changes in burst properties over days of training. For the coincidence of beta bursts between PFC and Hp, is there a change in the coincidence of bursts depending on the day or performance of the animal?

This is now reported in Fig. 3-3. After quantifying the proportion of independent and coincident bursts as function of experiment day or performance, we found a decrease in the proportion of coincident bursts within CA1, which was specific to the well-performed Y-maze.

(4) Regarding the changes in performance through days as well as variance of the beta burst frequency variance (Figures 3C and 4C); was there a change in the number of the beta bursts as animals learn the task, which might affect variance indirectly?

The difference in the burst count across days did not explain the results. We performed a permutation test (Fig. 4-4), where we randomly shuffled the day identity to control for the count difference across days. The change in variance remains significant.

(5) In the behavioral task, within a session, animals needed to alternate between two wells, but the central arm (1) was in the same location. Did the authors alternate the location of well number 1 between days to different arms? It is possible that having well number 1 in the same location through days might have an effect on beta bursts, as they would get more rewards in well number 1?

The central arm remained the same across days since we needed the animals to learn the alternation task. In our experience, the animal needs a few days to learn the alternation rule when we switch the central arm location. For this experiment, we were interested in the initial learning process, and we kept the central arm constant. Switching the central arm location is a great suggestion for a follow-up experiment where we can understand the effects of reward contingency change on beta bursts.

(6) The animals did not increase their performance in the F maze as much as they increased it in the Y maze. It would be more helpful to see a comparison between mazes in Figure 5 in terms of beta burst timing. It seems like in Y maze, unrewarded trials have earlier beta bursts in Y maze compared to F maze. Also, is there a difference in beta burst frequencies of rewarded and unrewarded trials?

We performed the analysis and found burst timing was similar between the two mazes (Fig. 4-2). Bursts on rewarded trials occurred later than those on unrewarded trials (Fig. 4). Interestingly, PFC bursts on rewarded trials were lower in frequency compared with unrewarded trials. CA1 bursts during rewarded and unrewarded trials had similar frequencies (Fig. 4-3).

(7) For individual cell analysis, the authors recorded from Hp and the behavioral task involved spatial learning. It would be helpful to readers if authors mention about place field properties of the cells they have recorded from. It is known that reward cells firing near reward locations have a higher rate to participate in a sharp wave ripple. Factoring in the place field properties of the cells into the analysis might give a clearer picture of the lack of modulation of HP cells by beta and sharp wave ripples.

As recommended, we quantified the mean speed, mean distance to goal locations, and spatial information for CA1 cells (Fig. 7 J-L). We found SWR reactivated CA1 cells had higher speed and were spiking further away from the goals compared with non-reactivated CA1 cells. This is consistent with prior work that shows SWR-associated reactivation in dorsal CA1 can correspond to trajectories taken as the animal moves towards goals. In intermediate CA1, the content of reactivations is biased toward place representations closer to goals (Jin et al., 2024). CA1 cells with or without phase locking to beta oscillations had similar spatial firing properties.

Reviewer #1 (Recommendations for the authors):

(1) Please make a figure representing what the goal entry means in Figure 1.

We have updated Fig. 1 to clearly show the definition of goal entry. We also edited the main text and methods to better explain the definition of goal entry.

(2) For Figure 1-1, please either change the contrast of the pictures, or define the lesioned areas, as it is a bit difficult to see the lesioned parts, especially in PFC.

We have increased the contrast for Fig. 1-1 for the histology to better show the lesions.

(3) For Figure 1-2, is it possible to show a beta burst from Hp?

Yes, we provided two examples of beta bursts from the hippocampus alongside burst examples from PFC in Fig. 1-2.

(4) Is it possible to make a supplementary table showing the number of tetrodes recorded from each animal per day, plus the number of isolated single cells?

Yes, we have included tetrode counts in Table 1-2 and cell counts in Tables 5-2 to 5-4.

Reviewer #2 (Public review):

(1) When presenting the power spectra for the representative example (Figure 1), it would be appropriate to display a broader frequency band-including delta, theta, and gamma (up to ~100 Hz), rather than only the beta band.

We agree the extended frequency range provides a better overview of the spectral characteristics during the goal period. We have now included example spectrograms up to 100 Hz to show the spectral content for a wider range of frequencies (Fig. 1-2). Further, we have included additional analyses to compare the spectral characteristics between periods when the animal was moving on the maze or immobile at the goal, for frequencies up to 100 Hz (Fig. 1C-H, Fig. 1-3 and 1-4). We used both Welch’s periodogram (Fig. 1-3) and continuous wavelet transform (Fig. 1-4) to demonstrate our findings on beta oscillations in both regions are robust and consistent.

What was the rat's locomotor state (e.g., running speed) after entering the reward location, during which the LFPs were recorded?

Because goal entry is defined as the time the animals break the infrared beam at the goal (response to Reviewer 1), the rat would have come to a stop. We have added the time-aligned speed profile to the spectra and raw data examples in the manuscript (Fig. 1B, Fig. 1-4, Fig. 2A, and Fig. 6A). In addition, we added the quantification of the animal’s speed at the time of beta bursts (Fig. 3-1, Fig. 4-1) and SWRs (Fig. 6-1D).

If the rats stopped at the goal but still consumed the reward (i.e., exhibited very low running speed), theta rhythms might still occasionally occur, and sharp-wave ripples (SWRs) could be observed during rest.

We typically find low theta power in the hippocampus after the animal reaches the goal location and as it consumes reward. Reviewer 2 is correct about occasional theta power at the goal. To compare differences in LFP characteristics between maze running and goal locations, we added additional analyses in Fig. 1-3 and 1-4. We did find SWRs during goal periods (Fig. 6) and we quantified SWR properties in an additional analysis in Fig. 6-1.

Do beta bursts also occur during navigation prior to goal entry? It would be beneficial to display these rhythmic activities continuously across both the navigation and goal entry phases.

We did not find consistent beta bursts in PFC or CA1 on approach to goal entry. We generated an additional goal entry-aligned spectrogram, and quantification (Fig. 1-4) to show that beta oscillations in both regions increased after goal entry. This was also supported by the Welch’s periodogram method (Fig. 1C-H, Fig. 1-3). Beta oscillations in the hippocampus during locomotion or exploration have been reported (Ahmed & Mehta, 2012; Berke et al., 2008; França et al., 2014; França et al., 2021; Iwasaki et al., 2021; Lansink et al., 2016; Rangel et al., 2015).

Additionally, given that the hippocampal theta rhythm is typically around 7-8 Hz, while a peak at approximately 15-16 Hz is visible in the power spectra in Figure 1C, the authors should clarify whether the 22 Hz beta activity represents a genuine oscillation rather than a harmonic of the theta rhythm.

We performed further spectral analysis comparing times when the animal is moving on the maze with times when the animal is immobile at the goal (Fig. 1-3 and 1-4). The results point to the beta frequency oscillations in both regions are unlikely to be harmonics of theta. The beta frequency bands in the spectrogram are independent of the theta band. We were initially concerned about the possibility that the 22 Hz power in CA1 may be a harmonic rather than a standalone oscillation band. If these are harmonics of theta, we should expect to find coincident theta at the time of bursts in the beta frequency. In Fig. 1B, Fig. 1-5, and Fig. 2A, we show examples of the raw LFP traces from CA1. Here, the detected bursts are not accompanied by visible theta-frequency activity. For PFC, we do not always see persistent theta-frequency oscillations like CA1. In PFC, we found beta bursts were frequent and visually identifiable when examining the LFP. We provided examples of the PFC LFP (Fig. 1B, Fig. 1-5, and Fig. 2A). In these cases, we see clear beta frequency oscillations lasting several cycles and these are not accompanied by any visible oscillations in the theta frequency in the LFP trace.

(2) The authors claim that beta activity is independent between CA1 and PFC, based on the low coherence between these regions. However, it is challenging to discern beta-specific coherence in CA1; instead, coherence appears elevated across a broader frequency band (Figure 2 and Figure 2-1D). An alternative explanation could be that the uncoupled beta between CA1 and PFC results from low local beta coherence within CA1 itself.

This is a legitimate concern, and we used three methods to characterize coherence and coordination between the two regions. First, we calculated coherence for tetrode pairs for times when the animal was at goals (Fig. 2B), which provides a general estimation of coherence across frequencies but lack any temporal resolution. Second, we calculated burst-aligned coherence (Fig. 2-2), which provides temporal resolution relative to the burst, but the multi-taper method is constrained by the time-frequency resolution trade-off. Third, we quantified the timing between the burst peaks (Fig. 2D), which described the timing differences but the peaks for the bursts may not be symmetric. Each method has its own caveats, but we drew our conclusion from the combination of results from these three analyses, which pointed to similar conclusions.

Reviewer 2 is correct in pointing out the uniformly high coherence within CA1 across the frequency range we examined. When we inspected the raw LFP across multiple tetrodes in CA1, they were similar to each other (Fig. 2A). This likely reflects the uniformity in the LFP across recording sites in CA1, which is what we saw with coherence values across the frequency range (Fig. 2B). We found that CA1 coherence between tetrode pairs within CA1 was statistically higher than tetrode pairs in PFC across the frequency range (Fig. 2B and C), thus our results are unlikely to be explained by low beta coherence within CA1 itself. The burst-aligned coherence using a multi-taper method also supports this. The coherence values within CA1 at the time of CA1 bursts were ~0.8-0.9.

(3) In Figure 2-1E-F, visual inspection of the box plots reveals minimal differences between PFC-Ind and PFC-Coin/CA1-Coin conditions, despite reported statistical significance. It may be necessary to verify whether the significance arises from a large sample size.

We will include the sample sizes in Table 2-1. We repeated the analyses based on average values per day for each animal (Fig. 2-2 E-F). The pattern of significance remains consistent.

(4) In Figure 3 and Figure 4, although differences in power and frequency appear to change significantly across days, these changes are not easily discernible by visual inspection. It is worth considering whether these variations are related to increased task familiarity over days, potentially accompanied by higher running speeds.

We agree with Reviewer 2 that familiarity increases across days, and the animal is likely running faster. The analysis for Fig. 3 (previously Fig. 3 and 4) includes only data from periods when the animal was at the goal and was not moving. We added a supplemental figure (Fig. 3-1), which shows that the speed of the animal was below 0.5 cm/s at the time of the analyzed bursts. We used linear mixed-effects models to quantify the relationship between power, frequency and day or behavioral quintile, which accounts for repeated measurements across animals.

(5) The stronger spiking modulation by local beta oscillations shown in Figure 6 could also be interpreted in the context of uncoupled beta between CA1 and PFC. In this analysis, only spikes occurring during beta bursts should be included, rather than all spikes within a trial. The authors should verify the dataset used and consider including a representative example illustrating beta modulation of single-unit spiking.

We agree with Reviewer 2 that the stronger modulation to local beta is another piece of evidence indicating uncoupled beta between the two regions. We appreciate this suggestion and have revised Fig. 5 (previously Fig. 6) to include examples illustrating beta modulation for single units. These are spike-phase raster and histograms. We want to clarify that in the revised manuscript, the spikes were only from periods when the animal was at the goal location (5 s after entry) and did not include the running period between goals. Although beta power fluctuates in bursts, our data show there are ongoing beta oscillations throughout this period (Fig. 1-3 and Fig. 1-4), which prompted us to examine the entire period.

(6) As observed in Figure 7D, CA1 beta bursts continue to occur even after 2.5 seconds following goal entry, when SWRs begin to emerge. Do these oscillations alternate over time, or do they coexist with some form of cross-frequency coupling?

This is a very helpful suggestion and led to some interesting findings. We performed two additional analyses: 1) burst/SWR cross-correlation on a shorter timescale and 2) SWR-aligned spectrogram for the beta frequency range. PFC beta burst timing and power appear to be anti-correlated with SWRs detected in CA1 (Fig. 6G and K). In contrast CA1 beta bursts were more likely to occur with SWRs (Fig. 6H and L). These results suggest there is temporal coordination between ongoing beta oscillations in PFC and SWRs in the hippocampus during waking. Cortical beta oscillations are reduced during hippocampal SWRs, perhaps to support the transient switch in global cortical states that accompanies SWRs.

To examine potential cross-frequency coupling between SWRs and beta oscillations, we computed the mean SWR band power (150-250).

Reviewer #3 (Public review):

Summary:

This paper explored the role of beta rhythms in the context of spatial learning and mPFC-hippocampal dynamics. The authors characterized mPFC and hippocampal beta oscillations, examining how their coordination and their spectral profiles related to learning and prefrontal neuronal firing. Rats performed two tasks, a Y-maze and an F-maze, with the F-maze task being more cognitively demanding. Across learning, prefrontal beta oscillation power increased while beta frequency decreased. In contrast, hippocampal beta power and beta frequency decreased. This was particularly the case for the well-performed and well-learned Y-maze paradigm. The authors identified the timing of beta oscillations, revealing an interesting shift in beta burst timing relative to reward entry as learning progressed. They also discovered an interesting population of prefrontal neurons that were tuned to both prefrontal beta and hippocampal sharp-wave ripple events, revealing a spectrum of SWR-excited and SWR-inhibited neurons that were differentially phase locked to prefrontal beta rhythms.

In sum, the authors set out to examine how beta rhythms and their coordination were related to learning and goal occupancy. The authors identified a set of learning and goal-related correlates at the level of LFP and spike-LFP interactions, but did not report on spike-behavioral correlates.

Strengths:

Pairing dual recordings of medial prefrontal cortex (mPFC) and CA1 with learning of spatial memory tasks is a strength of this paper. The authors also discovered an interesting population of prefrontal neurons modulated by both beta and CA1 sharpwave ripple (SWR) events, showing a relationship between SWR-excited and SWR-inhibited neurons and beta oscillation phase.

Weaknesses:

Moreover, there is little detail provided about sample sizes and how data sampling is being performed (e.g., rats, sessions, or trials), raising generalizability concerns.

We appreciate Reviewer 3’s thoughtful suggestions for making our claims convincing. We have included information about sample sizes in the revised manuscript.

The authors report on a task where rats were performing sub-optimally (F-maze), weakening claims.

Our experiment was designed to create a scenario in which one task was learned (Y-maze) and another was not (F-maze). This contrast allows us to determine differences in neural correlates of learning versus familiarity. The design produced a learned and not learned task with similar levels of familiarity over 5 days, within the same animal.

Likewise, it is questionable as to whether mPFC and hippocampus are dually required to perform a no-delay Y-maze task at day 5, where rats are performing near 100%.

We agree with Reviewer 3 that the mPFC and hippocampus may not be required when the animal reaches stable performance on day 5 (Deceuninck & Kloosterman, 2024). The data we collected spans the full range of early learning (day 1) to proficiency (day 5). We wanted to understand the dynamics of beta across these learning stages, which have not been reported previously.

Recent studies suggest mPFC and hippocampus are likely to be needed, in some capacity, for learning continuous spatial alternation tasks on a range of maze geometries. Lesions, inactivation or waking activity perturbation of hippocampus or hippocampus and mPFC on the W maze alternation task slowed learning (Jadhav et al., 2012; Kim & Frank, 2009; Maharjan et al., 2018). More recently, optogenetic silencing of mPFC after sharp wave ripples on the Y-maze alternation affected performance when the center arm was switched (den Bakker et al., 2023). The Y and F-mazes in our study both share the continuous alternation rule, where the animal needed to avoid visiting a previously visited location on the outbound choice relative to the center, and always return to the center location.

Further, the performance characteristics on the outbound and inbound components of our Y task are similar to the W task. We have analyzed the “inbound” and “outbound” performance of the animals on the Y-maze alternation task, and they are similar to the W maze alternation task. The “inbound” or reference location component is learned quickly whereas the “outbound”, alternation component is learned slowly.

There would be little reason to suspect strong oscillatory coupling when task performance is poor and/or independent of mPFC-HPC communication (Jones and Wilson, 2005) potentially weakening conclusions about independent beta rhythms.

Although many studies have examined the oscillatory coupling properties at the theta frequency between mPFC-HPC (Hyman et al., 2005; Jones & Wilson, 2005; Siapas et al., 2005), our understanding of beta frequency coordination between the two regions is less established, especially at goal locations. Our work suggests beta frequency coordination at goal locations does not share properties with those of theta frequency coupling between mPFC and HPC, which occurs primarily during movement on the maze. Our first novel finding is that beta oscillations occur at goal locations in mPFC and HPC; our second is that the beta frequency dynamics in these regions are surprisingly distinct. We are not aware of prior work describing these properties at goal locations in spatial navigation tasks, especially their temporal coordination.

Reviewer #3 (Recommendations for the authors):

(1) The conclusions from this article would be made much stronger if the authors (1) record from rats performing a task known to be dependent on mPFC-HPC communication (e.g. a spatial working memory task) or (2) record from the F-maze in well-trained rats (75-80% performance is common), or (3) show that Y-maze task performance is dependent on mPFC-hippocampal communication (see Maharjan et al., 2018, which used a W-track). It is possible that learning the Y-maze depends on mPFC-hippocampal communication, but that with asymptotic performance, this changes. This would put beta oscillation coupling findings into a nuanced perspective.

We appreciate the recommendation. The objective of this current manuscript is to present previously unknown properties of beta oscillations in hippocampal-prefrontal cortical networks. We agree that further investigation is required to fully dissect the functional contribution of beta dynamics in these networks. We are in the process of doing that.

The rule on the Y-maze in our experiment is identical to the continuous alternation rule on the W maze in Maharjan et al., 2018 and Kim and Frank, 2009 (Author response image 1), which showed the PFC and hippocampus are required for normal learning, respectively. The Y and W mazes share the same topology; there is one junction connecting three arms. The W maze has two 90-degree-angle turns which are not choice points. Thus, both maze tasks have one choice point and involve learning an alternation rule. Further the learning properties share similarities. For the Y-maze, the inbound portion (return to center) (Author response image 1) was quicker to learn than the outbound portion (alternation) (Author response image 1). The same pattern is observed in the W maze learning task (Maharjan et al., 2018, Fig. 3A and D, Kim and Frank, Fig. 4C). We agree an experiment is needed to show that Y-maze task learning is dependent on the function of hippocampal-prefrontal cortical networks. The shared topology, rule definition, and behavior profile suggest that the Y and W maze tasks engage similar learning processes.

Author response image 1.

W and Y maze alternation tasks share the same rule. Schematics illustrate a comparison of W and Y maze alternation task rules. The W maze has been inverted for visual comparison with the Y maze. Performance grouped by in- or outbound trials. Inbound trials originate from the side goals (2 or 3). Outbound trials originate from the center goal (1). Performance on the inbound trials was higher than that on the outbound trials, which is comparable to previously published alternation tasks on W-shaped mazes.

(2) Typically, when analyzing LFP profiles, experimenters include running velocity/speed. It should be ruled out whether spectral changes are confounded in any way by speed or time spent in the goal zones.

We appreciate this suggestion for better conveying our definition of goal period. This was raised by other reviewers. We have now included the speed profiles for the goal-entry-aligned spectrograms (Fig. 1B, 1-4, 2A, and 6A), as well as the speed quantification at the times of bursts (Fig. 3-1, Fig. 4-1) and SWRs (Fig. 6-1 D). We strictly define the goal period after entry into the sensor on the reward delivery device at the end of the arms. This is to ensure the animal is immobile for the period to avoid confounds related to speed. We also ensured the time periods are comparable between the trials we analyzed.

(3) The authors should describe in each analysis how data are sampled, and for extracellular electrophysiology experiments, cell counts from each rat reported. A table would be ideal. For example, it is unknown if entrainment analysis is performed on data collected from one rat or from all rats.

We have now added the missing information (Table 1-2, 5-1 to 5-3). We performed analyses using data from all animals.

(4) There was no profiling of mPFC neurons in terms of their behavioral correlates. The authors should strongly consider examining how individual neurons encode task variables (e.g., trial correctness, reward location...) and can do so using a generalized linear model. Adding an analysis of behavioral correlates could nicely tie into the beta-SWR analyses. For example, are SWR-beta rhythm-modulated neurons also behaviorally modulated?

This is a very helpful suggestion. We have added analyses on the behavioral correlates of PFC and CA1 neurons. We quantified three metrics: 1) whether PFC and CA1 spiking activity can distinguish goal location based on firing rate or phase preference, 2) firing distance to goal and 3) spatial information (Fig. 5 and 7). We also performed the same analysis based on SWR and beta modulation status. We found PFC cells that were both SWR- and beta-modulated showed the strongest task firing relationship (Fig. 7).

(5) Figure 2:

Are these data analyzed from well-trained rats? What is your N (rats/sessions/trials/epochs)?

These results in Fig. 2 are from all days. We added the breakdown in Table 2-1.

(6) Figure 3:

(a) The authors show that on the Y-maze, performance, beta oscillations power, and beta oscillation frequency change over days. However, for the F-maze, performance improves but appears to taper off at 60% and PFC beta frequencies do not change with learning. Do you have rats performing this task well above chance (e.g., 75-80%?), and if so, do beta oscillation frequencies in the mPFC gradually change?

We did not observe rats performing above chance on the F-maze over 5 days. They do not appear to learn the alternation rule on this task. This is why we used the F-maze as the “non-learner” control. The F-maze task design for this study appears to be difficult to learn and would likely require a much longer training period for the performance to exceed chance. We agree an important follow-up question is whether the effects reported here are generally observed across learning in different tasks. This is a future direction we are actively pursuing.

One existing data point may partially address the relevance of beta power change with learning. We analyzed the power as a function of performance quintile on the Y-maze (learned) or F-maze (not learned). The power changed with performance quintile on the Y-maze (learned) but not the F-maze (not learned) (Fig. 3D), pointing to the change in power being associated with learning status.

(b) Does the proportion of beta bursts change with learning? What about the proportion of coherent events?

We performed this analysis and the results are shown in Fig. 3-3. We found a decrease in the proportion of coincident bursts within CA1, which was specific to the well-performed Y-maze.

(c) Is the reduction in beta oscillation frequency in the mPFC related to running behavior? This should be ruled out. Same for the hippocampus.

The reduction is not due to running because we only included bursts when the animal was at the goal and immobile. We added a figure to show speed at the time of bursts (Fig. 3-1).

(d) Why don't you also show coherence as a function of learning?

This is a great suggestion. Coherence as a function of day or performance is now shown in Fig. 3-2. Overall, the trends were weak suggesting there was no strong change in coherence over days or as a function of learning. Although some of the linear mixed effects models were statistically significant, the marginal R² values (R²_m) were very low, indicating the effects of performance or day on coherence were small. Beta frequency coherence within each brain region either remained the same or slightly decreased across days (Fig. 3-2 D and F) or with performance (Fig. 3-2 J). For coherence across regions, we only found a weak but significant increase for the well-performed Y-maze task across days (Fig. 3-2 B).

(e) The statistics in the caption are great, but maybe consider using a table and a supplemental figure

We have now included the most relevant model output in the figure. These are the marginal R² (R²_m) and conditional R² (R²_c). These describe the contribution of the fixed or fixed and random effects on the model, respectively.

(7) Figure 4

(a) This figure is better suited as supplemental to Figure 3.

We have combined Fig. 4 with Fig. 3.

(b) Statistics might be better suited in a table and a supplemental figure.

We have now included the most relevant model output in the figure. These are the marginal R² (R²_m) and conditional R² (R²_c). These describe the contribution of the fixed or fixed and random effects on the model, respectively.

(8) Figure 5

(a) The shift of beta rhythms being linked to learning is interesting, albeit confusing, given that error trials were accompanied by even more 'precise' beta rhythm timing. Is it possible that beta rhythms are accompanied by an error signal? Or maybe instead related to running behavior?

We verified this observation was not due to speed since we only included periods when the animal was at the goal and immobile. We added a new supplemental figure (Fig. 4-1) to report this analysis.

(b) Changes in beta rhythm timing in the F-maze could be entirely explained by the data in the Y-maze, given that F-maze performance was, in general, very low. To test if beta rhythm shifting is a consequence of learning, the authors should show performance on the F-maze without the Y-maze, and vice versa.

As suggested, we also analyzed frequency variance data from each maze separately and found learning-related changes were more prominent on the Y-maze (learned) than on the F-maze (not learned) (Fig. 4-4 C). The F-maze data serves as a valuable control within the same animal for the learning-related effects we are reporting.

We agree that the Y-maze data is consistent with performance-related changes to beta timing. The addition of the F-maze provides a within-animal comparison for a task in which the animal performed less well. We agree an additional control group would provide further evidence to support learning-related changes in beta timing. However, the F-maze will likely take much longer to learn, therefore the additional days of exposure on the task will be a different confound. Thus, we wanted to ensure that we compared a “learned” with a “not learned” task within the same animal, with a comparable exposure period.

(9) Figure 6

(a) How many cells are analyzed here? Are they all pyramidal neurons? The authors should add the number of cells analyzed from each rat. A table would suffice.

We added tables (Table 5-2 to 5-4) to show this information. For our analysis, we included all units.

(b) The authors should include statistics about recorded units. Peak-to-trough timing and interspike interval are commonly used. Even demonstration of action potentials from individual cells is valuable for proof of concept.

We have added more information on the cell type classification (Table 5-2) and how we classified the cells (Methods) and also added example waveforms in Fig. 5 (previously Fig. 6).

(c) Why did the authors stop comparing the F-maze and Y-maze for entrainment analysis?

The data for each maze was comparable. Per the reviewer’s suggestion, we added the entrainment analysis for each maze separately (Fig. 5-1).

(10) Figure 7

(a) The authors should minimally include instantaneous velocity to show that during putative ripple events, the animal is in fact, quiescent.

We added the speed at the time of ripples in Fig. 6-1 D.

(b) Figure 7C: The authors should show how time spent in the reward zones varies over days. Is SWR power simply changing due to behavioral occupancy differences (e.g., a statistical sampling problem)? An analysis of the SWR rate over days would be valuable.

We agree and have added additional SWR properties across days in Fig. 6-1. This includes SWR power (Fig. 6-1 A), rate (Fig. 6-1 B), duration (Fig. 6-1 C) and speed (Fig. 6-1 D). To ensure we are comparing the equivalent time spent at goals, the SWR properties were calculated from the 10 s after goal entry (Fig. 6-1) and we only included goal visits that lasted at least 10 s. This controls for the potential confounds from occupancy differences across days.

(c) Why did the authors stop comparing the F-maze and Y-maze for SWR analysis?

The SWR properties were similar in both tasks. We have now added separate analyses for each maze in Fig. 6-1 and 6-2.

(11) Figure 8

(a) The authors discovered that mPFC neurons more strongly entrained to SWRs compared to their own beta rhythms, potentially indicating coordination of mPFC and hippocampus (lines 275:276). What about mPFC spiking to CA1 beta?

We quantified mPFC spiking entrainment to CA1 beta in Fig. 5 O and P. We found mPFC cells were more strongly entrained to the local beta within mPFC compared with CA1 beta.

(b) Lines 277:278: How did the authors arise at 8.6% being the expected proportion of neurons entrained by both SWRs and beta rhythms?

We have added a better explanation of how we arrived at the expected proportion. The calculation is based on the joint probability between the proportion of neurons modulated by SWRs (30%) and the proportion of neurons with phase locking to beta rhythms (22%). The expected proportion (6.6%) is calculated by multiplying the two values (30% ´ 22%) under the assumption that the two classes are independent. Deviations from the expected proportions are determined using a Fisher’s exact test.

We note that in the revised manuscript we reanalyzed the data with more stringent criteria. We now include SWRs with durations greater than 50 ms, rather than 15 ms, and we quantify beta spike phase-locking using only the spikes within the first 5 s after goal entry, for goal visits lasting at least 5 s.

With these more conservative criteria, the proportion of SWR- and beta-modulated cells (8%) is no longer significantly different from the expected proportion (6.6%, Fisher’s exact test p=0.09). Although this differs from our original significance test, the trend, and importantly, the distinct task correlates for this population (Fig. 7 C-D) still hold. We have updated the revised manuscript with these findings.

Our original result was: “The subset of PFC cells that are modulated by both SWR and beta (11%) is greater than the expected proportion (8.6%) under the assumption that SWR and beta can modulate the population independently (Fisher exact test, p=0.021), although the size of the difference is small.”

(c) The discovery that neurons modulated by both beta and SWRs are unique from those simply modulated by beta is really interesting. The authors discovered an interesting relationship among dually entrained mPFC neurons whereby SWR-excited mPFC neurons were entrained to the peak of beta, whereas SWR-inhibited mPFC neurons were entrained to the trough of beta. Then, in lines285:286, the authors write:

(12) "This relationship was not observed for PFC cells that are modulated by beta but not modulated by SWRs (Fig. 8C, right, Fig. 8-1A)."

Why would the authors expect this relationship to exist when the mPFC neurons were not SWR modulated? Were the authors referring to something else?

We should have phrased this more clearly. We edited the text to better convey the expected SWR and beta modulation patterns for the control populations. We wanted to ensure the relationship between SWR modulation direction and beta phase preference was not observed for the cells that were not SWR-modulated.

Furthermore, what about mPFC neurons that are SWR modulated but not beta modulated? For completeness, the authors should examine these.

We agree this is an important comparison and have now included this population in the new Fig. 7. As expected, we do not find any relationship between SWR modulation direction and spike preference to beta for the SWR-modulated and not beta-modulated population.

(13) Lines 294:296: The authors should elaborate on how they obtained an expected proportion of CA1 beta modulated neurons.

We have now added a description for calculating the expected number of beta-modulated CA1 neurons. These are the neurons that have a Rayleigh test p-value less than 0.05.

(14) What are these neurons doing to predict task information? Do they at all? Do they differ from beta-only neurons or non-phase-locked neurons?

This is a great suggestion, and we have added analyses to show the task correlates for the cells. We found brain region-specific differences for task correlates that depended on how the cell was modulated by SWRs or local beta oscillations. This is shown in the updated Fig. 7 and the accompanying supplemental Figs. 7-1 to 7-2.

For PFC cells, SWR and beta modulation status defined a subpopulation with a strong task structure correlation. There was a positive correlation between the direction of SWR modulation and the spiking distance relative to goals. SWR-excited cells were more active further away from goals, whereas the SWR-inhibited cells were active closer to goals. This correlation was not found for SWR-modulated PFC cells that were not beta-modulated. For PFC cells, the direction of SWR modulation is known to be correlated with movement speed, consistent with the hypothesis that movement-active cells become reactivated during SWRs, and immobility-active cells become suppressed (Jadhav et al., 2016; Yu et al., 2017). We found the same relationship, the direction of SWR modulation was positively correlated with mean spiking speed. Our results show beta modulation marks a subpopulation of PFC cells with stronger task-structure correlates.

For CA1 cells, we found the expected relationship between SWR modulation and task-structure correlates. SWR-excited CA1 cells spiked further away from the goal and when the animal was moving. This is consistent with the reactivation of trajectory-related spatial firing patterns during movement on the maze. This pattern was observed irrespective of beta modulation status.

Ahmed, O. J., & Mehta, M. R. (2012). Running speed alters the frequency of hippocampal gamma oscillations. J Neurosci, 32(21), 7373-7383. https://doi.org/10.1523/JNEUROSCI.5110-11.2012

Berke, J. D., Hetrick, V., Breck, J., & Greene, R. W. (2008). Transient 23-30 Hz oscillations in mouse hippocampus during exploration of novel environments. Hippocampus, 18(5), 519-529. https://doi.org/10.1002/hipo.20435

Deceuninck, L., & Kloosterman, F. (2024). Disruption of awake sharp-wave ripples does not affect memorization of locations in repeated-acquisition spatial memory tasks. Elife, 13. https://doi.org/10.7554/eLife.84004

den Bakker, H., Van Dijck, M., Sun, J. J., & Kloosterman, F. (2023). Sharp-wave-ripple associated activity in the medial prefrontal cortex supports spatial rule switching. Cell Rep, 42(8), 112959. https://doi.org/10.1016/j.celrep.2023.112959

França, A. S., do Nascimento, G. C., Lopes-dos-Santos, V., Muratori, L., Ribeiro, S., Lobão-Soares, B., & Tort, A. B. (2014). Beta2 oscillations (23-30 Hz) in the mouse hippocampus during novel object recognition. Eur J Neurosci, 40(11), 3693-3703. https://doi.org/10.1111/ejn.12739

França, A. S. C., Borgesius, N. Z., Souza, B. C., & Cohen, M. X. (2021). Beta2 Oscillations in Hippocampal-Cortical Circuits During Novelty Detection. Front Syst Neurosci, 15, 617388. https://doi.org/10.3389/fnsys.2021.617388

Hyman, J. M., Zilli, E. A., Paley, A. M., & Hasselmo, M. E. (2005). Medial prefrontal cortex cells show dynamic modulation with the hippocampal theta rhythm dependent on behavior. Hippocampus, 15(6), 739-749. https://doi.org/10.1002/hipo.20106

Iwasaki, S., Sasaki, T., & Ikegaya, Y. (2021). Hippocampal beta oscillations predict mouse object-location associative memory performance. Hippocampus, 31(5), 503-511. https://doi.org/10.1002/hipo.23311

Jadhav, S. P., Kemere, C., German, P. W., & Frank, L. M. (2012). Awake hippocampal sharp-wave ripples support spatial memory. Science (New York, N.Y.), 336(6087), 1454-1458. https://doi.org/10.1126/science.1217230

Jadhav, S. P., Rothschild, G., Roumis, D. K., & Frank, L. M. (2016). Coordinated Excitation and Inhibition of Prefrontal Ensembles during Awake Hippocampal Sharp-Wave Ripple Events. Neuron, 90(1), 113-127. https://doi.org/10.1016/j.neuron.2016.02.010

Jin, S. W., Ha, H. S., & Lee, I. (2024). Selective reactivation of value- and place-dependent information during sharp-wave ripples in the intermediate and dorsal hippocampus. Sci Adv, 10(32), eadn0416. https://doi.org/10.1126/sciadv.adn0416

Jones, M. W., & Wilson, M. A. (2005). Theta Rhythms Coordinate Hippocampal–Prefrontal Interactions in a Spatial Memory Task. PLoS Biology, 3(12). https://doi.org/10.1371/journal.pbio.0030402

Kim, S. M., & Frank, L. M. (2009). Hippocampal Lesions Impair Rapid Learning of a Continuous Spatial Alternation Task. PLoS ONE, 4(5). https://doi.org/10.1371/journal.pone.0005494

Lansink, C. S., Meijer, G. T., Lankelma, J. V., Vinck, M. A., Jackson, J. C., & Pennartz, C. M. (2016). Reward Expectancy Strengthens CA1 Theta and Beta Band Synchronization and Hippocampal-Ventral Striatal Coupling. J Neurosci, 36(41), 10598-10610. https://doi.org/10.1523/JNEUROSCI.0682-16.2016

Maharjan, D. M., Dai, Y. Y., Glantz, E. H., & Jadhav, S. P. (2018). Disruption of dorsal hippocampal-prefrontal interactions using chemogenetic inactivation impairs spatial learning. Neurobiol Learn Mem, 155, 351-360. https://doi.org/10.1016/j.nlm.2018.08.023

Rangel, L. M., Chiba, A. A., & Quinn, L. K. (2015). Theta and beta oscillatory dynamics in the dentate gyrus reveal a shift in network processing state during cue encounters. Front Syst Neurosci, 9, 96. https://doi.org/10.3389/fnsys.2015.00096

Siapas, A. G., Lubenov, E. V., & Wilson, M. A. (2005). Prefrontal Phase Locking to Hippocampal Theta Oscillations. Neuron, 46(1), 141-151. https://doi.org/10.1016/j.neuron.2005.02.028

Yu, J. Y., Kay, K., Liu, D. F., Grossrubatscher, I., Loback, A., Sosa, M.,…Frank, L. M. (2017). Distinct hippocampal-cortical memory representations for experiences associated with movement versus immobility. Elife, 6, e27621. https://doi.org/10.7554/eLife.27621

eLife Assessment

This study provides convincing evidence that the genetics of local adaptation in Arabidopsis is shaped by fluctuations in the environment and interactions with genotype and location. This is an important dataset contributing to the developing understanding of non-linear selection in plants and beyond.

Reviewer #1 (Public review):

[Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. The authors have addressed the comments raised in the previous round of review both by revising their original presentation where necessary and by providing responses to the reviewers in cases where they disagreed that revision of the original presentation would be necessary.]

Summary:

As a general phenomenon, adaptation of populations to their respective local conditions is well-documented, though not universally. In particular, local adaptation has been amply demonstrated in Arabidopsis thaliana, the focal species of this research, which is naturally highly selfing. Here, the authors report assays designed to evaluate the spatial scale of fitness variation among source populations and sites, as well as temporal variability in fitness expression. Further, they endeavor to identify traits and genomic regions that contribute to the demonstrated variation in fitness.

Strengths:

With many (200) inbred accessions drawn from throughout Sweden, the study offers an unusually fine sampling of genetic variation within this much-studied species, and through assays in multiple sites and years, it amply demonstrates the context-dependence of fitness expression. It supports the general phenomenon of local adaptation, with multiple nuances. Other examples exist, but it is of value to have further cases illustrating not only the context-dependence of fitness expression but also the sometimes idiosyncratic nature of fitness variation. I commend the authors on their cautionary language in relation to inferences about the roles of particular genomic regions (e.g.l.140-144; l.227)

Reviewer #2 (Public review):

Summary:

The goal of this study was to find evidence for local adaptation in survival and fecundity of the model plant Arabidopsis thaliana. The authors grew a large set of Swedish Arabidopsis accessions at four common garden sites in northern and southern Sweden. Accessions were grown from seed in trays, which were laid on the ground at each site in late summer, screened for survival in fall and the following spring, and fecundity was determined from rosette size and seed production in spring. Experiments were complemented by 'selection experiments', in which seeds of the same accessions were sown in plots, and after two years of growth, plants were sampled to determine fitness from genotype frequencies, providing a more comprehensive evaluation of lifetime fitness than can be gleaned from fecundity alone.

As the main result, southern accessions had higher mortality in northern sites in one of two years, but also suffered more slug damage in southern sites in one year, indicating a potential link between frost tolerance and herbivore resistance. Fecundity of accession was highest when growing close to the 'home' environment, but while accessions from one sand dune population in southern Sweden had among the lowest fecundities overall, they consistently had the highest fitness in the selection experiment. Accessions from this population had large seed size and rapid root growth, which might be related to establishment success when arriving in a new, partially occupied habitat. However, neither trait could fully explain the very high fitness of this population, suggesting the presence of other, unmeasured traits.

Overall, the authors could provide clear evidence of local adaptation in different traits for some of their experiments, but they also highlight high temporal and spatial variability that makes prediction of microevolutionary change so challenging.

Strengths:

A major strength of this study is the highly comprehensive evaluation of different fitness-related traits of Arabidopsis under natural conditions. The evaluation of survival and fecundity in common garden experiments across four sites and two years provides an estimate of variability and consistency of results. The addition of the 'selection experiment' provides an extended view on plant fitness that is both original and interesting, in particular highlighting potential limitations of 'fitness-proxies' such as seed production that don't take into account seedling establishment and competitive exclusion.

Throughout the study, the authors have gone to impressive depths in exploring their data, and particularly the discovery of 'native volunteers' in selection experiment plots and their statistical treatment is very elegant and has resulted in compelling conclusions. Also, while the authors are careful in the interpretation of their GWAS results, they nonetheless highlight a few interesting gene candidates that may be underlying the observed plant adaptations, and which likely will stimulate further research.

Overall, this study will likely make an important contribution to the field of evolutionary biology, and it is another very strong example of how the extensive molecular tools in Arabidopsis can be leveraged to address fundamental questions in evolution and ecology, to an extent that is not (yet) possible in other plant systems.

Reviewer #3 (Public review):

Summary:

The manuscript presents a large common garden experiment across Sweden using solely local germplasm. Additionally, there is a collection of selection experiments that begin investigating the factors shaping fecundity in these populations. This provides an impressive amount of data and analysis investigating the underlying factors involved. Together, this helps support the data showing that fluctuations and interactions are key components determining Arabidopsis fitness and are more broadly applicable across plant and non-plant species.

Strengths:

The field trials are well conducted with extensive effort and sampling. Similarly while the genetic analysis is complex it is well conducted and reflects the complexity of dealing with population structure that may be intricately linked to adaptive structure. This has no real solution and the option of presenting results with and without correction is likely the only appropriate option.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

As a general phenomenon, adaptation of populations to their respective local conditions is well-documented, though not universally. In particular, local adaptation has been amply demonstrated in Arabidopsis thaliana, the focal species of this research, which is naturally highly selfing. Here, the authors report assays designed to evaluate the spatial scale of fitness variation among source populations and sites, as well as temporal variability in fitness expression. Further, they endeavor to identify traits and genomic regions that contribute to the demonstrated variation in fitness.

Strengths:

With many (200) inbred accessions drawn from throughout Sweden, the study offers an unusually fine sampling of genetic variation within this much-studied species, and through assays in multiple sites and years, it amply demonstrates the context-dependence of fitness expression. It supports the general phenomenon of local adaptation, with multiple nuances. Other examples exist, but it is of value to have further cases illustrating not only the context-dependence of fitness expression but also the sometimes idiosyncratic nature of fitness variation. I commend the authors on their cautionary language in relation to inferences about the roles of particular genomic regions (e.g.l.140-144; l.227)

Weaknesses:

To my mind, the manuscript is written primarily for the Arabidopsis community. This community is certainly large, but there are many evolutionary biologists who could appreciate this work but are not invited to do so. The authors could address the broader evolution community by acknowledging more of the relevant work of others (I've noted a few references in my comments to the authors). At least as important, the authors could make clearer the fact that A. thaliana is (almost) strictly selfing and how this feature of its biology both enables such a study and also limits inferences from it. Further, it seems to me that though I could be wrong, readers would appreciate a more direct, less discursive style of writing, and one that makes the broader import of the focal questions clearer.

We agree that connecting the paper better to the broader field is desirable, and have tried to do this in the revision, including adding a brief description of A. thaliana that mentions the fact that it is highly selfing. That the availability of inbred lines influences the study design is fairly obvious. As for the inferences, we would argue that our main point—“know your organism”—is universal. For example, our selection experiments and common garden experiments produced diametrically opposite results in terms of fitness, and we attribute this to the importance of seedling establishment. It is hardly surprising that seedling establishment matters for an annual plant that favors disturbed habitats, but this is probably true for both outcrossers and selfers (although it is probable that selfing itself is an adaptation to such habitats), and could well be true for some long-lived trees as well. The Devil is in the details.

As a reader, I would value seeing estimates of the overall fitness of the accessions in the different conditions, i.e., by combining the survival and fecundity results of the common garden experiments.

Combining estimates would be possible in the common garden experiments, and would bring us somewhat closer to total fitness estimates, although as noted by another reviewer (and also emphasized by us), the time scale of our experiment is not sufficient to evaluate the trade-off between survival and fecundity. Furthermore, we would still be missing the establishment component of fitness, which we found to be extremely important. Therefore little would be gained by combining the estimates, while at the same time losing resolution to disentangle the fitness components. We thus decided to focus on the individual fitness components and leave a qualitative consideration of their joint effect for the Discussion.

Reviewer #2 (Public review):

Summary:

The goal of this study was to find evidence for local adaptation in survival and fecundity of the model plant Arabidopsis thaliana. The authors grew a large set of Swedish Arabidopsis accessions at four common garden sites in northern and southern Sweden. Accessions were grown from seed in trays, which were laid on the ground at each site in late summer, screened for survival in fall and the following spring, and fecundity was determined from rosette size and seed production in spring. Experiments were complemented by 'selection experiments', in which seeds of the same accessions were sown in plots, and after two years of growth, plants were sampled to determine fitness from genotype frequencies, providing a more comprehensive evaluation of lifetime fitness than can be gleaned from fecundity alone.

To clarify, fecundity was determined from total plant area using photos of the mature stems, not the rosettes or direct counting of seeds. That said, it is true that our fecundity estimate was well correlated with rosette area. Furthermore, we validate our fecundity estimates by showing they were highly correlated with seed production estimated by measuring and counting siliques on a separate set of plants grown under common garden conditions in one of our sites (Brachi et al. 2022).

As the main result, southern accessions had higher mortality in northern sites in one of two years, but also suffered more slug damage in southern sites in one year, indicating a potential link between frost tolerance and herbivore resistance. Fecundity of accession was highest when growing close to the 'home' environment, but while accessions from one sand dune population in southern Sweden had among the lowest fecundities overall, they consistently had the highest fitness in the selection experiment. Accessions from this population had large seed size and rapid root growth, which might be related to establishment success when arriving in a new, partially occupied habitat. However, neither trait could fully explain the very high fitness of this population, suggesting the presence of other, unmeasured traits.

Another clarification: the small set of “beach” accessions that performed well in the selection experiments came from several beaches on the Baltic coast of Skåne, and we also had one such accession from the island of Gotland, 400 km away as the bird flies. Furthermore, we sampled measured seed size in several hundred additional accessions and demonstrated that large seeds are indeed characteristic of this habitat.

Overall, the authors could provide clear evidence of local adaptation in different traits for some of their experiments, but they also highlight high temporal and spatial variability that makes prediction of microevolutionary change so challenging.

Strengths:

A major strength of this study is the highly comprehensive evaluation of different fitness-related traits of Arabidopsis under natural conditions. The evaluation of survival and fecundity in common garden experiments across four sites and two years provides an estimate of variability and consistency of results. The addition of the 'selection experiment' provides an extended view on plant fitness that is both original and interesting, in particular highlighting potential limitations of 'fitness-proxies' such as seed production that don't take into account seedling establishment and competitive exclusion.

Throughout the study, the authors have gone to impressive depths in exploring their data, and particularly the discovery of 'native volunteers' in selection experiment plots and their statistical treatment is very elegant and has resulted in compelling conclusions. Also, while the authors are careful in the interpretation of their GWAS results, they nonetheless highlight a few interesting gene candidates that may be underlying the observed plant adaptations, and which likely will stimulate further research.

Overall, the authors provide a rich new resource that is relevant and interesting both in the context of general evolutionary theory as well as more specifically for molecular biology.

Weaknesses:

While the repetition of the common garden experiments over two years is certainly better than no repetition (hence its mention also under 'strengths'), the very high variability found between the two years highlights the need for more extensive temporal replication. In this context, two temporal replicates are the bare minimum, and more repeats in time would be necessary to draw any kind of conclusion about the role of 'high mortality' and 'low mortality' years for the microevolution of Arabidopsis. It also seems that the authors missed an opportunity to explore potentially causal variation among years, as they did not attempt to relate winter mortality to actual climatic variables, even though they discuss winter harshness as a potential predictor.

We agree that two years is insufficient to understand how variation in selective pressures compound over time to generate micro-evolutionary change. The eight-year data in Oakley et al. (2023), which we discuss in the paper, support this. Our results are nonetheless sufficient to demonstrate the idiosyncratic nature of selection. In the revision, we further emphasize that far longer time series would be needed for definitive conclusions on long-term micro-evolutionary change.

Our short time series is one reason why we do not try to correlate with climate data, as this would amount to doing statistics with four data points (mostly two groups of accession N vs S, with mostly homogenous climates within groups, and two years). Another reason is that we simply have no idea how the kind of meteorological data that are publicly available would affect plant performance on a local scale. Hence we make no claims..

The low temporal variation also makes the accidental slug herbivory appear somewhat random. Potted plants are notoriously susceptible to slug herbivory, and while it is certainly nice that slug damage predominantly affected one group of accessions, it nonetheless raises the question whether this reflects a 'real' selection pressure that plants commonly face in their respective local environments.

Characterizing the plants as “potted” is not accurate. A more serious objection is having what is effectively an A. thaliana monoculture. But indeed we have no idea whether slugs exert a significant selective pressure on A. thaliana in Sweden, and we make no claims to that effect. The evidence for selection on glucosinolates by generalist herbivores such as slugs is fairly strong, but the precise agent is not known, and probably varies over time and space. Our results merely demonstrate one possibility.

The addition of the 'selection experiment' is certainly original and provides valuable additional insights, but again, it seems a bit questionable which natural process really has affected this outcome. While the genetic and statistical analysis of this experiment seems to be state-of-the-art, the experimental design is rather rudimentary compared to more standard selection experiments. Specifically, the authors added seeds from greenhouse-grown mothers to experimental plots and only sampled plants two years later. This means that, potentially, the first very big bottleneck was germination under natural conditions, which may have already excluded many of the accessions before they had a chance to grow. While this certainly is one type of selection, it is not exactly the type of selection that a 2-year selection experiment is set up to measure. Either initially establishing the selection experiment from plants instead of seeds, or genotyping the population over several generations, would have substantially strengthened the conclusions that could be drawn from this experiment.

We 100% agree that more data would have been beneficial, and hence we do not make any claims about the nature of selection. The selection experiment was an experiment per se, and we were lucky that very large fitness differences turned out to exist. As for initial selection on dormancy “set” incorrectly in greenhouse-grown seeds, we agree that this is a possibility, but we do not think it is likely to explain the data for several reasons. First, why would these differences favor a small set of beach accessions over every other accession in four very different field sites? Second, existing dormancy estimates do not predict fitness in our selection experiments. Third, the same seed batches germinated uniformly in the common-garden experiments with minimal stratification. Fourth, a much simpler explanation—seed size—exists. We clarify this in the revision while retaining our original message that further experiments are needed (and are underway).

Also, the complete lack of information on population density is a bit problematic. It is not clear if there were other (non-Arabidopsis) plants present in the plots, how many Arabidopsis plants were established, if numbers changed over the year, etc. Given all of these limitations, calling this a 'selection experiment' is in fact somewhat misleading.

Seeds were introduced into sites that appeared appropriate for A. thaliana, leaving the background community intact. We provided information on sowing density; the density of plants (A. thaliana and other species) that we obtained during the course of the experiments varied considerably between sites, much like in natural populations, although we lack systematic measurements. We provide more information (including photos) in the revision.

Despite these weaknesses, the authors could achieve their main goals, and despite the somewhat minimal temporal replication, they were lucky to sample two fairly distinct years that provided them with interesting variation, which they could partially explain using the variation among their accessions. Overall, this study will likely make an important contribution to the field of evolutionary biology, and it is another very strong example of how the extensive molecular tools in Arabidopsis can be leveraged to address fundamental questions in evolution and ecology, to an extent that is not (yet) possible in other plant systems.

Reviewer #3 (Public review):

Summary:

The manuscript presents a large common garden experiment across Sweden using solely local germplasm. Additionally, there is a collection of selection experiments that begin investigating the factors shaping fecundity in these populations. This provides an impressive amount of data and analysis investigating the underlying factors involved. Together, this helps support the data showing that fluctuations and interactions are key components determining Arabidopsis fitness and are more broadly applicable across plant and non-plant species.

Strengths:

The field trials are well conducted with extensive effort and sampling. Similarly while the genetic analysis is complex it is well conducted and reflects the complexity of dealing with population structure that may be intricately linked to adaptive structure. This has no real solution and the option of presenting results with and without correction is likely the only appropriate option.

Weaknesses:

A significant finding from this study was that fecundity is shaped more by yearly fluctuations and their interaction with genotype than it is by the main effect of location or genotype. Another significant finding is that the strength of selection can be quite strong, with nearly 5x ranges across accessions. It should be noted that there are a number of other studies using Arabidopsis in the wild with multiple years and locations that found similar observations beyond the Oakley citation. In general, the context of how these findings relate to existing knowledge in Arabidopsis is a bit underdeveloped.

We have tried to remedy this in the revision (see also comments by Reviewer #1).

The effects of the populations across the locations seem to rely on individual tests and PC analysis. It would seem to be possible to incorporate these tests more directly in the linear modeling analysis, and it isn't quite clear why this wasn't conducted.

We respond to this question below, in Recommendations for the Authors (first item from you).

I'm a bit puzzled by the discussion on how to find causative loci. This seems to focus solely on GWAS as the solution, with a goal to sequence vast individuals. But the loci that the manuscript discussed were found by a combination of structured mapping populations followed by molecular validation that then informed the GWAS. As such, I'm unsure if the proposed future approach of more sequencing is the best when a more balanced approach integrating diverse methods and population types will be more useful.

We are puzzled by this comment in return. Our statement about more sequencing (penultimate sentence of discussion) was referring to achieving a better understanding of the history of migration and selection rather than identifying causative loci.

Recommendations for the authors:

Reviewing Editor Comments:

The reviewers provide consistent and complementary recommendations that you should consider in a revision. In particular, you should try to better embed your results into the broader literature, both from Arabidopsis and other plant species. Also, try to make your text more accessible to readers outside the special topic.

Agreed!

Reviewer #1 (Recommendations for the authors):

(1) l.545: Here, the text states, importantly, that the accessions were randomized in the greenhouse. However, I have found no mention of this in the referenced Brachi et al. paper. I'm concerned that this be reported accurately and also that, if the accessions were not, in fact, randomized in the greenhouse, the potential for environmentally induced maternal influences to be confounded with genetically based differences be acknowledged, esp. in the case of the seed size difference found for B accessions.

Accessions were continuously randomized in the greenhouse although we stopped moving them when they flowered to reduce the risk of contamination. All the seeds for all the accessions were produced in the same greenhouse, in the same conditions, in one single planting. We state this clearly in the revised paper.

As for the seed size variation, note that the measurements we use in the paper were not taken on the seed we used in the experiments, so there is no reason they should be correlated with fitness if fitness was influenced by maternal effects. Moreover, we have plenty of data demonstrating that seed size variation is mostly genetic, and we added a supplementary figure showing that the measurements we used are strongly correlated with those of other experiments, including one done in the field. Finally, we already included a figure demonstrating that beach populations generally have much larger seeds.

(2) l.593: BLUPs are estimated with statistical uncertainty. I realize that it would be far from straightforward to take into account their sampling variances in the GWAS, but it should be acknowledged that ignoring the uncertainty of the BLUPs has an unknown impact on the findings from the GWAS.

All phenotypic measures come with error, and we had more replication than most GWAS studies (including essentially all human GWAS). The effect of such error is to reduce heritability and decrease the power of GWAS.

(3) l.48: As an earlier important reference demonstrating this point: Antonovics, Clay, Schmitt, 1987 Oecologia.

Indeed: added. Thanks!

(4) l.49: studied rapid evolution in a natural population[s] - delete 's'

Done.

(5) l.502: lead -> led

Corrected.

(6) l.53: The present study sought to gain insight into local adaptation in A. thaliana.; Vague, as is the rest of the intro.

Agreed. As noted above, the intro has been extensively reworked.

(7) l.533: What is alpha?

Alpha is the regularization parameter for the snmf algorithm. We added this clarification.

(8) l.553: Were the accessions also randomized in the field planting?

Yes, and we now state this.

(9) l.597: As I've noted in my public comments, the authors have done well to couch their genomic inferences with caveats. In line with such caution, I urge the authors to consider alternate phrasing to 'genetically determined' -> genetically influenced. Also l. 166: 'control of' -> influence on (among other instances).

No, “genetically determined” was correct on line 597, as this describes the model assumptions. But note that the liability-threshold model is by no means genetically deterministic: it is widely used in epidemiology to model genetic predisposition to diseases caused by the environment. For example, whether you get lung cancer or not primarily depends on luck and on your exposure to pollution (in particular smoking), but there are genes that influence how susceptible you are (to pollution; there are no genes that influence luck). There were some words missing in the sentence; hopefully things are clearer now.

Re l. 166, "control of” was changed to “effect on”, which is more accurate. Note, however, that there is nothing genetically deterministic here: the context is a variance-partitioning, and the results show very clearly that genetic factors play a minor role relative to environmental ones—and that most of the variance remains unexplained. (Luck?)

(10) l.595-602: Not clear.

As noted, some words were missing. Hopefully it is clearer now.

(11) l.744: Could have been confounded by cryptic native -- missing or extra word? Also next line.

Corrected.

(12) l.755-7: More direct phrasing would help readers understand this point.

Indeed. Sentence straightened out.

(11) l.129: 'The peak appears to involve a haplotype over 30 kb in length, which is consistent with a history of strong selection on this locus'. I don't understand the logic here. In what way should the length of the haplotype relate to the strength of selection? I would think this relates quite directly to the high degree of inbreeding.

Strong selection causes rapid allele-frequency change, leaving less time for recombination to break up associated haplotypes, causing increased linkage disequilibrium. This is standard population genetics (e.g., Maynard Smith and Haigh 1974). Inbreeding also causes increased haplotype sharing, but genome-wide. Telling them apart is difficult, hence we used “consistent”. However, Reviewer 3 points to the existence of a segregating inversion in this region, which is an even more likely explanation, and we focus on this in the revision.

(12) l.132: 'This suggests that the variation for slug damage seen in Figure 4 may be partly mediated by glucosinolate production.' I find the logic unclear here, as well.

“This” referred to several observations, which is poor English. We rearranged the sentences to make our meaning clearer.

(13) l.173: 'the accession-effect on fecundity' -> variation among accessions wrt fecundity.

We rewrote this clunky sentence.

(14) l.175: PCA unclear. Is this the same PCA described at l. 638? That is the only mention of PCA that I find in the methods, but I don't see that it connects here.

No this doesn’t refer to the same analysis. In line 175 we talk about a vanilla, textbook PCA: confronted with 200 fecundity estimates in 8 experiments, we used PCR to look for patterns across the 8 dimensions, and found that 3 dimensions captured most of the variation (as shown in the heatmap). We have added a few sentences about this in the methods. Line 638 describes the way images of plants were treated to estimate variation in rosette color.

(15) l.188: 'reveals what is causing them': Consider rephrasing to avoid language of causality, consistent with care taken elsewhere.

Well, the context here is very different: we are effectively doing a post hoc analysis, and there is nothing wrong with saying that “the significant value in the chi-square test is caused by an excess of…”, for example. The patterns we see in the PCA are caused by the kinds of patterns we go on to discuss—it captures these patterns, among other things. We changed “reveals” to “helps us understand” to be less biblical.

(16) l.219: downstream of -> conditional on.

Yes.

(17) l.235: Although native plants WERE growing nearby.

Yes.

(18) l.264: nearby: make this more explicit, i.e. within x km.

We changed the sentence to “Although native plants were growing within less than a hundred meters in most cases,...”

(19) l.285: none of our MAJOR conclusions depend on this.

We disagree. No conclusion in the paper could be confounded by potential natives.

(20) l.315: albeit it not as large as B accessions: delete 'it'.

Corrected.

(21) l.355: genetic basis of fitness -> genetic contribution to variation in fitness.

No, this is pretty much exactly Lewontin’s usage—the sentence (and section) is about changes in allele frequencies. The alternative suggested is less precise as we have no fitness estimates, and cannot say anything about variance.

(22) l.380-2: It seems to me that the authors could articulate a more compelling case.

Well, what we wrote is the truth. Of course all of this work also fits into a broader context, but those were the specific goals—and we achieved them. There are many papers that ask big questions, but actually answer much more limited ones.

(23) l.401: latitude vs. latitude???

Oops. Changed to “north vs. south”, which is hopefully less obscure!

(24) l.402: 'textbook local adaption,' -- the authors should acknowledge that rigid/simplistic thinking about LA has been recognized as a caricature for quite some time.

We do. Using “textbook” is meant to convey this: textbooks tend to be full of cartoonish simplifications, and most chapters on local adaptation will have a figure (cartoon or based on real data) showing reaction norms as two crossing lines.

(25) l.406: could maintain variation AMONG POPULATIONS [right?]

Could be either, but “among populations” is better in this context.

(26) l.419: Is there a possibility of a source envt maternal effect? See my comment on l.545.

As stated in response to the previous comment: In principle yes, but if so, then this environmental maternal effect happens to be strongly correlated with seed size, a trait we know to be genetically controlled and which is also extremely likely to influence seedling establishment. Experiments to confirm these results are underway—meanwhile we would be happy to accept bets against!

(27) l.434: 'very stable environments dating back to the last glaciation.' REF?

Ha! That would be Wikipedia references as the precise location of the “Littorina sea” and existence of obvious ancient beachlines now inland have entertained Swedish and Danish school children for generations. But the details are not important: the point is that beaches are vast (by A. thaliana standards) disturbed habitats maintained by the sea, and while they change, they do so on a geological time scale. Of course not all beaches harbor A. thaliana—the beaches of Hanö Bay are geologically unusual for Sweden—but this is beyond the scope of this publication. The sentence has been changed to something less specific making the relevant points.

(28) l.436: 'It is likely that the existence of S1 and S2 accessions is far more uncertain': unclear wording.

Yup. We rewrote the paragraph.

(29) l.446-7: 2nd person is jarring.

Reworded.

(30) ll.452-474: This is a welcome acknowledgement of the limits of molecular approaches to elucidating selection. It would be good scholarship to acknowledge earlier authors making such a case. I could suggest Rockman 2012. Evolution; Travisano and Shaw 2013. Evolution; Hoban et al. 2016. Am.Nat., and there are others.

True; added; thanks!

(31) l.490: 'might have to run a gauntlet of linkages to genes directly involved in local adaptation': This teleological wording is especially jarring.

(32) l.494: 'resistance allele to sweep': In concluding this manuscript, it does not seem appropriate to use as an example a single locus case. More broadly, I question the value/effectiveness of this paragraph as a conclusion to this manuscript.

We agree. The paragraph, gauntlets and all, has been replaced by discussion of why dissecting adaptive traits is hard.

Reviewer #2 (Recommendations for the authors):

(1) At the end of the introduction, the authors briefly outline the geographic scale of their study and mention both common garden and experimental evolution plots, but at this point in the manuscript, it is not clear how these differ, and as the methods only come at the very end, it would be better if some more details are provided here. Specifically, it could be mentioned explicitly here that common garden experiments consisted of placing greenhouse-grown plants in pots on the ground (but not burying/planting them), whereas selection experiments were established by sprinkling seeds in 1m 2 plots. This doesn't really become clear anywhere except in the methods, but this is fairly important for the interpretation of results.

We agree, and have expanded the description of the experiments in the legend to Figure 1, which also links to supplementary photos of the sites. We also note that the differences between the common-garden and the selection experiments should not be exaggerated. In particular, we did not use “greenhouse-grown plants”, but seedlings that had been allowed to establish outdoors in sheltered conditions, and we did not put “pots on the ground” but buried trays with holes in the bottom so that the plants could root in local soil. The main differences between the experiments are guaranteed establishment and lack of competition (from conspecific and other plants). This has also been clarified in the figure legend and in Methods.

(2) The authors discuss a difference in mortality between the two years of their common garden experiment and suggest that harsher winters could have been the cause of mortality in the north. However, they do not provide meteorological data to support this. Was the 2011-2012 winter harsher than the 2012-2013 winter, and did NM experience harsher conditions?

The problem is that we do not know what constitutes harshness from the point of view of the plant. Winters differ in many ways: temperature, duration, snow cover, etc. We have changed the relevant paragraph to make clear that our observations are consistent with those of Oakley et al., and that some aspect of winter weather is a plausible explanation.

(3) The authors find an indication for the involvement of the AOP cluster in overwinter survival, which, among other things controls the accumulation of hydroxyl/alkenyl/methylsulfinyl glucosinolates. Is the chemotype of the accessions in this study known?

Indeed they are! We had downloaded the data from Katz et al (2021), but the analysis did not make it into the first version of this paper. Thanks for the encouragement. We added a figure showing that their chemotypes are strongly associated with slug damage, supporting a causal relationship.

(4) L68: 'we added one field site in eastern Skane' - which site does this refer to? It seems odd to mention this before the common garden/selection experiments are mentioned. It should be clear that this is specifically referring to the latter.

Yes, this was out of place. The site is discussed later, when it becomes relevant.

(5) Figure 1: It would be useful if common garden and experimental evolution sites used different colors. In contrast, the use of colors for seasons in part C is unnecessary and distracting, as the red and blue colors for fall and winter are very similar to the colors for S1 and B.

Fixed.

(6) L90: The authors discuss mortality, but figures show survival. This seems an unnecessary complication for the reader, and the same unit should be used for discussion and presentation.

Survival is one minus mortality. We trust the readers to be able to do this conversion.

(7) L99: 'the converse was not true' - it is not immediately clear what this refers to. Referring to the non-linear relationship in panel 3a would make this clearer.

It refers to the result that the accessions with high mortality in NM did not necessarily have high mortality in NA. We think this is clear from the sentence in question.

(8) L108: 'S2 accession were also strongly affected' - this can be gleaned from the figures, but it is not immediately obvious as they are relatively complex. Could mean mortality/survival rates be provided to facilitate this?

We could but, throughout the paper, we have attempted to improve readability by not interrupting the text with numbers or repeating details that are presented in the figures.

(9) L117: 'however, an indirect association is likely' - what is meant by this?

We meant that it seems more likely that some accessions are more sensitive to stress regardless of source. This has been clarified.

(10) Figure 6: What are the two horizontal bars? What are dashed lines? Provide appropriate labels and a figure legend.

The top is a zoom-in of the bottom and the dashed lines outline the zoomed-in region. Clarified in legend.

(11) L174: Define 'BLUPS' here.

Done.

(12) L198: 'S1 and S2 accessions generally had higher fecundity in 2012-2013 than in 2011-2012' - this is not obvious from Figure 9. Especially for S1 (dark blue presumably), there appears to be no difference visible between years.

Changed text to note that differences are sometimes small, but that the stated pattern is seen in 13 out of 16 comparisons (which has p = 0.01 using a sign test).

(13) L401: 'latitude vs. latitude' - what is meant by this? Or is this a mistake?

Northern vs. southern. This has been clarified. Also noted by Reviewer 1.

(14) L436: 'the existence of S1 and S2 accessions is much more uncertain' - this is a bit of an odd expression. Could it be rephrased?

This has been clarified and the paragraph rewritten. Also noted by Reviewer 1.

(15) L673: Were plots tilled before sowing, or cleared of vegetation? If not, what other vegetation was present at sowing?

No clearing of vegetation was done except at the SR site, which was a weedy agricultural field. This and other information has been added. No vegetation surveys were made, but we added some more photos to give an idea of what the sites were like.

Reviewer #3 (Recommendations for the authors):

(1) In the section on lines 156-190 the large dataset is analyzed by accession in the linear model presented in Figure 6. Then, in Figure 8, the proposed populations appear to be tested individually in each experimental unit, leading to the probability argument in line 193. Is there a reason not to simply have accession nested within population in the model shown in Figure 6 as a way to directly test the between vs within population level components influencing the model? If the accessions are different, then it would be viable to treat this as random rather than fixed. Similarly, the field sites could be parsed into subsets as well.

Excellent question. We thought a lot about this. Fig. 6 presents a standard ANOVA that simply shows that the overall pattern is what we hoped for: very large effects of site, year, and accession, plus substantial interaction effects. This motivates the exploratory analysis presented in Figs. 7-9, where we focus on how the relative performance of accessions within experiments depended on the fixed effects (year and site) and whether this matched their (arbitrarily but independently defined) group designation—as would be expected under local adaptation.

We could explicitly add “group” to the model, as suggested, but this would give it a reality we do not think it deserves. Site, year, and accession are very much real, whereas group is the outcome of a somewhat arbitrary clustering of genotypes (i.e., accessions), analogous to race in human genetics, but without the sociological factors that sometimes warrant including race in a model not only because of direct genetic effects. Of course we could try to partition genotype and phenotype into within- and between-population variation using the classical quantitative genetics framework (i.e. Fst/Qst), but then we should have an a priori definition of “population”, which we don’t have. In addition to this fundamental objection, limited experimentation suggests that fitting a far more complex, nested mixed-effects model to our data is difficult in practice.

Thus we prefer our original approach. We have changed the writing to clarify our logic.

(2) Similarly, I'm not quite sure that the PC analysis is helping as the section is somewhat difficult to read given that the same impressions from Figure 7 are more explicitly shown in Figures 8 and 9.

Well, the PCA (Fig. 7) is what led us to the analyses in Figs 8-9, where we interpret the PCs in terms of group behavior. We have tried to clarify our logic in writing.

(3) Line 87-88 - An honest question, what is considered substantial divergence? The Fst values range from 0.05 to 0.27, suggesting that the divergences range across a spectrum and not all are substantial.

We removed the sentence. That the divergence is substantial enough to make GWAS difficult becomes clear later (although the extent to which this is due to selection rather than marker divergence is not clear).

(4) Line 129-130 - The 30kb AOP haplotype identified is likely the inversion associated with the major phenotypic variation identified in Sweden within Katz et al 2021. As such, the local LD structure may represent blocked recombination as much as linked selection.

Of course; we missed that. Thanks!

(5) Line 131-132 - I'm not quite sure what the evidence is that the MAM locus is more important.

The AOP and MAM loci are epistatic, and both have been found to have influence on fitness in Arabidopsis and Brassica ssp in the field, along with sequence signatures of selection for both loci in multiple Brassicaceae. I'm unsure if this statement is supported.

Indeed. This was a lazy and misleading reference to the fact that the MAM peak in Katz et al explains more of the variance than the AOP peak. It has been consigned to the dustbin of history. Instead we have added a figure showing that the glucosinolate profiles presented in that paper are highly correlated with slug damage in our study. Based on these results, I believe AOP explains more of the variation, but we leave pursuing this for those directly working on these pathways. The correspondence between our studies is certainly a nice confirmation of both.

(6) Line 131-132 - It should also be noted that the MAM locus is variable in Sweden, albeit having multiple independent haplotypes that convergently create the same phenotype. There is an indication from Gloss 2022 that these haplotypes may create small-effect phenotypic variation.

Correct again. What we meant to say was that the major polymorphism that is responsible for the highly significant MAM peak does not appear to segregate in Sweden, hence it is not surprising that we do not find an association at this locus either. We now say this. Needless to say, there could still be multiple variants at this locus that we do not have the power to detect.

eLife Assessment

This important study demonstrates that pyroglutamylation of peptide ligands alters G protein-coupled receptor (GPCR) activation in a subtype-specific manner and proposes a novel mechanism whereby the posttranslational modification steers the distal conformation of the peptide within the orthosteric binding site. The strength of the evidence supporting receptor-specific functional effects is convincing, but the strength of the evidence supporting the proposed structural mechanism is incomplete. This study will be of interest to pharmacologists and cell biologists interested in posttranslational modifications (PTMs) and activation mechanisms of GPCRs.

Reviewer #1 (Public review):

Summary:

The study identifies two previously uncharacterized endogenous receptors for Aplysia PRXamide peptides and shows that N-terminal pyroglutamylation can produce opposite effects on receptor activation. The authors further propose that this modification acts indirectly by altering peptide conformation and that receptor pocket properties determine the direction of its effect. The findings are potentially significant for understanding how peptide modifications influence receptor selectivity, but the evidence supporting the proposed molecular mechanism is not yet sufficiently strong.

Strengths:

The identification and functional characterization of the two receptors are valuable. The contrasting effects of pyroglutamylation, together with peptide and receptor mutagenesis and computational analyses, provide an interesting framework for investigating peptide-receptor selectivity.

Weaknesses:

Most of the statements about novelty and generality are stronger than warranted by the current data. The central mechanistic model relies heavily on computational predictions and indirect functional measurements, without direct structural or receptor-proximal evidence. In addition, the human receptor data provide only partial support for the proposed mechanism, particularly because the pyroglutamylated and non-pyroglutamylated peptides are not significantly different at human receptor 2. Thus, the data support differential effects of pyroglutamylation but do not yet fully establish the proposed distal conformational mechanism.

Reviewer #2 (Public review):

Summary:

The manuscript by Chang et al. presents the discovery and deorphanization of two bona fide PRXamide GPCRs, and their peptide agonists, in Aplysia californica (sea hare, a type of sea slug). Focusing on MMG2-DPb, the most potent peptide agonist, the authors demonstrate that pyroglutamination on the extracellularly facing N-terminus of the peptide reduces its agonist potency towards ApPRXa-R1 and increases potency towards ApPRXa-R2. Model-guided pocket and peptide mutagenesis demonstrate multiple differential dependencies and sensitivities of ApPRXa-R1 and ApPRXa-R2, supporting distinct mechanisms of peptide binding. The authors then show that a pyroglutaminated version of the dog Neuromedin U peptide has an increased potency towards human NMUR1, relative to the free-Gln version or the non-Gln-containing human peptide, whereas the potency towards NMUR1 is slightly reduced. These examples delineate N-terminal pyroglutamination of a peptide agonist as a receptor subtype selectivity controlling mechanism even when the peptide binds the receptor C-terminus.

Strengths:

(1) Identification of a class of GPCR-signaling peptides in a poorly characterized organism (Aplysia californica, a type of a sea slug).

(2) Deorphanization of two Aplysia californica GPCRs and the establishment of an Aplysia analog of the human Neuromedin U signaling system.

(3) A rigorous structure-function study of the Aplysia receptors and peptides.

(4) Identification of pyroglutamination as a mechanism controlling subtype selectivity in Aplysia PRXamide receptor system.

(5) Very clear and compelling graphics.

(6) Overall, the biochemical and pharmacological part of the study is strong, exciting, and well-presented.

Weaknesses:

(1) My biggest problem is with the emphasis on the 'non-contacting role' of the N-terminal pGlu: "purely on positional grounds, N-terminal pQ/Q is unlikely to form direct contacts with receptor residues", "Despite the absence of direct pQ/Q-receptor interactions...", "indirect PTM distal steering", etc. The absence of direct contacts between pGlu and the receptor residues is likely an artifact of limited precision in modeling and reflects insufficiently advanced modeling tools. The authors used Swiss-Model, Robetta, and HPEPDOCK: why not go for the state-of-the-art AI modeling software like Boltz, Chai, or AlphaFold? A quick AlphaFold3 generates a model where Gln and pGlu form perfect contacts with the receptor N-term and ECL2 (in ApPRXa-R2) or N-term and ECL3 (in ApPRXa-R1). The N-termini of both receptors are quite long, form well-defined tertiary structures, and fold onto the receptor extracellular loops; in the case of ApPRXa-R1, the N-terminal domain is stabilized by an intra-domain disulfide bond C15(NT)-C133(NT) and stapled to ECL2 via another disulfide bond C16(NT)-C322(ECL2), suggesting that these interactions are real. The 3D models generated by the authors are not available for review but based on figures, I don't see these N-terminal domains modeled at all, which would obviously affect the docked position of the peptide N-terminus and its contacts with the receptor. For example, in an AlphaFold model, ApPRXa-R1 F161, E317, R340 are in direct vicinity of the peptide's Gln / pGlu and are quite distinct in ApPRXa-R2 - direct contacts with these or nearby residues may easily explain the differential preferences for free Gln vs pGlu but none of them were tested via mutagenesis.

(2) Some conclusions and interpretations in the manuscript are overstated and lack precision. For example: "...a single N-terminal lactam modification converts a minimal chemical difference into bidirectional GPCR subtype outputs" makes the reader think about something as dramatic as an inversion of efficacy, an agonist becoming an antagonist, or at least the introduction of signaling bias, whereas in reality, the authors established that pyroglutamination reduces the potency of the peptide at one receptor but increases it at the other receptor. The use of PTMs and proteolytic processing is a broadly utilized mechanism for controlling GPCR peptide selectivity. The effects of pyroglutamination on the potency of MMG2-DPb towards its two Aplysia receptors very well align with this mechanism and should be described as such. Other examples of sentences that are similarly misleading: "... the same ligand, ..., produces opposite functional outcomes: pQ suppresses ApPRXa-R1 activation while enhancing ApPRXa-R2 activation" (in reality, both receptors are activated and the functional outcomes are the same, just achieved with different potency), "A single N-terminal lactam enables bidirectional GPCR subtype tuning...", "This bidirectional, subtype-specific effect of pQ is unprecedented in neuropeptide signaling..." (not bidirectional and not as unprecedented as the authors make it sound), etc.

(3) Overall, what the authors present as a new "indirect PTM distal steering" mechanism is really not supported by the data. The described systems fit the classical "lock-key" and "induced fit" paradigms rather than challenging them. I recommend the authors revisit their modeling approaches and model-guided interpretations of the biochemical/pharmacological data. Importantly, I do not think such revision would undermine the key strengths of the paper (listed above in Strengths). Based on experimental data alone, this is a very strong and interesting study and with more rigorous modeling and adequate interpretations, it can be made exceptional.

eLife Assessment

This important paper studies how adiposity affects brain development in the ABCD study. Major findings include: (1) central adiposity indices (BRI, WHtR) show stronger cross-sectional associations with cognition than BMI, partially mediated by frontotemporal cortical morphology; (2) the rate of adiposity accrual, rather than static adiposity at either timepoint, predicts follow-up cognition and altered (attenuated) cortical thinning; and (3) among overweight/obese adolescents, central fat reduction is accompanied by accelerated cortical thinning and catch-up in inhibitory control. The strength of evidence is solid, with the main limitation related to only looking at the cerebral cortex rather than examining the whole brain. The subgroup observation that fat reduction may be accompanied by normalization of both cortical trajectories and inhibitory control is potentially clinically meaningful, suggesting reversibility rather than fixed deficit, and would be of broad interest if it withstands more rigorous analysis.

Reviewer #1 (Public review):

Summary:

In this manuscript, the authors used the ABCD study to understand how changes in body fat composition affect brain development during the teenage years. They found that the key determinants of changes in the brain were changes in fat accumulation rather than just baseline measures. Less importantly, but still interesting, they reaffirmed that more complex measures, such as body roundness index, better captured the impact of obesity than body mass index.

Strengths:

This study uses an excellent open access dataset, asks an important set of questions, and is executed with diligence for proper methodology. The finding that trajectories in adiposity matter more than baseline differences is important both for understanding how the brain adapts to body composition changes and has potential impact for public health interventions.

Weaknesses:

While I am overall impressed with the study, there are a few weaknesses I would like to see the authors address:

(1) There is no reason to limit analyses to the cerebral cortex. Body composition changes are as likely to affect the subcortex or cerebellum as the neocortex. In some cases, like the hypothalamus, potentially even more likely.

(2) Confounders should be examined in more detail. How are adiposity changes mediated by (for example) socio-economic status?

(3) I'd like to see more rationale for excluding 483 kids with extreme adiposity indicators. Is it that they are untrustworthy entries? Otherwise, they could be particularly informative.

(4) Are there any blood measures of glucose/insulin available?

(5) Some of the figures could benefit from having raw data included rather than just showing the fitted trend lines.

(6) Are there any alternate explanations for baseline differences? Particular examples that came to mind are the influence that maternal adiposity has on offspring brain development.

Reviewer #2 (Public review):

This manuscript addresses an important public health question in adolescents using a large longitudinal ABCD cohort. It is generally well organized and presents a compelling story linking central adiposity, brain development, and cognition. The authors use baseline and 4-year follow-up data from the ABCD Study (N = 8,519 baseline; N = 1,873 longitudinal, the use of 4-year follow-up only should be justified) to examine how adiposity relates to cognitive performance and cortical structure in early adolescence. Major findings include: (1) central adiposity indices (BRI, WHtR) show stronger cross-sectional associations with cognition than BMI, partially mediated by frontotemporal cortical morphology; (2) the rate of adiposity accrual, rather than static adiposity at either timepoint, predicts follow-up cognition and altered (attenuated) cortical thinning; and (3) among overweight/obese adolescents, central fat reduction is accompanied by accelerated cortical thinning and catch-up in inhibitory control.

The use of large longitudinal data from ABCD is considered a strength as most prior neuroimaging work is cross-sectional. The focus on adiposity change velocity within adolescence is considered novel over single-timepoint designs, and the comparison of central-fat indices (BRI/WHtR) against BMI in a neurodevelopmental context is important given growing interest in these markers in cardiometabolic research. The subgroup observation that fat reduction may be accompanied by normalization of both cortical trajectories and inhibitory control is potentially clinically meaningful, suggesting reversibility rather than fixed deficit, and would be of broad interest if it withstands more rigorous analysis.

However, several major issues were identified. For example, the strength of the causal language is not supported by the design, the headline comparison between adiposity indices rests on coefficients that are described as standardized but evidently are not, and several analytic and reporting issues (attrition and selection, family clustering, regression to the mean in the subgroup analysis, internal inconsistencies in reported p values) must be resolved before the conclusions can be considered established.

Major Concerns

(1) ABCD provides more follow-ups and should be included. Also, there is a severe longitudinal attrition (8519 vs 1873) that needs to be addressed. No comparison of completers versus non-completers is provided. The cohort description of having 78.5% White seems to be wrong, suggesting either a coding error in the race variable or strong selection introduced by the exclusions. The exclusion of 483 participants for "extreme adiposity indicator values" is especially concerning in a study of obesity and should be justified with explicit criteria in the main text, with sensitivity analyses retaining these participants where possible.

(2) Adiposity is heavily influenced by socio-economic status, which itself contributes heavily to cognition and mental health as well. Many ABCD studies have reported this. Another important factor contributing to adiposity is sleep, which itself can contribute significantly to cognition and mental health. Many ABCD-based sleep studies have been published and can be reviewed.

(3) The manuscript repeatedly uses causal language that the observational design cannot support (e.g., "longitudinal fat accumulation ... drives cortical alteration," "fat reduction activated adaptive neural change," "neuroprotective management"). The cited literature (Likhitweerawong et al., 2022) is explicitly bidirectional: poor inhibitory control may promote weight gain rather than the reverse. The baseline mediation analysis is particularly problematic because exposure, mediator, and outcome were measured concurrently, so the temporal ordering required for mediation is assumed rather than established. Causal claims should be tempered throughout. Similarly, mediation analyses should be interpreted cautiously as statistical rather than causal.

(4) zBMI, WC, BRI, and WHtR are highly intercorrelated, as are follow-up adiposity and delta-adiposity (Table 3). Entering them simultaneously invites unstable estimates and sign flips. Please report pairwise correlations and variance inflation factors, and demonstrate that the key conclusions are robust.

(5) Subgroup analysis: regression to the mean, group labels, and internal inconsistencies. (a) Stratifying on delta-BRI >= 0 enriches the "decreasing" group for high baseline values, so the observed cognitive "catch-up" and accelerated thinning may partly reflect regression to the mean; the percentile-based sensitivity analysis mitigates but does not resolve this. Group comparisons should adjust for baseline BRI and baseline cognition. (b) zBMI >= 1 corresponds to overweight by WHO criteria, not obesity; labelling this group "obese adolescents" throughout (including the Abstract) is inaccurate. (c) The Abstract claims a Flanker slope difference of p < 0.05, but the Results report p = 0.058 versus controls (described, incorrectly, as significant) and p = 0.021 only versus the increasing-BRI group; these statements must be reconciled. (d) It is unclear whether subgroup p values were FDR-corrected and how t-tests/Wilcoxon tests produced "adjusted" p values; if covariate adjustment was intended, regression models are required.

Tables 2-4 mention that beta values are standardized, yet the magnitudes are clearly scale-dependent (e.g., WHtR beta = -7.5 vs. zBMI beta = -0.42 in Table 2; WHtR delta-beta = -780 in Table 3). This undermines the central claim that BRI/WHtR "outperform" BMI, since predictors cannot be compared on unstandardized coefficients. Please either truly standardize all predictors and outcomes, or compare indices formally (e.g., delta-R2, AIC/BIC, or non-nested model comparison tests), and report 95% confidence intervals for all estimates.

(6) Underspecified mixed models and family clustering. The random-effects structure of the LME models is never stated. ABCD includes twins and siblings, so family nesting must be modelled (and the treatment of site - fixed covariate versus random effect - clarified). As written, the analyses may be anti-conservative. Please provide full model specifications and confidence intervals.

(7) Interpretation of accelerated cortical thinning is overly strong. The manuscript equates faster thinning with better maturation and slower thinning with delay. This reading is contested: apparent cortical thinning also reflects myelination-related signal changes, and thickness-cognition associations are age- and region-dependent. Moreover, the present null finding for baseline adiposity contradicts Kaltenhauser et al. (2023), who reported that baseline adiposity attenuates thinning in ABCD; the discrepancy is cited but never reconciled and deserves direct discussion (differences in sample, covariates, or modelling) Unclear units and implausible mediation estimates. Table 1 gives delta-WC = 0.264, interpretable only if delta is per month (age in months), yet the text describes "annualized" change. Units for all delta variables must be stated explicitly and used consistently. Mediation effects reported as wide ranges (e.g., ACME = -29.95 to -0.18; "ACME = -0.1.17" is a typographical error) are uninterpretable and, at face value, implausibly large relative to the cognitive score scale; please report per-indicator ACMEs with confidence intervals and proportion mediated. Not sure how the ranges such as "beta = -0.29 to -0.00" cited as significant effects are informative.

(8) Pubertal confounding. Only baseline pubertal category is adjusted for, but pubertal tempo over the follow-up window confounds both adiposity change and cortical development. This should be adjusted for change in pubertal stage or acknowledged as a substantive limitation. Puberty interactions and sex-specific analysis should be performed too.

(9) Effect sizes should be discussed.

(10) Inclusion of both baseline adiposity and adiposity change in longitudinal models should be justified.

(11) Would there be nonlinear age/puberty effects?

eLife Assessment

This valuable study investigates the action of a floral homeotic transcription factor in different cell types. The results provide solid evidence that PhDEF shows more extensive binding and regulatory effects in the epidermis than in the mesophyll, supporting the conclusion that cell-layer context influences homeotic gene function. While there are some limitations, they do not undermine the principal observation of layer-dependent PhDEF activity, but the interpretation of the data is complicated by uncertainties in cell-type assignment and composition in the scRNA-seq data, the cell-type-enriched rather than strictly cell-type-specific nature of the ChIP-seq data, and differences in developmental stage between the single-cell and ChIP-seq experiments. The work will be of interest to developmental biologists studying transcriptional regulation and cell identity.

[Editors' note: this paper was reviewed by Review Commons.]

Reviewer #1 (Public review):

Summary:

The authors previously generated two cell-layer-specific mutants of petunia for the petal identity gene PhDEF. In this study, they profiled differential gene expression in those mutants through single-cell RNA sequencing (scRNA-seq). They found that more genes are highly and specifically expressed in the epidermal cell layer than in mesophyll cells. In addition, they identified cell-layer-specific and -aspecific PhDEF target genes. Using the extensive single-cell transcriptome and layer-specific target identification, the authors concluded that different cell identities affect homeotic regulator PhDEF, thereby influencing transcriptional regulation.

Major comments from first round of review:

This presented work provides comprehensive evidence, that pre-existing cell layer identity (epidermis and mesophyll) modulate transcriptional output of homeotic transcription factor, PhDEF.

However, a disconnection between PhDEF bindings to genome and transcriptional output undermines the robustness of their conclusion although some binding loci were shown to be correlated with DEG. This may indicate the chromatin state, the existence of interacting partners, and the non-productive binding of PhDEF, suggesting that PhDEF binding alone is not sufficient to predict transcriptional outcomes and additional regulatory mechanisms that shape gene expression in addition to the layer-specific regulatory mechanisms. This disconnection may also be due to the developmental timing. Indeed, it appears authors used different flower stages for ChIP-seq and scRNA-sequencing. In fully differentiated organs, PhDEF binding itself may be no longer transcriptionally productive, and differential gene expression results primarily from the pre-established cell identity rather than directly from the homeotic regulation of PhDEF. Therefore, the main question the authors asked-how homeotic identity works with cell-layer identity and how the homeotic gene, PhDEF, acts in mature organs-was not clearly explained by this study.

In Figure 2, the use of the term "target" is potentially misleading. It sounds like direct target genes (direct binding and differential expression) for PhDEF, but it refers only to DEGs.

Lines 496-497: When the authors state, "~ demonstrates for the first time that the regulatory function of homeotic factor is influenced by cell layer identity," it sounds overstated, as prior studies have shown that pre-existing tissue or cell identity can shape transcriptional activity and developmental output.

Significance:

This study is well-designed and technically sound. They utilize single-cell transcriptomics and ChIP-seq by using genetically well-defined genetic materials and layer-specific PhDEF deletion mutants. The analysis showed where PhDEF binds to genomic loci and which genes are differentially expressed in petal epidermis and mesophyll, providing evidence of cell-layer-specific function of homeotic gene in mature organs. Although certain mechanistic aspects were not elucidated, the data from the extensive genome-wide study contributed to drawing their conclusions.

Advances: This research goes beyond classical models of floral organ identity by showing that homeotic gene function is not uniform in the same floral organ. It represents a conceptual advance in our understanding by integrating cell layer identity into the framework of homeotic gene regulation.

Audience: This study will be of broad interest to scientists who study transcription networks, cell and organ identity in the context of plant development.

My field of expertise: Transcriptional regulation by transcription factor, epigenetic regulation of gene expression, plant development.

Comments on latest version:

Thank you for sharing the assessment. I am happy with the proposed eLife assessment. I do not have any further amendments to my review.

Reviewer #2 (Public review):

Summary:

This study from Cavallini-Speisser et al. cleverly leverages a tissue layer-specific mutant, single cell and bulk RNA-sequencing, and ChIP-sequencing to decipher tissue layer-specific regulation of petal development in petunia by the PhDEF transcription factor. The authors find common and unique targets of PhDEF between the epidermis and mesophyll and conclude that the activity of transcription factors like PhDEF are influenced by the pre-existing environment in the cell they are expressed in. Understanding when and how a given transcription factor drives expression of unique target genes in various contexts is an important aspect of developmental biology that can be elusive outside of highly tractable model systems. As such, I think this study is of high value and has strong potential to expand our understanding of how developmental specificity is mediated by commonly employed transcriptional regulators. However, I think there are some issues with possible over-interpretation and some places where documentation of experimental design and data quality control are lacking. I elaborate on these concerns below.

Major Comments From Review at Review Commons:

(1) Line 155: Assigning mesophyll clusters by default without any positive marker genes strikes me as problematic, especially as much of the analysis rests on comparing the transcriptomes of epidermis and mesophyll cells. Can the authors perhaps leverage published scRNA-seq datasets to find potential mesophyll markers, even homologs from other species, to improve confidence in the cluster assignment?

(3) Line 228: Through the description and interpretation of the ChIP-seq dataset, the authors use the fact that peaks are more abundant and bigger in the epidermis to conclude that binding of PhDEF is "stronger" in the epidermis. This implies a difference in physical interaction between the TF and the DNA that I don't think can be concluded from the data presented. This could be confounded by biology; if expression of PhDEF is more heterogeneous in the mesophyll than in the epidermis, the peaks from that tissue will be averaged out and appear smaller when in fact the binding is the same strength. This could also be a technical artifact if the ChIP was less efficient in one sample versus another. This conclusion requires reinterpretation. The authors have the power to address this at least partially with the scRNA-seq by measuring PhDEF heterogeneity. I believe assessing ChIP efficiency would have required a spike in control, but perhaps there is a computational way to address this. It is important to discuss these confounding factors in the text.

(3) Line 376: The authors risk overinterpreting a lack of differential gene expression detection in their analysis of PhDEF binding profiles. This can be affected by how deeply a library was sequenced or how many cells were analyzed per cell type. A gene might not be found to be DE if low depth or few cells resulted in noise or dropout. Lack of detection does not mean lack of differential regulation so the biological relevance of this portion of the analysis should be interpreted with caution.

(4) Line 476: The authors state there is a mismatch in developmental timing between the RNAseq and ChIP datasets. Why is this? This is mentioned briefly in the Discussion, but has the potential to be majorly confounding to the joint interpretation of the ChIP and RNAseq datasets. This experimental design choice should be justified more thoroughly and a consideration of the limitations it brings to data interpretation should be more prominent in the text.

Significance:

Strengths: The authors employ a unique and powerful mutant system to explore a fundamental developmental biology question. In addition, the datasets generated will likely be useful to other researchers working in petunia or flower development.

Limitations: While the mutant system employed here is a creative way to get at tissue-layer specific transcription factor activity, the ChIP samples still include heterogeneous cell types, which may impact the findings presented here.

Advance: This study uses a unique system to test the function of a transcription factor in distinct cell types. As stated above, understanding when and how a given transcription factor drives expression of unique target genes in various contexts is an important aspect of developmental biology.

Audience: I believe this work will be of primary interest to the plant development, single cell, and chromatin biology communities. These are specialized, basic research communities.

Reviewer Expertise: I am a plant developmental biologist who works with multiple modes of cell-type-specific NGS datasets including bulk and single cell RNAseq and ChIPseq among others.

Comments on latest version:

I find the authors' response to my review to be comprehensive and I think the revision looks solid. The eLife assessment works for me.

Reviewer #3 (Public review):

Summary:

This study addresses a fundamental but underexplored aspect of homeotic gene function: how regulators of cell identity act during late stages of organ development. The authors take advantage of layer-specific mutants of the MADS-box gene PhDEF in Petunia hybrida to dissect the roles of this floral identity regulator in the epidermis and mesophyll. By combining single-cell RNA sequencing with ChIP-seq analyses in wild-type and mutant chimeric petals, the work demonstrates that, although PhDEF is expressed at comparable levels in both petal layers, it binds to and regulates a substantially larger and more layer-enriched set of genes in the epidermis than in the mesophyll. The identification of both layer-specific and shared PhDEF binding sites supports a model in which pre-existing layer identity modulates the regulatory output of homeotic transcription factors.

Major comments from review at Review Commons:

Dissecting PhDEF binding preferences in the epidermis versus the mesophyll using the star and wico mutants is a clever and powerful approach. However, conclusions involving cell identity should be drawn with caution for two reasons. First, cell identities appear to be altered in the mutants: scRNA-seq data suggest that even cells assigned to the same cluster can be molecularly distinct across genotypes. For example, the transcriptomes of wild-type epidermis and mesophyll are highly similar (Pearson correlation R = 0.94), yet both show lower correlation with epidermal cells from star or wico petals. These results raise questions such as are the cells identified as epidermal cells really strictly epidermal in the mutants? Do you need to take into consideration cell composition of the mutants when you do differential expression analysis? Second, PhDEF is expressed in both epidermal and mesophyll clusters in all genotypes, albeit at different levels and in fewer cells in the mutants. As a result, the ChIP-seq profiles should be interpreted as cell type-enriched rather than cell type-specific. And there are a few things that need clarification:

a) Protoplasting and tissue dissection analyses suggest that mesophyll cells constitute more than 80% of the cells in wild-type petals, whereas the scRNA-seq data indicate a substantially lower proportion. Could this discrepancy reflect technical biases in cell recovery or capture efficiency, or issues related to cell identity assignment during clustering and annotation? Notably, the scRNA-seq data from star and wico petals show mesophyll proportions close to 80%. Is this difference due to an increased abundance of mesophyll cells in the mutants, or could it instead reflect differences in transcriptomic separability? In wild-type petals, the epidermal and mesophyll transcriptomes are highly correlated and express similar numbers of genes, with epidermal cells distinguished mainly by higher expression of a subset of genes. This raises the possibility that mesophyll cells in the wild type occupy a more plastic or less differentiated transcriptional state and may therefore be misclassified as epidermal cells, whereas disruption of regulatory mechanisms in the mutants enhances transcriptional divergence and alters cell clustering outcomes.

b) Cluster 0 appears to show internal heterogeneity, as the expression patterns of KCS3 and LLE2 are largely mutually exclusive. Do these patterns reflect the presence of distinct epidermal cell types within the limb that are currently grouped into a single cluster?

c) Clusters 0 and 7 both exhibit high expression of pigmentation genes, while cluster 7 additionally shows strong enrichment for cell division genes. Are cell cycle genes the primary features distinguishing these two clusters? If cell cycle effects are regressed out, would cluster 7 merge with cluster 0, potentially yielding a more continuous cell state trajectory and helping to resolve the pattern noted in point (b)?

d) Pearson correlation is largely driven by highly expressed genes and may therefore be insensitive to changes in cell identity markers. The conclusions that the less clear separation of mesophyll and epidermal cells in star is due to altered cell identity would be more convincing if supported by independent validation, such as in situ hybridization or reporter analyses, to directly visualize molecular alterations in the relevant cell types when comparing wild-type and mutant tissues.

Significance:

Overall, this work significantly advances our understanding of late homeotic gene function. It establishes compelling evidence for how developmental context constrains transcription factor activity and offers broadly relevant insights for studies of organ patterning. The combination of genetic mosaics with single-cell and chromatin-level analyses represents a powerful and generalizable strategy that will be of interest to both plant developmental biologists and researchers studying transcriptional regulation.

My expertise: single cell genomics, chromatin biology and plant development.

Comments on latest version:

The authors have addressed my concerns and questions. And I agree with your assessment of "valuable" for significance and "solid" for strength of evidence.

Author response:

General Statements

We wish to bring to the attention of the editor and reviewers that the title of the manuscript has been modified to more accurately reflect the conclusions of our study, i.e. that PhDEF has a major binding and regulatory role in the petal epidermis.

Point-by-point description of the revisions

We thank the three reviewers for their critical reading of our manuscript. We also appreciate that the three reviewers highlighted the conceptual advances and the broad findings that our study brings to the community. They have identified key limitations of our analyses, and we have either provided explanations for our choice, or performed new analyses to circumvent biases. We believe that the manuscript is greatly improved and that our main conclusion, that PhDEF has a major binding and regulatory action in the petal epidermis, is strongly supported by our data.

Reviewer #1 (Evidence, reproducibility and clarity):

Summary:

The authors previously generated two cell-layer-specific mutants of petunia for the petal identity gene PhDEF. In this study, they profiled differential gene expression in those mutants through single-cell RNA sequencing (scRNA-seq). They found that more genes are highly and specifically expressed in the epidermal cell layer than in mesophyll cells. In addition, they identified cell-layer-specific and -aspecific PhDEF target genes. Using the extensive single-cell transcriptome and layer-specific target identification, the authors concluded that different cell identities affect homeotic regulator PhDEF, thereby influencing transcriptional regulation.

Major comments:

This presented work provides comprehensive evidence, that pre-existing cell layer identity (epidermis and mesophyll) modulate transcriptional output of homeotic transcription factor, PhDEF.

However, a disconnection between PhDEF bindings to genome and transcriptional output undermines the robustness of their conclusion although some binding loci were shown to be correlated with DEG. This may indicate the chromatin state, the existence of interacting partners, and the non-productive binding of PhDEF, suggesting that PhDEF binding alone is not sufficient to predict transcriptional outcomes and additional regulatory mechanisms that shape gene expression in addition to the layer-specific regulatory mechanisms. This disconnection may also be due to the developmental timing. Indeed, it appears authors used different flower stages for ChIP-seq and scRNA-sequencing. In fully differentiated organs, PhDEF binding itself may be no longer transcriptionally productive, and differential gene expression results primarily from the pre-established cell identity rather than directly from the homeotic regulation of PhDEF. Therefore, the main question the authors asked-how homeotic identity works with cell-layer identity and how the homeotic gene, PhDEF, acts in mature organs-was not clearly explained by this study.

We thank the reviewer for raising this important issue. We agree that the difference in developmental timing between the scRNA-Seq and ChIP-Seq experiments might contribute to the disconnection that this reviewer pointed out.

First, we want to explain that the reason for performing scRNA-Seq on fully differentiated petals was purely technical, as we were initially aiming to obtain protoplasts at stage 8 (stage used for the ChIP-Seq) but could never retrieve enough of them for proper encapsulation in the 10X Chromium chips. We have now clearly explained this in the manuscript (lines 104-107).

A hypergeometric test shows that our ChIP-Seq and scRNA-Seq datasets overlap more than by chance (p = 0.000137); however, we agree that differences in developmental stages possibly bring a confounding effect to our conclusions. Therefore, we have decided to add to our manuscript the intersection between ChIP-Seq and bulk RNA-Seq data performed on WT, star and wico flowers at stage 8, that we published previously (Chopy et al., 2024). In that case, both datasets have been obtained with the same exact genetic material and at the same exact developmental stage.

This intersection confirms that very similar binding profiles are observed for PhDEF target genes, whether they are differentially expressed in the epidermis (star only), in the mesophyll (wico only) or in both layers (star and wico) (Figure 4A). However, star-specific DEGs were more often bound by PhDEF by epidermis+shared binding sites, and wico-specific DEGs more often with mesophyll-specific binding sites, suggesting a weak but significant association between binding and regulatory profiles. Performing similar tests for individual binding categories for DEGs identified by scRNA-Seq also revealed that epidermis-specific DEGs displayed more epidermis+shared binding sites than expected. Since the association between epidermal DEGs and shared+epidermal binding sites is found both in the intersection with RNA-Seq and scRNA-Seq data, we have now stated that “layer-specific binding and transcriptional regulation are partially linked, at least in the epidermis” (line 354).

In Figure 2, the use of the term "target" is potentially misleading. It sounds like direct target genes (direct binding and differential expression) for PhDEF, but it refers only to DEGs.

Indeed, the term "target" was referring to both indirect and direct targets of PhDEF. To avoid any possible confusion, we have replaced it by differentially expressed genes (DEGs) throughout the manuscript, when appropriate.

Lines 496-497: When the authors state, "~ demonstrates for the first time that the regulatory function of homeotic factor is influenced by cell layer identity," it sounds overstated, as prior studies have shown that pre-existing tissue or cell identity can shape transcriptional activity and developmental output.

We agree that previous studies have shown that cell identity influences transcriptional activity in general. While this might not have been specifically assessed in the context of different cell layers, we have rewritten this sentence accordingly.

Minor comments:

In the UMAP presentation, as depicted in Figures 2C, S3, and S5, the cells with zero expression can be colored in light gray (or an inverted color scheme). The purple hue masks the gene expressions of other cells, making it difficult to see the yellow or light green colored cells.

We have modified all UMAPs depicting gene expression as suggested, in Figures 2C and S5, and replaced UMAPs with DotPlots in Figure S3.

Reviewer #1 (Significance):

General assessment: This study is well-designed and technically sound. They utilize single-cell transcriptomics and ChIP-seq by using genetically well-defined genetic materials and layer-specific PhDEF deletion mutants. The analysis showed where PhDEF binds to genomic loci and which genes are differentially expressed in petal epidermis and mesophyll, providing evidence of cell-layer-specific function of homeotic gene in mature organs. Although certain mechanistic aspects were not elucidated, the data from the extensive genome-wide study contributed to drawing their conclusions.

Advances: This research goes beyond classical models of floral organ identity by showing that homeotic gene function is not uniform in the same floral organ. It represents a conceptual advance in our understanding by integrating cell layer identity into the framework of homeotic gene regulation.

Audience: This study will be of broad interest to scientists who study transcription networks, cell and organ identity in the context of plant development.

My field of expertise: Transcriptional regulation by transcription factor, epigenetic regulation of gene expression, plant development

Reviewer #2 (Evidence, reproducibility and clarity):

Summary:

This study from Cavallini-Speisser et al. cleverly leverages a tissue layer-specific mutant, single cell and bulk RNA-sequencing, and ChIP-sequencing to decipher tissue layer-specific regulation of petal development in petunia by the PhDEF transcription factor. The authors find common and unique targets of PhDEF between the epidermis and mesophyll and conclude that the activity of transcription factors like PhDEF are influenced by the pre-existing environment in the cell they are expressed in. Understanding when and how a given transcription factor drives expression of unique target genes in various contexts is an important aspect of developmental biology that can be elusive outside of highly tractable model systems. As such, I think this study is of high value and has strong potential to expand our understanding of how developmental specificity is mediated by commonly employed transcriptional regulators. However, I think there are some issues with possible over-interpretation and some places where documentation of experimental design and data quality control are lacking. I elaborate on these concerns below.

Major Comments:

Line 155: Assigning mesophyll clusters by default without any positive marker genes strikes me as problematic, especially as much of the analysis rests on comparing the transcriptomes of epidermis and mesophyll cells. Can the authors perhaps leverage published scRNA-seq datasets to find potential mesophyll markers, even homologs from other species, to improve confidence in the cluster assignment?

We thank the reviewer for raising this important point. Our statement of defining mesophyll identity by default was not entirely true (and we have now removed it), since it is supported by GO-enriched terms for cluster markers. For cluster "mesophyll 3", there is a strong enrichment for photosynthesis-related genes; and for cluster "mesophyll 2", there is a strong enrichment for water transport-related genes, both functions being fulfilled by the petal mesophyll. For cluster "mesophyll 1", the most enriched GO term is "glutathione metabolic process" that rather points to stress response. These three clusters also do not express any of the epidermal-identity genes, in contrast to the clusters that we assigned as epidermal. Cluster markers from the mesophyll display the lowest enrichment of all clusters (the best cluster markers have a log2FC between 2.8 and 4.9, in contrast to a log2FC between 8.1 and 9.9 for all other clusters), which is in line with our finding that the mesophyll expresses less specific genes than the epidermis, and suggests a basal identity with a transcriptomic signature that is less clear than in the epidermis. Therefore, it is not entirely trivial to find positive marker genes for the mesophyll with a strong specificity.

In order to be more transparent about the expression patterns of the genes we selected to assign cluster identity, we modified Figure S3 to include DotPlots of selected photosynthesis-related genes, histone genes, S-phase genes and vasculature genes in the same figure, to compare with DotPlots of epidermal genes and pigmentation genes from Figure 1D, to allow for an informed comparison. Our conclusions remain the same as previously: epidermal clusters are defined based on the specific expression of epidermal genes and/or pigmentation genes. We notice, however, that the cluster "upper limb epidermis" strongly expresses photosynthesis genes, which is likely why it is close in the UMAP space to the "mesophyll 3" cluster. We have no explanation for that, but the extremely high expression of pigmentation genes in this cluster, however, identifies it as epidermal without a doubt. We have also added in Figure S3 the DotPlots of expression levels of homologs of 15 epidermis-enriched and 9 mesophyll-enriched genes from tobacco petal scRNA-Seq published by Kang et al. (doi: 10.1111/nph.17992). This shows that our definition of epidermal and mesophyll clusters and the one from Kang et al. generally overlap.

Line 228: Through the description and interpretation of the ChIP-seq dataset, the authors use the fact that peaks are more abundant and bigger in the epidermis to conclude that binding of PhDEF is "stronger" in the epidermis. This implies a difference in physical interaction between the TF and the DNA that I don't think can be concluded from the data presented. This could be confounded by biology; if expression of PhDEF is more heterogeneous in the mesophyll than in the epidermis, the peaks from that tissue will be averaged out and appear smaller when in fact the binding is the same strength. This could also be a technical artifact if the ChIP was less efficient in one sample versus another. This conclusion requires reinterpretation. The authors have the power to address this at least partially with the scRNA-seq by measuring PhDEF heterogeneity. I believe assessing ChIP efficiency would have required a spike in control, but perhaps there is a computational way to address this. It is important to discuss these confounding factors in the text.

This is indeed another important point. We agree that the word "stronger", to describe PhDEF binding in the epidermis, was not appropriate and we have replaced it with "more frequent" which is a more factual interpretation of our results. We also agree that even this interpretation depends on potential ChIP artifacts that we have now evaluated.

We have used our WT scRNA-Seq data to explore the heterogeneity of PhDEF expression in the mesophyll and the epidermis, as suggested. The barplot in Figure S10B represents the number of cells (y-axis) with given PhDEF RNA counts (x-axis) in the clusters that we assigned as epidermis (left) and mesophyll (right). We have performed this analysis after removing cells that do not express PhDEF at all, which represents 47.7% and 47.4% of epidermal and mesophyll cells, respectively, hence very similar proportions. The distributions of expression of PhDEF in the epidermis and in the mesophyll are within the same ranges, with a slightly higher expression of PhDEF in the mesophyll than in the epidermis. The coefficients of variation (cv) of the two distributions are similar and slightly higher in the epidermis (cv = 0.34 in the mesophyll and cv = 0.36 in the epidermis, p = 0.00106 with Feltz and Miller’s asymptotic test). Therefore, it appears that the expression of PhDEF is actually higher and slightly less variable in the mesophyll than in the epidermis, meaning that it should not result in averaging out the peaks detected. This relies on the assumption that PhDEF protein levels are directly correlated with PhDEF RNA levels, which we have not explored in this study and remains a limitation. We have included this analysis lines 275-278 and Figure S10B.

Regarding ChIP efficiency, we had run different tests prior to sequencing: first, we have tested different amounts of chromatin, keeping the quantity of antibody unchanged, and tested ChIP enrichment by qPCR on a set of two positive (PhDEF and Pos2) and one negative (Neg1) control binding sites. Second, after selecting the best chromatin quantity, we have performed 4 independent ChIP replicates for each genotype (split between two assays named ChIP-1 and ChIP-2) and measured ChIP efficiency by qPCR. This is depicted in Figure S10A, with the replicates chosen for sequencing highlighted with an orange star.

We have now explained in greater detail in the Methods our preliminary tests. There is indeed variation of enrichment between replicates, and particularly between assays here (ChIP-1 vs. ChIP-2), which is inherent to the ChIP experiment. It might be particularly prominent in our case due to our custom antibody directed against PhDEF, in contrast to commercial antibodies that are commonly used in ChIP experiments with tagged transgenic lines. However, we see consistently lower enrichment for star as compared to wico and WT, in line with the more frequent epidermal binding of PhDEF. We have followed the ENCODE guidelines for our analysis pipeline, in particular applying the IDR. We have now added other mapping statistics in Table S5 including the FRiP (fraction of reads in peaks) metric, that is in the range of expected values but is lower for star (around 0.5%) than for WT and wico (0.7-1.2 %), again consistent with the more frequent epidermal binding of PhDEF.

Line 376: The authors risk overinterpreting a lack of differential gene expression detection in their analysis of PhDEF binding profiles. This can be affected by how deeply a library was sequenced or how many cells were analyzed per cell type. A gene might not be found to be DE if low depth or few cells resulted in noise or dropout. Lack of detection does not mean lack of differential regulation so the biological relevance of this portion of the analysis should be interpreted with caution.

We have added to this new version of the manuscript an intersection between ChIP-Seq and bulk RNA-Seq in WT, star and wico, as bulk RNA-Seq is much more sensitive than scRNA-Seq in detecting lowly expressed genes. We have also added the sentence that bulk RNA-Seq "better captures lowly expressed genes" than scRNA-Seq, line 333. This intersection revealed an association between epidermal DEGs (star-specific DEGs) and the presence of epidermal+shared binding sites for PhDEF. We have modified our conclusions accordingly.

Line 476: The authors state there is a mismatch in developmental timing between the RNAseq and ChIP datasets. Why is this? This is mentioned briefly in the Discussion, but has the potential to be majorly confounding to the joint interpretation of the ChIP and RNAseq datasets. This experimental design choice should be justified more thoroughly and a consideration of the limitations it brings to data interpretation should be more prominent in the text.

This concern has also been raised by the first reviewer, and we have now added to our study an intersection between ChIP-Seq and bulk RNA-Seq performed at the same stage. We have also explained the technical reasons for performing scRNA-Seq at a mature stage only (lines 104-107). Indeed, the intersection between bulk RNA-Seq and ChIP-Seq performed at the same stage revealed a significant association between epidermal DEGs (star-specific DEGs) and the presence of epidermal+shared binding sites for PhDEF. Testing for individiual binding categories, we could also detect an enrichment of epidermal+shared binding sites for epidermal-specific DEGs identified by scRNA-Seq. Therefore, we have now stated that “layer-specific binding and transcriptional regulation are partially linked, at least in the epidermis” (line 354).

Minor Comments:

Line 229: The authors compare correlations between pseudo-bulked transcriptomes and argue that in the star mutant the epidermis adopts a mesophyll-like identity. The correlation between epidermis and mesophyll in star is 0.97 and the correlations were 0.94 and 0.93 in the other genotypes tested. What is the meaningful cutoff for saying the transcriptomes are similar or not? Is 0.97 so much higher than 0.94 that this conclusion is supported?

The comparison of pseudo-bulk transcriptomes is a very global and exploratory approach. Given the high number of genes underlying these pseudo-bulk datasets, any difference in the correlation between them is statistically significant, which is not very informative. We agree with this reviewer that the interpretation of these correlation coefficients is somewhat arbitrary. We have simplified this part of the manuscript and have mostly focused on comparing star and wico pseudo-bulk transcriptomes to the WT ones, but not to each other's, which aligns well with our main message of a specific epidermal identity, easily shifting to a mesophyll-identity when PhDEF is missing or not entirely functional. We have also removed Figure 2E to a supplementary figure to give less emphasis to this analysis.

Line 264: What are the "manually chosen thresholds for differential expression"? Can the authors explain and justify this? There is very little detail on this in the materials and methods and this raises some concerns regarding how a threshold was chosen.

Seurat gives a default threshold of 0.25 for log2FC, which we found to be very permissive. In order to capture the most informative targets of PhDEF, but still to capture a meaningful number of targets, we empirically decided to increase this threshold to 0.75. On the WT scRNA-Seq dataset, we observed that the layer-specificity factor was also capturing meaningful differences in layer-specific expression (see Figure 1E). We chose a cut-off at 10% since it was the lowest to give a significant difference in the numbers of epidermis- vs. mesophyll-enriched genes in the WT petal. We have added these explanations in the methods.

Figure 3G: Could this plot be annotated with the classification of peak layer specificity? It is a little difficult for me as the reader to keep up with all the categories in the text, and showing them in the figure might make that easier to follow.

We have now annotated the Venn diagram in Figure 3G with the classifications of binding profiles.

Figure 4A: A comparison of only two cell categories should not use scaled expression, as this can over-emphasize small differences in gene expression. Can this be replaced with a dot plot that uses unscaled expression values?

We thank the reviewer for noticing this issue, we have built a DotPlot with unscaled values and replaced it in Figure 4A, which does not change our conclusions. We have also used unscaled values in Figure S14.

Figure 4C: Could this be represented more legibly with stacked, space-filled bar charts? As is, this is difficult to read, and might be impossible for someone who is color blind. In addition, the authors claim this analysis shows similar proportions across all categories, but I wonder if that would hold true if they performed an over-representation analysis normalized to the categories shown in "all genes expressed". This could allow them to statistically test whether there are real differences in representation among the categories.

We have now used a different and color-blind-friendly palette for pie charts of PhDEF binding profiles. Following reviewers' comments, we have analyzed the intersection of bulk RNA-Seq with ChIP-Seq (both performed at the same stage), and performed Chi2 goodness-of-fit tests that indeed support some association between binding profile and regulation profile, although this remains limited.

Supplemental Figure 2: It's great that the authors include these metrics, but it would be ideal to also include the plots of standard QC metrics for scRNA-seq such as those found here to give a better sense of per cell quality: https://satijalab.org/seurat/articles/pbmc3k_tutorial

In addition, it is important to include QC metrics for ChIPseq, which I did not find in the supplement. Metrics such as FRiP are important for interpreting ChIP library quality.

We have now added the standard QC plots (Feature number per cell and RNA counts per cell) to Figure S2, as suggested. Mitochondrial and ribosomal genes are not annotated in the Petunia axillaris nuclear genome that we used, therefore we could not compute mitochondrial or ribosomal gene counts. We have removed cells expressing less than 200 genes, but did not apply any upper thresholds as there were no obvious outliers.

FastQC reports for scRNA-Seq and ChIP-Seq have been deposited at https://entrepot.recherche.data.gouv.fr/dataverse/PhDEF_Flower_layer, as indicated in the Methods.

We have also added ChIP metrics in Table S5, including the number of reads, duplicated reads, mapped reads and computed the FriP score. This score ranges between 0.5 % and 1.2 %, which is satisfactory and above the minimum recommended score  by ENCODE of 0.3 %.

Line 654: What model was used for DESeq2?

We have used default parameters for DESeq2, ie a negative binomial GLM fitting and Wald significance tests. We have added this information in the Methods.

Line 752: Can the authors justify why peaks were called separately on input and ChIP samples rather than allowing MACS2 to call peaks in the ChIP sample over input background? That differs from the standard MACS2 pipeline and no explanation for this is provided in the text.

Our analysis pipeline has indeed been customized, in particular to detect peaks in our positive control PhDEF, for which a binding site of PhDEF in the promoter has been demonstrated experimentally by others in many different species. This binding has a strong experimental support, and we expected PhDEF to bind to its own promoter in the two cell layers. Our ChIP-Seq results show a posteriori that this particular peak is far from being the strongest one over the genome; therefore, we believe it represents a good control to detect binding enrichment for average targets. We have first tried the standard MACS2 pipeline that calculates the enrichment of IP over Input, but we could only detect PhDEF binding for one WT and one wico replicate, although the peaks were visually clear in the two replicates. Our input sample being generally noisy, we explored how separate peak calling between IP and Input would behave (as already done in e.g. Durand et al., 2023, doi: 10.1093/plcell/koad025). We also explored how thresholds for FDR in MACS2, and IDR thresholds for reproducibility between IP samples, would influence peak detection in IP and Input.

We found that calling peaks on the IP with a relaxed FDR (0.1), then applying the IDR at 0.1, allowed the capture of PhDEF binding to its own promoter in the two wico replicates (but still not in WT, because the peaks were lost after applying the IDR threshold). No peak was detected in the input with these settings, however for other genes we observed spurious peak detection in the input, therefore we decided to lower the FDR thresholds for input peaks to 0.05. Our choices have been made in an attempt to increase specificity at the risk of losing sensibility, and we probably lose true binding events. Considering that this ChIP has been performed on the endogenous PhDEF protein directly, and in chimeric flowers that only express PhDEF in half of the tissue, adapting the ChIP analysis pipeline was a necessary step.

We have now added a few lines in the Methods (lines 696-706) to explain our rationale.

Reviewer #2 (Significance):

General Assessment:

Strengths: The authors employ a unique and powerful mutant system to explore a fundamental developmental biology question. In addition, the datasets generated will likely be useful to other researchers working in petunia or flower development.

Limitations: While the mutant system employed here is a creative way to get at tissue-layer specific transcription factor activity, the ChIP samples still include heterogeneous cell types, which may impact the findings presented here.

Advance: This study uses a unique system to test the function of a transcription factor in distinct cell types. As stated above, understanding when and how a given transcription factor drives expression of unique target genes in various contexts is an important aspect of developmental biology.

Audience: I believe this work will be of primary interest to the plant development, single cell, and chromatin biology communities. These are specialized, basic research communities.

Reviewer Expertise: I am a plant developmental biologist who works with multiple modes of cell-type-specific NGS datasets including bulk and single cell RNAseq and ChIPseq among others.

Reviewer #3 (Evidence, reproducibility and clarity):

Summary:

This study addresses a fundamental but underexplored aspect of homeotic gene function: how regulators of cell identity act during late stages of organ development. The authors take advantage of layer-specific mutants of the MADS-box gene PhDEF in Petunia hybrida to dissect the roles of this floral identity regulator in the epidermis and mesophyll. By combining single-cell RNA sequencing with ChIP-seq analyses in wild-type and mutant chimeric petals, the work demonstrates that, although PhDEF is expressed at comparable levels in both petal layers, it binds to and regulates a substantially larger and more layer-enriched set of genes in the epidermis than in the mesophyll. The identification of both layer-specific and shared PhDEF binding sites supports a model in which pre-existing layer identity modulates the regulatory output of homeotic transcription factors.

Major comments:

Dissecting PhDEF binding preferences in the epidermis versus the mesophyll using the star and wico mutants is a clever and powerful approach. However, conclusions involving cell identity should be drawn with caution for two reasons. First, cell identities appear to be altered in the mutants: scRNA-seq data suggest that even cells assigned to the same cluster can be molecularly distinct across genotypes. For example, the transcriptomes of wild-type epidermis and mesophyll are highly similar (Pearson correlation R = 0.94), yet both show lower correlation with epidermal cells from star or wico petals. These results raise questions such as are the cells identified as epidermal cells really strictly epidermal in the mutants? Do you need to take into consideration cell composition of the mutants when you do differential expression analysis?Second, PhDEF is expressed in both epidermal and mesophyll clusters in all genotypes, albeit at different levels and in fewer cells in the mutants. As a result, the ChIP-seq profiles should be interpreted as cell type-enriched rather than cell type-specific.

We fully agree with this reviewer's comments, and alterations in cell identity in the star and wico mutants is indeed a main pitfall for our analysis. The phdef mutation alters the transcriptomic signatures of cells physically located in the epidermis or in the mesophyll, which can result in their artificial clustering with cells located elsewhere in the petal. It might be particularly true for the epidermis, since we see strong depletion in epidermal clusters in the star flowers (and even in the wico flowers), whereas the mesophyll is not much affected in wico flowers. As a result, it is likely that we lose many epidermal cells in star flowers that end up labeled as mesophyll cells, resulting in the under-estimation of DEGs in this layer. We had explored other possible ways to define epidermal or mesophyll cells in our dataset, based for instance on PhDEF or PhGLO1 expression, but since for the majority of cells PhDEF expression is simply not captured, we would have wrongly assigned phdef mutant identity to WT cells. In spite of this limitation, we find more DEGs in the epidermis than in the mesophyll, showing that even an underestimation of DEGs in the epidermis does not affect our main conclusion, which is that PhDEF has a major regulatory action in the epidermis. We have explicitly written this limitation in our manuscript, lines 234-236.

We agree that ChIP-Seq profiles are rather cell type-enriched than cell type-specific, which we have stated explicitly in the sentence line 308, saying that "differential binding between layers is quantitative". However, for simplicity we prefer to retain the term "specific" since our conclusions are based on the definition of peaks that can either be present or absent in a given genotype, and hence in a given cell layer.

And there are a few things that need clarification:

a) Protoplasting and tissue dissection analyses suggest that mesophyll cells constitute more than 80% of the cells in wild-type petals, whereas the scRNA-seq data indicate a substantially lower proportion. Could this discrepancy reflect technical biases in cell recovery or capture efficiency, or issues related to cell identity assignment during clustering and annotation? Notably, the scRNA-seq data from star and wico petals show mesophyll proportions close to 80%. Is this difference due to an increased abundance of mesophyll cells in the mutants, or could it instead reflect differences in transcriptomic separability? In wild-type petals, the epidermal and mesophyll transcriptomes are highly correlated and express similar numbers of genes, with epidermal cells distinguished mainly by higher expression of a subset of genes. This raises the possibility that mesophyll cells in the wild type occupy a more plastic or less differentiated transcriptional state and may therefore be misclassified as epidermal cells, whereas disruption of regulatory mechanisms in the mutants enhances transcriptional divergence and alters cell clustering outcomes.

Our protoplasting and tissue sections show that the mesophyll should represent 80% of cells in WT tissue, while we estimate it at 70% in our WT scRNA-Seq data based on our cluster assignment (mesophyll = 60% + vasculature = 10%, that we separated from the mesophyll cells but is actually embedded within this tissue). This is not a very strong difference, and considering the multiple steps that protoplasting and cell capture entail, we considered that this was reasonably close to the expected proportions.

In the star and wico flowers, we assign mesophyll identity to a greater proportion of cells, but as explained above, we believe that cells with altered epidermal identity are easily clustered as mesophyll cells, since they lose their specific transcriptomic signature. This is actually in line with one of the main messages of our article, that petal epidermis transcriptional identity is highly specific.

b) Cluster 0 appears to show internal heterogeneity, as the expression patterns of KCS3 and LLE2 are largely mutually exclusive. Do these patterns reflect the presence of distinct epidermal cell types within the limb that are currently grouped into a single cluster?

Indeed, there appears to be some internal heterogeneity within cluster 0. Since our main focus was to compare epidermal and mesophyll clusters, we did not explore further the heterogeneity within epidermal clusters and kept a coarse resolution.

c) Clusters 0 and 7 both exhibit high expression of pigmentation genes, while cluster 7 additionally shows strong enrichment for cell division genes. Are cell cycle genes the primary features distinguishing these two clusters? If cell cycle effects are regressed out, would cluster 7 merge with cluster 0, potentially yielding a more continuous cell state trajectory and helping to resolve the pattern noted in point (b)?

To answer one of Reviewer 1's comments, we have added additional DotPlots to better describe our clusters, in Figure S3. Clusters 0 (limb epidermis), 6 (upper limb epidermis) and 7 (replicating cells) exhibit high expression of pigmentation genes (in particular cluster 6), as depicted in Figure 1D. Cluster 7, consisting of only 30 cells, is the only cluster expressing histone genes and S-phase genes. It is possible that these few cells are cluster-6 cells that are replicating, but considering the very low number of cells involved, we have not explored any further their identity and decided to remove them, as it would only marginally affect the conclusions of our analyses. It is indeed surprising that cluster 6 (upper limb epidermis) is quite distinct in the UMAP space to the other epidermal clusters 0 (limb epidermis) and 4 (upper tube epidermis), and we have not observed similar situations in other scRNA-Seq studies. We speculate that this is due to the joint expression of anthocyanin-related and photosynthesis-related genes, which convey a very strong transcriptomic signature to these cells that distinguish them from other epidermal cells.

d) Pearson correlation is largely driven by highly expressed genes and may therefore be insensitive to changes in cell identity markers. The conclusions that the less clear separation of mesophyll and epidermal cells in star is due to altered cell identity would be more convincing if supported by independent validation, such as in situ hybridization or reporter analyses, to directly visualize molecular alterations in the relevant cell types when comparing wild-type and mutant tissues.

Following another reviewer's comments, we have now given less emphasis on the Pearson correlation analysis, that was to some extent subjective. Therefore, our conclusion that mesophyll and epidermal cells in star are less separated than in WT has been removed.

Minor comments:

(1) For color-coded figure legends (e.g., Fig. 1C), please also include the cluster numbers. This would facilitate interpretation, particularly for readers with reduced color sensitivity.

We have now used colour-blind-friendly palettes and we have added cluster numbers in Figure 1C.

(2) For figures containing abbreviations (e.g., Fig. 1D, st. / ca.), please explicitly define all abbreviations in the figure legend.

We have defined all abbreviations in the figure legends.

(3) For all UMAP figures, and for figures involving comparisons across clusters, please use consistent color schemes for the same clusters throughout the manuscript.

We have modified color schemes across the manuscript for colour-blind-friendly palettes, consistently used throughout the manuscript.

(4) In Fig. S6, the plot showing all genes does not exactly match Fig. 1C, although it appears to represent the same data. Please use the same version of packages, seed values and parameters for all UMAP plots to avoid such discrepancies.

Figure S6 is the result of integrating with Harmony the WT dataset (as shown in Figure 1C) with WT datasets after removing genes differentially expressed by the protoplasting process. Therefore these UMAPs are a result of integrating different datasets than in Figure 1C, which changes the shape of the UMAPs but cannot be controlled by the seed values, to our knowledge.

Reviewer #3 (Significance):

Overall, this work significantly advances our understanding of late homeotic gene function. It establishes compelling evidence for how developmental context constrains transcription factor activity and offers broadly relevant insights for studies of organ patterning. The combination of genetic mosaics with single-cell and chromatin-level analyses represents a powerful and generalizable strategy that will be of interest to both plant developmental biologists and researchers studying transcriptional regulation.

My expertise: single cell genomics, chromatin biology and plant development.

eLife Assessment

This important study uses the Cntnap2 mouse model of autism to investigate how reduced activity in the dorsal CA1 region of the hippocampus contributes to impaired temporal binding and memory flexibility. The authors combined trace conditioning and radial maze assays with fibre photometry, optogenetic rescue, and brain-wide c-Fos mapping to provide convincing evidence that dorsal CA1 hypoactivity limits the retention of temporally separated associations and biases learning towards less flexible strategies. The work advances understanding of hippocampal contributions to cognitive alterations associated with autism and will be of broad interest to researchers studying memory, hippocampal function, and neurodevelopmental disorders.

Reviewer #2 (Public review):

The authors investigate the contribution of dorsal CA1 hippocampal dysfunction to cognitive impairments in the Cntnap2 knockout mouse model of autism spectrum disorder. Using two complementary behavioral paradigms, trace fear conditioning and a relational/declarative memory radial maze task, together with fiber photometry, optogenetic manipulation, and cFos mapping, they examine whether altered CA1 function contributes to deficits in temporal binding and memory flexibility.

A major strength of the study is the combination of behavioral, recording, and causal manipulation approaches. The trace fear conditioning experiments show that Cntnap2 knockout mice retain associations across shorter temporal intervals but fail when the temporal gap is increased to 40 s. Fiber photometry reveals reduced dorsal CA1 activity under these conditions, and, importantly, optogenetic activation of dorsal CA1 pyramidal neurons during the trace interval rescues subsequent memory performance. This provides compelling evidence for a causal contribution of dorsal CA1 activity to the temporal binding deficit observed in this model.

The radial maze experiments extend these findings to a more complex form of relational/declarative memory. Cntnap2 knockout mice are able to acquire the task but show impaired flexibility when previously learned spatial relations must be recombined. Their behavior is also characterized by greater lateralization, consistent with increased reliance on an egocentric rather than an allocentric spatial strategy. The revised manuscript now explains more clearly how this paradigm distinguishes relational/declarative from procedural learning strategies and how the 20-s inter-trial interval introduces a temporal binding requirement. This clarification substantially improves the conceptual link between the two behavioral paradigms.

The accompanying cFos analyses further show reduced recruitment of hippocampal regions and increased engagement of striatal and prefrontal regions in Cntnap2 knockout mice. These findings are consistent with altered recruitment of memory systems accompanying the behavioral strategy shift. However, unlike the optogenetic experiments in the trace fear conditioning paradigm, the cFos measurements remain correlational. They therefore support, but do not by themselves establish, a causal reorganization from hippocampal-dependent declarative memory toward striatum-dependent procedural learning. The authors have appropriately moderated this interpretation in much of the revised manuscript, although some statements, particularly in the Abstract, could still be phrased more cautiously.

The revised manuscript also addresses the potential influence of the well-described hyperactivity of Cntnap2 knockout mice. The distinction between hyperactivity and impulsive-like behavior is now more explicitly discussed. In the trace fear conditioning experiments, comparable resting and freezing behavior under control conditions argues against locomotor activity accounting for the memory phenotype. In the radial maze, shorter decision latencies combined with longer post-choice running times are more consistent with reduced deliberation than with a simple increase in locomotor activity. Importantly, the authors acknowledge that the task was not specifically designed to measure impulsivity.

Overall, the revision has strengthened the manuscript and addressed my main concerns. The principal conclusion-that reduced dorsal CA1 activity contributes causally to impaired temporal binding in Cntnap2 knockout mice-is well supported by the converging behavioral, photometric, and optogenetic evidence. The broader proposal that hippocampal dysfunction is accompanied by greater reliance on egocentric/striatal learning strategies is also supported by the behavioral and cFos data, provided that it is interpreted as an association rather than a demonstrated causal reorganization of memory systems.

The study makes a valuable contribution by linking hippocampal circuit dysfunction to specific components of declarative memory in a widely used model of autism. More broadly, the combination of temporal binding and memory flexibility paradigms provides a useful framework for investigating how hippocampal dysfunction may alter the organization and flexible expression of memory in neurodevelopmental disorders.

Reviewer #3 (Public review):

Summary:

The manuscript evaluated behavioral phenotypes in the Cntnap2 knockout mouse using two behavioral paradigms: trace fear conditioning and a radial maze task. The trace fear conditioning training is normal, but memory generalization is impaired. The inflexibility is suggested to be related to low activity in dCA1 neurons, which can be rescued by ChR2. The radial maze task data suggested a similar conclusion. Brain-wide cFos mapping indicated impairments in the Cntnap2 knockout mouse. The brain-wide cFos mapping does not show direct correlations with Cntnap2, limiting the interpretation of these data in the context of this paper.

Strengths:

The behavior data are solid.

Comment on revised version.

The authors have addressed all my concerns.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

The uniqueness of this paper is the study of the formation of temporal binding-dependent memories in the cntnap2 mouse, a long-standing mouse model of autism that has been used to test therapeutic modalities.

Strengths:

I liked the combination of optical recordings and interventions and the backup of primary observations with control experiments.

Weaknesses:

(1) Fiber photometry recordings are too coarse to give salient clues to the underlying mechanism.

We acknowledge that fiber photometry provides population-level measurements and does not resolve the activity of individual neurons or synaptic mechanisms. Our aim was to identify alterations in the activity of defined neuronal populations during behaviour rather than to delineate the underlying cellular mechanisms. We agree that future studies employing higher-resolution approaches, such as two-photon calcium imaging, in vivo electrophysiology, or single-cell recordings, would provide important mechanistic insights into the circuit changes underlying the observed activity patterns.

(2) Are perturbed pyramidal cells causally responsible for the altered trace? What can be concluded about the possible role of inhibitory interneurons as potential drivers? The observations focus on abnormal regional activity as observed with fiber photometry and manipulated by optogenetics. The authors should state clearly the limits of their conclusions.

Our optogenetic experiments demonstrate a causal role for the targeted neuronal population in modulating the observed activity and behavioural phenotype. However, we do not conclude that this population is solely responsible for generating the altered fiber photometry signal, nor do we infer that it represents the exclusive driver of the underlying circuit dysfunction. Rather, our findings demonstrate that manipulating this population is sufficient to alter circuit activity, while acknowledging that the recorded signals likely reflect interactions between multiple neuronal populations. We have revised the manuscript to make these distinctions clearer.

(3) I found the "trace" nomenclature confusing. "....in which mice are required to memorize the association between a tone (Conditioned Stimulus) and a mild electric foot-shock (Unconditioned Stimulus), separated by a time interval called Trace (Sellami et al., 2017)." It seems that the conceptual model invokes the creation of an [eligibility] trace, characterized by its progressive disappearance over time. It may be a convention in the field or a matter of language, but it seems perverse to use "trace" to label the time interval rather than the entity that is decaying. If this is an accepted convention going back to Howard Eichenbaum, the authors should cite the paper that first introduced the convention.

We thank the reviewer for this comment. This is indeed an established convention in the classical/Pavlovian conditioning literature rather than terminology specific to our study or to Sellami et al. (2017). The term "trace conditioning" was coined by Pavlov (1927), who used "trace interval" to designate the empty period separating CS offset from US onset, precisely because — as the reviewer intuits — successful conditioning across this gap requires the organism to maintain a memory trace of the CS. In other words, the interval is named for the cognitive/neural entity (the decaying CS trace) that must be sustained across it, not because the interval itself is thought to be a physical or decaying object. So "trace interval" is shorthand for "the interval across which a trace must be maintained," analogous to how "delay conditioning" refers to a paradigm named for a temporal property of the procedure rather than the mechanism per se. We have added a citation to Pavlov (1927) at first use of the term, and clarified the phrasing to make the etymology explicit, as suggested.

(4) I would advocate for the addition of some discussion points for the authors to consider.

(a) Is the retention of activity in CA1 related to phenomena at the cellular or subcellular level in CA1 pyramidal cells? I'm thinking of dendritic, delayed, and stochastic CaMKII activation (DDSC) as defined by Yasuda's group or short-term and associative plasticity of calcium dynamics (STAPCD) as delineated by Caya-Bissonette and Beique.

We thank the reviewer for this insightful suggestion. Our study was designed to investigate network-level dynamics, and the approaches used do not allow us to determine whether the retained activity in CA1 arises from intracellular mechanisms, such as dendritic calcium dynamics or CaMKII-dependent signalling (including mechanisms such as DDSC or STAPCD), recurrent circuit interactions, or a combination of both. We therefore cannot directly assess the contribution of these cellular and subcellular processes. We have now added a paragraph to the Discussion acknowledging that persistent dendritic calcium signalling and CaMKII-dependent plasticity are plausible contributors to sustained CA1 activity and represent an important avenue for future investigation.

(b) Was the optogenetic intervention ever administered in a delayed fashion, capitalizing on the temporal advantages of optogenetics to probe dynamics?

This is indeed an important control. While we did not perform this intervention in the current study, this control was part of our seminal study demonstrating the causal role of the dCA1 in temporal binding (Sellami et al., PNAS, 2017, Fig. 1E doi: https://doi.org/10.1073/pnas.161965711). We showed that aged mice had reduced temporal binding capacity, associated with decreased dCA1 activity. We successfully rescued temporal binding capacity in aged mice through ChR2-induced activation of dCA1 pyramidal neurons during the trace interval, but not outside of the trace interval. Given the striking similarity between the temporal binding deficits observed in aged mice and those reported here in Cntnap2 KO mice, we did not repeat this previously established control in the present study. To acknowledge this point, we have also added a statement to the Discussion noting that confirming the temporal specificity of dCA1 optogenetic manipulation in the Cntnap2 KO model will be an important direction for future studies.

(c) Is the newfound reliance on corticostriatal pathways something more than compensation at the behavioral level? Could it be driven in part by the ASD-related genetic changes?

We agree that the increased reliance on corticostriatal pathways could reflect both compensatory recruitment at the behavioral level and a direct consequence of ASD-related genetic alterations affecting circuit development and function. Our current data demonstrate a shift in circuit engagement but do not allow us to distinguish whether this represents an adaptive compensation or a primary consequence of Cntnap2 deficiency. However, given that ASD-associated mutations can alter the development, connectivity, and plasticity of corticostriatal circuits, it is possible that the observed changes reflect intrinsic circuit reorganization rather than solely a compensatory behavioral strategy. We have revised the Discussion to acknowledge this possibility and to clarify that altered corticostriatal recruitment may represent a direct consequence of the genetic disruption that subsequently shapes behavioral strategies.

Reviewer #2 (Public review):

The authors investigate the contribution of dorsal CA1 hippocampal dysfunction to cognitive impairments in the Cntnap2 knockout mouse model of autism spectrum disorder. Building on previous evidence implicating the hippocampus in episodic and relational memory processes, they combine trace fear conditioning, fiber photometry, optogenetic manipulation, a relational/declarative memory radial maze task, and cFos mapping to test whether altered CA1 function contributes to deficits in temporal binding and memory flexibility.

The study has several important strengths. First, the work addresses a relatively understudied aspect of autism-related cognition, namely hippocampal-dependent memory processes, whereas much of the literature has focused on social behavior, cortical circuits, or striatal dysfunction. Second, the authors employ multiple complementary approaches that converge on a coherent mechanistic hypothesis. The behavioral data demonstrate a reduced ability of Cntnap2 knockout mice to retain associations across long temporal gaps. Fiber photometry recordings reveal reduced dorsal CA1 activity during conditions that challenge temporal binding, and optogenetic activation of dorsal CA1 neurons during the trace interval is sufficient to rescue memory performance. Together, these findings provide strong support for a causal contribution of dorsal CA1 activity to temporal binding deficits in this model.

The second major strength of the manuscript is the extension of these findings to a more complex hippocampus-dependent memory task. The radial maze experiments indicate that Cntnap2 knockout mice show impaired memory flexibility and a greater reliance on egocentric learning strategies. The accompanying cFos analyses suggest altered recruitment of hippocampal and striatal networks during learning, providing a systems-level framework that may explain the observed behavioral phenotype.

Overall, the main conclusions regarding impaired temporal binding and reduced dorsal CA1 engagement are well supported by the data. The optogenetic rescue experiments are particularly compelling because they move beyond correlation and directly test causality. The manuscript therefore makes a meaningful contribution to our understanding of how hippocampal dysfunction may contribute to cognitive abnormalities associated with autism.

Weaknesses:

Some conclusions are necessarily more inferential than others. In particular, the interpretation that the observed behavioral phenotype reflects a broader shift from hippocampal-dependent declarative memory toward striatum-dependent procedural learning is supported primarily by cFos activity patterns and behavioral strategy measures. While the data are consistent with this interpretation, they do not directly demonstrate a causal reorganization of memory systems. Similarly, although the findings identify a mechanism in the Cntnap2 model, caution is warranted when extrapolating these conclusions to autism spectrum disorder more broadly; but I believe this caution is addressed in the discussion.

We agree that our data do not directly demonstrate a causal reorganization of memory systems with cFOS activity patterns. Our intention was to propose that the behavioural strategy together with the brain-wide cFos activation patterns are consistent with a shift in the relative engagement of hippocampal- and striatal-dependent networks. We have revised the manuscript to moderate our interpretation throughout, replacing causal language with wording that reflects an association between the observed behavioural changes and altered recruitment of these memory-related circuits.

Despite these limitations, the study presents a coherent and well-executed body of work that provides novel mechanistic insight into hippocampal contributions to cognitive dysfunction in a widely used autism model. The findings should be of considerable interest to researchers studying hippocampal function, memory systems, and neurodevelopmental disorders.

Reviewer #3 (Public review):

Summary:

The manuscript evaluated behavioral phenotypes in the Cntnap2 knockout mouse using two behavioral paradigms: trace fear conditioning and a radial maze task. The trace fear conditioning training is normal, but memory generalization is impaired. The inflexibility is suggested to be related to low activity in dCA1 neurons, which can be rescued by ChR2. The radial maze task data suggested a similar conclusion. Brain-wide cFos mapping indicated impairments in the Cntnap2 knockout mouse. The brain-wide cFos mapping does not show direct correlations with Cntnap2, limiting the interpretation of these data in the context of this paper.

We agree that brain-wide cFos mapping does not directly identify the molecular or cellular mechanisms by which Cntnap2 deficiency alters circuit function. Rather, we use cFos as a functional readout of network recruitment during behaviour. Our interpretation is therefore limited to identifying differences in activity patterns associated with the behavioural phenotype, rather than establishing direct mechanistic links to Cntnap2 function. We have clarified this point in the revised manuscript and moderated the text accordingly.

Strengths:

The behavior data are solid.

Weaknesses:

The underlying mechanism is not fully investigated.

Major points：

(1) The authors should thoroughly check their manuscript as there are many typos in the current version that affect the readability.

(2) In trace fear conditioning, the tone test impairment can be rescued by ChR2. Have the authors tried rescue experiments with Cntnap2? Rescue experiments in the radial maze task are also essential, either with ChR2 or Cntnap2.

We agree that rescue experiments in the radial maze would provide additional mechanistic insight by testing whether restoring dCA1 activity optogenetically in Cntnap2 KO mice is sufficient to rescue declarative memory flexibility. However, we have already established the causal role of dCA1 in temporal binding and relational/declarative memory in the radial maze task in our earlier work (Sellami et al., PNAS, 2017, Figure 2E). Indeed, optogenetic inhibition of dCA1 in the radial maze task, specifically during the inter-trial interval, prevented flexible relational/declarative memory formation. In the present study, we confirmed the causal contribution of dCA1 activity to temporal binding in Cntnap2 KO mice in the trace fear conditioning paradigm (Figure 1J-L). Therefore, repeating the optogenetic experiment in the radial maze would provide only limited additional information, while requiring dedicated experimental cohorts and additional validation. Moreover, as the team is based in 2 different institutions and countries (France and Australia), we unfortunately do not have the capacity to perform these experiments.

Regarding a rescue with the Cntnap2 protein, here, we used the Cntnap2 KO as a well-accepted model of autism spectrum disorder (ASD), rather than understanding the role of this protein in temporal binding capacity and memory formation. The present study was hence designed to determine how dCA1 activity relates to temporal binding and memory performance in ASD, building on the causal evidence established in our previous work. We therefore consider Cntnap2-specific rescue experiments an important follow-up to further test the sufficiency of dCA1 activation, rather than an essential experiment for establishing the conclusions of the present study.

We have highlighted these points as an important direction for future investigations and revised the manuscript to better distinguish our findings from this additional mechanistic question.

(3) The quality of the cFos example image in Figure 3 is too low. The authors should also provide example images for the other brain regions in the supplementary data, if possible.

The low quality of the cFos image in Figure 3 resulted from a formatting issue during figure uploading. This has now been corrected, and a higher-resolution image has been included in the revised manuscript. We have also added representative images for the other brain regions analyzed to the Supplementary Information, as requested.

(4) The causal link between the brain-wide cFos mapping and the Cntnap2 knockout is weak. How to explain the increase of cFos cell densities in some brain regions, but the decrease in others?

We agree that cFos mapping cannot explain the mechanisms underlying the regional increases and decreases in activity. We interpret these bidirectional changes as reflecting differential recruitment of distributed brain networks rather than direct effects of Cntnap2 deficiency on individual brain regions, and have clarified this in the revised manuscript.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) Improve clarity of the Figure layout: Figure 1, panel I legend. I think the identification of experimental groups is in the wrong place. It really applies to panels K, N, and L, not to panel I. This is important because the genetic stimulation in panels K, N, and L is the key experiment in this whole figure.

We agree that the placement of the experimental group identification in the original figure legend could be confusing. We have revised the figure layout and legend so that the description of the experimental groups is now associated with panels K and L, where the optogenetic stimulation experiments are presented, thereby improving the clarity of the figure.

(2) Fix Figure references. "This indicates impaired retention of trace fear memory and temporal binding ability, relative to WT mice (Figure 2I)....: [there is no Fig. 2I, so I'm guessing you mean Fig. 1I; subsequent Figure references also seem to confuse Fig. 1 and 2].

We thank the reviewer for identifying these errors. We have carefully checked and corrected all figure citations throughout the manuscript, including the reference to Figure 1I-L in this section, to ensure that each citation now refers to the appropriate panel.

(3) Missing words? In both T20 and T40 conditions, regardless of genotype, [the response to] tone frequency significantly decreased.

We thank the reviewer for spotting this omission. The sentence has been corrected to read: "In both T20 and T40 conditions, regardless of genotype, the response to tone frequency significantly decreased."

(4) Jargon that should be declared at an appropriately early point. R/DM = relational/declarative memory.

We have now introduced the term as relational/declarative memory (R/DM) at its first appearance in the manuscript and have ensured that the abbreviation is used consistently thereafter.

(5) Confusing statement? "In the 40s trace-conditioned group, while the frequency of the calcium transients did not differ between WT and Cntnap2 KO mice throughout conditioning (Figure S1G), their amplitude was significantly reduced both during the presentation of the second tone and consistently across the three trace intervals in Cntnap2 KO mice (Figure 1I)."

We have revised this sentence to improve its clarity and explicitly distinguish between the frequency and amplitude of calcium transients. The revised text now makes clear that, although the frequency of calcium transients did not differ between genotypes throughout conditioning, their amplitude was significantly reduced in Cntnap2 KO mice during the second tone presentation and across the three trace intervals.

Reviewer #2 (Recommendations for the authors):

(1) The manuscript would benefit from a clearer distinction between conclusions directly supported by the data and broader interpretations. In particular, statements suggesting a shift from declarative to procedural memory systems could be presented more cautiously, as the cFos analyses provide indirect rather than causal evidence for such reorganization.

We thank the reviewer for this constructive comment. We have revised the manuscript throughout to more clearly distinguish conclusions that are directly supported by our data from broader interpretations. In particular, statements referring to a shift from declarative to procedural memory systems have been moderated.

(2) Additional clarification of the behavioral interpretation of the radial maze task would be valuable for readers who are less familiar with this paradigm, particularly regarding its relationship to relational/declarative memory and its distinction from procedural learning strategies. In addition, the results' interpretation in this paradigm is unclear to me. What is the significance of the 20-second delay in this task (why not use 10 sec or 60 sec)? I believe the paradigm tests spatial rather than temporal distant items (in contrast with trace fear conditioning that refers to temporally distant events)? Please explain further the link between the two behavioral paradigms.

We thank the reviewer for these comments.

Regarding the relationship between relational/declarative and procedural learning: The R/DM task was designed by our group (Marighetto and colleagues; Mingaud et al., 2007; Sellami et al., 2017, 2018) specifically to dissociate two learning/memory systems that can support the same behavioral output (correct arm choice) but that differ fundamentally in their underlying representations and, critically, in their flexibility.

During acquisition, an animal can solve each of the three arm-pair discriminations either by forming a flexible, relational representation of the whole spatial configuration (i.e. an allocentric/hippocampus-dependent "cognitive map" strategy, in which the reward's location is encoded relative to distal cues and to the other pairs) or by learning a set of rigid, response-based rules (i.e. an egocentric/striatum-dependent "turn left/turn right" procedural strategy tied to each specific pair).

Both strategies can produce equivalent accuracy during initial acquisition, which is why acquisition performance alone cannot distinguish them. The flexibility probe (recombining pairs A and B into a novel pair AB, without moving the reward) is the critical dissociation: only an animal that encoded the reward's location relationally, within a broader spatial map, can generalize correctly to this untrained configuration; an animal that relied on a rigid stimulus– response rule for each pair individually would be unable to solve the recombined pair above chance. Performance on pair AB is therefore the read-out of hippocampus-dependent relational/declarative memory, while the degree of left–right lateralization during acquisition (quantified by a lateralization index) is a converging behavioral signature of reliance on the egocentric/procedural system.

We have clarified this distinction in the Methods and Results to facilitate interpretation of the paradigm.

Regarding the 20-s delay and its relationship to trace fear conditioning: The reviewer is correct that the R/DM task and TFC differ in the information being integrated: the R/DM task involves spatially distinct discrimination episodes, whereas TFC involves temporally separated stimuli presented in the same context. However, both tasks require information separated by an interval to be integrated into a unified memory representation, a process referred to here as “temporal binding” (Sellami et al., 2017). The 20-s inter-trial interval in the R/DM task was based on our previous work, in which this interval was shown to be sensitive to age-related deficits in temporal binding and relational memory (Sellami et al., 2017). Similarly, our TFC experiments established a longer temporal-binding capacity in young mice (up to 40 s) compared with aged mice (20 s). Thus, although the two paradigms involve different types of information, they share the requirement to maintain and integrate information across a temporal gap. The different intervals used in the two paradigms reflect their distinct task structures and demands.

(3) Although the authors address potential locomotor confounds, a brief discussion of how hyperactivity and impulsivity might influence behavioral performance in the different paradigms would strengthen the interpretation of the results.

We thank the reviewer for raising this point. We agree that distinguishing hyperactivity from impulsivity strengthens the interpretation of our behavioral phenotypes, and we have added a discussion paragraph making this distinction explicit. Briefly: in the TFC paradigm, our existing open-field data (Figure S1D) show that while Cntnap2 KO mice travel more distance and move faster, their resting time is unaffected relative to WT; so, the reduced freezing/temporal binding phenotype at the 40 s trace cannot be attributed to a general locomotor confound, since freezing (immobility) was actually comparable between genotypes during acquisition. In the R/DM radial maze task, however, we agree the phenotype is better read as impulsivity than hyperactivity per se: KO mice show shorter decision latencies specifically at the choice point (Figure 2G), with no corresponding genotype effect on their post-choice running speed for correct trials (Figure 2H), indicating that the effect is on deliberation before response rather than on general motor output. We have revised the Discussion to make this distinction, and its implications for interpreting the egocentric/procedural bias, explicit.

(4) Minor presentation issues:

- Ensure that all statistical tests, sample sizes, and post hoc comparisons are reported consistently throughout the manuscript and figure legends.

We have carefully reviewed the statistical reporting throughout the manuscript and figure legends to ensure consistency. We have clarified sample sizes, statistical tests, and post hoc comparisons where required and in the methods.

- Carefully proofread the manuscript for minor typographical and grammatical errors. For instance, a) several figures numbers indicated in the main text do not correspond to the figures the authors refers to (page 5 Figure 2I, page 6 Figure 2J and 2K, page 7 "data in fig S2"... and so on); b) rephrase (not clear) page 9: "Differences in the relative contribution of individual substructures between trained WT and Cntnap2 KO mice involved dCA1, dCA2, and dCA3, and favored the DMS, IL, and ACC."

We thank the reviewer for highlighting these issues. We have carefully proofread the manuscript and corrected all figure reference errors and typographical inconsistencies throughout the text. We have also revised the unclear sentence on page 9 to improve clarity and accuracy.

- Page 9, an additional sentence is needed to explain the significance of the bibliographical reference: "Atypicalities in declarative memory have been reported in ASD, with specific impairments in relating items (Minor et al., 2023)."

We have revised this section to clarify the significance of the cited study and its relevance to our findings. Specifically, we have expanded the sentence to highlight that impairments in relational processing in ASD may reflect difficulties in integrating individual experiences into coherent memory representations, a process that relies on hippocampal function.

Reviewer #3 (Recommendations for the authors):

Major Points:

As stated in the public review, the authors need to thoroughly check their manuscript before submission. There are too many typos in the current version that affect the readability. For example, but not limited to:

(1) Page 5, "Figure 2I" should be "Figure 1I"; "Figure 2J" should be "Figure 1J"; "Figure 2K" should be "Figure 1K".

(2) The figure legends of Figure 1J-L are missing.

(3) Page 8, "In contrast, Cntnap2 KO mice showed no learning-induced dCA1 activation together with hypoactivation of dCA3 in both naïve and trained groups compared to WT mice (Figure S2A)", should it be dCA2 instead of dCA1? Please double-check the manuscript.

We thank the reviewer for highlighting these issues. We have thoroughly checked the manuscript and corrected all typos, figure reference errors, and fixed figure legend information. Regarding the point (3), we did mean to describe the lack of training-induced activation of dCA1, but failed to reference back to Figure 3, likely causing the confusion. This has now been clarified. The manuscript has also been carefully proofread to improve clarity and readability.

Minor points:

There are two duplicate rows (two wt) in their raw datasheet, Fig3BCD & S2 and Fig3EF. Please correct them in case of further problems.

WT Naive 178.736 103.821 250.703 185.229 437.575 50.535 266.254 306.187 306.187 WT Naive 178.736 103.821 250.703 185.229 437.575 50.535 266.254 306.187 306.187

WT Naive 46.87 37.94 15.19 8.57 4.98 12.02 8.88 20.98 2.42 12.77 14.68 14.68 WT Naive 46.87 37.94 15.19 8.57 4.98 12.02 8.88 20.98 2.42 12.77 14.68 14.68

We thank the reviewer for noticing this mistake. These duplicated rows have been corrected.

eLife Assessment

This study provides valuable evidence that glycogen phosphorylase is unlikely to represent an effective insecticidal target in Plutella xylostella and that diflubenzuron does not directly inhibit this enzyme. The combination of biochemical, molecular and physiological approaches provides convincing data to support these conclusions, although the proposed metabolic compensation mechanism is supported primarily by indirect evidence and would benefit from direct demonstration.

Reviewer #1 (Public review):

[Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers.]

Summary:

In this study, the authors investigate whether glycogen phosphorylase represents a molecular target of benzoylphenylurea insecticides and evaluate the physiological consequences of suppressing glycogen phosphorylase activity in the diamondback moth Plutella xylostella. The authors combine recombinant protein biochemistry, enzyme inhibition assays, RNA interference, structural modelling, metabolite profiling, gene expression analyses, and physiological measurements to determine whether diflubenzuron directly inhibits glycogen phosphorylase and whether suppression of this enzyme is sufficient to impair insect development. Based on these experiments, the authors conclude that diflubenzuron does not directly inhibit glycogen phosphorylase and that insects tolerate substantial suppression of this enzyme through compensatory metabolic responses.

Strengths:

This study addresses an important question in insect toxicology by systematically evaluating glycogen phosphorylase as a potential insecticidal target. The authors combine complementary biochemical, molecular, physiological, and structural approaches, including recombinant enzyme characterization, inhibitor assays, RNA interference, metabolite profiling, structural modelling, and measurements of fitness-related traits. This integrative approach provides a comprehensive evaluation of the biological consequences of glycogen phosphorylase suppression. In particular, the biochemical evidence that diflubenzuron does not inhibit glycogen phosphorylase, together with the observation that strong suppression of glycogen phosphorylase produces only transient physiological effects without measurable impacts on development or reproduction, provides strong support for the conclusion that glycogen phosphorylase is unlikely to represent an effective standalone insecticidal target.

Weaknesses:

The main limitation concerns the proposed mechanism underlying metabolic compensation. The observed increases in gluconeogenic gene expression, changes in metabolite abundance, and reductions in total protein are consistent with activation of compensatory metabolism, but are insufficient to directly demonstrate increased gluconeogenic flux or establish that amino acid-derived carbon is incorporated into newly synthesized glucose. Similarly, although the analyses of glycogen-associated enzymes strengthen the discussion of alternative metabolic pathways, changes in gene expression alone do not demonstrate that these pathways contribute to glycogen utilization in vivo.

Some mechanistic interpretations therefore extend beyond the data presented. For example, decreases in total protein are interpreted as evidence of protein catabolism fuelling gluconeogenesis, yet they do not directly demonstrate amino acid mobilization or incorporation into glucose. Likewise, increased expression of gluconeogenic genes is interpreted as evidence of increased pathway activity, although transcriptional changes do not necessarily reflect metabolic flux. Finally, the absence of major developmental defects following glycogen phosphorylase suppression is attributed primarily to metabolic compensation, but an alternative explanation is not fully considered. Such explanation could be that glycogen phosphorylase is not rate-limiting for glucose homeostasis under the nutrient-rich experimental conditions, where dietary carbohydrates are continuously available. Consequently, the proposed compensatory mechanism remains plausible and well supported by indirect evidence, but several aspects would benefit from more cautious interpretation.

Overall, the authors successfully achieve their primary objective of evaluating glycogen phosphorylase as a candidate insecticidal target. The study provides useful biochemical and physiological evidence that this enzyme is unlikely to represent an effective target for insecticide development in P. xylostella, while highlighting the importance of metabolic plasticity when assessing metabolic targets. The experimental approaches and datasets presented here should be valuable to researchers studying insect metabolism, insecticide mode of action, and target validation, although the precise mechanisms underlying the proposed metabolic compensation remain an important subject for future investigation.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Strengths:

This study addresses an important question in insect toxicology by systematically evaluating glycogen phosphorylase as a potential insecticidal target. The authors combine complementary biochemical, molecular, physiological, and structural approaches, including recombinant enzyme characterization, inhibitor assays, RNA interference, metabolite profiling, structural modelling, and measurements of fitness-related traits. This integrative approach provides a comprehensive evaluation of the biological consequences of glycogen phosphorylase suppression. In particular, the biochemical evidence that diflubenzuron does not inhibit glycogen phosphorylase, together with the observation that strong suppression of glycogen phosphorylase produces only transient physiological effects without measurable impacts on development or reproduction, provides strong support for the conclusion that glycogen phosphorylase is unlikely to represent an effective standalone insecticidal target.

We thank the reviewer for the positive assessment of our study and for recognizing the value of our integrative approach, including recombinant enzyme characterization, RNAi, metabolite profiling, structural modelling, and fitness measurements. We are also grateful for the acknowledgement that our biochemical evidence and phenotypic observations provide strong support for the conclusion that GP is unlikely to be an effective standalone insecticidal target.

Weaknesses:

(1) The proposed metabolic compensation mechanism is supported by indirect evidence.

We agree with the reviewer that our data demonstrate correlation with, rather than direct proof of, increased gluconeogenic flux. We have revised the manuscript throughout to moderate our interpretations and to clearly frame the compensation model as a plausible interpretation supported by multiple lines of indirect evidence, rather than an established mechanism. Specific revisions are detailed below.

Specific revisions:

(1) Abstract (Lines 35–39):

“We provide evidence that insects compensate through a multi-layered metabolic response: upregulation of gluconeogenic enzymes (PEPCK, G-6-Pase), selective upregulation of glycogen branching enzyme (GBE) but not α-amylase, and changes in protein content suggestive of catabolic substrate mobilization”

(2) Abstract (Lines 41–42):

“Indicating that the metabolic compensation response is ultimately effective in sustaining development”

(3) Abstract (Lines 42–44):

“These findings suggest that GP is functionally non-essential under these conditions, likely through a combination of gluconeogenic compensation and the availability of alternative carbon sources”

(4) Introduction (Lines 88–90):

“Our findings reveal that PxGP is functionally non-essential for larval development under the tested conditions, reflecting a previously uncharacterized gluconeogenic compensation mechanism”

(5) Results – Gene expression (Line 265):

“Gene expression analysis is consistent with gluconeogenic activation”

(6) Results – Protein content (Line 278):

“Changes in protein content suggestive of catabolic substrate mobilization”

(7) Results – Protein decline (Line 283):

“This protein decline is consistent with substrate mobilization”

(8) Results – GBE expression (Line 389):

“PxGP knockdown selectively upregulates glycogen branching enzyme expression”

(9) Results – GBE differential response (Lines 400–403):

“This differential expression pattern—selective GBE upregulation with unchanged α-amylase expression—indicates that the compensatory response to GP suppression involves targeted remodeling of glycogen structure rather than a generalized upregulation of all glycogen-degrading enzymes.”

(10) Results – Fitness assessment (Lines 411–412):

“To determine whether the metabolic changes observed following PxGP knockdown are associated with measurable physiological consequences”

(11) Results – Fitness interpretation (Lines 426–430):

“GP suppression likely triggers protein catabolism to supply amino acids for gluconeogenesis, contributing to transient weight loss. In the presence of continuous dietary carbohydrate supply, the compensatory response appears sufficiently effective to restore metabolic homeostasis before developmentally critical transitions”

(12) Discussion – Glycogen accumulation paradox (Line 520-521):

“a paradox that could be explained by the activation of GP-independent glycogen catabolism via alternative enzymes like...”

(13) Results – Trehalose and G6P (Lines 299–303, 315–317):

“Trehalose levels remained stable at 48-72 h but increased substantially by 96 h (7.44-fold elevation, P < 0.05) (Figure 9F). This increase coincided with the significant upregulation of gluconeogenic genes (e.g., PEPCK and G-6-Pase), consistent with the idea that gluconeogenesis contributes to trehalose synthesis and ensures adequate carbohydrate reserves for the upcoming pupation”

And: “Together, the coordinated behavior of G6P and trehalose is consistent with gluconeogenesis-derived glucose being converted into storage and transport carbohydrates”

(14) Results – Integrated interpretation (Lines 330–332):

“Collectively, these metabolite and gene expression data reveal a biphasic metabolic adaptation that provides a coherent explanation for why substantial GP suppression causes no mortality or developmental defects”

(15) Discussion – Definitive proof (Lines 527–531):

“At 96 h, trehalose and G6P levels in dsGP-treated larvae were maintained at markedly higher levels than in dsGFP controls, which had declined by this time point, coinciding with 3–4‑fold upregulation of PEPCK and G‑6‑Pase. These changes strongly support, albeit indirectly, de novo glucose synthesis as the primary rescue mechanism.”

(16) Discussion – Protein catabolism confirmation (Lines 505):

“provides independent biochemical support for protein catabolism”

(2) Some mechanistic interpretations extend beyond the data presented.

We accept this criticism and have systematically revised the manuscript to distinguish more carefully between observed transcriptional/metabolic changes and functional interpretations. We now consistently frame these as correlative evidence consistent with—but not proving—the proposed mechanisms.

Specific revisions:

The revisions listed under Weakness 1 above—particularly those modifying language around protein decline (Lines 278, 283), gene expression (Lines 265, 389), and trehalose/G6P changes (Lines 299–303, 314–317)—also directly address this weakness by removing the implication that transcriptional or protein-level changes constitute functional proof of pathway activation. Additionally, we have made the following revisions:

(17) Additional discussion of GBE/α-amylase limitation (Line 405-408)

We have added explicit acknowledgement that transcriptional changes alone do not demonstrate functional pathway activation:

“However, as these observations are limited to the transcript level, further enzymatic activity assays or glycogen structure analyses would be required to determine whether this transcriptional change translates into functional alterations in glycogen mobilization.”

(18) Fitness section – clarification (Lines 426–430):

We have revised the interpretation of transient larval weight loss as described above, introducing “likely” and clarifying that the data are consistent with the model rather than conclusively demonstrating it.

(3) An alternative explanation for the limited phenotype is not fully considered.

We thank the reviewer for this important point. We agree that the continuous availability of dietary carbohydrates under our experimental conditions represents a plausible alternative explanation for the lack of phenotype. We have added explicit discussion of this alternative interpretation in the Abstract, Introduction, and Discussion sections.

Specific revisions:

(19) Abstract (Lines 42–44):

As shown above, we have revised the abstract to include: “These findings suggest that GP is functionally non-essential under these conditions, likely through a combination of gluconeogenic compensation and the availability of alternative carbon sources.”

(20) Introduction (Lines 88–90):

As shown above, we have revised the introduction to include: “under the tested conditions” and to acknowledge that the observed phenotype reflects a combination of factors.

(21) Discussion – Biphasic response (Lines 531–539):

We have revised the discussion of the biphasic response to moderate the causal language and acknowledge alternative interpretations:

“This biphasic response—initial metabolic stress followed by delayed (48-72 h) and robust compensation—creates a temporal buffer by 96 h and is consistent with the absence of overt phenotypes despite substantial GP suppression. This temporal coordination raises the possibility of a developmentally programmed response, potentially involving hormonal regulation, that anticipates energy demands at critical transitions. Critically, this pattern suggests a potential vulnerability window (~72 h) during which combined inhibition of glycogenolysis and gluconeogenesis might overcome compensation, although this remains speculative and would require experimental validation.”

(22) Discussion – Alternative explanation paragraph (Lines 540–547):

We have added a following paragraph explicitly addressing the alternative explanation:

“We also acknowledge that the absence of a severe phenotype may reflect an additional, non‑mutually exclusive explanation: under our experimental conditions, with continuous dietary carbohydrate availability, GP activity may not be rate‑limiting for maintaining glucose homeostasis. The observed transient larval weight loss could thus reflect both active metabolic compensation and the inherently low demand on GP for glucose supply in a feeding larva. Distinguishing between active compensation and the non‑limiting nature of GP will require future studies under nutrient‑restricted conditions or during fasting intervals, where GP's role is likely to become more critical.”

Additional revisions to moderate language throughout the manuscript

Beyond the specific revisions addressing each weakness, we have also made the following broader language modifications to ensure consistent and cautious interpretation throughout:

(23) Changes from “ablation” to “suppression” (Lines 180, 195–197, 357):

“Ablation” to “suppression” throughout.

(24) Conclusions – Fundamental principle (Lines 619):

“our findings uncover a fundamental principle of...”

(25) Conclusions – Strategic path (Lines 621–623):

“This metabolic plasticity renders GP non-viable as a standalone insecticidal target but illuminates a potential strategic path forward”

(26) Discussion – Trehalase inhibitors (Lines 556–558):

“For example, trehalase inhibitors such as Validamycin A are potent insecticides [44, 45], suggesting that trehalase inhibition may not be subject to the same degree of metabolic compensation”

(27) Discussion – Metabolic pincer (Line 561):

“a 'metabolic pincer' attack that might overcome adaptive compensation”

(28) MM/GBSA and structural predictions (Lines 27–32, 450, 455–457):

Abstract (Lines 27–32)

“Molecular docking and MM/GBSA analysis predict that this selectivity reflects differential side-chain engagement: GPI is predicted to occupy the allosteric site at the dimer interface via contacts with seven residues spanning both subunits (ΔG = −34.63 kcal/mol), whereas DFB's difluorobenzoyl moiety is predicted to remain solvent-exposed without productive protein contacts (ΔG = −29.29 kcal/mol).”

Line 450

“MM/GBSA analysis indicated”

Lines 455–457

“These structural predictions are consistent with established structure–activity relationships of acyl urea compounds [15], in which target selectivity is governed by the side-chain substitution pattern rather than by the shared acyl urea core”

(29) Scope of conclusion (Lines 471–472):

“This provides the first direct biochemical evidence excluding GP as a candidate molecular target for diflubenzuron”

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Major recommendations

(1) Moderate the interpretation of the proposed compensatory mechanism throughout the manuscript.

We fully agree with the reviewer's assessment. Throughout the manuscript, we have systematically moderated the language used to describe the compensatory mechanism, ensuring that our interpretations are framed as plausible inferences supported by multiple lines of indirect evidence rather than as established conclusions. The specific revisions are detailed in items (1)–(7), items (10)–(16) of our response to Weakness 1 above, and items (24)–(25) of our response to Weakness 3 above.

Additional language modifications addressing the reviewer's concern about assertive terminology are detailed in our response to Minor Recommendation 1 below.

Line 495

“Our investigation shows that this tolerance reflects a robust”

Line 498

“As insects elicited a compensatory gluconeogenic pathway.”

Line 518

“Direct metabolite quantification reflected this compensation”

(2) Clarify the evidence supporting alternative glycogen degradation pathways.

We thank the reviewer for this valuable suggestion. We have revised the GBE and α-amylase sections to more clearly distinguish between transcriptional changes and functional pathway activation. The revised wording now explicitly acknowledges that our observations are limited to the transcript level and that functional confirmation would require additional experiments. The specific revisions are detailed in items (8)–(9) and item (12) of our response to Weakness 1 above. We have also added a sentence acknowledging that these observations are limited to the transcript level (item 17) of our response to Weakness 2 above.

(3) Discuss alternative explanations for the limited physiological phenotype.

We thank the reviewer for this important point. We agree that the continuous availability of dietary carbohydrates under our experimental conditions represents a plausible alternative explanation for the lack of phenotype. We have incorporated this alternative interpretation into the Abstract, Introduction, and Discussion sections, as detailed in items (3)–(4) of our response to Weakness 1 above, and items (21)–(22) of our response to Weakness 3 above.

We have also revised the subsequent conclusion (Lines 553–554) to align with this alternative interpretation, changing “GP fails as a standalone target due to compensation via gluconeogenesis” to “GP appears to fail as a standalone target, at least in part because of compensation via gluconeogenesis.” This ensures consistency with the preceding discussion of GP’s potentially non‑rate‑limiting role.

Minor recommendations

(1) Review the manuscript for statements using terms such as "demonstrates," "confirms," "proves," or "establishes."

We have systematically reviewed the entire manuscript and replaced overly assertive terms (e.g., “demonstrates,” “confirms,” “proves,” “establishes”) with more cautious language (e.g., “suggests,” “is consistent with,” “provides evidence that,” “indicates”) wherever they refer to the proposed metabolic mechanism. Specific revisions are listed under Major Recommendation 1 above. In addition:

MM/GBSA and structural predictions (Lines 27–32, 450, 455–457): Abstract (Lines 27–32): Changed “Molecular docking and MM/GBSA analysis reveal that...” to “Molecular docking and MM/GBSA analysis predict that...”

Results (Line 450): Changed “MM/GBSA analysis confirmed” to “MM/GBSA analysis indicated.”

Results (Lines 455–457): Changed “These structural data confirm that...” to “These structural predictions are consistent with...”

Conclusion scope (Lines 471–472): Changed “This provides the first direct biochemical evidence excluding GP as a candidate molecular target for BPUs” to “This provides the first direct biochemical evidence excluding GP as a candidate molecular target for diflubenzuron.”

Changes from “ablation” to “suppression” (Lines 180, 195–197, 357): Changed “ablation” to “suppression” throughout the manuscript where referring to GP knockdown.

(2) Ensure that the distinction between changes in transcript abundance, enzyme activity, and metabolic flux is maintained consistently throughout the Results and Discussion.

We have carefully reviewed the Results and Discussion sections to ensure that we consistently distinguish between transcript abundance (measured by RT-qPCR), enzyme activity (measured by activity assays), and inferred metabolic flux (not directly measured). The revisions listed under Major Recommendations 1 and 2 above directly address this point. Key changes include:

Consistently using “transcriptional upregulation” or “expression” when referring to qPCR data, rather than “activation” or “pathway activity.”

Explicitly acknowledging that transcriptional changes do not necessarily reflect metabolic flux.

Using “is consistent with” rather than “demonstrates” when linking gene expression changes to functional outcomes.

Additional clarifications and corrections to data presentation

During the preparation of the revised manuscript, we also made several corrections and clarifications to the data presentation:

RNAi knockdown efficiency (Line 364–367): We have added a clarification that the knockdown efficiency in the cohort used for enzyme activity and fitness assays (54.66% at 48 h) was lower than that achieved in the dose–response experiment (87.59% at 48 h), reflecting batch-to-batch variation between independently injected cohorts. We have also noted that the enzyme-activity data should be interpreted against the transcript reduction measured in this same cohort.

Total protein data comparability (Lines 374–380): We have added a note clarifying that the total-protein data shown in Figure 9D and Figure 10–figure supplement 2 derive from independent experimental cohorts processed at different homogenization ratios; absolute protein concentrations are therefore not directly comparable between the two panels. Both datasets nonetheless show a transient decline of approximately 30% in total protein in dsGP-treated larvae within the first 72 h.

Trehalose and G6P data interpretation (Lines 304–309, 314–315, 345–353): We have revised the description of trehalose and G6P levels at 96 h to clarify that the large fold-differences primarily reflect a pronounced decline in dsGFP control values at this time point, rather than a net increase in absolute metabolite content in dsGP-treated larvae. The revised wording now indicates that GP-suppressed larvae maintain their trehalose and G6P pools at a stage when control larvae are actively depleting them.

Gene expression recovery kinetics (Lines 251–252): We have corrected the description of PxTre and PxHex expression at 96 h to state that they “returned to, and modestly exceeded, control levels” rather than “returned to near baseline levels.”

Cell line description update (Lines 644–654): In response to editorial requirements, we have updated the cell line description to include supplier authentication, mycoplasma testing status, and confirmation that the cells were used experimentally within one year of purchase.

Figure corrections:

Figure 2 legend: revised to state “maximum inhibition did not exceed 55.07 ± 7.05%” to match the main text.

Figure 4 legend: corrected normalization description from “control set to 1.0 at each time point” to “calibrated to the 24 h control sample (set to 1.0).”

Figure 6 legend: corrected reference sample from “L1 set to 1.0” to “Egg set to 1.0.”

Figure 8D: corrected y-axis label from “Adult emergency (%)” to “Adult emergence (%).”

Figure 9C: corrected in-figure label from “G-6-P” to “G-6-Pase.”

Terminology correction: We have standardized the use of “similarity” (rather than “identity”) when referring to sequence comparisons (Lines 190, 1075).

PROSITE motif name correction (Lines 181–184): We have corrected “phosphatase-pyridoxal phosphate linkage site” to “phosphorylase pyridoxal-phosphate attachment site.”

Figure 11 statistical description (Lines 1138–1142): We have revised the figure legend to specify that data in (B) and (C) are mean ± SEM of three biological replicates, while data in (D–F) are shown as box plots with sample sizes indicated, and the statistical comparison is dsGP vs. dsGFP (independent samples t‑test). These revisions do not affect any results or conclusions.

Gene expression peak description (Line 301): Changed “coincided temporally with peak expression of gluconeogenic gene” to “coincided with the significant upregulation of gluconeogenic genes” to avoid implying a single defined peak.

Minor textual corrections

We have also corrected the following textual issues:

Sentence structure (Lines 433–435): We have revised the sentence structure to correct a comma splice and clarify the logical relationship. The original sentence “To provide structural insight into the observed selectivity, GPI potently inhibits PxGP (IC₅₀ = 2.96 nM) while DFB does not, we performed...” has been revised to “To provide structural insight into the observed selectivity—in which GPI potently inhibits PxGP (IC₅₀ = 2.96 nM) while DFB does not—we performed...”

Additional minor corrections: We have corrected a small number of typographical errors and stylistic inconsistencies throughout the manuscript.

All of these corrections are limited to data presentation, figure labeling, and textual clarity. They do not alter any experimental results, quantitative conclusions, or the overall interpretation of the study.

eLife Assessment

The paper presents a valuable finding that the human brain and models that incorporate sentence structures can capture sentence-level semantics beyond word meaning, while large language models behave differently. The evidence supporting the authors' claims is solid, though the stimuli are highly controlled and some analyses could be more thorough. This work will be of interest to researchers in language neuroscience and those developing language models.

Reviewer #3 (Public review):

Summary:

Large Language Models have revolutionized Artificial Intelligence and can now match or surpass human language abilities on many tasks. This has fueled interest in cognitive neuroscience in exposing representational similarities between Language Models and brain recordings of language comprehension. The current study breaks from this mold by: (1) Systematically identifying sentence structures for which brain and Large Language Model representations diverge. (2) Using models structured by semantic roles to help account for divergences. As such the study may now fuel interest in characterizing how Large Language Models and brain representations differ, which may prompt new more brain like language models.

Strengths:

(1) This study challenges a literature trend that has touted similarities between Transformer models and human cognition based on representational correlations with brain activity. This challenge is substantiated by identifying sentences for which brain and model representations of sentences diverge.

(2) This study conducts a rigorous pre-registered analysis of a comprehensive selection of the state-of-the-art Large Language Models, on a controlled sentence comprehension fMRI dataset. The analysis is conducted within a Representation Similarity framework to support similarity comparisons between graph structures and brain activity without needing to vectorize graphs. Transformer language models are predicted and shown to diverge from brain representations on subsets of sentences with similar word-level content but different sentence structures.

(3) The study introduces a 7T fMRI sentence comprehension dataset and accompanying human sentence similarity ratings which may be a fruitful resource for developing more human-like language models. Unlike other model-based sentence datasets, the relation between grammatical structure and word-level content is controlled, and subsets of sentences for which models and brains diverge are identified.

Weaknesses:

(1) The interpretation of findings is nuanced. Although Transformers underperform as brain models on the critical subsets of controlled sentences, a Transformer outperforms all other models when evaluated on the union of all sentences when both word-level content and structure vary. Transformers also yield equivalent or better models of human behavioral data. Thus, although Transformers have demonstrable flaws as human models which are pinpointed here, when tested on all sentences (some) Transformers are more human-like than the other models considered.

(2) There may be confounds between the critical sentence structure manipulations and visual processing. This is inconvenient because activation in brain regions that process semantics tends to partially correlate with low-level representations of sentence surface features encoded in visual cortex. Although the study commendably controls for confounds associated with sentence length, correlations with key sentence structure models are most salient in visual cortex and diminish in other brain networks when V1-V4 activation is controlled for.

(3) Sentence similarity computations are emphasized as the basis for unifying comparative analyses of graph structures and vector data. A strength of this approach is that correlation is not always the ideal similarity metric. However, a weakness is that similarity computations are not unified across models. This has practical consequences because different similarity metrics applied to the same model produce positive or negative correlations with brain data.

Comments on revised version.

Thanks again for the responses. In particular, the new sentence-length control is helpful for interpreting the anomalous outcomes from the DIEM analysis, and the amendments to the discussion are appreciated.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

(1) The rationale behind averaging sentence embeddings across multiple transformer models (with different architectures and training objectives) is unclear. These transformer-based models have different training paradigms and model architectures, which may result in misaligned semantic spaces. The averaging operation may dilute the distinct sentence representations learned by each model, potentially weakening the overall semantic encoding for sentences. Please clarify this choice or cite supporting methodology.

The reviewer questions the rationale for averaging sentence embeddings across different models. However, our method involves computing correlations separately for each model, then averaging the correlations. We apologize for the confusion. We have clarified this on page 3:

“Results for the ‘Transformers’ model are computed by computing correlations separately for five different transformer models and then taking a simple average of these correlations. Results for each individual transformer are presented in Supplementary Information Figure S2.”

(2) All structure-sensitive models discussed incorporate semantics to some extent. Including a purely syntactic baseline, such as a model based on context-free grammar, would help confirm the importance of syntactic structures.

Following the suggestion, we have implemented two syntactic models and discuss the results on page 10:

“We also found that purely syntactic models based on constituency parses (see Benepar and CFG) show poor correlations with brain activity (see Supplementary Information Figure S2). Examining the corresponding RSA matrices (see Figure S1), this seems to be due to such models being overly sensitive to syntactic form, and relatively insensitive to which words are assigned to different nodes within the syntactic tree. This is most evident for the edit-distance similarity metric, and to a lesser extent also for the subtree similarity metric. This finding highlights the value of hybrid approaches designed to appropriately balance sensitivity to lexical, syntactic, and compositional information in representing semantic information at the sentence level.”

(3) In Figure 2, human behavioral judgments show weak correlations with neural data, and even fall below those of computational models, suggesting the behavioral judgments may not reflect the sentence structures in a brain-like way. This discrepancy between behavioral and neural data should be clarified, as it affects the interpretation of the results.

While the behavioural judgements are made by different participants and involve a different task than the neuroimaging results, nonetheless we agree the difference is surprising and warrants more detailed consideration. We have included a more detailed discussion of this issue on page 11:

“Our study has several limitations. First, we found a surprisingly low correlation between behavioural ratings and brain activations (see Figure 2). This may be partly explained by differences in task structure. In the behavioural experiment, participants viewed many pairs of related sentences, and were explicitly asked to pay attention to differences in the words of each sentence. In contrast, in the fMRI task, participants read one sentence at a time without an explicit comparison. In addition, we suspect that presentation of so many sentence pairs with highly similar structures may have biased the way in which participants rated sentence similarity. Modifications to the behavioural task to mitigate these aspects may reduce the divergence between behavioural and brain findings.”

(4) To better contextualize model and neural performance, sentence similarity should be anchored to a notion of semantic "ground truth", such as the matrix shown in Figure 1a. Comparing this reference with human judgments, brain responses, and model similarities would help establish an upper bound.

While our design matrix served as the basis for constructing a set of stimuli with systematic modifications, we respectfully suggest that it should not be regarded as a ‘semantic ground truth’. Sentence pairs within each category will not have the same degrees of semantic similarity since the words and context differ across sentences in a graded manner. Furthermore, while we anticipated ‘different’ sentence pairs would be less similar than ‘swapped’ sentence pairs, and that within each of the six block diagonals the ‘modified’ or ‘substituted’ sentence pairs would be the most similar, we did not have any prediction about the magnitude of these differences. Our goal was to construct a set of sentence pairs which spanned a range of semantic similarities, and allowed for dissociation between lexical similarity and overall similarity in meaning. The design matrix is not intended to represent a ‘ground truth’ that human judgements or brain representations would be expected to conform with.

(5) The structure of this paper is confusing. For instance, Figure 5 is cited early but appears much later. Reordering sections and figures would enhance readability.

We agree that placement of figures was not ideal in the previous draft. We have reworked the manuscript so that all figures appear closer to their mention in the text, and the figure (now Figure 3) appears in the correct order. We have also substantially revised the discussion, and included subheadings to help guide the reader through the various different issues we include.

(6) While the analysis is broad and comprehensive, it lacks depth in some respects. For instance, it remains unclear what specific insights are gained from comparing across brain regions (e.g., whole brain, language network, and other subregions). Similarly, the results of simple-average and group-average RSA appear quite similar and may not advance the interpretation.

We included both analyses in line with our preregistration, and also because we believe the fact that two distinct approaches to analyzing the data yield similar results strengthens our conclusions.

(7) While explaining the grid-like pattern due to sentence length is important, this part feels somewhat disconnected from the central question of this paper (word order). It might be better placed in supplementary material.

We believe that the grid-like pattern in the RSA results is an important unexpected finding that warrants discussion in the main manuscript.

Reviewer #3 (Public review):

(1) The interpretation of findings is nuanced. Although Transformers underperform as brain models on the critical subsets of controlled sentences, a Transformer outperforms all other models when evaluated on the union of all sentences when both word-level content and structure vary. Transformers also yield equivalent or better models of human behavioral data. Thus, although Transformers have demonstrable flaws as human models, which are pinpointed here, in the general case, (some) Transformers are more human-like than the other models considered.

The reviewer argues that we overstate some of our conclusions, as several transformers achieve higher brain correlations than the hybrid model when computed over all sentence pairs, as well as on the behavioural data. In response, we first note that our primary interest in this paper is on the block-diagonal sentence pairs, as these were specifically designed to interrogate how different models represent sentence structure. The comparison with all sentence pairs is presented for comparison but is not our primary focus on this paper, as also reflected in the pre-registered prediction that our VerbNet-CN hybrid model would show higher brain correlations than transformers over this block diagonal subset.

Second, we have included a new analysis in the revised manuscript (Figure S9) where we compute brain correlations controlling for the pattern of similarities observed in the primary visual cortex (averaged over participants), as a way to control for visual similarity. This added control substantially reduces the brain correlations of the transformers, such that they all have lower correlations than VerbNet-CN and AMR-smatch even over the set of all sentence pairs. We provide interpretation of this result in the discussion.

Third, we would like to note one of the disadvantages of transformers as a model of mind or brain representations is that they are largely a ‘black box’ whose workings are poorly understood. One advantage of hybrid models like our simple semantic role model is that they can be much easier to interpret, thereby enabling them to be used to determine which features are most important for brain representations of sentence meaning, and what mechanisms are used to combine individual words into a full sentence. Given their relative simplicity and interpretability, we believe hybrid models have considerable value as scientific tools, even in cases where they achieve comparable correlations to transformers. We have added a short discussion of this issue in the revised manuscript (page 10).

(2) There may be confounds between the critical sentence structure manipulations and visual representations of sentence stimuli. This is inconvenient because activation in brain regions that process semantics tends to partially correlate with visual cortex representations, and computational models tend to reflect the number of words/tokens/elements in sentences. Although the study commendably controls for confounds associated with sentence length, there could still be residual effects that remain. For instance, the Graph model correlates most strongly with the visual cortex despite these sentence length controls.

We agree with the reviewer that this is a potential confound. As noted in the previous response, we have implemented a new control analysis in which we directly control for visual similarities as reflected in participant-averaged similarities of primary visual cortex activations in response to all stimuli. These results are shown in Figures S8-S11 in the SI. We show that transformer correlations are reduced much more than graph and hybrid models with this control. Also, we note that the AMR-smatch graph model shows high correlations with other brain regions even after removing correlations with the visual cortex (Figure S10). This indicates that the model represents a range of sentence features, including both superficial visual or length-related features, as well as semantic features that are represented in common with language and other cortical regions.

(3) Sentence similarity computations are emphasized as the basis for unifying comparative analyses of graph structures and vector data. A strength of this approach is that correlation is not always the ideal similarity metric. However, a weakness is that similarity computations are not unified across models. This has practical consequences here because different similarity metrics applied to the same model produce positive or negative correlations with brain data.

The reviewer notes that the method for computing similarities differs between the vector-based (mean and transformer) models, and the hybrid and syntax-based models, thereby potentially adding an additional confound to our results. We agree that this is a potential limitation, and our correlations should always be understood as applying to a model paired with a similarity metric. However, we believe that this is mostly unavoidable when comparing different formalisms. In the revised manuscript we have incorporated an entirely new similarity metric for vector-based models (DIEM similarity), as well as an extended discussion of the effect of different similarity metrics for graph and hybrid models.

eLife Assessment

This is a potentially valuable study in the area of connectomics, comparing across species to obtain a uniform synaptic density of ~1 synapse per micron. While the methods add to the field, the datasets are limited and filtered, and the role of cell-type on spacing is passed over. Additionally, some of the analyses could be tightened; hence, overall, the findings remain incomplete.

Joint Public Review:

Summary:

Brain sizes vary by orders of magnitude between different organisms, while neuronal circuits maintain essential computational functions. Castro and Cardona explore invariant features across animal species that lend themselves as fundamental constraints on scaling mechanisms. The authors leverage a number of existing connectomics datasets for their study, from which they extract morphologies and synaptic connections for numerous neurons, resulting in a set of neuronal features such as neuron lengths and synapse counts.

Their core result is a rather stable synapse density of 1 synapse per 1 µm of dendritic path length across neurons from different datasets, in line with prior results in mouse (Turner et al., 2022) and human (Lomba et al., 2022). The authors show that this finding also holds for the selected neurons in Drosophila (adult and larva) and zebrafish larva, and further aim to explore mechanisms creating this density and its consequences on neuronal excitability. They show that their data is in agreement with previously reported scaling rules of dendritic arborization (Cuntz et al., 2012). Through exploration of synapse sizes and dendrite radii, the authors identify a compensatory effect of dendrite radius for fluctuations of synaptic densities that can provide voltage response stabilities.

Overall, this study is an addition to the burgeoning field of comparative connectomics, and as such, the presented analyses are important methodological contributions. The conclusions presented here are, however, too general. The analyzed datasets are reduced to a few cell types due to technical challenges, and cell type identities are ignored even though they have large ramifications for the analyses presented (e.g., E vs I for stability analyses, and cell type correlations with synapse sizes).

Strengths:

(1) Novelty.

The authors provide new ways to compare neuronal measurements between brains and species and how to interpret the results. In particular, their conclusions for normalization of connection strength (by total number of synapses onto a neuron) are an important contribution to the analysis of neuronal circuits, especially in Drosophila. The finding of the density is important, but similar numbers have been reported by prior studies of some of the same datasets (Loomba, Turner).

(2) Wiring optimization.

The presented study is an important validation of the scaling law across organisms and cell types.

(3) Scale.

The authors collected neurons from several connectomics datasets from different organisms that were created by different labs, requiring standardizing reconstructions and extracting features across a large number of neurons

Weaknesses:

(1) Data preparation and presentation. The authors describe how the data was processed and manually corrected in the methods, but at no point other than in Figure S4 (which shows a single neuron from afar) is an actual neuron skeleton shown to convince the reader that this was done well. The resolution of the skeleton has important ramifications for the measured path lengths. Further, it is not clear how spine heads, very short branches, and twigs were handled. Tracing out spine heads could double the path length, similar for twigs. How these were handled should be discussed and justified. Radius measurements were taken at the trunk of the neuron, but for synapses on spines, the spine size (especially neck (Harnett et al., 2012)) is critical for estimating input resistance. This should at least be discussed and justified. Further, it is not clear how synapses onto the soma were handled. What was chosen as the radius there?

(2) Data filtering. The authors excluded substantial numbers of neurons from the analyzed datasets, leaving only cells from a few cell types. However, the manuscript gives the impression in many places that the analyses were run across the entirety of those datasets. Given the number of neurons used (e.g. only 60 from the Drosophila adult!), it would be much more prudent to talk about the specific cell types and not the datasets/animals. Further, the way the included neurons were chosen is likely to introduce a bias. E.g., in Drosophila datasets, neurons with more synapses on the backbone rather than twigs automatically have higher postsynaptic completion ratios. Overall, given the number of neurons, the results cannot be easily generalized even for the datasets used here.

(3) Ignoring cell types. Especially for the mammalian datasets, presynaptic cell type identity is correlated with synapse size and location. Ignoring this information removes important context from the analysis. For instance, inhibitory synapses are more prevalent close to the soma, etc.. This clearly impacts the analyses presented, yet is ignored and not discussed. Further, the sign of the synapses and, in particular, the balance of E vs I synapses, is important for the analysis of voltage response stability.

(4) Wiring optimization rule. The presented analysis convincingly confirms the scaling law from Cuntz, et al. (Figure 3b). However, taken together with the prior finding of the 1 synapse / 1 µm density along dendrites, the sensitivity analysis appears to be circular (Fig. 3c). Since the inputs are length and synapse count, the same that were put into the density analysis, the result rather reconfirms the scaling law.

(5) Synapse size and PSD area. Two measurements for the synapses were used for the MICrONS datasets, but in some places, this is confusing. Why were the volumetric synapse sizes not fully replaced with the PSD measurements, as these are likely a superior measurement of the "size" of the synapse? Annotations of the PSD should also be shown somewhere to convince the reader of their quality.

(6) Zebrafish dataset. The validity of the corrections used for the zebrafish dataset is difficult to judge for someone not familiar with that dataset. It appears that there are substantial problems in the reconstruction of the dendritic tips. Other datasets were ignored for seemingly similar reasons. Why was this one kept? The correction factor introduces an arbitrariness to the analysis, and it directly affects the core result of this study.

eLife Assessment

This in vitro study provides important evidence that the mitochondrial intermembrane space protein HAX1 functions as a direct activating cofactor of the AAA+ chaperone CLPB, resolving a longstanding question arising from the strong genetic, proteomic, and disease associations between these proteins. The convincing work advances our understanding of mitochondrial proteostasis by demonstrating that HAX1 promotes CLPB oligomerization and will potentially even enable the generation of small molecules that mimic the function of HAX1 in CLPB activation. The study will be of general interest to biochemists working on chaperones and/or mitochondrial biology.

Reviewer #1 (Public review):

The mitochondrial intermembrane space (IMS) is a compartment under constant proteostasis stress. During the development of multicellularity, the IMS acquired the AAA+ ATPase CLPB, a disaggregase whose absence from cells results in aggregate formation in the IMS and whose mutation in humans leads to rare but severe human diseases.

The precise molecular function of CLPB in the human IMS remains unresolved and is addressed in this study, in particular the crosstalk with the IMS protein HAX1, which is a prominent interaction partner of CLPB.

The authors find HAX1 to serve as an activating cofactor of CLPB as a disaggregase and refoldase in a purified system. This is well in line with the similar phenotypes of HAX1 and CLPB loss, and is an exciting finding as it not only assigns a direct role for HAX1 in the IMS but also yields the potential for CLPB activity regulation by regulating amounts of HAX1. The experimental support for this finding is strong and acquired by a combination of different and carefully executed in vitro enzyme activity assays.

One potential weakness of this study is that it does not consider recently identified players in the CLPB-HAX1 axis, FAM136A (now called TIMCC) and MIA40, that might additionally modulate/regulate CLPB activity. This might constitute an exciting route for future research.

Reviewer #2 (Public review):

The authors sought to determine the mechanistic relationship between the mitochondrial intermembrane space proteins CLPB and HAX1, whose genetic, proteomic, and disease associations have long suggested a functional partnership. Using biochemical reconstitution with purified proteins, they provide evidence that HAX1, an intrinsically disordered protein, acts as a direct activating cofactor of CLPB, stimulating its oligomerisation, ATPase activity, disaggregase activity, and refoldase activity. They further identify a region of HAX1 required for CLPB interaction and propose a model in which HAX1 promotes formation of a distinct active CLPB assembly.

A major strength of the work is the comprehensive biochemical approach. The authors combine activity assays, mutational analysis, interaction studies, and oligomerisation measurements to address the central question from multiple angles. The data provide convincing evidence that HAX1 is not simply a client of CLPB but instead functions as a positive regulator.

The principal weakness concerns the mechanistic model of oligomer remodelling. While the SEC and stoichiometric analyses are consistent with the formation of a smaller HAX1-bound CLPB complex, the proposed transition from a dodecameric to a hexameric assembly is inferred rather than directly demonstrated. Additional structural or biophysical evidence would strengthen this aspect of the study.

Overall, the authors largely achieve their aims. The evidence strongly supports the conclusion that HAX1 is a direct stimulatory cofactor of CLPB and provides an important mechanistic framework linking two proteins associated with overlapping mitochondrial and hematological disease phenotypes. Some aspects of the proposed oligomeric remodeling mechanism should be interpreted more cautiously, as they remain supported primarily by indirect evidence.

eLife Assessment

This valuable large-scale comparative study of odorant receptor (OR) genes across more than 100 insect species, combining sequence- and structure-based approaches, is a significant step in the exploration of the evolution of this large gene family involved in the detection of odorant signals by olfactory neurons. This work uncovers a structural feature unique to the odorant receptor co-receptor Orco that reduces ligand binding affinity. However, the strength of evidence is incomplete relative to the authors' claims about the inferred number of OR genes per species, the odor response spectra and the evolutionary and ecological interpretations.

Reviewer #1 (Public review):

Objectives of the study and impact of the work

The authors of this article primarily aim to reconstruct the evolutionary history of the insect odorant receptor (OR) family, which is responsible for the detection of odorant signals by olfactory neurons. Due to the lack of phylogenetic signal present in the sequences of this multigene family, which evolves very rapidly, phylogenetic analyses have so far never made it possible to precisely retrace how ORs diversified prior to the appearance of present-day insect orders, and what the drivers of this diversification were. For example, one may suspect that the adaptation of ORs to odors emitted by plants constituted a critical step in insect evolution during the "angiosperm terrestrial revolution," which occurred at the end of the Cretaceous, but nothing currently allows this to be asserted.

There are very nice examples, notably in drosophilids, derived from comparisons between closely related species and documenting mechanisms of OR adaptation to certain signals. However, what the authors attempt to do in this work is to produce a macroevolutionary analysis at the scale of insects as a whole, based almost exclusively on bioinformatic analyses. To do this, they annotated OR genes in about one hundred insect species and developed pipelines for analyzing sequence similarity, structural similarity and functional similarity, the latter being estimated through a molecular docking approach. An important element in the evolution of insect ORs is the appearance of a unique co-receptor, called Orco, which appears to be an OR that has lost the ability to bind odorants. In addition to the large-scale bioinformatic analysis, the authors also aim to explore more specifically the factors that favored the emergence of Orco and the selective advantage conferred by the existence of OR-Orco complexes.

Given the importance of odorant receptors in insect biology and in their adaptation to different environments and lifestyles, retracing their evolutionary history is indeed a major question in evolutionary biology. In principle, this type of work therefore has the potential to become a reference in the field and to provide a basis for significant scientific advances.

Major strengths and weaknesses

The sampling chosen for collecting OR sequences is very impressive, with more than 100 insect families represented, covering most of the major orders. This sampling appears appropriate for the question being addressed. The analysis pipeline used to collect the sequences makes sense, relying on homology-based annotation tools coupled with a structure-based filter. Nevertheless, one can note aberrant numbers of ORs for certain species (much lower than reality). A lower number of OR genes is somewhat expected, as the authors chose to apply a fairly stringent filter on sequence quality (based on predicted 3D structure), which reduces the number from 14,000 to 9,000. This choice seems logical given the subsequent use of these data, but it inevitably leads to data loss. However, the low number of genes also results from the fact that the pipeline did not function correctly for all genomes.

In the revised version of the manuscript, the authors included a benchmarking step, which is a good point. They compared their OR gene annotations with previous reports in the same species. Unfortunately, this comparison is irrelevant because the chosen reference OR repertoires are actually a mix of transcriptome and genome annotations. Furthermore, comparisons with entire OR gene repertoires essentially demonstrate that their annotation was of good quality for species in which OR sequences were already present in the query OR database, but poor for species in which they were not. Therefore, these supplementary analyses made by the authors are not particularly in favor of an overall good quality of OR gene repertoires in the >110 species studied. The fact that some OR genes may be missing and that the total number may not be exact for each species is not prohibitive for studying the evolution of the family on a broad scale. However, it does call into question the correlation between the number of ORs and lifestyles and diets.

From the dataset collected, the authors attempted to categorize ORs in several ways, starting with the reconstruction of sequence similarity networks. The approach is interesting, but fails to reveal homology relationships between ORs from species belonging to different insect orders. So it is unclear what the advantage of this approach is compared with the "classical" phylogenetic approach.

The clustering based on structure also leads to the identification of a majority of "order-specific" clusters, which does not provide major insights into the evolution of ORs. However, the authors highlight a group of ORs in flies that appear to possess an unusual intracellular region, as well as a cluster of OR shared across many insect orders that exhibit a larger binding cavity. This is really interesting, although more relevant to OR structure than to their evolution.

The analysis of structural diversity then leads the authors to focus on the Orco co-receptors, which are characterized by modifications of the binding pocket and the appearance of an extracellular loop that could explain the loss of the ability to bind odorant molecules. This part, which relies on in vitro experiments, is interesting and constitutes the most striking result of this study, which could in itself have been the subject of a separate manuscript.

The rest of the manuscript is based on the prediction of OR response spectra using molecular docking. The work that has been carried out is extremely substantial, and the objective of linking clusters based on sequence similarity or 3D structural similarity with functional categories is entirely relevant. The docking score threshold used was chosen thoughtfully, which is very good, and according to the calculation performed should ensure a true positive rate of more than 20%, which is excellent in such a docking analysis. But in the absence of functional validation, this 20% true positive rate is not sufficient to extrapolate OR function, and docking-derived binding breadth measures used in the remaining of the manuscript have to be taken with caution, as acknowledged by the authors themselves. Consequently, the chance that results of this part of the work will enlighten the evolution of OR on a broad scale is low. For example, the fact that insect lineages that emerged after the Permian-Triassic extinction have more broadly-tuned OR is an interesting observation, yet remains highly speculative.

In summary, despite the large number of analyses performed, the authors do not really succeed in achieving the stated objective of reconstructing the evolutionary history of insect ORs, and the results obtained do not strongly support all the conclusions regarding the links between OR repertoires and environment or lifestyle.

Reviewer #2 (Public review):

The remarkable evolvability of the olfactory system enables animals to rapidly adapt to dynamic and chemically complex environments. Over the past two decades, substantial effort has been devoted to uncovering the evolutionary principles that drive the diversification of odorant receptors (ORs), yielding key insights into the forces shaping their striking variability in both vertebrates and insects. In this manuscript, Zhang and colleagues analyze the OR repertoires of over 100 insect species, leveraging sequence and structural similarity to infer patterns of gene family evolution within this diverse and ecologically important clade. By integrating sequence-based and structure-based comparisons, their study builds on a compelling and recently emerging line of research made possible by the advent of AlphaFold, which has previously clarified the phylogenetic relationship between insect Ors and the gustatory receptor gene family and revealed the unexpectedly deep evolutionary origins of this ancient structural fold.

Applying this approach to a large set of ORs derived from species throughout the insect phylogeny, the authors confirm many previously reported patterns of OR evolution. More importantly, with their large-scale approach the authors found new structural features of different Or clades, resulting in intriguing testable hypotheses about Or evolution that have the potential to support future directions in studying this important gene family. For example, the authors identify a structural feature mostly unique to the OR co-receptor ORco, a beta-sheet in EL2, which they functionally show reduces odorant binding affinity - a key aspect of ORco, which does not bind ligands in the ancestral ligand-biding site. This is a particularly strong part of the manuscript, since the authors support their in silico-derived hypothesis with functional data.

In an attempt to assess the relationship between sequence identity and structure on one hand and function on the other, the authors perform an in-silico structure prediction and chemical docking analysis. While not suitable to assess the actual function of individual receptors without functional testing, this strategy revealed potential receptor clades with differences in the number and type of docked chemicals. As a response to the previous reviews, the authors now changed their writing to refer to these results as 'docking-derived function' instead of just 'function'. While I appreciate that the authors changed their wording, this semantic change still can be read as referring to docking-derived predictions as conclusions of actual biological function. However, this is not the case. Any functional claims based solely on the docking approach need to be interpreted with caution, as in silico-only approaches are currently not well enough predictors for actual receptor function. Accordingly, I strongly recommend to remove the parts that connect receptor function to structure or ecology.

Assessing their docking analyses with respect to species ecology, the authors identify correlations suggesting that Or repertoires might follow predictable evolutionary trajectories broadly aligned with species ecology. These are interesting hypotheses. However, due to the inability to predict actual receptor function the author's claims about ecology-function relationships are not well supported. Accordingly, the authors should change the respective parts of the manuscript in way that it is clear these are currently untested hypotheses. This includes adjusting the title, since the work does not find any experimental support for ecological tuning.

I believe that future studies combining functional testing with an explicit evo-eco framework could benefit from this work and the explicit hypotheses it generates. But the authors should be careful to not overstate their in-silico derived results.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Objectives of the study and impact of the work:

The authors of this article primarily aim to reconstruct the evolutionary history of the insect odorant receptor (OR) family, which is responsible for the detection of odorant signals by olfactory neurons. Due to the lack of phylogenetic signal present in the sequences of this multigene family, which evolves very rapidly, phylogenetic analyses have so far never made it possible to precisely retrace how ORs diversified prior to the appearance of present-day insect orders, and what the drivers of this diversification were. For example, one may suspect that the adaptation of ORs to odors emitted by plants constituted a critical step in insect evolution during the "angiosperm terrestrial revolution," which occurred at the end of the Cretaceous, but nothing currently allows this to be asserted.

There are very nice examples, notably in Drosophilids, derived from comparisons between closely related species and documenting mechanisms of OR adaptation to certain signals. However, what the authors attempt to do in this work is to produce a macroevolutionary analysis at the scale of insects as a whole, based almost exclusively on bioinformatic analyses. To do this, they annotated OR genes in about one hundred insect species and developed pipelines for analyzing sequence similarity, structural similarity, and functional similarity, the latter being estimated through a molecular docking approach. An important feature in the evolution of insect ORs is the emergence of a unique co-receptor, called Orco, which appears to be an OR that has lost the ability to bind odorants. In addition to the largescale bioinformatic analysis, the authors also aim to explore more specifically the factors that favored the emergence of Orco and the selective advantage conferred by the existence of OR-Orco complexes.

Given the importance of odorant receptors in insect biology and in their adaptation to different environments and lifestyles, retracing their evolutionary history is indeed a major question in evolutionary biology. In principle, this type of work therefore has the potential to become a reference in the field and to provide a basis for significant scientific advances.

Major strengths and weaknesses:

The sampling chosen for collecting OR sequences is very impressive, with more than 100 insect families represented, covering most of the major orders. This sampling appears appropriate for the question being addressed. The analysis pipeline used to collect the sequences makes sense, relying on homology-based annotation tools coupled with a structure-based filter. Nevertheless, one can note aberrant numbers of ORs for certain species (much lower than reality), which indicates that the pipeline probably did not function correctly for all genomes. In the absence of a validation step comparing the results with already known OR repertoires, it is difficult to estimate the overall quality of the data. The authors chose to apply a fairly stringent filter on sequence quality (based on predicted 3D structure), which reduces the number from 14,000 to 9,000. This choice seems logical given the subsequent use of these data, but it inevitably leads to data loss. The fact that some OR genes may be missing and that the total number may not be exact for each species is not prohibitive for studying the evolution of the family at a broad scale; however, it calls into question certain results that rely on this total number, such as the correlation between the number of ORs and genome size, lifestyle, and diet.

We thank the reviewer for raising this important concern. To objectively evaluate how much our strict structural filtering may have reduced OR counts, we collected published OR annotations for species included in our study and compared those values with our structurally intact OR counts (Author response table 1). For most previously annotated species, the difference was within approximately ten ORs, indicating that our counts are generally comparable to published annotations after structural filtering. Diabrotica virgifera virgifera was a clear exception. Previous work, using transcriptomic evidence and manual annotation, reported 193 ORs, including 124 complete ORs and three pseudogenes, whereas our original pipeline retained only nine structurally intact ORs. We found that this discrepancy was mainly caused by insufficient query representation. The previous D. virgifera virgifera annotation was published in 2025, whereas our original annotation was performed in 2022, so our query set likely did not adequately represent this lineage. When we re-annotated this species using new queries, 78 complete ORs were retained after structural filtering. We have updated the D. virgifera virgifera annotation in the revised dataset.

We also considered genome size as a possible contributor to low automated recovery. Among the 115 species analyzed, D. virgifera virgifera has the third-largest genome, after Locusta migratoria and Thermobia domestica. Large genomes contain many repetitive regions, which can make automated OR annotation more difficult and may reduce the number of recovered intact gene models. The two larger-genome species in our dataset had previously been manually annotated with transcriptomic support, suggesting that automated annotation alone may be less reliable for large, repeat-rich genomes. At the same time, published OR annotations are not always free from error. For example, Propsilocerus akamusi was previously reported to have 17 ORs, whereas our pipeline retained ten complete ORs. After downloading and modeling the 17 published sequences, we found that PaOR1, PaOR5, PaOR6, PaOR9, and PaOR14 differed substantially from canonical OR structures or appeared to be incorrectly annotated (Author response image 1). Truncated genes such as PaOR4, which lacks part of the intramembrane region of TM2-4 but still clearly resembles an OR, would be retained by our strategy; sequences with more severe structural inconsistency would be excluded. Because many previous OR annotations have not been experimentally validated, it is difficult to treat all published OR counts as exact ground truth.

To test whether possible OR loss affected OR-count-based conclusions, we first removed all potential GR sequences from the dataset. For species with published OR counts, we then repeated the pGLS analyses using the larger of our structurally intact OR count and the published OR count (Fig. S1b-e). The main ecological associations were retained: OR count remained associated with larval diet, adult diet, and habitat, and was not associated with circadian rhythm. This indicates that the main OR count ecological analyses are reasonably robust to moderate underestimation caused by missed ORs or strict structural filtering.

In contrast, the association between OR count and genome size was no longer supported in this sensitivity analysis (Author response image 2a and b). We therefore removed the analysis and discussion of a negative relationship between genome size and OR count. The original pattern likely reflected the difficulty of annotating ORs in large, repeat-rich genomes rather than a reliable biological relationship.

The following revision has been added to the revised manuscript (lines 122-126): " We then repeated the phylogenetic generalized least squares (pGLS) analyses using the larger of our structurally intact OR count and the published OR count, and the main ecological associations were retained. This indicates that the main OR-count ecological analyses are reasonably robust to moderate underestimation caused by missed ORs or strict structural filtering."

These new benchmark and sensitivity analyses directly strengthen the evidence supporting our dataset and its ecological interpretations.

From the dataset collected, the authors attempted to categorize ORs in several ways, starting with the reconstruction of sequence similarity networks. The approach is interesting, but in the end, the results do not seem to be sufficiently exploited, and it is not obvious what the advantage of this approach is compared with the "classical" phylogenetic approach, which generally fails to reveal homology relationships between ORs from species belonging to different insect orders. Here again, the majority of the clusters identified are "order-specific," and when this is not the case, the authors did not attempt to exploit the results. For example, clusters SeqC26 or SeqC28, which appear to be shared by many insects, are potentially very interesting. It might have been relevant to combine this similarity-based clustering approach with phylogenetic reconstructions within each shared cluster.

We thank the reviewer for this insightful suggestion. We agree that the previous version did not sufficiently explore the evolutionary information contained in shared sequence communities such as SeqC26 and SeqC28. Following the suggestion, we extracted the sequences from these key shared communities and reconstructed within-community gene trees to search for orthologous groups.

Because shared SeqCs may contain conserved genes across insect orders, we used the gene trees of SeqC26 and SeqC28 to identify candidate cross-order orthologous groups. In SeqC26, we identified 56 orthologous groups, 11 of which included more than ten species. Among these 11 larger groups, 4 spanned multiple insect orders. One example, OGG54, includes sequences from Orthoptera, Blattodea, Trichoptera, and Hymenoptera; after excluding the possibility of GR contamination, this group includes locust LmigOR5 and LmigOR4. LmigOR5 has been reported to bind geranyl acetone and to be associated with avoidance behavior in the migratory locust (Chang et al. 2023). The functions of the corresponding receptors in other insect orders remain to be tested, and we now present these as candidate conserved modules rather than confirmed functional orthologs. In contrast, we did not identify clear cross-order orthologous groups in SeqC28. This suggests that the ORs in SeqC28 are more similar at the sequence-community level, but do not necessarily reflect traceable orthologous relationships.

The following revision has been added to the revised manuscript (lines 156-164): "SeqC26 and SeqC28 are large shared clusters that may contain cross-order orthologous relationships. In SeqC26, we identified 56 orthologous groups, four of which spanned multiple insect orders. OGG54 included ORs from Orthoptera, Blattodea, Trichoptera, and Hymenoptera. This group contains the locust receptors LmigOR5 and LmigOR4. LmigOR5 has been reported to bind geranyl acetone and to be associated with avoidance behavior in the migratory locust. The functions of the corresponding receptors in other insect orders remain to be tested, and we now present these as candidate conserved modules rather than confirmed functional orthologs. In contrast, we did not identify clear cross-order orthologous groups in SeqC28."

The clustering based on structure also leads to the identification of a majority of "orderspecific" clusters, but once again, the clusters shared by several orders are not truly exploited, which does not provide new insight into the evolution of ORs. However, the authors highlight a group of ORs in flies that appear to possess an unusual intracellular region. This is interesting, although it is a result more relevant to OR structure than to their evolution. The function of these ORs in Drosophila melanogaster, if it is known, is not discussed.

We thank the reviewer for this useful suggestion. Similar to the sequence communities, the structural communities also contain broadly shared groups. In the revised analysis, we focus in particular on StrC23, the only structural community present in 12 insect orders. StrC23 accounts for 46.3% of all structurally intact ORs in our dataset (4124 of 8905 ORs).

Given the extremely low sequence similarity among insect ORs, the existence of such a broad conserved structural community is important. We found that StrC23 has a significantly larger binding-pocket volume than other ORs and Orco (Fig. S3 d). From a structural perspective, this suggests that StrC23 receptors may have greater potential to accommodate diverse VOCs. We therefore interpret StrC23 not as sequence-level conservation, but as conservation at the level of OR structural evolution: many species appear to retain a large set of ORs with relatively large binding pockets, which may provide a structural basis for broad docking-derived VOC binding potential before lineage-specific OR diversification occurs.

Following the reviewer’s suggestion, we also examined available functional data for the long-IL3 receptors in StrC17. The long-IL3 orthologous group includes Drosophila melanogaster DmOr13a (droMel44 in our dataset) and Bactrocera dorsalis BdorOR13a (bacDor_18 in our dataset). Previous studies indicate that both receptors respond to 1-octen-3-ol (Kreher et al. 2008; Xu et al. 2023). Earlier work also identified a group of Dipteran fly OR homologs, including DmOr13a and BdorOR13a, as 1octen-3-ol-specific responsive receptors associated with oviposition behavior(Liu et al. 2023). When we modeled these homologous receptors, they also showed an extended IL3 region.

We therefore propose that the long-IL3 ORs identified here likely correspond to the previously reported fly homologs involved in 1-octen-3-ol responses. However, 1-octen-3-ol has different behavioral functions in different species: it acts as an attractant in blood-feeding mosquitoes and tsetse flies (Hall et al. 1984; Kline et al. 2007), contributes to plant-host localization in parasitoids(Morawo and Fadamiro 2016), functions as an aggregation cue in some beetles(Pierce et al. 1989), and is often associated with oviposition regulation in flies (Kreher et al. 2008; Liu et al. 2023). Mosquito receptors known to detect 1-octen-3-ol, such as AgOR8, AaOR8, TaOR8, and CquiOR118b(Lu et al. 2007; Xu et al. 2015; Dekel et al. 2016; Frunze et al. 2024), do not contain a long IL3 region (Fig. S3 e). This raises the possibility that the long IL3 structure in fly homologs may relate to the oviposition-associated role of 1-octen-3-ol in flies.

Because OR ligand binding is primarily mediated by the binding pocket, IL3 may not directly determine ligand binding. Instead, it may influence downstream regulation of olfactory responses. Mutational analysis of Orco suggests that IL3 can regulate channel activation and Orco-dependent olfactory responses, indicating that variation in intracellular loops may affect response efficiency or sensitivity(Turner et al. 2014; Bobkov et al. 2021). We now discuss this only as a plausible mechanism requiring future experimental validation.

The following revision has been added to the revised manuscript (lines 196-202): "StrC23 is the only structural community present in 12 insect orders and accounts for 46.3% of all structurally intact ORs in the dataset. We found that StrC23 has a much larger binding-pocket volume than other ORs and Orco. From a structural perspective, this suggests that StrC23 receptors may have greater potential to accommodate diverse VOCs. This reflects conservation at the level of OR structural evolution: many species appear to retain a relatively large set of ORs with larger binding pockets, which may provide a structural basis for structure-derived volatile organic compound binding potential before species-specific OR diversification."

The following revision has also been added (lines 408-419): "The long-IL3 OR orthologous group includes Drosophila melanogaster DmOr13a and Bactrocera dorsalis BdorOR13a, both of which respond to 1-octen-3-ol and belong to a group of dipteran fly OR homologs associated with oviposition behavior. Modeling of these homologous ORs showed that they also contain a long IL3 region. We therefore propose that long-IL3 ORs correspond to these fly homologs that recognize 1-octen-3-ol. However, 1-octen-3-ol has different behavioral functions in different species: it acts as an attractant in blood-feeding mosquitoes and tsetse flies, whereas in flies it is usually associated with oviposition regulation. Mosquito receptors known to detect 1-octen-3-ol, such as AgOR8, AaOR8, TaOR8, and CquiOR118b, do not contain a long IL3 region. This suggests that the long IL3 structure in fly homologs may be related to the oviposition-associated role of 1-octen-3-ol in flies, which requires future experimental validation."

The analysis of structural diversity then leads the authors to focus on the Orco co-receptors, which are characterized by modifications of the binding pocket and the emergence of an extracellular loop that could explain the loss of the ability to bind odorant molecules. This part, which relies on in vitro experiments, is interesting and constitutes the most striking result of the study, which could in itself have been the subject of a separate manuscript. However, the molecular dynamics modelling does not add anything in the way it is conducted (5 ns is too short).

We thank the reviewer for the positive assessment of our Orco results and for raising this important concern. We agree that 5 ns is insufficient to evaluate the conformational stability or long‑term dynamics of the receptor complex. However, our simulations were designed specifically to examine whether the EL2 β‑sheet could sterically affect the early movement of VOCs near the extracellular entrance of the binding pocket. Importantly, rather than relying on a small number of long trajectories, we performed 60 independent short simulations with the explicit goal of capturing a statistical trend in how VOCs respond to this local steric constraint. Within 5 ns, VOC trajectories were visibly altered by the EL2 β‑sheet. Across the 60 replicates, VOCs reached the binding pocket less frequently in Orco proteins containing this structure, revealing a consistent trend (Fig. 4C and Movie S1). We therefore consider this timescale sufficient to detect the early steric response of small molecules, and the large number of repeats provides statistical confidence in this conclusion.

The following revision has also been added (lines 688–691): “It should be noted that the molecular dynamics simulations were designed to examine the early trajectories of VOCs near the extracellular entrance of the binding pocket, rather than to assess the conformational stability of the receptor complex. A duration of 5 ns is sufficient for small molecules to respond to local steric constraints.”

The rest of the manuscript is based on the prediction of OR response spectra using molecular docking. The work that has been carried out is extremely substantial, and the objective of linking clusters based on sequence similarity or 3D structural similarity with functional categories is entirely relevant. Nevertheless, I see two major problems with this in silico functional analysis:

(1) The docking score threshold used was chosen thoughtfully, which is very good, and according to the calculation performed, should ensure a true positive rate of more than 20%, which is excellent in such a docking analysis. But in the absence of functional validation, this 20% true positive rate is not sufficient to extrapolate OR function as the authors do in the remainder of the manuscript. The risk of error remains too high to compare in such detail the function of ORs from insects with different lifestyles or diets.

We thank the reviewer for pointing out this important limitation of our computational functional interpretation. We fully agree that, although the threshold calibrated from available experimental data improves enrichment, a true-positive rate of approximately 20% is not sufficient to make deterministic claims about individual OR-VOC pairs. We therefore do not equate docking predictions with true OR response spectra. To prevent any misinterpretation, we have undertaken a major revision of our terminology and framing throughout the manuscript:

The main purpose of docking in our study is not to predict the real ligand of every OR or to compare the exact functional properties of individual receptors across lifestyles. Because the study includes 115 insect species, thousands of ORs, and a large VOC set, systematic experimental validation of all ORVOC combinations is currently not feasible. Instead, we use docking as a high-throughput and internally comparable theoretical screening method to estimate the potential binding tendencies of OR repertoires toward different functional-group VOCs.

Under this framework, we focus on repertoire-level trends generated by a uniform structural modeling, docking, and scoring pipeline. The value of this analysis is to provide candidate receptors, VOC categories, and ecological associations for future experimental testing. In the revised manuscript, when the results are based solely on docking analyses, we avoid directly referring to them as OR functions. Instead, we add the prefix “docking-derived” to prevent potential misunderstanding. We also explicitly state that confirming specific OR ligands and behavioral functions will require heterologous expression, electrophysiology, calcium imaging, genetic perturbation, or behavioral assays.

The following limitation statement has been added to the revised manuscript (lines 285-288): "Although this threshold enriches experimentally responsive OR-VOC pairs, the resulting hit rate is not sufficient to support deterministic inference for individual receptor-ligand pairs. Therefore, subsequent analyses were interpreted at the OR repertoire level rather than as experimentally validated OR response spectra.”; (lines 292-293) “It should be noted that dFunCs represent docking-derived functional communities rather than experimentally confirmed functional classes.”

(2) The six functional clusters identified are only slightly different from one another, with similar detection of all chemical families except acids and amines (which was expected, given that these families are a priori detected by IRs rather than ORs). This shows that even though the approach is relevant and deserves to be tested, it cannot be used to establish a link between groups/lineages of ORs and response spectra at the scale of insects as a whole. This is reflected in the final analysis by the fact that there is no visible link between sequence or structural clusters and functional clusters. Given the uncertainty surrounding the docking results, the entire subsequent analysis of the relationship between the Binding Breadth Index and ecological variables is highly questionable.

We thank the reviewer for this key comment. We agree that the differences among dFunCs for any single VOC functional group are often modest. However, the dFunC classification was not based only on whether an OR strongly recognizes one specific functional-group category. Rather, it was based on each OR’s overall docking-derived binding profile across a large VOC set.

In other words, the differences among dFunCs mainly reflect combinations of relative binding tendencies across multiple VOC categories, rather than a strong difference for one chemical family alone. Therefore, even if the predicted binding level for a single functional group differs only slightly among dFunCs, their multivariate profiles can still form distinct dFunCs.

We also agree that SeqCs, StrCs, and dFunCs do not show a simple one-to-one correspondence. This is expected because the three classifications are based on different information: sequence similarity, overall structural similarity, and docking-derived binding-profile similarity. At present, there is no direct evidence that insect OR sequence, structure, and potential binding profile maintain a conserved one-to-one relationship across the insect class. We also cannot exclude the possibility that false positives in docking reduce the resolution of the functional classification.

To avoid overinterpretation, we now refer to FunCs as docking-derived functional communities (dFunC) and define BBI as docking-derived BBI. This metric is intended for macro-scale relative comparisons, not for assigning true ligands or specific ecological functions to individual ORs. Ecological interpretations based on dFunC or BBI have been substantially toned down and are presented as candidate trends for future functional validation.

The following limitation statement has been added to the revised manuscript: "Ecological associations based on BBI should be interpreted as repertoire-level predicted binding trends, not as direct evidence for lifestyle-specific OR function or behavioral olfactory capacity."

This reframing ensures our conclusions are appropriately supported by the computational evidence and aligns the manuscript's claims with its methodological strengths.

Finally, the evolutionary analysis proposed to conclude that the work suffers from an incorrect interpretation: ORs of non-holometabolous insects cannot be considered equivalent to those of species that existed before the Permian-Triassic extinction. The fact that a locust or a cockroach has more narrowly tuned ORs than holometabolous insects does not mean that this was also the case for ancestral insects. To advance this type of conclusion, it would be necessary to conduct a phylogenetic analysis and reconstruct ancestral states, which is not the case here.

In summary, despite the large number of analyses performed, the authors do not succeed in achieving the stated objective of reconstructing the evolutionary history of insect ORs, and the results obtained do not sufficiently support the conclusions regarding the links between OR repertoires and environment or lifestyle.

We thank the reviewer for this important point. We agree that a rigorous test of changes in ancestral OR functional spectra before and after the EPME would require ancestral reconstruction within a phylogenetic framework based on reliable orthologous groups. For insect ORs, this is currently limited by rapid sequence divergence, very restricted cross-order and even within-order orthology, and frequent gene duplication, loss, and lineage-specific diversification.

Therefore, our current data cannot reliably reconstruct complete ancestral OR repertoires for different insect-order nodes, nor can they accurately trace extant ORs back to ancestral OR states before and after the EPME. The analysis is better understood as a comparison of OR repertoire composition among extant insect lineages, rather than as a comparison of ancestral states.

If major environmental transitions influenced the long-term evolution of insect ORs, such effects may appear either as traceable orthologous changes or as broad differences in extant repertoire composition and binding-profile composition. However, differences observed among extant lineages cannot be directly attributed to the EPME because they may also reflect later lineage-specific expansions, gene losses, and ecological adaptations.

Accordingly, we have revised the conclusion. We no longer state that ancestral insect OR functional spectra changed before and after the EPME. Instead, we state that comparisons among extant insect repertoires reveal differences in docking-derived binding-profile composition among lineages. These patterns may provide hypotheses for exploring links between major geological/ecological transitions and long-term olfactory evolution, but they require broader phylogenetic sampling, ancestral reconstruction, and functional validation.

The following revision has been added to the revised manuscript (lines 427-430): "We also detected differences in dFunC composition between extant lineages whose order-level origins fall before and after the EPME. Lineages originating after the EPME showed a higher proportion of broad-tuned ORs, likely influenced by multiple factors."

We also added the following clarification (lines 444-450): “However, because insect ORs evolve rapidly and cross-order orthologous relationships are limited, the current dataset does not allow reliable reconstruction of complete ancestral OR repertoires at deep insect nodes. The EPME-related comparison should therefore be interpreted as a descriptive comparison among extant lineages, not as a direct test of ancestral OR functional changes before and after the EPME. This hypothesis will require further testing through broader taxon sampling, reliable orthology assignment, ancestral-state reconstruction, and functional assays.”

This revision directly addresses the reviewer's concern by removing the unsupported causal claim and reframing the analysis within the appropriate, evidence-based scope of our study, thereby strengthening the manuscript's contribution as a source of robust comparative patterns and testable macroevolutionary hypotheses.

Reviewer #2 (Public review):

The remarkable evolvability of the olfactory system enables animals to rapidly adapt to dynamic and chemically complex environments. Over the past two decades, substantial effort has been devoted to uncovering the evolutionary principles that drive the diversification of odorant receptors (ORs), yielding key insights into the forces shaping their striking variability in both vertebrates and insects. In this manuscript, Zhang and colleagues analyze the OR repertoires of over 100 insect species, leveraging sequence and structural similarity to infer patterns of gene family evolution within this diverse and ecologically important clade. By integrating sequence-based and structure-based comparisons, their study builds on a compelling and recently emerging line of research made possible by the advent of AlphaFold, which has previously clarified the phylogenetic relationship between insect Ors and the gustatory receptor gene family and revealed the unexpectedly deep evolutionary origins of this ancient structural fold.

Applying this approach to a large set of ORs derived from species throughout the insect phylogeny, the authors confirm many previously reported patterns of OR evolution. Unfortunately, the way these results are presented lacks clarity in what is already known from previous work in the field versus what is a novel finding based on the analysis of this dataset.

We thank the reviewer for pointing this out. We agree that the original manuscript did not distinguish previously established findings from the novel results of the present study with sufficient clarity. Rapid OR evolution, lineage-specific expansion, and associations between OR repertoires and ecological traits have been reported in several insect groups. We have revised the manuscript to acknowledge these studies more explicitly and to indicate which results confirm known patterns, which extend them across a broader phylogenetic scale, and which arise from our new analyses.

Our study contributes more than a confirmation of previous observations. To our knowledge, it provides the first integrated analysis of OR evolution across 115 insect species that combines sequence similarity, structural conservation, and docking-derived binding profiles within a unified macroevolutionary framework. This analysis identifies class-wide repertoire patterns and previously unrecognized structural features, including the conserved structural cluster StrC23, that could not be evaluated using narrower taxonomic datasets.

The study also provides a mechanistic insight into Orco specialization. We experimentally demonstrate that the distinctive EL2 β-sheet of Orco reduces ligand binding affinity, supporting its contribution to the reduced odorant binding capacity of this conserved coreceptor. In addition, because exhaustive experimental characterization of the large insect OR family is currently impractical, our experimentally calibrated computational pipeline provides a framework for identifying repertoire-level trends, generating testable hypotheses, and prioritizing receptors for future functional studies.

The reviewer raises several specific examples of the distinction between previous knowledge and novel findings in the comments below. We address each of these points in the corresponding responses. We believe these revisions more clearly position our contribution as testing, extending, and integrating previously reported patterns while also identifying new class-wide structural and functional features.

It is unclear how complete the odorant receptor sets are. I recommend benchmarking the pipeline by comparing its output to a gold standard and a frequently vetted complete OR set, such as that of Robertson and Wanner 2006 or similar.

We thank the reviewer for this suggestion. We agree that without benchmarking, readers cannot adequately evaluate the completeness and accuracy of the OR annotation workflow. Following the suggestion, we performed detailed comparisons using three relatively well-curated repertoires: Drosophila melanogaster, Anopheles gambiae, and Bombyx mori. Because the species included in our study did not include Apis mellifera from Robertson and Wanner 2006, we did not use that dataset as the benchmark.

The benchmark addresses three questions. First, how well does the automated genome-based OR annotation workflow recover full-length ORs reported in previous studies? Second, does the structural filtering step improve the structural quality and reliability of the OR dataset? Third, for candidates in tandem-repeat regions where exon-merging errors may occur, do the candidate ORs share genomic positions with the most similar literature ORs, or are they more likely to represent distinct candidates? For Drosophila melanogaster, the accepted repertoire contains 60 OR genes and 65 protein products including splice variants. At the gene level, the automated workflow recovered most benchmark ORs before structural filtering, but Or47b and Or98b were not annotated. After structural filtering, fragmented or structurally incomplete ORs were removed, leaving 56 ORs that otherwise correspond one-to-one with the benchmark ORs. Two ORs that had been annotated before filtering, DmOr59a and DmOr85e, were removed by the structural filter because DmOr59a was fragmented and DmOr85e lacked TM7 in the predicted structural model (Author response image 3).

For Anopheles gambiae, the published repertoire includes 79 OR genes. Our automated workflow initially annotated 77 ORs, including 72 published ORs. After removing fragments and five full-length literature ORs, the final structurally filtered set contained 67 ORs. We inspected the five removed full-length ORs and found that AgOr52, AgOr47, and AgOr64 lacked a complete transmembrane region, whereas AgOr58 and AgOr6 contained all transmembrane regions but showed local deletions in some regions (Author response image 3). This indicates that structural filtering removes clearly incomplete structures, but can also be conservative enough to exclude a small number of true ORs with atypical transmembrane-region length variation.

For Bombyx mori, the published repertoire contains 66 OR genes including two pseudogenes. Our automated workflow initially annotated 103 ORs, including 60 previously reported ORs. After removing fragments and seven literature ORs, the final set contained 64 ORs, of which 53 corresponded to literature annotations. For structurally retained candidates that did not map one-to-one to literature ORs, we checked genomic locations (Table R2). Most were not located on the same chromosome or in the same tandem region as their most similar literature ORs, making exon-merging artifacts less likely. They may therefore represent previously unannotated OR candidates, although transcriptomic evidence or manual annotation will be needed for confirmation.

Overall, the benchmark shows that the automated workflow recovers most known OR repertoires and that structural filtering produces a high-confidence set suitable for structural comparison and docking. We now explicitly describe the final dataset as a set of structurally intact ORs rather than a complete OR repertoire for every species, and we discuss how strict filtering may underestimate OR counts. We added a benchmark description to the main text (lines 100-104) and included the detailed comparison strategy and results in Supplementary Text S1: "Benchmarking against existing OR repertoires from Drosophila melanogaster, Anopheles gambiae, and Bombyx mori showed that the annotation workflow provides high-confidence annotations suitable for downstream structural comparison and docking (see Supplementary Text S1, Table S1-3)."

Using their structural clustering approach, the authors identify a structural feature mostly unique to the OR co-receptor ORco, a beta-sheet in EL2, which they functionally show reduces odorant binding affinity - a key aspect of ORco, which does not bind ligands in the ancestral ligand-binding site. This is a particularly strong part of the manuscript, since the authors support their in silico-derived hypothesis with functional data.

We thank the reviewer for the positive assessment of the Orco-related work.

Lastly, in an attempt to assess the relationship between sequence identity and structure on one hand and function on the other, the authors perform an in silico structure prediction and chemical docking analysis. As it stands, this part is on the more speculative side since the docking approach has not been verified with available functional datasets.

We thank the reviewer for this comment and agree that the docking analysis is predictive. Molecular docking cannot be treated as experimental evidence for OR response spectra or as a definitive assignment of ligand specificity for individual OR-VOC pairs.

In the revised manuscript, we clarify that docking is used as a unified, comparable theoretical framework for macro-scale OR repertoire comparison. It is meant to estimate docking-derived binding potential across thousands of ORs and many VOCs, not to replace heterologous expression, electrophysiology, calcium imaging, or behavioral validation.

We also clarify that the docking framework was not used without any empirical calibration. We used existing OR-VOC experimental response data to examine the relationship between docking score and experimental hit rate, and we evaluated the ability of docking-derived labels to recover functional tendencies in available datasets. These analyses support the use of docking for repertoire-level enrichment and hypothesis generation, but they do not justify deterministic conclusions for individual receptor-ligand pairs. The revised manuscript now states this limitation explicitly.

The following revision has been added to the revised manuscript (lines 285-288): "Although this threshold enriches experimentally responsive OR-VOC pairs, the resulting hit rate is not sufficient to support deterministic inference for individual receptor-ligand pairs. Therefore, subsequent analyses were interpreted at the OR repertoire level rather than as experimentally validated OR response spectra."

Summary of Major Revisions:

In direct response to the eLife Assessment and reviewer comments, our revisions have systematically strengthened the evidence and clarified the interpretation:

(1) Added critical validations: Benchmarking of the annotation pipeline and sensitivity analyses for ecological correlations.

(2) Deepened evolutionary analysis: Phylogenetic exploration of shared clusters and structural characterization of StrC23.

(3) Clarified the scope and limitations of the computational functional analysis: Adopted “docking-derived” terminology and specified that the results represent hypothesis-generating, repertoire-level comparative trends rather than experimentally validated receptor functions.

(4) Corrected overinterpretations: Revised conclusions regarding the EPME and holometabolous/non-holometabolous comparisons to be descriptive and hypothesis-generating. '

(5) Improved scholarly accuracy: Updated framing to properly acknowledge prior work and corrected minor points throughout.

We believe that these revisions address the concerns underlying the assessment of incomplete evidence and provide stronger, more appropriately qualified support for the manuscript’s main conclusions.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

The annotation pipeline produces a number of genes that is slightly lower than reality, which is expected given the presence of a fairly stringent filter. However, the number is far too low for certain species, for example, Harmonia or Diabrotica. Whenever possible (for example, for Drosophila, Bombyx, Anopheles, etc.), adding a benchmarking step by comparing the results obtained with already identified repertoires would be a real improvement. Moreover, the dataset contains sequences of gustatory receptors that must be removed in order not to bias the analysis (CO₂ receptors in Drosophila melanogaster and Bombyx mori, for example, based on what I was able to verify). It might therefore be necessary to add a specific step in the pipeline to check for the presence of GRs in the data, since the sequences and 3D structures can be quite similar to those of ORs and therefore difficult to separate.

We thank the reviewer for this careful observation. To evaluate OR annotation counts, we compared our results with published OR annotations for previously annotated species. Except for Diabrotica virgifera virgifera, most species differed from published counts by fewer than ten ORs, supporting the general reliability of our annotation counts. For D. virgifera virgifera, the very low original count was likely caused by a combination of insufficient query representation and the difficulty of annotating ORs in a large genome. The previous D. virgifera virgifera annotation was published in 2025, whereas our original annotation was performed in 2022, so our query set likely did not adequately represent this lineage. When we re-annotated this species using new queries, 78 complete ORs were retained after structural filtering. We have updated the D. virgifera virgifera annotation in the revised dataset. Large-genome insects over 2 Gb are rare in our dataset, and the sensitivity analysis indicates that this issue does not substantially affect the main ecological associations based on OR count. Regarding gustatory receptor contamination, we found that a small number of sequences labeled as olfactory receptors but actually corresponding to GRs had been retained as queries because of an operational error during OR annotation. To remove potential GRs systematically, we collected 1,418 insect GR genes from the literature and databases and combined them with the OR library to create a receptor reference set. We then compared all annotated OR candidates against this reference set using BLASTP and removed candidates whose best match was a GR. In total, we identified and removed 76 potential GRs distributed across 38 species. These included the Drosophila CO² receptor genes Gr63a and Gr21a, as well as Bombyx mori BmGr10 and BmGr9, which the reviewer specifically noted. All downstream analyses have been repeated using the corrected dataset. We are grateful for this careful observation, which substantially improved the rigor of the dataset.

Both Reviewer 1 and Reviewer 2 requested benchmarking against known OR repertoires, and we agree that this is necessary. We therefore benchmarked the workflow against Drosophila, Anopheles, and Bombyx repertoires, evaluating recovery of known ORs, the effect of structural filtering, and potential exon-merging artifacts in tandem regions. We further added sensitivity analyses using the larger of structurally intact OR counts and published OR counts for species with available literature data. These analyses are now included in the revised Results and supplementary materials.

The revised text is as follows (lines 100-104): "Benchmarking against existing OR repertoires from Drosophila melanogaster, Anopheles gambiae, and Bombyx mori showed that the annotation workflow provides high-confidence annotations suitable for downstream structural comparison and docking (see Supplementary Text S1, Table S1-3)."

For the SSN approach, the validation by comparison with the phylogeny of hymenopterans is a good idea, but the result is not as convincing as presented in the manuscript since the cluster SeqC25 does not correspond to a monophyletic group. This, therefore, raises the question of the evolutionary information that can be drawn from this clustering approach. The validation with lepidopteran PRs is also a good idea, but it would be necessary to verify exactly to which PR clades the sequences used for the analysis belong, in order to determine which clade corresponds to each of the identified SeqC clusters.

We thank the reviewer for this important point. For the Hymenoptera comparison, we agree that SeqC25 does not correspond to a strict monophyletic clade in existing Hymenoptera OR phylogenies. We have clarified that the SSN approach is not intended to replace traditional phylogenetic trees or to reconstruct ancestor-descendant relationships among OR communities.

SeqC25 should be interpreted as a sequence-similarity community that is strongly expanded in Hymenoptera and partially corresponds to previously defined Hymenoptera OR subfamilies, but not as a strict phylogenetic clade. SeqCs describe modular boundaries in OR sequence-similarity space, whereas phylogenetic trees infer branching relationships under a specific evolutionary model. For insect ORs, where sequence similarity is often very low, duplication and loss are frequent, and tree support is often limited, the two approaches need not produce identical groupings.

We now state that the evolutionary information provided by SSNs mainly concerns sequence-similarity diversification patterns: highly expanded modules, order-specific or cross-order communities, community boundaries, and differences in repertoire composition among lineages. These results complement phylogenetic analysis, but detailed relationships within close taxa or OR subfamilies still require traditional phylogenetic methods.

For the Lepidoptera pheromone receptor (PR) validation, we agree that the previous analysis mixed experimentally validated PRs and candidate PRs, making it unclear which PR branch each SeqC corresponded to. We reorganized the PR dataset by separating classic PRs from recently reported new PR branches and mapped these sequences to SeqCs using BLASTP. Classic PRs were assigned almost entirely to SeqC6 (93.9%), whereas the new PR branch mapped to SeqC35 (Fig. S2f). Both SeqC6 and SeqC35 are shared by Trichoptera and Lepidoptera, suggesting that they may correspond to distinct PR-related sequence communities.

The following revision has been added to the revised manuscript (lines 144-150): "This comparison indicates that SSN-based communities are broadly consistent with established phylogenetic subfamilies, while capturing sequence-similarity modules rather than strictly monophyletic clades. We further evaluated the SSN using curated lepidopteran pheromone receptors (PRs). To distinguish different PR lineages, experimentally validated classic PRs and the recently reported new PR lineage were analyzed separately. Classic PRs were predominantly assigned to SeqC6 (93.9%), whereas the new PR lineage was assigned to SeqC35 (fig. S2F and Data S5), supporting the hypothesis of multiple independent origins of PRs.”

(3) In Figure 3, the cluster StrC17 appears to be present in both Lepidoptera and Diptera, whereas when looking at Figure S3, it seems to be present only in Diptera. There may be an error somewhere.

We thank the reviewer for identifying this inconsistency. After rechecking the data, we confirmed that StrC17 contains 47 ORs, of which 46 are from Diptera and one is from Lepidoptera. We have corrected Figure S3. This correction does not affect the main conclusions regarding the long-IL3 ORs, because all downstream analyses were conducted using the correct StrC17 composition.

(4) The phylogeny presented in Figure S4 does not help clarify the phylogenetic context of the Orco study. The main reason is that it is not correctly rooted: the ORs of M. rhabei should have been used as an outgroup, or the two GRs of D. melanogaster that are present (perhaps by mistake) in this analysis and that clearly diverge strongly from the rest of the sequences. In any case, this phylogenetic analysis does not seem to add much compared with what was presented in the article by Thoma et al. in 2018.

We thank the reviewer for pointing this out. The original purpose of the tree was to show that early-diverging insect ORs lack clear orthologous relationships, making it difficult to reconstruct ancestral OR states before and after Orco emergence. However, because the figure contributes little to the core conclusions and could introduce confusion, we have removed it from the revised manuscript. The revised text instead directly cites previous phylogenetic studies for the early OR/Orco evolutionary background.

(5) In Figure 5A, many VOCs appear unclassified when using molecular descriptors. Have you tried classifying VOCs using molecular fingerprints instead of selected molecular descriptors? And can you explain how those 32 descriptors were chosen? Furthermore, it is surprising not to see a "terpenoids" category given the importance of these molecules in insect chemical ecology. One would expect them to cluster together in a chemical space based on molecular descriptors; where are they in Figure 5A?

We thank the reviewer for this question. We first clarify that unclassified in Figure 5A does not mean that these VOCs were excluded from the analysis. It only means that they were not assigned to one of the nine predefined major functional-group categories. All VOCs were included in molecular descriptor calculation, chemical-space visualization, and downstream docking.

The 32 molecular descriptors were not arbitrarily selected by us. They come from the optimized odorant metric proposed by Haddad et al. 2008. That study began with 1,664 Dragon molecular descriptors, represented odorants as multidimensional physicochemical vectors, and selected 32 descriptors that best explained similarity in odor-induced neural responses across multiple published datasets. Haddad et al. further showed that this optimized descriptor set performed well across different animals, recording methods, and levels of olfactory-system organization. We therefore adopted these 32 descriptors to represent odor physicochemical space.

We agree that molecular fingerprints are another useful representation, especially for capturing substructures and scaffold information. We chose molecular descriptors rather than fingerprints because the purpose of Figure 5A is to visualize VOC distribution in a continuous physicochemical odor space, and because the Haddad descriptor set was optimized using neural response data. We have added this rationale to the revised manuscript.

We also agree that terpenoids are important in insect chemical ecology. In the original classification, VOC categories were based mainly on functional groups, whereas terpenoids are defined by biosynthetic origin and carbon skeleton rather than by a single functional group. Therefore, terpenoid VOCs are distributed across several functional-group categories, such as terpene alcohols, terpene aldehydes, and terpene ketones, and are not expected to form one independent cluster in a two-dimensional odour space.

Following the reviewer’s suggestion, we added an extra annotation for common terpenoid-related compounds and mapped them onto the odor space shown in Figure 5A. The result shows that terpenoids are distributed across multiple regions and functional-group categories rather than forming a single compact cluster (Fig. S5 c). We therefore retain the major functional-group classification for the main analyses and add terpenoids as an additional annotation, while explaining why they are not treated as a main category parallel to alcohols, aldehydes, ketones, and esters.

In addition, because the downstream analyses in this study rely on large-scale docking predictions, and because docking itself contains inherent uncertainty, we chose to use relatively basic and chemically explicit major functional-group categories. This classification reduces interpretive instability that could arise from excessive subdivision of VOC classes. By contrast, treating terpenoids as an independent primary category would group together molecules with substantially different functional-group properties, thereby increasing the complexity of interpreting docking-derived binding profiles. Therefore, in the revised manuscript, we retain the major functional groups as the core VOC classification scheme and add terpenoids as an additional annotation.

The specific revision in the revised manuscript is as follows (lines 268-271): "Although terpenoids are important in insect chemical ecology, they are distributed across multiple regions of odor space. To reduce the docking instability that could result from excessive VOC subdivision, we did not treat terpenoids as a category parallel to the nine basic functional-group categories."

(6) The methodology used to calculate hit rates with respect to functional data on known ORs is not sufficiently explained. The result could be shown for the different species (drosophila, mosquito, butterfly). In addition, ORs from the locust should also have been used for this comparison.

We thank the reviewer for this suggestion. In the revised Methods, we now describe the workflow in detail. We first compiled receptor-odorant combinations with clear response relationships from published functional experiments. For each OR-VOC pair, we performed docking using the same pipeline as in the global analysis and extracted the corresponding docking score. We then labeled each pair as response or non-response according to the original experimental data. Next, we grouped pairs into docking-score intervals and calculated, within each interval, the proportion of experimentally positive pairs, which we define as the hit rate. To reduce fluctuation from random sampling, each score interval was sampled three times. Finally, following the hit-rate modeling strategy of Lyu et al. 2019, we fitted the relationship between docking score and hit rate with a Bayesian curve-fitting approach and used the posterior expectation of the parameters to draw the hit-rate curve. The threshold was determined from the docking score at which the hit rate reached a plateau.

Following the reviewer’s request, we calculated hit-rate curves separately for Drosophila melanogaster, Anopheles gambiae, Lepidoptera, and Locusta migratoria OR data (Fig. S5). Drosophila, Anopheles, and Lepidoptera showed similar trends: the hit-rate plateau was approximately 20%, and the corresponding docking-score threshold was around -8 kcal/mol. This is consistent with the original combined analysis and indicates that the threshold was not driven by a single species dataset. The locust dataset behaved differently. Its experimental positive rate was low, approximately 5.2%, and most reported locust ORs are narrowly tuned. In such a highly imbalanced dataset, docking score was less able to enrich experimental positives to the same level as in the other datasets; most score intervals had hit rates below 10%. When the locust data were combined with Drosophila, Anopheles, and Lepidoptera, the overall plateau decreased to about 18%, and the corresponding threshold shifted to about -11 kcal/mol.

We consider this difference informative because it shows that OR tuning properties and experimental positive rates can affect the relationship between docking score and hit rate. The locust result suggests that stricter thresholds may be required for systems dominated by narrowly tuned ORs and low positive rates. However, our goal is to establish a unified empirical threshold for large-scale repertoire comparison, not to optimize a separate threshold for every lineage. We therefore retain -8 kcal/mol from the Drosophila, Anopheles, and Lepidoptera datasets as the main threshold, present the locust analysis as a sensitivity test, and note that this threshold may have lower positive-enrichment ability in narrowly tuned OR systems.

The specific revision in the revised manuscript is as follows (lines 281-285): "In addition, our singletaxon hit-rate sensitivity analysis showed similar trends for Drosophila, Anopheles, and moths. Because the locust OR repertoire had an extremely low positive rate, we did not include the locust functional data in the final threshold assessment (see Supplementary Text S2 for details)."

Methods section (lines 748-755): "We first compiled published functional datasets containing explicit OR-VOC response relationships. These datasets included ORs from Drosophila melanogaster, Anopheles gambiae, Helicoverpa armigera, Spodoptera littoralis, and Locusta migratoria. For each dataset, OR-VOC combinations were assigned response or non-response labels according to the original experimental results. Each OR-VOC pair was then docked using the same structural modeling, docking, and scoring workflow used in the global OR-VOC analysis. Pairs were grouped into docking-score intervals, and the hit rate for each interval was defined as the proportion of experimentally positive pairs among all pairs in that interval. Each score interval was sampled three times."

(7) It is not clear how the ROC curves were constructed; they appear to be drawn with very few points. The methodology needs to be explained in more detail here, since this is an important point.

We thank the reviewer for this suggestion. We have added a detailed description of ROC construction. The ROC analysis was designed to evaluate whether the dFunC labels are directionally consistent with available experimental functional data.

Specifically, a positive dFunC label for a functional-group VOC indicates that ORs in that docking-derived community are predicted, relative to the all-insect OR background, to bind more VOCs from that functional group. A negative label indicates a lower predicted binding tendency. We mapped Drosophila ORs to the dFunC classification and assigned each OR a positive or negative label for each functional group according to its dFunC. We then used the Drosophila experimental functional matrix to count measured responses of each OR to different functional-group VOCs. For each functional group, we plotted ROC curves and calculated AUC values by comparing the dFunC-predicted labels with experimental response counts. We agree that the curves are based on limited data points for some functional groups. This is because comprehensive OR-VOC functional matrices are still sparse, and many insect ORs have been deorphanized using limited odor panels that do not cover many functional groups.

The specific revision in the revised manuscript is as follows (lines 794-798): "Drosophila melanogaster ORs were mapped to dFunCs to evaluate the consistency between dFunC labels and experimental response data. For each VOC functional group, ORs were assigned positive or negative labels according to their dFunC labels, and these labels were compared with response counts from the Drosophila functional matrix to generate ROC curves and AUC values."

(8) What the BBI represents is not very clear, even though the calculation method is presented in detail. This should be clarified in the main text and/or in the figure legends.

We thank the reviewer for this suggestion. We now define BBI more explicitly in the main text. BBI, or binding breadth index, is a relative repertoire-level metric calculated from docking-derived dFunC labels. A dFunC label indicates whether ORs in a docking-derived binding-profile community show higher or lower predicted binding tendency toward a functional-group VOC category relative to the all-insect OR background. When we calculate BBI for a set of ORs, we summarize the predicted binding tendencies of the dFunCs to which those ORs belong, thereby estimating the relative potential binding breadth of that OR repertoire for the VOC category.

Thus, BBI should be interpreted only as a relative docking-derived metric. When comparing two OR groups for the same VOC functional group, a higher BBI means that the group is predicted, under our docking-derived framework, to have broader potential binding breadth. It does not directly represent experimentally confirmed olfactory breadth, odor perception, or behavioral function. We have revised the terminology accordingly to docking-derived potential binding breadth.

The following revision has been added to the revised manuscript (lines 319-321): "BBI is a docking-based, repertoire-level relative metric used to summarize the potential binding breadth of a set of ORs toward a given functional-group VOC category."

(9) Insects with saprophagous larvae in your dataset are almost exclusively flies. Therefore, the conclusions concerning the saprophagous diet could just as well result from phylogenetic constraints rather than from an adaptation to the diet.

We thank the reviewer for pointing this out. We agree that, because the currently available saprophagous larval species are mainly concentrated in Diptera, this result may be influenced by lineage effects. We have therefore toned down the interpretation and present it as a candidate association. Future inclusion of more saprophagous species from non-dipteran lineages will allow this relationship to be tested more rigorously.

The following revision has been added to the revised manuscript (lines 493-496): "This interpretation should be treated with caution, as the saprophagous larval species currently available in our dataset are mainly concentrated in Diptera and may therefore reflect lineage effects. Future inclusion of saprophagous species from broader insect lineages will help test this association more rigorously."

Reviewer #2 (Recommendations for the authors):

Main feedback

(1) This study presents an interesting approach to unify the exceptionally large OR multigene family in a single macroevolutionary framework. While this is a non-trivial task, given the low sequence similarity of ORs, the inferences broadly overlap with previous findings, suggesting that the approach works. Unfortunately, the line between previously identified patterns of OR biology and evolution and new insights derived from this study is somewhat blurry throughout the text at the moment and needs to be sharpened. As it reads, the manuscript is prone to overselling the results.

A few examples:

(a) Lines 37-41 - contrary to the claim in this sentence, previous work has described the principles of OR function, structure, and evolution to an extent that would warrant concluding that the fundamental logic of insect olfaction is actually comparatively well understood.

We thank the reviewer for this comment. We agree that previous studies have already established important principles of insect OR biology, including OR-Orco complex structure, ion-channel function, ligand recognition, and rapid OR evolution. Our intention was not to imply that the basic logic of insect olfaction remains unknown, but to emphasize that an insect-class-scale framework integrating OR sequence divergence, structural variation, and docking-derived binding-profile diversity has been lacking. We have revised the Introduction to clarify this point and to better acknowledge the contribution of previous studies.

We therefore revised this sentence in the revised manuscript as follows (lines 38-42): "However, a classwide integrative framework linking OR sequence divergence, structural variation, and docking-derived binding-profile diversity is still lacking. Such a framework is needed to distinguish broadly conserved patterns from lineage-specific features and to place species-level OR diversification in a broader macroevolutionary context."

(b) Lines 47-48 - work in several systems has emphasized the "evolutionary interplay between receptor family diversification and macroecological adaptation", both in species with highly specialized ecologies (plant host specialists, slave making ants -> these are even cited in the manuscript) and in a broader macroevolutionary context (e.g. bee diet breadth, Singh et al. 2025), which does not really justify the claim that OR evolution with respect to ecology is a profound enigma. The present study indeed presents the first one combining data throughout the entire insect tree of life, which is an impressive feat. The fact that inferences drawn from smaller datasets with smaller phylogenetic breadth are supported is a useful finding. I recommend emphasizing this.

We agree that the relationship between OR repertoire evolution and ecological adaptation has been investigated in multiple insect systems and should not be described as an unresolved “profound enigma.” The novelty of our study is to test, integrate, and extend these previously reported patterns across 115 insect species under a unified sequence-structure-docking profile framework. We have revised the text to emphasize this class-wide extension rather than overstating the unknowns in the field. The following revision has been added to the revised manuscript (lines 46-47): "Although several studies have linked OR repertoire evolution to ecological adaptation in specific insect lineages, a unified class-wide comparison remains lacking."

(c) Lines 347 - 348 - what is meant by the statement that the manuscript 'revealed the mechanism of insect olfactory perception underlying macroenvironmental influences' is unclear to me. What mechanism is referred to here? What do the authors mean by 'macroenvironmental influences'? It reads as if the authors claim to have shown how olfactory perception works in insects, which is not an accurate statement.

We thank the reviewer for pointing out that this statement was unclear. We did not intend to claim that our study reveals the mechanism of insect olfactory perception itself. Rather, our results identify class-scale associations between OR repertoire variation, ecological traits, and docking-derived binding profile composition. We have revised the Discussion.

In the revised manuscript, this sentence has been changed to the following more accurate wording (lines 399-402): "Our results reveal class-level associations between OR repertoire variation and ecological traits and identify differences in docking-derived binding-profile composition among extant insect lineages that may be relevant to long-term macroevolutionary transitions."

(d) Lines 348-350: I do not agree with the statement that the authors "describe the evolutionary process through which Orco emerged". The data that the beta-sheet in EL2 impedes ligand binding in non-orco ORs is compelling and suggests an intriguing contribution to why Orco lack ligand-binding function, but is not sufficient to comprehensively describe the evolutionary process of Orco evolution from ancestral ORs. This would require more sequences of early-branching insect species. Accordingly, this statement should be toned down.

We agree that the original wording overstated the extent to which our data describe Orco emergence. Our results identify structural features, including the EL2 beta-sheet and specialized binding-pocket properties, that may have contributed to the loss of ligand-binding function during Orco specialization. However, a complete reconstruction of Orco origin will require broader sampling of early-diverging insect lineages and more reliable ancestral-state reconstruction. We have revised the relevant statements accordingly.

We therefore toned down this statement in the revised manuscript and described it more accurately as follows (lines 402-405): "We further examined Orco specialization within the insect 'conserved chassisdiversified sensors' model and identified structural features, including the EL2 beta-sheet and specialized binding-pocket properties, that may have contributed to the loss of ligand-binding function during Orco evolution." We also revised another sentence as follows (lines 522-525): "Overall, we systematically explored the relationships among sequence, structure, and docking-derived function in insect ORs, identified structural features potentially associated with Orco specialization, and provided a framework for testing how OR repertoire evolution may relate to macroenvironmental and lifestyle variation."

It is understandable that the authors want to emphasize the value of their work, but this can't be achieved by inaccurately portraying the state of the field. The framing should be adjusted throughout the manuscript.

We thank the reviewer for this important suggestion. We agree that the manuscript should emphasize its contribution without overstating gaps in the field. We have therefore revised the framing throughout the manuscript to better acknowledge previous advances in insect OR structure, function, evolution, and ecological adaptation. The revised text presents our main contribution as an insect-classscale integration and extension of these findings using a unified sequence, structure, and docking-derived binding-profile framework.

(2) It is unclear how complete the odorant receptor sets are, because the annotation pipeline used is not benchmarked. The high similarity of ORs often located in clusters of up to 50 genes on the genome represents a major obstacle for purely automated annotation, leading to artifacts such as the erroneous joining of exons across genes. The lack of manual verification in this study, combined with stringent filtering of annotations that do not meet structural criteria, does likely lead to a set of sequences that present a correct fold, but at the same time, the approach may discard misannotated genes and cannot distinguish between correct gene models and artifacts where exons of more than one gene could have been merged. Please benchmark the pipeline by comparing its output to a gold standard and frequently vetted complete OR set, such as that of Robertson and Wanner 2006 or similar. That would allow assessing the presented work better and reveal how complete the OR counts are. This is important since the authors use OR counts to derive conclusions about the evolutionary dynamics of the gene family.

We thank the reviewer for raising this concern. Because this point overlaps with the benchmark issue raised in the public review, we provide the detailed benchmark results in our response above and in the revised supplementary text. Briefly, we compared the annotation and structural-filtering pipeline with curated OR repertoires from Drosophila melanogaster, Anopheles gambiae, and Bombyx mori. The results show that the workflow recovers most known ORs, while the structural filter removes fragmented or structurally incomplete models. We also added sensitivity analyses using published OR counts where available, which supported the main OR-count ecological associations.

(3) The in silico functional docking analysis is intriguing and bears the potential to produce testable hypotheses on the functional evolution of ORs. Docking can produce a wide range of results, including artifacts, and thus should be benchmarked with existing datasets derived from functional experiments (there are several large datasets available). While the authors have used previous functional data to constrain their model, they should benchmark it by running it on a set of ORs with known ligand-binding profiles to convincingly show that their predictions are an accurate assessment of the actual functional properties of ORs.

We thank the reviewer for this suggestion. We agree that docking-derived results should be evaluated using ORs with known ligand-response data. In this study, however, the key prediction used for downstream analyses is the dFunC functional label derived from each OR’s overall docking-derived binding profile, rather than the exact ligand assignment of each OR-VOC pair. Therefore, we evaluated the dFunC labels using the Drosophila melanogaster experimental OR response matrix.

Drosophila ORs were mapped to dFunC communities, and for each VOC functional group, ORs were assigned positive or negative labels according to the predicted binding tendency of their dFunC. These labels were compared with experimentally measured response counts from the Drosophila OR functional matrix. ROC curves and AUC values were calculated for each VOC functional group. Most functional-group labels achieved AUC values above 0.75, indicating that the dFunC classification is broadly consistent with the Drosophila experimental response matrix.

We did not use individual OR-VOC pair accuracy as the primary validation criterion because the hit rate analysis showed that the experimental positive rate in the enriched docking-score range is approximately 23%. This indicates that docking is more appropriate for evaluating dFunC-level binding profile tendencies than for exact ligand assignment. We have revised the manuscript to clarify this validation logic. The functional labels of dFunCs should be further tested with experimental response data from more insect species and can be updated as prediction methods improve.

The following details have been added to the revised manuscript (lines 794-798): " Drosophila melanogaster ORs were mapped to dFunCs to evaluate the consistency between dFunC labels and experimental response data. For each VOC functional group, ORs were assigned positive or negative labels according to their dFunC labels, and these labels were compared with response counts from the Drosophila functional matrix to generate ROC curves and AUC values."

(4) The authors conclude that holometabolous and non-holometabolous insects exhibit distinct OR differentiation patterns, stating that sequence diversity "increases progressively with increasing degree of insect order divergence, suggesting a gradual accumulation of sequence variation during insect evolution" (Lines 332-333). If the pattern is indeed gradual, is the phylogenetic splitting of holometabolous vs. non-holometabolous insects not arbitrary? Couldn't it be split at any point along the phylogeny and lead to a similar difference?

We thank the reviewer for pointing out this important logic issue. We agree that, if OR sequence variation accumulates gradually with phylogenetic distance, the difference between holometabolous and non-holometabolous insects should not be interpreted as two completely separate OR differentiation modes.

Our result is better understood as a comparison of extant deep insect lineages. In this comparison, holometabolous insects show higher OR sequence-community diversity than non-holometabolous insects. This pattern is still informative because it suggests that, during long-term evolution, OR repertoires in holometabolous lineages have accumulated broader sequence-community diversity. This may be related to lineage-specific OR duplication and loss, ecological diversification, life-history differences, and differences in repertoire size.

We have revised the manuscript accordingly. We no longer describe the result as two distinct OR differentiation patterns. Instead, we describe it as a difference in OR sequence-community diversity between extant holometabolous and non-holometabolous lineages.

The following revision has been added to the revised manuscript (lines 370-371): "Additionally, extant holometabolous insects showed higher OR sequence-community diversity than non-holometabolous

insects."

(5) Further, if I understand correctly, the assessment of the differentiation patterns seems to rest on total counts of ORs in 'SeqCs', quantified in Figure S9. These counts, however, are not normalized and thus seem influenced by the total number of sequenced genomes belonging to a specific clade in the dataset. These have a heavy bias towards holometabolous species. If this is true, the analyses should be repeated based on normalized counts. OR is this assessment based on the number of 'SeqCs' across the phylogeny? If yes, does this correlate with OR numbers? Please clarify.

We thank the reviewer for pointing this out. We agree that unnormalized SeqC counts can be affected by unequal sampling, because holometabolous insects include more sampled species and more ORs in our dataset.

To address this issue, we added normalized analyses. First, we calculated SeqC diversity at the species level (Fig. S9a). Second, because OR number was significantly correlated with SeqC count (pGLS p-value = 0.001035), we performed equal-OR random sampling (Fig. S9 b). Specifically, we randomly sampled the same number of ORs from holometabolous and non-holometabolous insects, 100 ORs per group, calculated how many SeqCs were covered by the sampled ORs, and repeated this procedure 100 times to obtain the distribution of SeqC diversity under equal OR sampling depth. These analyses showed that holometabolous insects still exhibited higher OR sequence-community diversity after controlling for OR sampling depth. We have revised the manuscript to clarify the calculation and now interpret the result as a normalized difference in sequence-community diversity, rather than as an uncorrected difference in total SeqC number.

We added the following normalized analysis to the revised manuscript (lines 371-379): "Because total SeqC counts may be affected by unequal species sampling and OR numbers, we added normalized analyses. First, we calculated SeqC numbers at the species level and found that holometabolous insects contained more SeqCs on average. Second, because OR number was significantly correlated with SeqC count (P=0.001035), we performed equal-OR random sampling between holometabolous and nonholometabolous insects. Under the same OR sampling depth, holometabolous insects still covered more SeqCs, indicating higher OR sequence-community diversity after controlling for OR number. Therefore, this result supports a genuine difference in SeqC diversity rather than a difference driven by unequal species sampling or OR number. "

(6) Phylogenetic differences in OR evolutionary patterns have been described previously for Paleoptera compared to Neoptera. Are the authors simply picking up on these patterns? Is the holometabolous vs non-holometabolous hypothesis supported when only Neoptera are taken into account?

We thank the reviewer for raising this point. Differences in OR evolutionary patterns between Paleoptera and Neoptera have already been described in previous studies, so it was necessary to test whether our pattern simply reflected this older phylogenetic division.

Following the suggestion, we restricted the comparison to Neoptera and compared Holometabola with non-holometabolous Neoptera. After excluding Paleoptera and earlier-diverging lineages, Holometabola still showed higher OR sequence-community diversity (Fig. S10). Additional normalized analyses indicate that this trend is not explained only by unequal species sampling or OR counts (Fig S9a-b).

Therefore, our result is not simply a rediscovery of the known Paleoptera versus Neoptera difference. Instead, within Neoptera we observe an additional difference in SeqC diversity between Holometabola and non-holometabolous Neoptera.

We added the following comparison of neopteran insects to the revised manuscript (lines 379-382): "The higher SeqC diversity observed in Holometabola was retained when the comparison was restricted to Neoptera, indicating that this pattern is not simply a restatement of previously reported Paleoptera-Neoptera differences."

Minor comments

(1) Line 10: "all the insect orders were found to contain fully functional OR repertoires". It is unclear what the authors mean by this. What is a fully functional OR repertoire? Is it the number of receptors? The number of functional receptors? Given this lack of clarity and the fact that the manuscript does not present functional data on Or binding profiles, this part should be altered.

We have revised this sentence. The intended meaning was that all sampled insect orders include ORs assigned to all six docking-derived functional communities.

(2) Lines 17-18: I am not convinced the presented data explain the adaptive relationship between insect olfactory potential and diverse ecological environments. It suggests that few aspects of insect ecology (larval diet and terrestriality) correlate with receptor number and predicted ligand binding breadth. This should be toned down.

We have toned down the statement. The sentence has been revised as follows: " Our findings provide a class-level framework for investigating insect OR evolution and generate testable hypotheses about how receptor repertoire diversification may relate to ecological adaptation."

(3) Line 44: the low amino acid sequence identity among ORs has been described before. Please add appropriate references.

We have added the requested references.

(4) Lines 52 - 55: The 1:3 stoichiometry is not yet confirmed in vivo - indeed, other work suggests a 2:2 stoichiometry as a possibility. Please adjust this sentence to reflect this uncertainty.

We have revised the sentence to reflect this uncertainty (lines 51-56): Unlike vertebrate G protein-coupled ORs, insect ORs function as ligand-gated ion channels by assembling with the conserved co-receptor Orco, with recent cryo-EM structures supporting a 1:3 OR–Orco heterotetrameric model. However, the exact in vivo stoichiometry remains to be fully resolved, and alternative arrangements such as 2:2 may also be possible.

(5) Lines 94-95: Please cite studies that previously annotated ORs.

We have added the relevant citations.

(6) Lines 99-100: It is well established that hymenopterans have the largest sets of ORs among insects. Please add appropriate references.

We have added the relevant citations.

(7) Lines 101-102: It has been shown before that Odonata have very few ORs. However, there are other insects with even smaller sets of ORs. Please add appropriate references.

We have clarified that the statement refers only to our intact OR dataset. The sentence has been revised as follows: "In our OR dataset, Odonata had the fewest intact ORs, with a mean of 4 (N = 3)."

(8) Lines 109-111: This sentence appears to be illogical. Please revise.

We re-evaluated analyses that depend on OR counts. Because the association between genome size and OR count was sensitive to annotation completeness, structural filtering, and a few extreme species, we removed this result and the related discussion from the revised manuscript.

(9) Line 113: What is ecological behavior? Do you mean ecology?

We have replaced ecological behavior with ecological traits.

(10) Lines 148 - 150: This has been described previously. Please add appropriate references.

We have added the appropriate citation.

(11) Lines 161-162: This has been previously described. Please add appropriate references.

We have added the appropriate citation.

(12) Lines 199-201: Could it be that a lack of orthologs is a result of sampling bias? Only very few genomes of early-branching lineages have been sequenced.

We thank the reviewer for this helpful comment. We agree that the apparent lack of direct orthologs among early-diverging insect ORs may be influenced by limited genome sampling from these lineages. We have revised the manuscript to clarify that, with the currently available data, we cannot reliably determine direct orthologous relationships among early-diverging insect ORs or reconstruct ancestral OR states. We now state that broader sampling of high-quality genomes and transcriptomes from early-diverging insect lineages will be needed to test this question more rigorously.

The revised text is as follows (lines 218-220): "With the currently available early-diverging insect genomes, we could not reliably identify direct orthologous relationships among early ORs, which limits the reconstruction of ancestral OR states."

(13) Line 204-205: Do you mean TdomOR1-8 may be ancestral to Orco? This is not possible, given that Orco and TdomOR1-8 co-exist in one genome. But previous phylogenetic analyses of multiple silverfish genomes suggest that Orco is closely related to TdomOR1-8 and similar receptors in other species. Please revise.

We thank the reviewer for pointing this out. We did not mean that TdomOR1-8 are the ancestors of TdomOrco. Because currently available OR data from early-diverging insects are limited, direct orthologous relationships cannot be reliably identified and the ancestral OR state of Zygentoma cannot be reconstructed with confidence. Therefore, we can only search among extant zygentoman ORs for comparators that are closely related to the Orco clade.

The current gene tree shows that TdomOR1-8 are closely related to TdomOrco. Thus, they can serve as structural references for comparing Orco-related receptors and for understanding structural changes that may have been involved in Orco specialization.

In the revised manuscript, we have changed the relevant sentence to (lines 223-225): “Therefore, we speculate that Orco recruitment in the ancestral zygentoman lineage may have been accompanied by expansion of a set of candidate ORs, and that TdomOR1-8 represent extant retained members of this receptor set.”

(14) Lines 387-388, this was already known. Please cite accordingly

We have added the appropriate citation.

(15) Line 389: Please include a reference for the finding that early ORs have a homologous tetrameric configuration.

We have added the appropriate citation.

(16) Lines 403-407: This hypothesis has been previously put forward by del Marmol and colleagues. Add reference.

We have added the appropriate citation.

Author response table 1.

Number of OR genes in previously annotated species.

Author response table 2.

ORs in this study that did not show one-to-one correspondence with published Bombyx mori ORs.

Author response image 1.

Structurally erroneous ORs in Propsilocerus akamusi.

Author response image 2.

Sensitivity analysis of OR counts. (a-b) Correlations between updated species-level OR counts and genome size.

Author response image 3.

OR protein structures removed by structural filtering from the automated annotations of Drosophila melanogaster and Anopheles gambiae.

Author response image 4.

Normalized analysis of SeqC diversity between holometabolous and non-holometabolous insects within Neoptera at the species level (a) and OR-count level (b). Welch's t-test was used. ***p < 0.001.

Reference

Bobkov YV, Walker Iii WB, Cattaneo AM. 2021. Altered functional properties of the codling moth Orco mutagenized in the intracellular loop-3. Sci Rep 11: 3893.

Carey AF, Wang G, Su CY, Zwiebel LJ, Carlson JR. 2010. Odorant reception in the malaria mosquito Anopheles gambiae. Nature 464: 66-71.

Chang H, Unni AP, Tom MT, Cao Q, Liu Y, Wang G, Llorca LC, Brase S, Bucks S, Weniger K et al. 2023. Odorant detection in a locust exhibits unusually low redundancy. Curr Biol 33: 5427-5438 e5425.

Dekel A, Pitts RJ, Yakir E, Bohbot JD. 2016. Evolutionarily conserved odorant receptor function questions ecological context of octenol role in mosquitoes. Sci Rep 6: 37330.

Frunze O, Lee D, Lee S, Kwon HW. 2024. A single mutation in the mosquito (Aedes aegypti) olfactory receptor 8 causes loss of function to 1-octen-3-ol. Insect Biochem Mol Biol 167: 104069.

Guo S, Kim J. 2007. Molecular evolution of Drosophila odorant receptor genes. Mol Biol Evol 24: 11981207.

Hall DR, Beevor PS, Cork A, Nesbitt BF, Vale GA. 1984. 1-Octen-3-ol. International Journal of Tropical Insect Science 5: 335-339.

Kline DL, Allan SA, Bernier UR, Welch CH. 2007. Evaluation of the enantiomers of 1-octen-3-ol and 1octyn-3-ol as attractants for mosquitoes associated with a freshwater swamp in Florida, U.S.A. Med Vet Entomol 21: 323-331.

Kreher SA, Mathew D, Kim J, Carlson JR. 2008. Translation of sensory input into behavioral output via an olfactory system. Neuron 59: 110-124.

Liu WB, Li HM, Wang GR, Cao HQ, Wang B. 2023. Conserved Odorant Receptor, EcorOR4, Mediates Attraction of Mated Female Eupeodes corollae to 1-Octen-3-ol. J Agric Food Chem 71: 1837-1844.

Lu T, Qiu YT, Wang G, Kwon JY, Rutzler M, Kwon HW, Pitts RJ, van Loon JJ, Takken W, Carlson JR et al. 2007. Odor coding in the maxillary palp of the malaria vector mosquito Anopheles gambiae. Curr Biol 17: 1533-1544.

Morawo T, Fadamiro H. 2016. Identification of Key Plant-Associated Volatiles Emitted by Heliothis virescens Larvae that Attract the Parasitoid, Microplitis croceipes: Implications for Parasitoid Perception of Odor Blends. J Chem Ecol 42: 1112-1121.

Paddock KJ, Corcoran JA. 2025. Life-stage dependent behavior mimics chemosensory repertoire diversity in a belowground, specialist herbivore. G3 (Bethesda) 15.

Pierce A, Pierce H, Borden J, Oehlschlager C. 1989. Production Dynamics of Cucujolide Pheromones and Identification of I-Octen-3-o1 as a New Aggregation Pheromone for Oryzaephilus surinamensis and O. mercator (Coleoptera: Cucujidae). Environmental Entomology 18: 747-755.

Qiu L, Tao S, He H, Ding W, Li Y. 2018. Transcriptomics reveal the molecular underpinnings of chemosensory proteins in Chlorops oryzae. BMC Genomics 19: 890.

Rondoni G, Roman A, Meslin C, Montagne N, Conti E, Jacquin-Joly E. 2021. Antennal Transcriptome Analysis and Identification of Candidate Chemosensory Genes of the Harlequin Ladybird Beetle, Harmonia axyridis (Pallas) (Coleoptera: Coccinellidae). Insects 12.

Tanaka K, Uda Y, Ono Y, Nakagawa T, Suwa M, Yamaoka R, Touhara K. 2009. Highly selective tuning of a silkworm olfactory receptor to a key mulberry leaf volatile. Curr Biol 19: 881-890.

Tian Z, Sun L, Li Y, Quan L, Zhang H, Yan W, Yue Q, Qiu G. 2018. Antennal transcriptome analysis of the chemosensory gene families in Carposina sasakii (Lepidoptera: Carposinidae). BMC Genomics 19: 544.

Turner RM, Derryberry SL, Kumar BN, Brittain T, Zwiebel LJ, Newcomb RD, Christie DL. 2014. Mutational analysis of cysteine residues of the insect odorant co-receptor (Orco) from Drosophila melanogaster reveals differential effects on agonist- and odorant-tuning receptor-dependent activation. J Biol Chem 289: 31837-31845.

Wang Q, Smid HM, Dicke M, Haverkamp A. 2024. The olfactory system of Pieris brassicae caterpillars: from receptors to glomeruli. Insect Sci 31: 469-488.

Xu L, Jiang HB, Yu JL, Pan D, Tao Y, Lei Q, Chen Y, Liu Z, Wang JJ. 2023. Two odorant receptors regulate 1-octen-3-ol induced oviposition behavior in the oriental fruit fly. Commun Biol 6: 176.

Xu P, Zhu F, Buss GK, Leal WS. 2015. 1-Octen-3-ol - the attractant that repels. F1000Res 4: 156.

Xu Q, Wu Z, Zeng X, An X. 2020. Identification and Expression Profiling of Chemosensory Genes in Hermetia illucens via a Transcriptomic Analysis. Front Physiol 11: 720.

Yan C, Sun X, Cao W, Li R, Zhao C, Sun Z, Liu W, Pan L. 2020. Identification and expression pattern of chemosensory genes in the transcriptome of Propsilocerus akamusi. PeerJ 8: e9584.

Zhang S, Zhang Z, Wang H, Kong X. 2014. Antennal transcriptome analysis and comparison of olfactory genes in two sympatric defoliators, Dendrolimus houi and Dendrolimus kikuchii (Lepidoptera: Lasiocampidae). Insect Biochem Mol Biol 52: 69-81.

Zhang Y, Wang B, Zhou Y, Liao M, Sheng C, Cao H, Gao Q. 2023. Identification and characterization of odorant receptors in Plutella xylostella antenna response to 2,3-dimethyl-6-(1-hydroxy)-pyrazine. Pestic Biochem Physiol 194: 105523.

eLife Assessment

This fundamental study presents experimental evidence on how geomagnetic and visual cues are integrated in a nocturnally migrating insect. The evidence supporting the conclusions is compelling. The work will be of broad interest to researchers studying animal migration and navigation.

Reviewer #1 (Public review):

[Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. The authors have addressed the comments raised in the previous round of review.]

Summary:

The manuscript by Ma et al. provides robust and novel evidence that the noctuid moth Spodoptera frugiperda (Fall Armyworm) possesses a complex compass mechanism for seasonal migration that integrates visual horizon cues with Earth's magnetic field (likely its horizontal component). This is an important and timely study: apart from the Bogong moth, no other nocturnal Lepidoptera has yet been shown to rely on such a dual-compass system. The research therefore expands our understanding of magnetic orientation in insects with both theoretical (evolution and sensory biology) and applied (agricultural pest management, a new model of magnetoreception) significance.

The study uses state-of-the-art methods and presents convincing behavioural evidence for a multimodal compass. It also establishes the Fall Armyworm as a tractable new insect model for exploring the sensory mechanisms of magnetoreception, given the experimental challenges of working with migratory birds. Overall, the experiments are well designed, the analyses are appropriate, and the conclusions are generally well supported by the data.

Strengths:

• Novelty and significance: First strong demonstration of a magnetic-visual compass in a globally relevant migratory moth species, extending previous findings from the Bogong moth and opening new research avenues in comparative magnetoreception.<br /> • Methodological robustness: Use of validated and sophisticated behavioural paradigms and magnetic manipulations consistent with best practices in the field. The use of 5 min bins to study a dynamic nature of magnetic compass which is anchored to a visual cue but updated with latency of several minutes is an important finding and a new methodological aspect in insect orientation studies.<br /> • Clarity of experimental logic: The cue-conflict and visual cue manipulations are conceptually sound and capable of addressing clear mechanistic questions.<br /> • Ecological and applied relevance: Results have implications for understanding migration in an invasive agricultural pest with expanding global range.<br /> • Potential model system: Provides a new, experimentally accessible species for dissecting the sensory and neural bases of magnetic orientation.

Weaknesses:

Overall, this is a strong study, and the authors have completed an excellent major revision.

Reviewer #2 (Public review):

Summary:

The work titled "Geomagnetic and visual cues guide seasonal migratory orientation in the nocturnal fall armyworm, the world's most invasive insect" provided experimental evidence on how geomagnetic and visual cues are integrated, and visual cues are indispensable for magnetic orientation in the nocturnal fall armyworm.

Strengths:

It has been demonstrated that the Australian Bogon moth could integrate global stellar cues with the geomagnetic field for long distance navigation. However, data are lacking for other insects. This study suggested that the integration of geomagnetic and visual cues may represent a conserved navigational mechanism broadly employed across migratory insects.

Weaknesses:

The visual cues used in the indoor experimental system designed by the authors may have some limitations in ecological relevance. The author may need more explanations on this experimental system.

In the revised manuscript, the authors have added explanations in the discussion section. I am fine with the revision.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer # 1 (Public review):

(1) Structure and Presentation of Results

• I recommend reordering the visual-cue experiments to progress from simpler conditions (no cues) to more complex ones (cue-conflict). This would improve narrative logic and accessibility for non-specialist readers. The authors have chosen not to implement this suggestion, which I respect, but my recommendation stands.

Thank you for this suggestion. We understand your point that presenting the experiments from simpler to more complex conditions may seem more intuitive. However, we have kept the original order because it better reflects the logic of the study itself. Our work first asked whether fall armyworms, like the Bogong moth, use a magnetic compass that is integrated with visual cues. Only after establishing this behavioral feature did we go on to test whether visual cues are required to maintain magnetic orientation. To make this reasoning clearer to readers, we have explicitly stated in the Introduction that magnetic orientation in the Bogong moth depends on the integration of visual cues, which provides clearer context for the experimental design.

(2) Ecological Interpretation

• The authors should expand their discussion on how the highly simplified, static cue setup translates to natural migratory conditions, where landmarks are dynamic, transient, or absent. Specifically, further consideration is needed on how the compass might function when landmarks shift position, become obscured, or are replaced by celestial cues. Additionally, the discussion would benefit from a more consolidated section with concrete suggestions for future experiments involving transient, multiple, or more naturalistic visual cues. This point was addressed partially in one paragraph of the Discussion, which reads as follows:

"In nature, they are likely to encounter a range of luminance-gradient visual cues, including relatively stable celestial cues as well as transient or shifting local features encountered en route. Although such natural cues differ from our simplified laboratory stimulus, they may represent intermittently sampled visual inputs that can be optimally integrated with magnetic information, with the congruency between visual and magnetic cues likely playing a key role in maintaining a stable compass response. Whether the cues are static or changing, brief periods without them may still allow the subsequent recovery of a stable long-distance orientation strategy. Determining which types of natural visual cues support the magnetic-visual compass, and how they interact with magnetic information, including how their momentary alignment or angular relationship is integrated and how such visual cue-magnetic field interactions may require time to influence orientation, together with elucidating the genetic and ecological bases of multimodal orientation, will be important objectives for future research." While this paragraph is informative, the wording remains lengthy, somewhat unclear, and vague. Shorter, clearer statements would improve readability and impact. For example:

• How could moths maintain direction during periods when only the magnetic field is present and visual landmarks are absent?

• Could celestial cues (e.g., stars) compensate, and what happens if these are also obscured?

• What role does saliency play when multiple visual landmarks are present simultaneously?

• How might a complex skyline without salient landmarks affect orientation?

Including simple, concise sentences that pose concrete open questions and suggest experimental designs would strengthen the discussion without creating space issues. In my view, a comprehensive discussion of how the simplified, static cue setup relates to natural migratory conditions-where landmarks are dynamic, transient, or absent-would add significant value to the paper.

Thank you for this constructive and insightful comment. You correctly point out that our articulation of the ecological relevance of the simplified, static cue setup was not sufficiently clear. We also agree that the original wording in the Discussion remained overly general. In the revised Discussion, we updated the manuscript to incorporate recently published findings on the use of light–dark gradients for orientation in fall armyworms. However, we explicitly note that it remains unclear whether fall armyworms can exploit naturally occurring luminance gradients, such as those generated by the moon, for orientation under natural conditions. We further emphasize that during natural migration the visual environment is dynamic, with celestial cues available intermittently and local visual features changing continuously during flight. In this context, we outline several key unresolved questions, including whether celestial cues can compensate when local landmarks are absent; how multiple visual cues are weighted and integrated with geomagnetic information; how transient visual cues (like moving clouds or changing illumination) influence orientation; and how luminance gradients that are common in natural nocturnal environments interact with the geomagnetic field to support orientation. For each of these issues, we briefly suggest experimental approaches to guide future research.

(3) Methodological Details and Reproducibility

• The lack of luminance level measurements should be explicitly highlighted.

Thank you for your helpful suggestion. You are right that luminance level is an important experimental parameter. We have stated this information in the Methods section under Behavioral apparatus: “The ambient light level in the experimental environment was measured to be below 1 lux using a Testo 540 lux meter (Testo SE & Co. KGaA, Titisee-Neustadt, Germany). Further work is still required to compare the illuminance used in this study with that under natural conditions, which are inherently variable.” This point is also clarified in the legend of Figure S3 in the supplementary material.

• The authors chose not to adjust figure legends by replacing "magnetic South" with "magnetic North." While I believe this would be more conventional and preferable, this is ultimately a minor stylistic issue.

Thank you very much for your suggestion. We understand your point and agree that using “magnetic North” would be more conventional. However, because our experiments focus on the orientation behavior of the autumn population, magnetic South is aligned with the landmark direction representing the potential migratory direction, which we believe makes the figures more intuitive for readers. We therefore consider this a minor stylistic issue.

(4) Conceptual Framing and Discussion

• Although the authors made a good attempt to explain the limitations of using an artificial visual cue, I believe there is room or a more explicit argument. For example, it could be stated clearly that this species is unlikely to encounter a situation in nature where a single, highly salient landmark coincides with its migratory direction. Therefore, how these findings translate to real migratory contexts remains an open question. A sentence or two making this point directly would strengthen the discussion.

Thank you for your helpful suggestion. We now address this point explicitly in the Discussion, noting that fall armyworms are unlikely to experience a natural visual environment dominated by a single, static, and highly salient landmark coinciding with their migratory direction. Consequently, how these findings translate to real migratory contexts remains an open question.

(5) Technical and Open-Science Points

• Sharing the R code openly (e.g., via GitHub) should be seriously considered. The code does not need to be perfectly formatted, but making it available would be highly beneficial from an open-science perspective.

Thank you for the suggestion. We agree that making code openly available is valuable from an open-science perspective. The MMRT script used in this study is Moore’s Modified Rayleigh Test, available from the original publication by Massy et al. (2021; https://doi.org/10.1098/rspb.2021.1805). In the previous version, we only cited this reference in the Materials and Methods section; we have now added a direct link to the script to improve clarity and accessibility. We have also provided a public link to the data-recording scripts used in the Flash Flight Simulator (https://doi.org/10.17632/6jkvpybswd.1). This repository additionally includes a map-based optical flow script that was not used in the present study but is shared for completeness.

Reviewer #1 (Recommendations for the authors):

• LL. 133-137 (end of paragraph starting with "The fall armyworm is a migratory crop pest native to the Americas"): Suggest splitting into shorter, clearer sentences. The limitations of this method could be better articulated here and elaborated in the Discussion.

Thank you for this suggestion. We have revised this paragraph by splitting it into shorter, clearer sentences and by articulating the limitations of this method more explicitly. These limitations are further elaborated in the Discussion.

• LL. 181-185 (end of paragraph starting with "To examine if fall armyworms integrate geomagnetic and visual cues for seasonal migratory orientation"): It would be helpful to state explicitly that season-specific headings have been confirmed in the lab using a flight simulator, but destination regions remain unknown without further tracking experiments.

Thank you for this helpful suggestion. We have now clarified in the revised manuscript that season-specific orientation headings have been confirmed in the laboratory using a flight simulator, while the actual migratory destination regions remain unclear in the absence of tracking experiments.

• LL. 230-234 (start of paragraph "Our previous research showed that fall armyworms reared under artificially simulated fall conditions…"): Clarify which migratory season is being referenced.

Thank you for this helpful suggestion. We have clarified in the text that the migratory season referenced here is the autumn migratory season. In addition, we have added information in the Methods to specify the actual calendar season during which the insects were reared under the simulated conditions.

• LL. 270-272 (middle of Fig. 2 caption): Suggest explicitly mentioning that for this population, the seasonally appropriate direction is southbound in autumn and northbound in spring, as this may not be clear to non-specialists.

Thank you for this helpful suggestion. We have now explicitly stated the seasonally appropriate migratory directions for this population, indicating southbound migration in autumn and northbound migration in spring, to improve clarity for non-specialist readers.

• LL. 421 (middle of paragraph starting with "We also considered the limitations of the Rayleigh test…"): Add that the groups lacking visual cues exhibited "lower directedness as per lower vector length (r)" in addition to lower flight stability.

Thank you for this helpful suggestion. We further note that the conclusions drawn from the flight stability analysis are consistent with those based on individual r-value analyses.

• LL. 499-501 ("unlike some vertebrates that can rely solely on magnetic information (Mouritsen, 2018)"): This point is slightly downplayed. It should be emphasized that nearly all tested vertebrates and invertebrates (e.g., birds, mole rats, fish, frogs, and other insects) demonstrate a magnetic compass without requiring visual landmarks. Moths are the only tested invertebrates so far that show landmark-magnetic field dependency for their magnetic compass to be manifested in a behavioural orientation response in Flight Simulator.

Thank you for this important comment. We agree that this point represents a key synthesis in the Discussion, as it concerns how our findings relate to, and differ from, magnetic orientation demonstrated in other animal groups. We have therefore expanded the Discussion to note that studies have shown that some animals can exhibit directional preferences in simplified visual environments solely in response to changes in the magnetic field, and we now cite representative examples from birds and mole rats. At the same time, we also acknowledge important methodological and phenotypic differences among taxa. In particular, moths’ magnetic orientation has been assessed using a flight simulator, a setup in which stable directional behavior must be actively maintained during continuous movement. This is an important difference from orientation assays in birds during take-off or in terrestrial mammals such as mole rats. Moreover, whether birds and other animals rely on visual input to detect or calibrate magnetic information under certain conditions remains an open question. We therefore emphasize here both the phenotypic differences observed across experimental systems and the methodological considerations.

• LL. 560-565 (paragraph starting with "Our flight simulator system (Dreyer et al., 2021) …"): Suggest clarifying what the Flash flight simulator system is and how it differs from the Mouritsen-Frost flight simulator.

Thank you for this suggestion. We have added a brief clarification of the Flash flight simulator and how it differs from the Mouritsen–Frost system.

• LL. 605-608 ("Spectral measurements …"): Explicitly mention that total illuminance was not measured and that further work is required to compare the illuminance used with natural conditions which of course vary.

Thank you for this helpful suggestion. We agree that total illuminance is an important factor. We have now added a statement noting that the ambient light level in the experimental environment was measured to be below 1 lux using a Testo 540 lux meter, and we further acknowledge that additional work is required to compare the illuminance used in this study with that under naturally variable conditions.

• LL. 628-641 (end of paragraph starting with "Electromagnetic noise at the experimental site ... "): Explain why this matters for interpreting behavioural responses. Highlight that although conditions were somewhat magnetically noisy which based on the past work may disrupt magnetic compass as it was shown in birds (eg Engels et al. 2014 Nature), the observed magnetic response under certain conditions indicates that the magnetic sense remained functional when landmark and magnetic field were aligned. This way you can pre-empt this criticism of your magnetic conditions being not ideal and noise on the left handside of the spectrum measured (which is not uncommon).

Thank you for this helpful suggestion. We have now cited Engels et al. (2014, Nature) in this section and expanded the text to explain why electromagnetic noise at the experimental site is relevant for interpreting the behavioural responses. We also clarify the rationale for measuring electromagnetic noise and discuss the observed low-frequency (“left-hand side”) noise in the spectrum.

• Fig. 51: Suggest adapting Y-axes and using violin or box plots (e.g., panels A/B starting from 30 up to 50, etc.).

Thank you for this helpful suggestion. We have revised Fig. 5 accordingly by adapting the Y-axis scaling and replacing the original plots with box plots, as suggested.

eLife Assessment

Davis and colleagues present fundamental findings on pressure mechanosensation in lymphatic vessels, with broad relevance to mechanosensory physiology. Using an impressive array of knockout mouse models and rigorous pressure myography, they provide compelling evidence that mechano-activation of GNAQ/GNA11-coupled GPCRs drives lymphatic vessel contractility via an IP3-IP3R1-ANO1 Ca²⁺ release and depolarization pathway. While the study falls short of pinpointing the specific GPCR(s) mediating pressure-induced chronotropy - having ruled out candidates such as AT1R and G12/13-coupled receptors - their single-cell RNA sequencing profiling establishes a benchmark candidate list (including adhesion and frizzled GPCRs) that provides a foundation for future target identification.

Reviewer #1 (Public review):

Summary:

Davis and co-authors used many mouse models to investigate the mechanisms which regulate the contractility of mouse popliteal collecting vessels, primarily chronotropy. The authors use prior literature from the vasculature as a framework to test concepts in lymphatic vessels. The mouse models they used provide evidence for and against the involvement of multiple proteins in regulating chronotropy and other contractile properties in lymphatic vessels. The large amount of data indicate that mechano-stimulation of GNAQ/GNA11-coupled GPCRs generates IP3, which induces intracellular Ca2+ release through IP3R1 to drive depolarization through the activation of ANO1 Cl- channels.

Strengths:

Major strengths of the study are the vast number of mouse knockout models which were used to test the importance of ion channels and G protein signaling pathways in the regulation of lymphatic vessel contractility. The study is a valiant effort and contains a large amount of data. The authors achieved several objectives to find that ANO1 and IP3R1 regulate chronotropy and many other potential proteins do not regulate chronotropy. This study will have a major impact on the field.

Weaknesses:

The authors did an excellent job with the revision. There are no weaknesses.

Reviewer #2 (Public review):

Summary:

In this study, Davis et al., embarked on the quest of the molecular elements responsible for the regulation of lymphatic phasic contractile activity in response to variation of transmural pressure, a mechanism (termed pressure-induced lymphatic chronotropy by the authors). This contractile activity is critical for drainage of interstitial fluid from the tissue and transport of lymph and its cellular and molecular constituents back to the blood circulation. The authors' aim was to investigate the mechanism(s) involved in the pressure-induced regulation of lymphatic pumping, and test whether activation of cation channels, shown in other systems to play mechanosensitive roles are directly at play, and/or whether mechano-activation of Gq/11-coupled GPCRs is necessary to generate second messengers to activate those channels, as it has been suggested for the regulation of myogenic tone in arteries. To achieve their goal, the authors used their well-described, highly reliable protocols of mouse lymphatic vessel isolation, pressure myography, and data acquisition to obtain frequency-pressure relationships and other contractile parameters from transgenic mice where specific channels or molecular elements of interest have been ablated. They combined these data with scRNAseq analysis of these gene targets to determine their respective role and level of expression in lymphatic muscle cells. Their findings demonstrated that, with the exception of ANO1 Cl channels, part of the contractile pacemaker mechanism, none of the exhaustive list of tested ion channels was critical to pressure-induced lymphatic chronotropy, but only ablation of Gq/11 genes or IP3R1 gene in smooth muscle cells led to chronotropy inhibition. To strengthen these findings, they then repeated the myography experiments in the presence of a Gq/11 inhibitor and screened frequency-pressure relationships in vessels from transgenic mice lacking each of the 7 most expressed GPCR according to their scRNAseq dataset. While none of the GPCR ablation showed significant inhibition, they could conclude that transmural pressure activates GPCRs coupled to Gq/11 which generate IP3 to induce SR Ca2+ release through IP3R1 and activate ANO1-mediated depolarization and increase contraction frequency.

Strengths:

The manuscript's strengths reside primarily on very robust, clean and unequivocal pressure myography data and analysis. The research team is mastering these techniques they developed more than a decade ago and have implemented to mouse lymphatics to study their contractile properties, with consistent and convincing outcomes. They also provide data from an impressive list of transgenic mice in order to determine the role of the targeted genes in pressure-induced lymphatic chronotropy, relaying on pharmacological small molecule inhibitors to confirm their findings. Finally, the use of scRNAseq analysis they gathered from previously published dataset brings novelty with respect to expression of the genes of interest in all populations of cells comprising the lymphatic vessels, but more critically to validate or contrast the potential impact of genetic alteration of the given gene in the ability of lymphatic muscles to respond to change in pressure.

Comments on revised version.

The new experiments, additional data, and the now extensive discussion better temper the authors' initial conclusion, the new title as well. The authors should be commended for such an impressive amount of work.

Reviewer #3 (Public review):

In this manuscript, Davis and colleagues aimed to identify the molecular sensors and signaling cascade that enable collecting lymphatic vessels to increase their spontaneous contraction frequency in response to intraluminal pressure (pressure-induced chronotropy). They tested whether the process is similar to blood vessel myogenic constriction by relying on cation channels (TRPC6, TRPM4, PKD2, PIEZO1, etc.) or instead require the activation of G-protein-coupled receptors (presumably mechanosensitive GNAQ/GNA11-coupled receptors), using ex vivo pressure myography of mouse popliteal lymphatics, smooth muscle-specific conditional knockouts, quantitative PCR validation, and single-cell RNA sequencing for target prioritization. The authors convincingly demonstrate that pressure-induced chronotropy does not require the cation channels implicated in arterial myogenic tone but is blunted by deletion of GNAQ/GNA11 or IP3 receptor 1, supporting a model of GPCR > IP3 > Ca2+ release > Cl⁻ channel activation > depolarization. The core conclusion is robust. The work redefines lymphatic pacemaking as G-protein-coupled receptor-dependent mechanotransduction, distinct from arterial mechanisms, and provides a genetically validated toolkit that is useful for studying lymphatic function and dysfunction.

Strengths:

(1) The data are of high quality and highly sensitive functional readouts

(2) The systematic genetic targeting is a major strength that overcomes pharmacological artifacts

(3) Careful quantitative analyses of frequency-pressure slopes

Comments on revised version:

The authors have satisfactorily addressed my comments. No additional comments.

Reviewer #4 (Public review):

Reviewing Editor:

In this revised study Davis and colleagues provide novel findings that are fundamental to the understanding of pressure mechanosensation in lymphatic vessels and are of critical importance to other areas of mechanosensory physiology. The authors used an impressive number of knockout mouse models and employed rigorous state-of-the-art methods, including pressure myography recordings. They provide compelling evidence that mechano-activation of GNAQ/GNA11-coupled GPCRs generates IP3, which induces Ca2+ release from internal stores through IP3R1 and drives depolarization through the activation of ANO1 Cl- channels to induce lymphatic vessel contractility. To uncover the specific GPCR(s) involved in this response, the authors compiled and ranked a list of candidates using single cell RNA sequencing (scRNAseq). In addition to ruling out all major mechanosensitive ion channels previously implicated in this response, the authors also ruled out the involvement of G12/13-coupled GPCRs and 7 most consistently expressed GPCRs in lymphatic smooth muscle along with AT1R. The latter is considered the most likely mechanosensory GPCR in vascular smooth muscle. Although the authors' data fall short of identifying the specific GPCR(s) that mediates pressure-induced chronotropy, they provide compelling evidence that Gq/11-coupled GPCRs are of critical importance to this function. Furthermore, their scRNAseq analysis identified 136 GPCRs in lymphatic smooth muscle, including strongly or moderately expressed candidates such as an olfactory GPCR, three frizzled GPCRs and several adhesion GPCRs (aGPCRs) that encode their own agonist along when many GPCRs with low expression levels. This list represents a foundation for future studies using knockout animal models and aimed at identifying the molecular identity of the GPCR or GPCRs that mediate mechanosensation in lymphatic vessels."

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

(1) Davis and co-authors used many mouse models to investigate mechanisms that regulate the contractility of mouse popliteal collecting vessels, primarily chronotropy. Many of the mechanisms studied were previously shown to regulate pressure-induced constriction in small arteries. The authors use prior literature from the vasculature as a framework to test similar concepts in lymphatic vessels. The mouse models used provide evidence for and against the involvement of multiple proteins in regulating chronotropy and other contractile properties in lymphatic vessels. They propose that mechano-activation of GNAQ/GNA11-coupled GPCRs generates IP3, which induces Ca2+ release through IP3R1 and drives depolarization through the activation of ANO1 Cl- channels. Major concerns include the author's major conclusion that GNAQ/GNA11-coupled GPCRs contribute to chronotropy. This conclusion is not supported by the data presented.

Although we have not yet identified a specific GPCR in LMCs that mediates pressure-induced chronotropy, our experiments have established that Gq/11-coupled GPCRs are critical for this response. The almost identical phenotypes of GNAQ/GNA11 double KO vessels with ANO1 KO and IP₃R1 KO vessels point to an upstream, mechanosensitive GPCR that regulates ANO1-regulated pacemaking activity through calcium release by IP₃R1. New experiments show that the highly specific GNAQ/GNA11 inhibitor YM254890 decreases contraction frequency to the same degree as, or even stronger than, that observed in Gq/11 double KO mice (Fig. 22). Additional data assessing the responses of inguinal-axillary lymphatic vessels (IALVs) from G11^-/- and Gq/11 double KO further confirm our findings and conclusions (Suppl. Fig. 1). To address your concern about identifying a specific GPCR, we have performed new scRNAseq analyses to identify and rank the most prominently expressed GPCRs in LMCs (Fig. 21) along with new experiments to test the most likely GPCR candidates that emerged from that analysis (Fig. 22). Please see our specific responses below and the additional data that we now provide.

While our results still fall short of identifying a specific Gq/11-coupled GPCR involved in this LMC mechanosensitive response, they rule out all of the major mechanosensitive ion channels previously implicated in this process (Figs. 4-15), rule out G12/13-coupled GPCRs (Fig.20) and strongly support a critical role for the canonical Gq/11-coupled mechanism (Fig. 18 and Suppl. Fig. 1). New data to eliminate contributions of the 7 most consistently expressed LMC GPCRs, along with the most likely mechanosensory GPCR in VSMCs (AT₁R) (Fig. 22), out of the 136 GPCRs detected in our LMC scRNAseq dataset, raise interesting GPCR candidates that require future testing. The remaining list of 11 strongly or moderately expressed candidates includes 3 frizzled GPCRs that typically do not signal through G proteins, an olfactory GPCR, and several adhesion GPCRs (aGPCRs) that encode their own agonist (containing the Stachel peptide sequence). Some of these adhesion GPCRs recruit and signal through Gq/11, and aGPCRs have been implicated in mechanosensation in other cell types (PMID: 25937282, PMID: 40395494). However, pharmacological tools to specifically test most of these aGPCRs are currently lacking. Another possibility, of course, is that the key LMC protein may be a GPCR with a very low expression level.

At present, the only way to test most of those candidates would be to generate new mice with smooth muscle KO of the specific GPCRs, as Offermann’s group has done in a few other cases (PMID: 40623996, PMID: 38597112, PMID: 39920161), but we are without the resources to embark on such an expedition. We are therefore unable to resolve this last issue with currently available resources and/or technology but now acknowledge this as a shortcoming of our study and a future direction. We have adjusted our title to be more congruent with the data presented in the revised manuscript.

Strengths:

(2) One major strength of the study lies in the vast number of mouse knockout models that were used to test the importance of ion channels and G protein signaling pathways in the regulation of lymphatic vessel contractility. In this regard, the study is a valiant effort. The authors achieved several objectives to find that ANO1 and IP3R1 regulate chronotropy, and many other potential proteins do not regulate chronotropy. This study will have a major impact on the field if additional support for G proteins is provided.

Please see our specific responses below and the additional data that we now provide to strengthen support for a GNAQ/GNA11 mechanism.

Weaknesses:

(3) Major conclusions concerning the involvement of G proteins are drawn from the global Gna11 knockout mouse models. This conclusion is weak. Global Gna11 knockout mice are highly likely to have a multifactorial phenotype that could create significant differences in the data.

We agree that global GNAQ/11 DKO mice likely have a multifactorial phenotype and that, IF our hypothesis involved testing some aspect of in vivo function with multiple cell types or organs, this would complicate our conclusions. However, because our approach focuses on ex vivo testing of the F-P relationship of isolated lymphatic vessels, a response mediated specifically by LMCs (i.e., independent of other circulating factors, endothelium, neural control, etc.), our results are NOT likely to be influenced by such extrinsic factors. Stefan Offermanns and colleagues have used similar mice (single floxed gene + Cre + global KO) and ex vivo assays of arteries in multiple peer-reviewed studies (e.g. PMID 31549965), so there is clear precedent for our approach. An alternative approach to the global GNAQ/11 DKO mice would be to use Myh11Cre; GNAQ^f/f; GNA11^f/f double floxed mice, but then those would likely suffer from the potential limitation of competing recombination between the two floxed genes, as we and others have encountered this problem in previous studies using a single Cre with multiple floxed alleles (PMID: 29336844, PMID: 36868507).

(4) Control experiments need to be performed on vessels from the global knockout mice if these major conclusions are to be made.

Control experiments were conducted on popliteal vessels from Myh11Cre; GNAQ^f/f; GNA11^+/+ mice (uninduced littermates) and from GNAQ^f/f; GNA11^+/+ mice (lacking Myh11Cre), the combination of which had a normal F-P relationship (Fig. 18). We have also conducted experiments on Myh11Cre; GNAQ^f/f; GNA11^-/- mice that were not induced with tamoxifen (GNA11^-/-). It is not clear if the reviewer is suggesting that we include GNAQ^-/- mice, but we did not have access to those mice. Gnaq^-/- mice exhibit significant developmental abnormalities (PMID: 22723772), although according to Offermanns and colleagues (PMID: 9687499), only GNAQ homozygous deficient mice displayed “obvious phenotypic defects”.

(5) Similarly, pharmacological tools or alternative approaches to manipulate G proteins should be used to support the data from these mouse models to draw these major conclusions.

To address this comment, we have performed new experiments on WT mice using the potent GNAQ/11 inhibitor YM254890 (at 100 nM, the concentration used in many previous publications by other groups), which had nearly the same effect on abrogating the F-P relationship as induced Myh11Cre; GNAQ^f/f; GNA11^-/- mice (see Fig. 22). The specificity of YM-254890 for the Gq family has recently been verified by excellent work out of David Yule’s lab (PMID: 40563241).

(6) The Gnaq smKO mice are the most specific G protein model studied here. However, there is no phenotype. Do not discuss trends in the data. If the data are not significant, conclude so. If more experiments are required to reach significance, provide more data in the manuscript.

With respect to the Gq smKO data:

(1) We have added more data to the control group.

(2) We have now listed the p value (0.0819 in Fig. 18) for the Gq/11 controls vs Gq smKO, instead of simply designating it “n.s.” with a cut-off of p<0.05; the readers can now judge for themselves how close the values come to being significant.

(3) Although the F-P slope does not reach significance (Fig. 18B), the frequency is significantly different between the Gq/11 controls and Gq smKO at 3 pressures (see the frequency graph in Fig. 18C), so significant differences ARE indeed detected in Gq smKO vessels.

(4) The F-P slope for the DKO is lower than that of the G11 KO and it appears the two knockouts are cumulative, suggesting that Gnaq deletion is contributing to the lower values of the Gq/11 DKO. As further support for this conclusion, we have performed new experiments on IALVs (in part also to address concern #8 below) and that analysis is shown in Suppl. Fig. 1. We include those data as a Supplemental Figure because the contraction data in all the other figures are for popliteal lymphatic vessels. Note that the F-P analysis is not appropriate for IALVs because the control vessels typically show only a 2-2.5x change in frequency over the entire pressure range and frequency essentially reaches a plateau at P = 2 cm H₂O. The frequency plots for control, G11^-/- and Gq/11 DKO IALV vessels show the same trend as for popliteal vessels (significantly lower than control vessels at all pressures, with only a single exception), but in this case the DKO vessels have significantly lower frequencies than G11^-/- vessels at the lower pressures (1-2-3 cm H₂O), further supporting the results from popliteal vessels.

(7) The conclusions repeatedly refer to a signaling pathway wherein the upstream component is GPCRs, which activate G proteins. While this may be the case, no GPCRs were identified here, and the involvement of G proteins is questionable, as the authors outline in lines 693-695 and noted above. The conclusions should be tempered, including in the abstract, unless additional experiments are performed to support the involvement of G proteins. Perhaps then the authors may be able to infer that GPCRs are involved.

Please see our response to your comment #1. We have now included additional experiments using the Gq/11 inhibitor YM254890 (at a concentration of 100 nM that is reputed to be “selective” for Gq/11, according to the literature (PMID: 40563241). The blunting of the F-P relationship in the presence of that compound is comparable to, and perhaps even more significant than, the blunting observed for the DKO (the summary analysis is shown for P= 3 cm H₂O in the middle of Fig. 22, in a format that matches the other analyses in that figure, but the frequencies were significantly attenuated over the entire lower pressure range). These results strengthen support for a ligand-independent mechanotransduction mechanism through one or more GPCRs of the Gq/11 family. However, because 1) there are at least 136 possible specific GPCR candidates and 2) our new specific tests of the 7 most abundantly expressed GPCRs were negative, pinning down the exact GPCR(s) involved will require much more work. Given the amount of data already presented in this study, and the possibility that the unidentified GPCR is an orphan receptor or other GPCR with very low expression, we think it is appropriate to defer that line of experimentation to a future study. We now clearly state that this is a limitation of the present work and a future direction.

(8) Line 318. The point regarding the choice to use popliteal vessels versus IALVs will be unclear to the uninitiated, particularly as the authors previously used IALVs. Including additional justification in the text and/or data from IALVs in Figure 1, which compares IALVs to popliteal vessels, would better explain the logic.

We now include additional justification in the text by citing Zawieja 2018 and the similarities to Ano1 KO and IP₃R1 KO findings in IALVs in our other papers. Note also the very close agreement between the Gq/11 data for popliteal vessels vs IALVs in Fig. 18 C and Suppl. Fig. 1).

(9) The conclusions drawn for TRPC6 and TRPC3 are less convincing. Germline global knockout mice, which are known to undergo compensation, were used, and high data variability is apparent. Using TRPC3 and TRPC6 blockers in the mouse models studied in Figure 4 would strengthen the arguments made regarding these proteins.

Our primary reason for using the global KO mice is that pharmacological inhibitors of TRP channels, particularly TRPC6, are quite non-specific. We did not have access to Trpc6 floxed mice but our ex vivo assay focuses a specific aspect of LMC function in isolation from extrinsic influences (see answer to comment #3 above), and of the 18 cell subtypes detected in these vessels by scRNAseq, Trpc6 is only detectable in LMCs (Fig. 5 panel A). With regard to consequences of global knockouts and possible compensation, that is why we generated the Trpc6/Trpc3 DKO mice, as upregulation of Trpc3 is a known compensatory mechanism for Trpc6 deletion (PMID: 16055711); to our knowledge only one other group has taken the trouble to test this (PMID: 16055711) with respect to vascular smooth muscle function.

(10) Did you perform power analysis to ensure that experimental numbers were sufficient to conclude that no statistical difference exists between datasets? If not, this needs to be done. For example, data shown in Figure 5C for tone and 6C for frequency and tone appear to be significantly different, but are concluded not to be so.

With respect to Fig. 5, there are significant differences in contraction frequency between the TrpC6/C3 DKO mice and their SV129 controls at 5 of 7 pressures, but the frequencies of the KO mice were actually HIGHER—the opposite of the result expected if TRPC6 or TRPC3 were critical for pacemaking. With respect to the effects of TRPC6 KO on tone, there were no significant differences in tone between TRPC6^-/- vessels and their SV129 controls, but there were indeed significant differences between the TrpC6/C3 DKO mice and controls at 4 of 7 pressures, but with tone actually being HIGHER than normal (opposite to what would be predicted if those channels were critical for the development of myogenic tone). We point out how that observation is at odds with the conclusions of at least one study on arterial myogenic tone (PMID: 11861411), but that topic is not the focus of our study.

(11) At the end of each result section, a concluding statement is made regarding the effects on pressure-induced chronotropy. In many cases, there are additional effects of manipulating protein expression on other contractile properties. One example is for TRPC3 and TRPC6 (lines 414-416), but others are TRPV4, TRPV3, ENaC, Kir, Cav3.1/3.2, etc. Some interpretation is in the Discussion, but the concluding statements at the end of each result section should be expanded to summarize what the authors think the other significant differences in the data represent.

Thank you for allowing us the flexibility to address this at the end of each relevant section in the Results. We have now expanded several of these sections as you suggest.

(12) Kv7.4 channels. You state you have data (not shown) with linopiridine and XE991. Why not show those results here to support the experiments with the Kcnq4 smKO mice? Otherwise, I suggest you remove the statement from the unpublished data.

We have removed the comments about the unpublished data with other Kv7.4 inhibitors.

(13) Figure 13A. Kcnj2 is modestly expressed in LECs, but very little is present in LMCs. This likely underlies the effect of barium. If you remove the endothelium, does the effect of barium disappear? While this is not the major focus of the study, the effects of barium are dramatic, and it should be made clear whether this is due to inhibition of Kir channels in smooth muscle or endothelial cells.

Thank you for making this very good point. We cannot rule out an effect on LECs in this context and denuding popliteal LVs is nearly impossible without compromising their contractile function because of their small size and the presence of intraluminal valves. The valves prevent easy passage of an air bubble (the opening is only 1/4 the area of the lumen) and attempts at mechanical removal result in extreme valve damage and/or residual LECs at valve sites. If there is an influence of inhibiting Kir channels in LECs, it is not via electrical communication between LECs and LMCs, which is minimal or non-existent as we have shown in previous studies (PMID: 30143234, PMID: 28994159, PMID: 30355030). Thus, a potential LEC influence would be mediated by release/production of vasoactive factors, which is a point that we now discuss.

(14) Figure 18C tone. Several values for losartan look different but are not labelled as such. Please clarify and discuss if different.

All the significantly different values were labeled as such, but that panel is no longer in the revised paper as the AT₁R frequency data have been included in a different form (Fig. 22).

(15) The manuscript should include raw data traces in figures that show the major pathways that you conclude regulate chronotropy.

Thank you for this suggestion. Three new figures have now been added to illustrate the blunting of F-P relationship in Ano1 smKO (Fig. 3), IP₃R1 smKO (Fig. 17) and GNAQ/11 DKO vessels (Fig. 19).

Reviewer #2 (Public review):

Summary:

In this study, Davis et al. embarked on the quest for the molecular elements responsible for the regulation of lymphatic phasic contractile activity in response to variation of transmural pressure, a mechanism (termed pressure-induced lymphatic chronotropy by the authors) critical for drainage of interstitial fluid from the tissue and transport of lymph back to the blood circulation. Their aim was to investigate the mechanism(s) involved in the pressure-induced regulation of lymphatic pumping, and test whether activation of cation channels, shown in other systems to play mechanosensitive roles are directly at play, and/or whether mechano-activation of GNAQ/GNA11-coupled GPCRs is necessary to generate second messengers to activate those channels, as it has been suggested for the regulation of myogenic tone in arteries. To achieve their goal, the authors used their well-described, highly reliable protocols of mouse lymphatic vessel isolation, pressure myography, and data acquisition to obtain frequency-pressure relationships and other contractile function parameters from transgenic mice where specific channels or molecular elements of interest have been ablated. They combined these data with scRNAseq analysis of these gene targets to determine their respective role and levels of expression in lymphatic muscle cells. Their conclusion is that none of the exhaustive list of tested ion channels was critical, except ANO1 Cl channels, part of the contractile pacemaker mechanism, but that transmural pressure activates GNAQ/GNA11-coupled GPCRs, which generate IP3 to induce SR Ca2+ release through IP3R1 and activate ANO1-mediated depolarization.

Strengths:

The manuscript's strengths reside primarily in very robust, clean, and unequivocal pressure myography data and analysis. The research team is mastering these techniques they developed more than a decade ago and have implemented in mouse lymphatics to study their contractile properties, with consistent and convincing outcomes. They also provide data from an impressive list of transgenic mice in order to determine the role of the targeted gene in pressure-induced lymphatic chronotropy, relying on pharmacological small molecule inhibitors only when necessary. Finally, the use of scRNAseq analysis they gathered from previously published datasets brings novelty with respect to the expression of the genes of interest in all populations of cells comprising the lymphatic vessels, but more critically, to validate or contrast the potential impact of genetic alteration of the given gene on the ability of lymphatic muscles to respond to a change in pressure.

Weaknesses:

(1) The main weakness may reside in the fact that while the authors provide a convincing demonstration that GNAQ/GNA11 are involved in the regulation of the F-P relationship, they give little evidence of the involvement of "upstream" receptors. Indeed, inhibition of AT1R, shown to be involved in myogenic regulation of arteries (a phenomenon the authors rightfully compare to pressure-induced lymphatic chronotropy), didn't lead to a similar effect (decrease in F-P) in lymphatic vessels. Arguably, other GPCRs might be involved in lymphatic vessels, but as such information is not provided in the manuscript, the author's conclusions should be dampened. More in-depth discussion would be required. In fact, it can be argued that the discussion is very restricted with respect to the amount of data and information the manuscript provides.

To address these valid concerns, we performed a detailed scRNA-seq analysis of the GPCRs expressed in mouse LMCs. Approximately 136 GPCRs were identified as being expressed in over 0.5% of LMCs, the top 20 of which are expressed at moderate levels in a substantial percentage (>36%) of LMCs. We have added a new Figure showing bubble plots of these 20 GPCRs in the various cell populations of IALVs, ranked according to their expression level (Fig. 21); we also show their expression in the other cell types. We then experimentally tested 7 of the most likely candidates for which inhibitors were available by determining whether blocking the GPCR would alter pacemaking by lowering the contraction frequency at P=3 cm H₂O. This was a streamlined assay compared to measuring the complete F-P relationship but produced similar results; to demonstrate this point, we also performed the streamlined analysis for the ANO1 inhibitor, Ano1 smKO, IP₃R1 smKO and Gq/11 DKO, all of which showed significantly reduced frequencies, compared to their respective controls at P=3 cm H₂O (see Fig. 22, black symbols). The alignment of statistical significance for the latter results with the corresponding F-P analyses shown in Figs. 2, 16 and 18) demonstrates that this assay is sufficiently powerful to detect a significant effect on pacemaking. Unfortunately, the results for each of the GPCRs tested were negative (Fig. 22, red symbols). We conclude that none of those particular GPCRs (NPY1R, ET_AR, NPR3, TBA₂XR, S1PR1, AT1_AR, AT1_BR), or two additional GPCRs (HT-2R, HRH1) that were tested in previous studies (PMID: 12770929, 28453392), are critical for pacemaking or the F-P relationship in LMCs. Of the remaining 11 GPCRs, several are adhesion receptors (Adgrl1, Adgrl2, Adgrl3, Adga2), two are orphan receptors (Gprc5b, Olfr1033), one codes for the GABA-B1 receptor (Gabbr1, with unknown function in the lymphatic system), two code for proteins in the Wnt signaling pathway (Fzd4, Fzd2), one codes for a protein in the Hedgehog signaling pathway (Smo) and one encodes a beta-amyloid binding protein (Tm2d1). The possible roles of these GPCRs, and the other 100+ GPCRs expressed at much lower levels remain to be tested in future studies. At any rate, systematic investigation of these GPCRs is well beyond the scope of this study, and in many cases not possible due the lack of known soluble ligands/antagonists. We have significantly expanded the Discussion with regard to this topic and stated that our inability to identify a specific Gq/11-coupled mechanosensory GPCR in LMCs is a limitation of our study.

(2) Overall, the authors convincingly achieved their aim by performing an impressive number of technically challenging experiments, leading to solid datasets. While these support their main conclusions, a more elaborate discussion might be required to refine them.

We hope this limitation has been tempered by the addition of the new data and the expanded Discussion section mentioned above.

(3) This study is likely to have an important impact on the field as it provides some answers to the lingering question of how lymphatic vessels regulate their contractile activity to variation in transmural pressure and certainly proposes an experimental means to further explore and address that question.

Thank you.

Reviewer #3 (Public review):

In this manuscript, Davis and colleagues aimed to identify the molecular sensors and signaling cascade that enable collecting lymphatic vessels to increase their spontaneous contraction frequency in response to intraluminal pressure (pressure-induced chronotropy). They tested whether the process is similar to blood vessel myogenic constriction by relying on cation channels (TRPC6, TRPM4, PKD2, PIEZO1, etc.) or instead require the activation of G-protein-coupled receptors (presumably mechanosensitive GNAQ/GNA11-coupled receptors), using ex vivo pressure myography of mouse popliteal lymphatics, smooth muscle-specific conditional knockouts, quantitative PCR validation, and single-cell RNA sequencing for target prioritization. The authors convincingly demonstrate that pressure-induced chronotropy does not require the cation channels implicated in arterial myogenic tone but is blunted by deletion of GNAQ/GNA11 or IP3 receptor 1, supporting a model of GPCR > IP3 > Ca2+ release > Cl⁻ channel activation > depolarization. The core conclusion is robust. The work redefines lymphatic pacemaking as G-protein-coupled receptor-dependent mechanotransduction, distinct from arterial mechanisms, and provides a genetically validated toolkit that is useful for studying lymphatic function and dysfunction.

Strengths:

(1) The data are of high quality and highly sensitive functional readouts

(2) The systematic genetic targeting is a major strength that overcomes pharmacological artifacts

(3) Careful quantitative analyses of frequency-pressure slopes

Weaknesses:

(1) The use of inguinal-axillary vessels for single-cell RNA sequencing rather than the popliteal segment studied functionally.

Agreed. The need to use IALVs derives from the much smaller amount of tissue needed for scRNAseq analysis of individual popliteal vessels (and 3-5x longer time required for cleaning) before pooling them. We have tried to clearly state this limitation. However, functional comparisons between popliteal and IALV function are quite similar in almost all aspects that we have studied. In the majority of our protocols, preliminary experiments using IALVs were performed side by side with popliteal vessels (for Ncx1 smKO, Trpc6^-/-, Trpc6/3 DKO, G_12/13 DKO, Piezo1 smKO, Trmp4 smKO) without detection of a significant difference in phenotype. In addition, we have performed new experiments using IALVs from G11^-/- and GNAQ/GNA11 DKO mice to support the major positive findings drawn from the popliteal vessel studies (compare Fig. 18 with Suppl. Fig. 1).

(2) No direct testing of the specific G-protein-coupled receptor involved.

Agreed. As described in our response to the other reviewers (see point 1 to Reviewer 1 and point 1 to Reviewer 2), to address these concerns we have added a new figure (Fig. 21) ranking the top-expressed GPCRs in LMCs and have experimentally tested several of the most likely candidates (Fig. 22). Unfortunately, our results were still negative, allowing us to rule out those particular GPCRs but not to pinpoint a specific GPCR. The possible roles of other 100+ GPCRs in LMCs remain to be investigated in future studies. We have adjusted our title to better reflect our results.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) The authors state multiple times that several cation channels have been shown to contribute to vascular tone. Arterial smooth muscle cell ANO1 channels also contribute to arterial myogenic tone (PMID: 22872152). An ANO1-dependent mechanism is proposed here in lymphatic vessels, and the same ANO1 mouse models are studied. PMID: 22872152 is not discussed or cited and should be.

Now cited.

(2) Lines 88-97. It is important to note that the Bulley paper studied tamoxifen-inducible smooth muscle cell-specific knockout mice, whereas the Sharif-Naeini study used germline SM22-driven PKD1 knockout mice. The mice used in the Sharif-Naeini paper were subsequently found to have an ADPKD-like phenotype due to non-specific knockout of PKD1 in other cell types (PMID: 20856231). That should also be noted. The authors also state "PKD1 channel" in this paragraph when PKD1 does not form a channel.

Corrected and clarified. A new paper on PKD1 by Jaggar’s group is now cited in that context.

(3) Line 119. Perhaps you mean PKD1/2 channels?

Corrected.

(4) Line 312. It is unusual that data from Figure 2A is described before that in Figure 1. Can you reorder the figures to fit the text?

Corrected.

(5) Line 130. Can you provide references that show that ANO1 is not intrinsically mechanosensitive?

Added.

(6) Line 160. I believe the Pkd2f/f mice are available from the Maryland PKD Core and were not generated by the Jaggar lab.

Corrected.

(7) Line 319 needs a reference.

Added.

(8) The color choice for the symbols in many figures, particularly Figures 2, 3, and 4, makes it difficult to know which condition is which.

Corrected.

(9) I may have missed it, but the Figure 1A legend seems to need a definition for every cell type abbreviation.

It is shown in panel A (Figure 2).

(10) In several places, the authors state that they deleted gene expression in LMCs, when the mouse model deletes expression from all smooth muscle cells. Please refine wording.

Clarified.

(11) Line 664-666. Sharif-Naeini did not propose that PKD2 channels contribute to myogenic tone. That was Bulley et al.

Corrected.

Reviewer #2 (Recommendations for the authors):

(1) The role of TRPV4: It is interesting to note that despite a quasi-non-expression of the TRPV4 gene in LMC, KO this gene in these cells results in an increase in contraction frequency and a strong decrease in amplitude at low pressures, suggesting the channel is expressed and active in LMCs. It is not too clear how macrophages could be indirectly involved in this process by releasing products. Are they also sensitive to pressure? How? And why should they be more involved than any other cell type present in the vessel wall (i.e, LECs)?

Macrophages definitely release products that affect LMC activity (PMID: 38826322) and they may be sensitive to pressure but that has not yet been demonstrated. However, as shown in the paper cited, selective KO of Trpv4 in macrophages does not abrogate the F-P relationship in lymphatic vessels.

(2) The role of AT1R in pressure-induced chronotropy: The lack of effect of AT1R inhibition on F-P doesn't mean other GPCRs could play a similar role in lymphatics. A more in-depth discussion could be provided. Furthermore, could G-proteins be activated by pressure independently of coupled receptors?

We have extensively expanded our discussion of possible mechanosensitive GPCRs in LMCs. We have also cited at least one paper providing evidence for G-proteins being activated by pressure independently of coupled receptors (PMID 28148497), although we believe our added assessment of LMC GPCR expression reveals novel targets to explore for mechanotransduction in LMCs.

(3) Figure 1: Please explain the apparent discrepancy between the described protocol and the experiment displayed in A. Please explain and confirm that the experiments presented in the study followed the protocol described in the Methods section. Add text to Methods.

The example does indeed follow the stated protocol in the “METHODS section” for the sequence and duration of pressure steps. Note that we stated 2 min was typical at each pressure level but for some of the lower pressures 4-5 min were required to get a sufficient number of contractions, as in this case. The diameter and pressure traces at P=8, 10 cm H₂O are not shown in the example so that we could expand the time axis for better resolution.

It can also be inferred from the trace that the F-P curve was built from frequencies obtained when pressure was lowered and increased. Are values at a given pressure identical if recorded when pressures were lowered or increased? Does it influence the relationship?

There is indeed some variability in the response to stepping pressure up vs down because of rate-sensitive effects, as described in one of our previous studies (PMID: 19001046). In the protocol we implemented here, the FREQ at P=3 cm H₂O upon step up from 0.5 cm H₂O was usually slightly higher than the initial FREQ at P=3 cm H₂O. Those two values were averaged together and this is now stated.

The FREQ graph in B displays values up to 10 cm H2O, which are not shown in A.

The data at P=8, 10 cm H₂O were obtained for that same vessel. The complete range is shown to illustrate that the relationship reaches a plateau above 5 cm H₂O.

(4) Figure 1: Is there a need to repeat/duplicate the FREQ graph in B and D?

It has now been replaced with Ejection Fraction.

(5) Figure 19: With respect to point #2, and whether G-proteins are involved/important in the process, given the current state of understanding, that part of the illustration might need to be amended.

The figure has been revised.

Reviewer #3 (Recommendations for the authors):

Below are some suggestions for the authors to consider:

(1) Test specific GPCRs pharmacologically (e.g., AT1R) on wild-type vessels to determine if pressure-induced chronotropy is blunted.

Done, Figure 22.

The authors could also rank Gq/11 receptors based on the expression in LMCs (e.g., qPCR using FACS-sorted cells).

Done, Figure 21.

(2) Residual chronotropy in ANO1 knockout mice: Quantify the remaining frequency-pressure slope in Myh11-CreERᵀ²;Ano1ᶠ/ᶠ vessels and test whether it is sensitive to low-chloride buffer or chloride channel blockers to probe an alternative Cl⁻ conductance.

The role of chloride in lymphatic pacemaking is well appreciated and the reviewer offers some intriguing experiments for future experimentation. The use of low Cl^- solutions was part of the fundamental discovery of lymphatic muscle pacemaking and the role of a calcium-activated chloride channel as the basis for spontaneous transient depolarizations (STDs) by Dirk Van Helden in 1993 who reported that “Lymphatic STDs reversed at potentials near -35 mV (range -40 to -25 mV; n = 4)” and were “suppressed by exposure to low chloride (replaced with sodium isethionate) solution [(> 5 minutes)]”, although more details were not provided. The effect of low Cl^- and lymphatic muscle excitability was followed up in a 2008 paper by von der Weid et al., which showed that STD frequency and amplitude increased acutely (within 2-3 min) after the bath solution was changed to a solution containing 10% of the initial Cl^- concentration (replaced with methane sulphonate). This result likely suggests that the acute reduction of extracellular Cl^- increased the reversal potential for Cl^- and that within 2-3 minutes the intracellular Cl- content had not quite run down. Work by Boedtkjer’s group (Mohanakumar et al. 2018) also demonstrated that human lymphatic pacemaking was also heavily dependent on the bath Cl^- concentration. Using wire myography preparations, they showed that human thoracic rings and mesenteric vessels ceased spontaneous contractions when Cl^- was removed (replaced with aspartate), and that contractions returned upon washout with a normal Cl^- containing solution. In that study, the authors noted that there was some variability in the timing required for contractions to cease with the mesenteric LVs. In some cases, an acute large constriction immediately followed the exchange with Cl^- free solution, which would also point to an acute depolarizing/stimulatory effect of a rapid resetting of E_Cl to a more positive value when external Cl^- was removed acutely removed. A limitation to the approach is the use of wire myographs as opposed to pressurized vessels, as the wire method does not allow the vessels to be stretched in a physiological manner. Similarly, diastolic depolarization was not typically observed by Van Helden and von der Weid, as they used pinned-out tissue sections (required for recording STDs from short vessel sections), which highlights the necessity of using a physiological stretch stimulus.

On the face of it, the complete cessation of contractions reported in the Boedtkjer paper is at odds with our previous and present results using Ani9 and Ano1 smKO and IP₃R1 smKO vessels, which retain some baseline level of pacemaking, albeit largely pressure-independent. Thus, the hypothesis that a separate Cl^- channel may be active and important holds some merit. However, it is worth noting that we do not expect this basal pacemaker drive to be calcium-dependent, as IP₃R1 calcium oscillations are maintained. IP₃R1 smKO vessels, which largely lack subcellular calcium transients in diastole, also have a lower-than-expected contraction frequency compared to Ano1 smKO vessels, although this is in part due to an elongation of the action potential plateau. We also do not think the residual activity is simply due to the inability of inducible Cres to drive complete recombination, as we performed experiments with Ani9-treated Ano1 smKO vessels (Fig. 2B) and did not observe a significant additional reduction in the concentrations at which maximal inhibition is assumed, 2-10 mM (see also Harlow et al. 2025). We do not see a robust presence of Ano2 (TMEM16b) and while Ano6 (TMEM16) is expressed, it is a calcium-activated scramblase with permeability to both cations and Cl^- (Ye et al 2019).

We also have personal experience with Cl^- substitution (replaced to 10 mM by aspartate), originally obtained as part of our work identifying Ano1 as the primary calcium-activated chloride channel, although the inability to ascertain the intracellular Cl^- at any point in time while the bath exchange is occurring significantly limited the interpretability of that data. During review of that manuscript, we were asked to remove the Cl^- substitution experiments but spontaneous contractions still persisted in the majority of the vessels (n=6). Notably this was not a complete removal of Cl^- as was done in Mohanakumar et al. 2018. However, the complete removal of Cl^- is a much more significant intervention than the inhibition of a single channel, likely altering HCO₃^- and Na⁺ flux through the associated Cl^- antiporters and symporters. How Cl^- removal affects LMC intracellular pH regulation via impaired bicarbonate transport also remains unknown. Nor is it well appreciated how Cl^- removal affects mitochondrial function and ROS production in LMCs, which regulate potentially significant confounding pathways related to lymphatic muscle pacemaking. Without control of these significant confounding circumstances, the hypothesis that the baseline depolarization is still a Cl^- channel is far from a certainty.

In Author response image 1, we provide the reviewer with the expression data for other members of the Anoctamin family, the potential Anoctamin-regulating CLCA family, and other documented chloride channels in LMCs (clusters 5-6).

Author response image 1.

(3) I understand this would be a huge undertaking, but aligning scRNAseq with functional vessels using popliteal LMCs or inguinal-axillary LMCs could help rule out potential tissue-specific differences.

We understand the reviewer’s concerns and, owing to the small amount of cells that can be recovered from popliteal vessels, we instead now provide new data using pressure myography of inguinal axillary vessels (IALVs) for the data with positive findings (Gq/11 inhibition) that support our findings in popliteal vessels (Suppl. Fig. 1). This points to a fundamental role for Gaq, Ga11, and possibly Ga14, in signaling upstream of IP₃R1 mobilization and Ano1 activation. We are aware of the length of this manuscript (now further increased in order to respond to the reviewers’ suggestions) and, given the negative data for the vast majority of putative mechanosensory pathways, we have omitted some of the parallel data collected in IALVs. In the majority of our protocols, preliminary experiments using IALVs were tested side by side with popliteal vessels (from Ncx1 smKO, Trpc6^-/-, Trpc3/6^-/-, G12/13 DKO, Piezo1smKO, Trmp4 smKO mice) without detection of a significant difference in phenotype.

eLife Assessment

This study presents important findings on the structure and functional potential of microbial communities along a vertical sediment profile of a salt marsh. The authors examine hypotheses about the burial microbial assembly process and functional complementarity at depth and provide solid evidence to support their claims. Overall, this work will be broadly of interest to microbial ecologists.

Reviewer #1 (Public review):

Summary:

In this revised manuscript, Vineis et al. examined the functional capacity for organic matter decomposition among cooccurring microbes in Spartina patens dominated deep salt marsh sediment cores using genome reconstruction, cooccurrence networks, and genome scale metabolic modeling. They attempted to test a couple of hypotheses that include 1) microbial communities are structured according to sediment depth, 2) deeper communities are functionally streamlined, 3) the community found within deeper sediments is more likely to contain cooccurring members, 4) there is metabolic complimentary enabling sequential decomposition of complex carbon among microbial communities located in deeper sediments, and 5) the population structure is depth-dependent. They identified depth-dependent structure of microbial communities and populations, environmental filtering of microbes with depth, and metabolic complimentary among members of the Bathyarchaeia BA1 subnetwork that possibly enables sequential decomposition of organic carbon in the deep sediment. Overall, the authors have achieved their aims, with the results supporting their main conclusions. The findings of this work will contribute to our understanding of organic matter transformation in salt marsh sediments and, more broadly, microbial metabolism in energy-limited systems.

Strengths:

(1) Two long sediment cores (down to 240 cm deep) were collected in this study, allowing investigation of the less well characterised subsurface microbiome in salt marsh.

(2) A genome-resolved metagenomic approach was employed here, which provides information on both the structure and functional potential of the salt marsh sediment microbiome, which is not possible in commonly performed 16S rRNA-based surveys.

(3) Metabolic complementarity analysis and metatranscriptomics were used to address the likelihood of metabolic handoffs, providing evidence of potentially active microbial interactions.

Weaknesses:

(1) No geochemical data are available to provide context for the genomic analysis here. Without such information, readers cannot even tell whether the surface sediment samples were oxic or anoxic.

(2) A single metagenomic binning tool, CONCOCT, was used in this study, which very likely has resulted in a limited number of MAGs recovered. More (high-quality) MAGs are expected with the use of additional binners and a bin consolidation procedure. A manual bin refining process was used to improve the quality of the bins obtained though.

Reviewer #2 (Public review):

This work provides a detailed metabolic reconstruction of sediment microbiomes along a depth profile in a Spartina patens salt marsh in Massachusetts, USA. Using a combination of genome reconstruction, co-occurrence network analysis, and metabolic profiling, the authors describe the metabolic potential of co-occurring microbial consortia in understudied deep sediments.

Major strengths of this study include the detailed metagenomic and metatranscriptomic characterization of the understudied deep marsh sediments. The authors recovered genomes representing a substantial portion of the deep sediment microbiome (up to ~60%) and provided an initial explanation of pathways related to the potential for organic carbon decomposition in this environment. Of particular interest is the genomic capability of the deep sediment microbiome to process complex organic compounds, highlighting the need for a collaborative consortium to carry out their decomposition. Improved understanding of the microbial transformation of deep sediment organic carbon in blue carbon ecosystems is vital to better understand the fate of this large carbon pool in the face of climate change.

After assessing a revised version of the manuscript, I only have one methodological comment that readers may find useful as going through the manuscript:

The authors normalize the relative abundance of MAGs by dividing the reads mapping to a MAG by the reads mapping to their whole set of recovered MAGs. They provide a rationale for following this approach in the Methods section. However, I still argue that these values are misleading due to differences in read recruitment levels across samples (and importantly, across depths). In particular, the relative abundance of MAGs in the surface layers (0-40 cm) is overestimated, since the higher diversity in these samples recruited fewer reads due to poor assembly in samples with high sequence-space diversity (shown in Supp. Fig 2). A better normalization approach would be to divide MAG abundance by genome equivalents (e.g., using MicrobeCensus) or any other approach leveraging the abundance of universal single-copy genes in prokaryotes. These approaches are assembly-independent and leverage existing databases of universal single-copy genes. That being said, these methodological limitations do not seem to greatly impact the biological findings of this manuscript.

All other comments have been addressed.

Author response:

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

eLife Assessment:

This study provides a valuable genome-centric characterization of microbial communities across deep sediment cores from a Spartina patens salt marsh. The study provides claims on the metabolic capabilities of the deep sediment microbiome as well as on a burial microbial assembly process and functional complementarity at depth. However, some of these claims remain incomplete and would benefit from further supporting evidence. Overall, this work will be of interest to microbial ecologists working on wetlands.

We appreciate all of the efforts of the reviewers and editors. Additional analysis and extensive edits were made to the original manuscript to address the comments of both reviewers and editor. We believe this effort has significantly improved the manuscript.

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this manuscript, Vineis et al. examined the structure and functional potential of microbial communities along a vertical sediment profile of a salt marsh, using a genome-centric metagenomic approach. They attempted to test whether (1) the microbial communities within dynamic upper layers contain genomes with diverse functional potential, (2) the energy limited deeper sediments contain microbial consortia assembled to metabolise complex carbon, and (3) microbial compositional changes in the low energy sediments mirror the burial processes observed in marine environments with similar energetic limitations. Results revealed a core microbial consortia that contains a collective metabolic potential for complex carbon and aromatics degradation, suggesting putative syntrophic interactions. Besides, the recovery of MAGs assembled independently from multiple depths in the same core and the consistent relative abundance structure of MAGs within co-occurrence network modules together suggest burial process as a likely mechanism for microbial assembly.

Strengths:

(1) Two long sediment cores (down to 240 cm deep) were collected in this study, allowing investigation of the less well characterised subsurface microbiome in salt marsh.

(2) A genome-centric metagenomic approach was employed here, which provides information on both the structure and functional potential of the salt marsh sediment microbiome, which is not possible in commonly performed 16S rRNA-based surveys.

Weaknesses:

(1) In both the abstract and conclusion, the authors claimed that results from this study provide a "mechanistic understanding" of the assembly and distribution of the microbial communities in salt marsh sediment (P2, L31 and P35, L645-649). However, both claims are speculative and not supported by solid evidence. Firstly, the genomic data presented in this study and supplementary physical properties of sediments in the broader area are not enough to make a solid claim (that appears in the title) on microbial assembly being governed by a burial process. Alternative explanations include residual bioturbation, slow porewater advection, etc. Therefore, this remains an interesting hypothesis unless additional evidence is provided to rule out the alternative explanations. Similarly, the claim on the detailed syntrophic interactions among members within a co-occurrence network module (e.g. P36, L649-652) is purely speculative and warrants functional validation experiments to prove.

The reviewer makes two points in section 1 of their review, and we have addressed each point as follows: 1) We have removed the term “mechanistic understanding” from the manuscript and instead focused on the co-occurring group of microbes and the evidence for overall community and population structure with depth. Statistical analyses were added to the manuscript including a test for the influence of depth on the overall community structure (Fig. 3) and sample-specific single nucleotide variants (SNVs) within a collection of MAGs that were abundant and prevalent throughout the subsurface (Fig. 7). The text now contains additional text in the discussion (lines 702-720 and 843-845) that addresses alternative explanations of our observations including the potential for advection and residual bioturbation. 2) We agree that to prove syntrophy, functional validation through experimentation is required and we have added a statement to this effect on line 858-864.

(2) A major aim of this work was to study complex carbon degradation. However, neither CAZymes, the first-line carbon degradation enzymes, nor peptidases, which can be important contributors to carbon degradation at depth, was examined here. METABOLIC, which the authors used for functional annotation of MAGs, by default generates peptidases outputs and can be easily integrated here.

We expanded our analysis to include CAZymes and peptidases including an analysis of the number of peptidase pathways among our network modules (Table S3, Fig. S6), and a supplemental table of glycoside hydrolase and polysaccharide lyase (Table S2). Methods regarding the CAZyme and peptidase analysis are included on lines 282-296, results on 490-509, and 581-588.

(3) No geochemical data is available to provide context for the genomic analysis here. Without such information, readers cannot even tell whether the surface sediment samples were oxic or anoxic. A reference to a PhD thesis is provided (P6, L126) but it would be most helpful to extract relevant data from there and provide as a supplementary table.

Oxygen concentrations were not measured as part of that study. Although oxygen concentrations are heterogeneous, due to bioturbation, and radial oxygen loss from roots in marsh systems similar to ours they are generally below detection at the shallowest depth examined in our study. We have added additional citations relevant to this point on lines 747-753.

(4) A single metagenomic binning tool, CONCOCT, was used in this study, which very likely has resulted in a limited number of MAGs recovered. More (high-quality) MAGs are expected with the use of additional binners and a bin consolidation procedure.

We agree that multiple binning tools can potentially identify more bins, contigs can be mistakenly binned when using purely automated processes. To sidestep the chance of including erroneous contigs in our collection, we chose to invest a large amount of time and effort required to bin manually. CONCOCT was used as an initial guide to identify some of the most easily reconstructed MAGs but all contigs within MAGs were subjected to visual inspection of the sequencing coverage profile over several samples, sequence composition congruency, and real-time completion and contamination estimation as contigs were manually added and/or subtracted from the collection in each bin. This approach is not more widely applied because of the expertise and time required to manually reconstruct MAGs from each sample individually. Many researchers find that human involvement is often required to improve the accuracy of automated binning tools, which has given rise to several tools in addition to Anvi’o, BinaRena, ICoVeR, and ggKBase. To elaborate on this procedure, we made a video tutorial of our approach to binning manual-binning-approach. An example of the manual bin refining process can be found here (merenlab-MAG-refinement)

(5) Several terminologies are misleading here. Firstly, the term "co-occurring" or "co-located" microbes or MAGs (e.g. P1, L19 and P31, L537) can be misleading as it could imply a close spatial relationship. However, co-occurrence networks rely on correlations of (relative) abundance and show statistical associations instead of direct spatial or physical relationships. I would suggest alternative names such as co-abundant or statistically associated microbes.

We have added text to address the important point raised by the reviewer regarding the term “cooccurrence”. We added additional language in the methods section to define co-occurrence and to make clear that the term should not be interpreted to infer a direct spatial or physical relationship (lines 121-123). However, the term “cooccurrence” is commonly used to describe the significant correlations identified molecular ecological networks and we think the introduction of additional terminology such as “statistically associated” could lead to additional confusion.

Secondly, the term "persistent conversion of soil organic carbon" (P36, L654) in the conclusion is also misleading as it implies an active process, which cannot be tested without metatranscriptomics or metaproteomics data.

We agree and thus we added metatranscriptomic data to the manuscript to assess the activity of the microbes in the subnetwork of commonly co-occurring MAGs within the subsurface. Our results indicate that the pathways for complex carbon are both present and active at the time of sampling. The methods section describing this additional analysis is included on lines 155-168 and 314-324 and we created a new figure to communicate the present and active MAGs in the Bathyarchaeia BA1 subnetwork (Fig. 7). Even with the additional metatranscriptome data, we agree that the term “persistent conversion of soil organic carbon” is still not supported by our data because we sampled at a single timepoint and we have removed this text from the manuscript.

(6) Based on a NMDS plot of KEGG IDs (Figure 4B), the authors claimed that the functional potential among MAGs in modules 1, 2 and 7 was very similar (P18, L346). However, the dispersions of modules 1 and 2 were just too large. A proper statistical test, such as PERMANOVA, should be used to support the claim.

We replaced the analysis of the KEGG modules with a more detailed characterization of the functions identified in the MAGs within each of the modules (Fig. 5) (including the CAZy and peptidase analysis described above) (Table S2 and S4). This revised approach provides more resolution on the higher-order functions that are differentially detected among the modules. This allows for greater clarity on the functions that are specific to the MAGs within each module.

(7) Genome-scale metabolic networks was analysed using Metag2Metabo (M2M) and results were discussed in detail (P26, L453-466). However, the source data should be provided in a supplementary table to show what metabolites are producible by which MAGs.

The M2M analysis generates a map of thousands of genes and reactions for each MAG which is stored in an XML-formatted file. Files for each MAG are available upon request so the metabolic pathways can be explored in Fluxer (https://fluxer.umbc.edu/) or Escher (https://escher.github.io/). The metabolites produced through complimentary reactions within the Bathyarchaeia BA1 subnetwork are included in Table S6 and the complimentary gene content for the Benzoyl-CoA pathway is shown in Table S7.

Reviewer #2 (Public review):

This work provides a detailed metabolic reconstruction of sediment microbiomes along a depth profile in a Spartina patens salt marsh in Massachusetts, USA. Using a combination of genome reconstruction, co-occurrence network analysis, and metabolic profiling, the authors describe the metabolic potential of co-occurring microbial consortia in understudied deep sediments.

Major strengths of this study include the detailed metagenomic characterization of the understudied deep marsh sediments. The authors recovered genomes representing a substantial portion of the deep sediment microbiome (up to ~60%) and provided an initial explanation of pathways related to the potential for organic carbon decomposition in this environment. Of particular interest is the capability of the deep sediment microbiome to process aromatic organic compounds, highlighting the need for a collaborative consortium to carry out their decomposition. Improved understanding of the microbial transformation of deep sediment organic carbon in blue carbon ecosystems is vital to better understand the fate of this large carbon pool in the face of climate change.

However, I have a few concerns in the interpretation of the results, and in the case of the surface sediments there is a lack of strong evidence in my opinion.

(1) A stronger ecological interpretation is needed regarding the meaning of the co-occurrence network analysis. The authors correctly note that their analysis identifies groups of co-occurring genomes, which may indicate shared niche space, not necessarily interspecific ecological interactions (as the authors imply for instance in lines 423-425). When performing network analysis using samples from the entire sediment profile (0-240 cm), they identified consortia that co-vary in relative abundance along the depth gradient most likely because of shared environmental filtering forces, such as changes in redox potential and sediment chemistry. Supplementary Figure S4 showing that different modules have distinct abundance distributions along the sediment profile supports this idea. Being that the case, I would like the authors to define the ecological significance of the "connector hub". Is it merely taxa that is prevalent in the whole sediment profile? Since the modules are physically separated (in different sediment depth layers), they are not really interacting between each other. As it stands, it is not clear why the authors decide to study connector hubs in greater detail, along with their subnetworks.

These are all very good points raised by the reviewer and we have taken the following steps to clarify the role of connector nodes. Additional text regarding the environmental role of a connector node is included on lines 273-278, 440-442. We added additional text regarding our decision to study the Bathyarchaeia BA1 MAGs that were classified as connectors in greater detail (lines 514-534).

(2) I question if the lack of network modules in the surface sediment is really a consequence of non-significant interspecific ecological interactions and not the result of methodological biases. The low MAG recovery and thus short read recruitment in surface-level metagenomes may hinder the ability of the authors to identify co-varying microorganisms in the surface sediment. The high diversity of the surface sediment prevents proper assembly of the surface microbiome. I would also argue that as redox potential declines sharply in salt marsh sediments just below the root surface, the microbial community in the first few centimeter's changes rapidly and is significantly different from the more stable deep sediment microbiome. Due to the sampling design, the study has less representation of the surface layer (only 0-30 cm, while the cores extend down to 240 cm). Grouping sediment microbiomes by depth based on similarity in their sequence space (e.g., Mash) or functional profile (e.g., KEGG annotation) before performing network analysis could help to better infer ecological relationships within the distinct ecological niches of the marsh sediment profile, rather than performing a single network analysis of all samples combined.

We agree with rationale that the low MAG recovery from surface sediments inhibits our ability to build a reliable network that would capture interactions within this region of the sediment profile. While the reviewer’s idea of grouping samples based on functional similarity or depth is and interesting idea, this decreases the number of samples used for estimating co-occurrence. Because we had few MAGs derived from surface sediments, it is unlikely that this recommended approach would yield additional connections. Additional analysis of surface sediments is certainly warranted in order to tease apart the interactions and their relevance to the biogeochemistry of the sediment. The reviewer comments may be useful to others attempting to identifying niche spaces and connections in the surface. See lines 775-783.

(3) Normalizing the relative abundance of MAGs by dividing by the total reads mapping to a particular sample can be misleading due to differences in recruitment levels across samples (and depths). A better approach would be to normalize by metagenome library size, or preferably by genome equivalents (e.g., using MicrobeCensus) or a similar approach.

We decided to normalize to the number of reads mapped to our non-redundant collection of MAGs instead of to the total number of sequences in each sample for several reasons. 1.) We do not know what the proportion of unmapped reads represents and could include virus and eukaryotic life. Comparing the amount of microbial DNA to the eukaryotic fraction can be misleading because changes in the relative abundance from these sources would skew the microbial relative abundance. In human microbiome studies, subtraction of human reads is often carried out prior to calculating relative abundance, but we lack any such reference in our system. We added text to explain this in the methods (lines 219-232). 2.) We are not comparing the total abundance of MAGs across samples or comparing individual genes across samples, so tools such as MicrobeCensus would not offer an improved analysis. Additionally, many of these tools rely on reference genomes within the workflow which is challenging in these sediments due to the many novel taxa. There are challenges with any relative abundance calculation and clear explanation of our read mapping strategy and Fig. S2 allows the reader to see the potential bias.

Recommendations for the authors:

Reviewing Editor Comments:

It appears that the main issue is that some of the claims in the paper are not adequately justified. You can find the detailed assessment of the two reviewers in their reviews, and their recommendations are listed below. We encourage you to examine these comments and incorporate the recommendations in your manuscript. In particular, after consulting the reviewers, we encourage you to prioritize the following:

(1) Addressing alternative hypotheses.

(2) Improving MAG binning and assembly.

(3) Revising statistical analyses.

(4) Including CAZymes and peptidases in the analysis.

(5) Revising the terminology and interpretation (e.g. regarding co-occurrence).

Reviewer #1 (Recommendations for the authors):

(1) Title: consider shortening it.

We agree and shortened the title

(2) P3, L49-52: incomplete sentence.

The corrected sentence in on lines 45-47.

(3) P3, L54: consider adding a citation for "blue carbon stocks" for a broad readership.

We added a citation. Line 54.

(4) P7, panel A: consider adding coordinate axes; panel B: would be helpful to add more details on the tools (e.g. CONCOCT for binning), key parameters (e.g. dereplication at 95% ANI), and/or key statistics used (e.g. the number of high-quality and medium-quality MAGs).

We added additional details to figure 1.

(5) P8, L131 (and L148): define MAG at its first use. Are they bins with >50% completeness and <10 contamination MAGs?

MAGs are first defined on line 24. Bins are a minimum of 50% complete and maximum 10% redundant (lines 200-205).

(6) P9, L168: the default setting for dRep uses -comp 75, which won't result in MAGs with a completeness <75%; however, some MAGs in Table S1 have a completeness <75%.

The methods on this point are clarified on lines 191-206.

(7) P10, L173: by the "75% threshold", do you mean for completeness/completion?

We have clarified that sentence to specify “completion” on line 203.

(8) P10, L175: provide the version number for GTDB used.

The version is now provided in the text on line 205.

(9) P10, L185: information in the GitHub repository is incomplete; e.g. the code for running CONCOCT binning is missing.

The GitHub is now up to date including a link to a video explaining how binning and MAG refining were conducted for one of the samples.

(10) P14, L268: how many MAGs are novel at each taxonomic ranks? Table S1 does not include genus level data.

We have included additional text regarding the novelty of MAGs on line 370-375. Table S1 now has genus-level classification.

(11) P14, L271, 274, and 276: please provide the exact numbers and/or percentages.

The percentages are now reported on lines 372-375.

(12) P14, L272: however, Figure 2 only shows phylum data.

We have reworked this paragraph and corrected this error (lines 384-406)

(13) P14, L281: deep instead of shallow?

The paragraph was rewritten and this point now appears on lines 392-394.

(14) P16, L298-302: would be helpful to provide a supplementary table showing these values.

These values were added to table S1.

(15) P16, L314: aren't they 30% and 50%, respectively?

This error was corrected. See lines 378-379.

(16) P18, L339: five Dehalococcoidales and no GIF9 MAGs?

We revised this section and the details are now included on lines 438-440.

(17) P21, L385: module 5 instead of 4?

This section was revised and the corrected details can be found on lines 460-462.

(18) P21, L386: Figure S4 is irrelevant here.

This section was revised and the corrected details can be found on lines 460-461.

(19) P22, L394: Figure 5 instead?

This section was removed during revision

(20) P24, L413: to me, these MAGs were present along the entire core instead.

The section regarding “bimodal” MAGs was removed during revision.

(21) P24, L411: there are multiple typos in the scientific names in Figure S6.

We have removed this supplemental.

(22) P25, L430: Fig. 6A instead?

The correction of this error is on line 525 and references figure S7A.

(23) P25, L440: module 1?

The GIF9 classification was replaced with the family-level classification AB-539-J10 and we have corrected any error assigning these MAGs to the incorrect module. The original sentence containing this error was revised, but the section including these results can be found on lines 514-534.

(24) P28, L501: these pathways were present in not even half of the MAGs.

The details of selenate and arsenate are now presented in figure 5.

(25) P30, L508: also provide a key for the size of nodes.

The original figure 6 is now figure S7 and was revised to include a key for genome size in the network.

(26) P31, L553: please provide a supplementary table showing the metabolic potential of each MAG to support this claim.

We created figure 5 to address this comment in addition to table S9. The associated text can be found on lines 739-745.

(27) P32, L564: I would suggest changing the word "consistency" to "correlated pattern" for clarity.

We are trying to describe the correlated pattern as being consistent across multiple samples. See lines 730-732 for the revised use of this term.

(28) P32, L567: also Bacteroidales.

We have clarified the genomic content of MAGs containing benzoyl-CoA reductase subunits and production of benzoyl-CoA from phenol on lines 630-635

(29) P32, L575: please tone down this statement; some MAGs are capable of both (Figure 6C).

We have toned down the statement and discussion of the metabolic handoffs and syntrophy. See lines 801-817.

(30) P33, L582-585: aromatics are not products of fermentation so their decomposition doesn't "follow" the initial fermentation.

We revised the statement on lines 815-817.

Reviewer #2 (Recommendations for the authors):

(1) Performing co-assemblies could help improve the recovery of microorganisms from the surface sediment layer.

We have responded to this comment in response to reviewer 1, point #4

We agree that multiple binning tools can potentially identify more bins, contigs can be mistakenly binned when using purely automated processes. To sidestep the chance of including erroneous contigs in our collection, we chose to invest a large amount of time and effort required to bin manually. CONCOCT was used as an initial guide to identify some of the most easily reconstructed MAGs but all contigs within MAGs were subjected to visual inspection of the sequencing coverage profile over several samples, sequence composition congruency, and real-time completion and contamination estimation as contigs were manually added and/or subtracted from the collection in each bin. This approach is not more widely applied because of the expertise and time required to manually reconstruct MAGs from each sample individually. Many researchers find that human involvement is often required to improve the accuracy of automated binning tools, which has given rise to several tools in addition to Anvi’o, BinaRena, ICoVeR, and ggKBase. To elaborate on this procedure, we made a video tutorial of our approach to binning manual-binning-approach. An example of the manual bin refining process can be found here (merenlab-MAG-refinement)

(2) Performing diversity analysis of metabolic pathways in surface vs deep sediments could help prove the idea of more versatility in the surface marsh sediment.

To address this comment, we created a figure detailing the proportion of MAGs containing central microbial functions among the modules with distinct depth distributions (Fig. 5), analyzed the number of complete pathways (Fig. 4, Fig. S5) and differences in CAZyme and peptidase genes (Fig. S6).

(3) To strengthen the idea that burial is key in deep marine sediments, looking at the abundance of mobility structures (i.e., flagella) along the sediment profile could help make a case.

This is an interesting idea, however, many of the same genes used for motility are also used to form biofilms. This analysis could be an interesting aspect of these sediments that warrants an independent study.

(4) Lines 60-63, but oxygen leaked from macrophytes could also exacerbate microbial respiration rates.

This point is clarified on lines 59-64.

(5) Line 49-52: Check this sentence. It does not read properly. "2this" -> "this".

We corrected this error on lines 45-47

(6) Line 173: I assume >75% is only for completion, not contamination, right? I would rephrase to make this clear.

This point is clarified on lines 201-206.

(7) Line 179, what is Bacteria_71? Citation is needed.

A citation has been added for this single-copy gene collection on line 209.

(8) Line 201: When performing log transformations, how are 0s, which are highly abundant in microbiome data, handled?

The details of matrix transformation are clarified on line 252-254. However, unlike amplicon data (ASVs and OTUs), the read recruitment data contained very few zeros in the dataset.

(9) Line 417: The Pseudolabrys MAG most likely utilizes the dsr and apr genes in reverse, as is typical for other Alphaproteobacteria, not for sulfate reduction. Additionally, sat can also function in assimilatory sulfate reduction.

Yes, we agree with the reviewer’s assessment of Pseudolabrys. However, during revision, the section regarding MAGs with bimodal distribution was removed.

(10) Line 414 uppercase mags.

During revision, the section containing this error was removed.

(11) Line 605: "examined".

This error was corrected on line 682.

(12) Line 610: Are independent MAGs from adjacent sediment layers more similar to each other than those from layers that are farther apart? Demonstrating such a pattern could provide stronger evidence that burial acts as a major force shaping the assembly of the deep sediment microbiome.

We conducted a community-level analysis based on MAG relative abundance (Fig. 3) and a population analysis of SNVs for six abundant MAGs to test for an effect of depth (Fig. 8). While we no longer focus our analysis on a mechanistic explanation of the depth-dependent patterns, with the exception of lines 769-772. As requested by the reviewers we discuss the observed patterns in light of several other potential mechanisms (lines 838-845)

(13) While acknowledging that selection pressure following burial is a major driver of microbial community assembly in marine sediments, I would ask whether other biogeomorphic forces could also influence the assembly of sediment microbiomes over millennial timescales, such as bioturbation or sediment redistribution within the salt marsh environment.

Based on our community and population-level analysis, we suggest that bioturbation and sediment redistribution are likely to shape the upper 40 cm. Beyond this depth, we did not find evidence for major restructuring events. This topic is discussed on lines 830-845, but the formation of distinct relative abundance peaks for the modules primarily found in deeper sediments remains an outstanding question (lines 772-774).

eLife Assessment

This important submission from Ambler and colleagues brings new insights into how torpor conditions may confer resilience in cases of cardioprotection. The study is backed by solid evidence and represents a unique interoceptive mechanism of interest.

Reviewer #1 (Public review):

[Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. After previous evaluation by three competent reviewers, the authors have provided an extensive response to their criticisms and recommendations. I believe that authors have done an adequate job in addressing the initial reviews and recommend that the paper be processed for the Version of Record without further work at this time.]

Summary:

Torpor can be induced by chemogenetic activation of the medial preoptic area. This activation leads to protection from myocardial infarction in an isolated heart preparation despite normalization of the ambient temperature, thus, in principle, uncoupling hypothermia from torpor-induced neuroprotection. Putative pathways of protection are suggested by proteomic studies.

Strengths:

(1) Elegant strategy for inducing torpor in rats.

(2) Appropriate controls for verifying the neuron transducer.

(3) Cardiac protection is significant and appears independent of hypothermia.

(4) Interesting omic strategy to begin to find established and novel pathways mediating organ autonomous torpor-induced protection.

Reviewer #2 (Public review):

Summary:

Elley and colleagues induced a synthetic torpor-like state in rats (a non-hibernating species) by chemogenetically activating neurons in the medial preoptic area of the hypothalamus. They show that this state substantially reduced cardiac infarct size in an ex vivo ischemia-reperfusion model. They further report that protection persisted when ambient temperature was raised to prevent hypothermia and used exploratory phosphoproteomics to identify candidate cardioprotective signaling pathways.

Strengths:

This is the first demonstration that a torpor-like state is cardioprotective in a species that does not naturally enter torpor, which meaningfully advances the potential clinical utility of synthetic torpor. The experimental design is logical, and the controls are generally appropriate. The characterisation of the responsible neuronal population using ISH against QPLOT markers adds mechanistic depth and supports the cross-species conservation argument. The phosphoproteomic analysis, though exploratory, generates plausible and biologically coherent hypotheses grounded in the hibernation literature.

Reviewer #3 (Public review):

Summary:

The manuscript by Elley and colleagues describes experiments on the effects of synthetic torpor on ex vivo heart ischemia. The key aspect of the study was the use of viral-vector mediated manipulation of the hypothalamic medial preoptic area (MPA) in rats. They used AAV-CaMKIIa-hM3D(Gq). The authors report that chemogenetic activation of the MPA prior to an ex vivo heart ischemia-reperfusion insult induces cardio protection against infarct size that is independent of prior in vivo hypothermia. Phosphoproteomic analysis of cardiac tissue suggested changes in cell survival and death pathways.

Strengths:

This study has important strengths. The idea is novel. The experimental design is appropriately rationalized and fascinating. The manuscript is written and presented concisely.

Reviewer #4 (Public review):

From the Reviewing Editor:

The core finding and implication of this article is that reduced body temperature and hibernation conditions may have value in understanding basic physiology and during medical conditions. The subject of this submission should be considered as a new interoception topic, which is of current high interest that promises to link CNS regulatory mechanisms to the cardiovascular system and vice versa. But there is also substantial relevance to sleep, the immune system, cancer and protective mechanisms to other organ systems after injury and disease.

This submission demonstrates that a torpor-like state can be related to a cardioprotective effect in a model of ischemia-reperfusion injury. Torpor is characterized by hyperthermia and reduced heart rate and oxygen consumption. As such, it has complicated and unpredictable physiological impacts upon a variety of organ systems. The study contends that the effect of torpor is mediated by the medial preoptic region of the hypothalamus. Protein analysis in a rat model suggests the effects of a torpor-like state may involve detectable effects upon intrinsic cell death and stress related pathways. A further understanding the mechanisms of torpor research will likely reveal new protein and genetic targets that are conserved targets. The authors propose a hypothesis that torpor conditions may reflect an evolutionary mechanism that will have implications for considering application to clinical trials and therapeutic protocols.

This is a provocative study that raises many interesting questions and suggestions for further study. For the most part, the authors acknowledge that there are certainly many future experiments that should be carried out on the basis of the results from the current study. Notwithstanding the need for more experimentation to extend our understanding of torpor, the authors respond to the reviews with additional supplementary data and the recognition of important caveats for alternative brain and physiological mechanisms in play.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Torpor can be induced by chemogenetic activation of the medial preoptic area. This activation leads to protection from myocardial infarction in an isolated heart preparation despite normalization of the ambient temperature, thus, in principle, uncoupling hypothermia from torpor-induced neuroprotection. Putative pathways of protection are suggested by proteomic studies.

Strengths:

(1) Elegant strategy for inducing torpor in rats.

(2) Appropriate controls for verifying the neuron transducer.

(3) Cardiac protection is significant and appears independent of hypothermia.

(4) Interesting omic strategy to begin to find established and novel pathways mediating organ autonomous torpor-induced protection.

We thank the reviewer for their positive feedback and review of our study.

Weaknesses:

(1) The study would benefit from using inhibitory chemogenetics of the same neurons to demonstrate that this might make cardiac response to ischaemia worse.

This is an interesting experiment proposal and is something we will consider for the future. It is not immediately clear to us whether the hypothesis would be that inhibiting these neurons would make cardiac ischaemia worse, or that it would be similar to the control groups as we would not be inducing synthetic torpor.

(2) Infecting an area of the brain not known to be involved in torpor would be a useful control.

We chose to infect neurons in the same brain region without the presence of the HM3DGq DREADD construct as our control. This is a well-established control for this type of experiment.

(3) In vivo cardio protection seems essential as the validation of the strategy requires support that is in the intact animal.

We agree that an in-vivo model is important to establish whether the same cardioprotection occurs in the intact animal and whether it is observed when synthetic torpor is induced following an ischaemic injury. Our data suggests that synthetic torpor pre-conditions the cardiac tissue, prior to isolation of the heart, as the protection remains post removal of the heart and without any ongoing in-vivo systemic signal/drive. This provides some evidence that this may be used as a pre-conditioning model for organ protection.

(4) The assumption that the positive effects of torpor are mediated via a phosphoproteomic change rather than a translational or transcriptional control mechanism is not established.

Cardiac tissue for the proteomics experiment was collected following 90 minutes of synthetic torpor. This time point was selected for several reasons, including: in-vivo body temperature and heart rate reach their nadir at this time point, and we collected the hearts for the Langendorff ischaemia-reperfusion injury data set at 90 minutes and wanted to keep this consistent.

While we tested for both total protein and phosphoprotein changes, the only significant differences occurred in the phosphoprotein levels. At 90 minutes, the predominant mechanisms for cellular responses are phosphorylation changes – somewhat too early for changes in transcription or translation. In future, it would be of interest to look at later time points to determine whether changes have occurred at the transcription/translation level.

(5) A 40 percent reduction in infarct size may work for genetically identical rats with no co-morbidities, but is unlikely to be significant enough to weather the variability that emerges in humans because of these differences and more. The question is not what the mechanism is, but how do we make it more robust? Overall, this is at best a preliminary data set that requires more experiments to deliver on its immense promise.

We agree with the reviewer that these are preliminary findings, as this is the first demonstration of a torpor-like state in a species that does not naturally enter torpor is protective. However, understanding the mechanisms by which the cardioprotection occurs is going to be important if we want to replicate this state in humans and to understand whether this is specific to cardiac tissue or if it extends to multiple different organs. Furthermore, by understanding the mechanisms, this may reveal avenues that would allow us to enhance these protective effects and make the method more robust.

Reviewer #2 (Public review):

Summary:

Elley and colleagues induced a synthetic torpor-like state in rats (a non-hibernating species) by chemogenetically activating neurons in the medial preoptic area of the hypothalamus. They show that this state substantially reduced cardiac infarct size in an ex vivo ischaemia-reperfusion model. They further report that protection persisted when ambient temperature was raised to prevent hypothermia, and used exploratory phosphoproteomics to identify candidate cardioprotective signaling pathways.

Strengths:

This is the first demonstration that a torpor-like state is cardioprotective in a species that does not naturally enter torpor, which meaningfully advances the potential clinical utility of synthetic torpor. The experimental design is logical, and the controls are generally appropriate. The characterisation of the responsible neuronal population using ISH against QPLOT markers adds mechanistic depth and supports the cross-species conservation argument. The phosphoproteomic analysis, though exploratory, generates plausible and biologically coherent hypotheses grounded in the hibernation literature.

We thank the reviewer for their positive assessment of our work.

Weaknesses:

The primary weakness is that the central conclusion - that hypothermia is not necessary for cardioprotection - exceeds the evidence. The thermoneutral groups were not demonstrably normothermic (36.4 vs 37.05{degree sign}C, p=0.44 with n=6), core temperature telemetry was absent in the majority of control animals contributing to the infarct endpoint, and the decisive test, i.e., a correlation between individual nadir temperature and infarct size, was never performed. Additional weaknesses include the absence of sex-stratified analysis despite known estrogenic contributions to torpor

The core body temperature in the thermoneutral synthetic torpor animals was ~ 0.6°C cooler than the control animals housed at room temperature, which was not significantly different. A 0.6°C body temperature reduction is akin that the changes observed around a sleep-wake cycle and we consider it to be biologically implausible that this degree of temperature reduction in the ‘thermoneutral synthetic torpor’ group is sufficient to account for the observed cardioprotection. We have since added a graph and correlation analysis of individual surface temperature against infarct size in the synthetic torpor animals (see supplementary figure 5). This demonstrates no correlation between nadir temperature during synthetic torpor and infarct size, supporting the hypothesis that the protective effect is not driven by cooling.

Reviewer #3 (Public review):

Summary:

The manuscript by Elley and colleagues describes experiments on the effects of synthetic torpor on ex vivo heart ischaemia. The key aspect of the study was the use of viral-vector mediated manipulation of the hypothalamic medial preoptic area (MPA) in rats. They used AAV-CaMKIIa-hM3D(Gq). The authors report that chemogenetic activation of the MPA prior to an ex vivo heart ischaemia-reperfusion insult induces cardio protection against infarct size that is independent of prior in vivo hypothermia. Phosphoproteomic analysis of cardiac tissue suggested changes in cell survival and death pathways.

Strengths:

This study has important strengths. The idea is novel. The experimental design is appropriately rationalized and fascinating. The manuscript is written and presented concisely.

We thank the reviewer for their positive evaluation of our paper.

Weaknesses:

The study has important weaknesses in the experimental design and validation of the model.

(1) The study is based on the use of a DREADD-designed viral vector (AAV-CaMKIIa-hM3D(Gq) -mCherry) that is activated by 2 mg/kg IP injection of CNO. The rationale is to putatively activate the MPA. The authors show no evidence for chemogenetic activation of neurons in the MPA. This could be done using a variety of different approaches, even phosphoproteomics.

Chemogenetics is a very well-established technique and we demonstrate histological data showing expression of the DREADD in the MPA region and large physiological changes in body temperature, heart rate and oxygen consumption that we feel is sufficient to demonstrate that these neurons are activated. The field of chemogenetics is sufficiently established that it is not common practice to demonstrate neuronal activation in the context of large induced physiological responses.

(2) The stereotaxic injections are difficult to precisely and locally place, particularly bilaterally. Figure 2F is only a schematic. It would be better to show actual low magnification brain sections (bregma +0.12 to -0.48) from a representative rat to show the placement of the AAV.

Figure 2F is a schematic produced from imaging data, showing the regions within the MPA that expressed the AAV across all animals. We mapped the areas of viral transfection in each animal and then mapped this onto the brain atlas demonstrating which regions were consistently transduced across all animals. We have added a supplementary figure (supplementary figure 2) showing overlaid sections from 6 animals that underwent detailed vector mapping, which shows the full extent of vector transduction in these animals.

(3) The control rats were injected with AAV-CaMKIIa-EGFP. Why was EGFP used instead of mCherry for the control?

We prioritised using the same serotype and promoter for our experiments and would not anticipate the colour of fluorophore to have an effect on experimental results.

(4) Ideally, a mutant non-activatable variant of AAV-CaMKIIa-hM3D(Gq) should have been used for a better control.

We thank the reviewer for this suggestion, however, to the best of our knowledge, a viral vector supplier Addgene, does not stock this variant of virus, nor have we seen one reported in the literature.

(5) The authors should comment on whether there is any neurotoxicity in the MPA associated with the forced AAV expression of hM3D-Gq.

This is not something we have assessed, however, administration of AAV into the CNS is not associated with a robust inflammatory response or neurotoxicity [1]. We used a viral titre of 1.7x10¹³ viral genome copies per ml and injected 200 nL at each POA site, which is within reported guidelines by other users of the Addgene construct.

(1) Mastakov MY, Baer K, Symes CW, Leichtlein CB, Kotin RM, During MJ. Immunological aspects of recombinant adeno-associated virus delivery to the mammalian brain. J Virol. 2002;76(16):8446-8454. doi:10.1128/jvi.76.16.8446-8454.2002

(6) Is there any inflammatory pathology seen in the MPA with AAV transduction?

This is not something that we assessed.

(7) There are no experiments to show that the systemic torpor is specifically associated with the MPA region. Experiments should be done with injections of AAV-CaMKIIa-hM3D(Gq)-mCherry placed in other brain regions, for example, the nearby nucleus accumbens.

The medial preoptic area is widely believed to be key for triggering torpor. Data from our lab and other groups has demonstrated that neurons within the preoptic area are sufficient to trigger torpor entry in mice. However, we concede that it is possible that other brain regions form part of the same circuit and that activation of the circuit at this / these points could also trigger a synthetic torpor bout. Should this be the case, our key message would remain the same: animals that do not naturally enter torpor can be induced into a similar state through targeted CNS activation, and this state is cardioprotective.

(8) The mapping of the distribution of neurons responsible for synthetic torpor is not mechanistic enough and is not directly to the point. While excitatory and inhibitory markers are examined, a more interesting and deeper approach would have been to use glutamate receptor antagonists to manipulate the torpor response.

We thank the reviewer for this interesting suggestion. This type of experiment is not commonly performed and would be extremely technically challenging. A more common approach would be to genetically access glutamatergic neurons in the region and silence them with chemo or optogenetics.

(9) The ischaemia and reperfusion aspects of the Langendorff method need to be clarified. The isolated hearts are already ischaemic after their removal from the rat. The reperfusion aspect is caused by reflow of blood to generate oxidative stress, but in the ex vivo model, is there really reperfusion injury?

The Langendorff model is a well-established ex-vivo technique where the excised heart is perfused retrogradely through the aorta. This allows oxygenated buffer that also contains metabolic fuel to be pumped through the aorta, forcing the aortic valve closed and diverts the buffer directly into the coronary arteries to sustain the organ outside of the body. By temporarily stopping buffer perfusion then reinstating it as we did here, we can mimic a full ischaemia-reperfusion injury. It has been commonly used since the late 19th century to study heart physiology and pathophysiology including ischaemia-reperfusion injury.

We agree with the reviewer that the hearts may have experienced a degree of ischaemia when being removed, unfortunately that is one of the confounds associated with ex-vivo preparations such as the Langendorff model. However, it should be noted that both the control and synthetic torpor hearts were exposed to the same conditions following removal from the animal. To minimise ischaemic damage, all hearts were placed in ice-cold buffer immediately following excision. Preparations were excluded from our data if a) cannulation took longer than 3 minutes, b) more than 2 cannulation attempts were made and c) if the heart rate was slower than 200 bpm during the baseline 30-minute recording. This criteria is in accordance with other laboratories and is frequently reported in the literature.

(10) The authors show that whole animal oxygen consumption is reduced in the torpor state. The measurement is crude and most likely reflects the inactivity of the animal's skeletal muscle in the torpor state. A more relevant and direct experiment would be to do oxygen consumption (or Seahorse) assays on extracts of the isolated hearts.

We did not simultaneously measure oxygen consumption alongside animal activity so we cannot determine whether muscle inactivity is a contributing factor in this data. However, previous work on mice has shown that VO₂ to be decreased by 15% during the light phase (mouse inactivity/sleep period) compared to the dark phase (their active phase) [1]. Whereas during fasting-induced torpor, a ~45% decrease in VO₂ has been reported [2], which is consistent with what we observed during synthetic torpor in rats (~39% decrease in VO₂).

Furthermore, we simultaneously recorded activity of our rats while recording core body temperature and heart rate before and during synthetic torpor. We have since updated the manuscript to include this data as a supplementary figure (supplementary figure 6), showing that there is no significant difference between the activity of animals in synthetic torpor compared to controls (a curious contrast with natural torpor).

We also agree with the reviewer regarding the seahorse experimental suggestion and plan to use the seahorse assay to measure oxygen consumption in isolated cardiomyocytes and cardiac tissue punches in the future.

(1) Nie Y, Gavin TP, Kuang S. Measurement of Resting Energy Metabolism in Mice Using Oxymax Open Circuit Indirect Calorimeter. Bio Protoc. 2015;5(18):e1602. doi:10.21769/bioprotoc.1602

(2) Hrvatin S, Sun S, Wilcox OF, et al. Neurons that regulate mouse torpor. Nature. 2020;583(7814):115-121. doi:10.1038/s41586-020-2387-5

(11) The authors report that the synthetic torpor induces bradycardia. There is no follow-up on this important observation. The MPA-heart connection is not analyzed. (A) Is the link through cardiovascular centers in the brainstem? (B) Is the torpor-induced bradycardia mediated through increased parasympathetic or decreased sympathetic autonomic tone? Pharmacological experiments could also be done.

We thank the reviewer for these suggestions and agree that these will be important follow-up experiments to understand how the cardioprotection is mediated. We currently have a grant under review that will perform these important experiments.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Suggestions for improved or additional experiments or analysis:

(1) Chemogenetic inhibition of the MPO.

This is an interesting experiment proposal, we chose to focus on the activation of these neuronal populations to establish whether activation would be sufficient to drive synthetic torpor in a species that does not naturally enter torpor. Inhibiting these neurons would be an interesting avenue to explore in the future.

(2) In vivo validation of torpor-induced protection.

We plan to use an in-vivo cardiac ischaemia-reperfusion injury model in future experiments and this forms a work package in a grant we currently have under review.

(3) A search for the dose and timing of torpor that provides the most robust protection.

Using the in-vivo cardiac ischaemia-reperfusion injury model described above, in future we aim to determine whether synthetic torpor is protective following the ischaemic episode, which would better model a clinical scenario such as when a patient presents with a heart attack.

(4) Use of inhibitors of transcription and translation as a strategy to understand what are the nodes of regulation that mediate neuroprotection.

An interesting suggestion. We might consider this in future studies but feel it is beyond the scope of this manuscript.

(5) Testing in males, females, and rats of different genetic backgrounds.

This data set includes both male and female rats, however we did not power our experiments to determine sex differences. In future, we will plan to power our studies to allow determination of sex differences, and to use rats from different genetic backgrounds.

(6) Use of chemical or molecular tools to assess the relevance of phosphoproteomic data.

Our future experimental plans, which are currently under review in a grant application include the use of inhibitors and activators of the kinases we identified in our phosphoproteomic data, to determine whether these can modulate the infarct size in the ischaemia-reperfusion injury model and modulate the cardioprotective effect from synthetic torpor.

Reviewer #2 (Recommendations for the authors):

(1) Clarifying core body temperature measurement:

The authors report post-CNO core body temperature (mean, 90-100 minute window) for four groups: room temperature MPAGq (31.64 {plus minus} 1.02{degree sign}C, n=10), room temperature MPAEGFP (37.05 {plus minus} 0.43{degree sign}C, n=6), thermoneutral MPAGq (36.4 {plus minus} 0.73{degree sign}C, n=6), and thermoneutral MPAEGFP (37.05 {plus minus} 0.43{degree sign}C, n=6). Please clarify the provenance of the MPAEGFP temperature data, as the room temperature and thermoneutral MPAEGFP groups report identical means to two decimal places (37.05{degree sign}C) with identical standard deviations ({plus minus}0.43{degree sign}C). It should also be noted in the limitations that continuous core temperature telemetry was available in only a subset of animals - specifically, 10 of 12 room temperature MPAGq animals and 6 of 16 room temperature MPAEGFP animals that contributed to the primary Langendorff infarct endpoint. It is correct that this means that for the majority of room temperature MPAEGFP control animals (10 of 16), no core temperature measurement of any kind was paired with their infarct data? If so, this should be stated as a limitation. Additionally, pre-CNO baseline temperatures are not reported numerically for any group, and individual nadir temperatures across the 90-minute induction period are not provided, meaning the full thermal exposure of individual animals cannot be characterized from the data as presented.

We thank the reviewer for these suggestions, and we have since corrected the in-text error for the MPA^EGFP thermoneutral and room temperature data. While we did not have the in-vivo core body temperature in all of our animals, we did record surface body temperature if core body temperature was absent and we have since plotted this surface body temperature data against the infarct size and performed a correlational analysis. We found no significant correlation between the two. We have since also updated our supplementary figure 1 to include the pre-CNO baseline temperature and heart rates for all animals in the synthetic torpor and control groups and those exposed to room temperature or thermoneutral environments.

(2) Conclusions exceed evidence:

While the demonstration that chemogenetic MPA activation under thermoneutral conditions is associated with reduced infarct size is an interesting and valuable finding, the conclusion, as stated, that "the hypothermic component of torpor is not necessary for the protective effects of synthetic torpor" (lines 303-304), goes beyond what the evidence can support. The cardioprotective effect observed in the thermoneutral condition may indeed reflect the contribution of other (non-temperature) factors - but the involvement of a temperature effect cannot be ruled out on the basis of the data presented. The thermoneutral MPAGq group, although not significantly different from the thermoneutral MPAEGFP control in core temperature (36.4 {plus minus} 0.73{degree sign}C vs. 37.05 {plus minus} 0.43{degree sign}C, p=0.44), had a mean core temperature that was in fact 0.65{degree sign}C lower than controls, and with n=6 in each group the non-significance of this difference reflects limited statistical power to detect equivalence as much as true normothermia. Given the limitations in telemetric coverage described above, a non-significant difference between group means is not sufficient to conclude that hypothermia is categorically uninvolved in mediating the cardioprotective effect. The appropriate test would be to demonstrate an absence of correlation between individual core body temperature nadir and individual infarct size across all animals in both the room temperature and thermoneutral conditions - a continuous analysis that the current telemetry coverage may not fully support, but that would directly address the question. We would therefore ask the authors either to provide such a correlation analysis if the paired data permit it, or to temper their conclusion accordingly - stating that factors other than hypothermia are sufficient for cardioprotection, rather than that hypothermia is not involved, which is the stronger and less well-supported claim.

We agree with these suggestions and have adjusted our conclusions within the text. We have also performed a correlational analysis between the surface temperature of the animals and the infarct area, which we have included in the supplementary figures. We found no significant correlation between the two, which further supports our conclusion that other factors are likely to contribute to the cardioprotection observed, rather than the hypothermia. We feel that it is biologically implausible to suppose that the 0.6°C drop in body temperature observed in the thermoneutral synthetic torpor group is sufficient to recapitulate the cardioprotection observed.

(3) Use of the word torpor: The statement:

"this cardioprotection induced by synthetic torpor persisted in the absence of hypothermia"

Presupposes that the thermoneutral condition is still synthetic torpor, just without one feature. But the authors have no basis for that framing - they have not demonstrated that the thermoneutral condition preserves the metabolic suppression component, and they have in fact abolished the temperature component that is part of the very definition they invoke. A more accurate statement would be:

"Chemogenetic activation of MPA neurons under thermoneutral conditions, which prevents hypothermia and whose effect on oxygen consumption was not assessed, still produced a reduction in infarct size."

We thank the reviewer for this suggestion, and we have amended the text in our manuscript to reflect this.

(4) Lack of data addressing possible sex differences. Given the known role of estrogen in contributing to torpor, a more thorough analysis of sex differences beyond underpowered stratification is needed. In the absence of this analysis, this should be noted as a limitation.

The preoptic area of the hypothalamus is enriched with estrogen receptors and previous work by our group has demonstrated that the estrous cycle modulates fasting-induced torpor. Our current work did not aim to investigate sex differences and is not powered to do this. As a result, we have now added this limitation to our discussion.

(5) QPLOT marker characterization is incomplete. The QPLOT classification requires co-expression of markers including Qrfp and Tacr3, but the authors only tested three of the five (Ptger3, Lepr, Opn5). Ptger3 was expressed in 100% of transduced cells, but the question is whether Ptger3 expression alone is sufficient evidence of QPLOT identity without testing Qrfp and Tacr3. The conclusion that these neurons are "analogous" to mouse QPLOT neurons is reasonable but somewhat circular, given the marker selection. The assertion that 29% expressed "all three QPLOT markers" should be contextualized as three of five.

We have updated our manuscript to state that our assessment is for three of the five QPLOT neuron markers.

Reviewer #3 (Recommendations for the authors):

Other comments:

(12) The figure legends should state the group sizes (animal numbers) for the graphical data.

We have added this information to the figure legends.

(13) The control and experimental groups do not appear to be balanced. Were any animals removed from the treatment groups?

Some MPA HM3DGq injections failed and the animals did not enter synthetic torpor following CNO injection. These were excluded from the study (the success rate for inducing synthetic torpor was 80%). Some animals were excluded from the ischaemia-reperfusion study, due to time from heart excision to cannulation on the Langendorff apparatus being more than 2 minutes, more than 2 cannulation attempts or a heart rate of less than 200 bpm at the end of the equilibration period. 8 hearts were excluded from the heart infarction data set based on these criteria. We have updated the methods section of our manuscript to emphasise this.

(14) With small group sizes, it is better to show the data set variation with standard deviations rather than SEM.

We thank the reviewer for noting our mistake and have corrected it and/or reported individual data points.

(15) The precise P value for the infarct size reduction with torpor should be stated rather than p<0.05.

We have updated the manuscript and the figure legends to include the precise P values.

(16) Is there translational relevance of this work using an in vivo model of bona fide heart ischaemia-reperfusion injury?

We absolutely plan to complete those experiments, but feel they are beyond the scope of this initial reporting of the principle that synthetic torpor in the rat is cardioprotective.

eLife Assessment

This work by Pyne and Pandey et al. addresses DNase X (DNase1L1) activity at the macrophage phagocytic cup, using an innovative imaging approach that couples visualization of cup formation to spatially resolve DNA degradation. The methodology is technically sound, and the central finding that DNA digestion begins prior to phagolysosomal maturation is well supported. Overall, the study is solid and provides a valuable framework for investigating early events at the phagocytic cup that may shape responses to pathogens and inflammatory disease.

Reviewer #1 (Public review):

Pyne and Pandey et al. report the observation of early DNA degradation at the phagocytic cup during macrophage engulfment. Using an elegant experimental system that combines actin staining to visualise cup formation with direct monitoring of DNA degradation, the authors identify rapid recruitment of the membrane-bound nuclease DNase X (DNase1L1) to nascent phagocytic cups. This recruitment occurs within minutes of cup formation, is independent of DNA presence at the substrate, and appears to originate from intracellular membrane structures rather than from the extracellular environment. The results support the conclusion that DNase X activity is present at the phagocytic cup and that DNA digestion can begin prior to phagolysosomal maturation.

The study is technically strong. The experimental system is clean, specific, and allows precise spatial and temporal detection of DNA degradation. The imaging-based approaches are carefully executed and enable convincing visualisation of DNase X recruitment and activity. The use of an alternative substrate beyond the primary SNS system strengthens the core observation, and the data broadly support the authors' central claim.

However, several limitations temper the physiological interpretation. The system relies largely on short, free DNA substrates, leaving open how efficiently DNase X processes more complex or physiologically relevant DNA structures, such as nucleosome-bound DNA or neutrophil extracellular traps (NETs). It remains unclear whether DNase X deficiency would alter macrophage responses to larger nucleic acid structures, influence engulfment efficiency, or modify downstream inflammatory signalling pathways such as TLR9 or STING activation. Moreover, the experimental setup prevents full phagocytic cup closure, potentially prolonging DNase activity compared with physiological phagocytosis, which typically proceeds rapidly to cargo internalisation. For example, the peak signal observed in Figure 5 occurs approximately 90 minutes after phagocytic cup formation, a time point at which many phagocytic cups would be expected to have already closed under physiological conditions. Additional work using fully engulfed cargo in more physiological contexts would clarify whether early DNase X activity meaningfully contributes to overall DNA clearance kinetics.

Mechanistically, the signal that triggers DNase X recruitment remains unresolved. Although actin rearrangement was excluded as the primary driver, the upstream cues that direct DNase X-containing membrane structures to the forming cup are not yet defined.

In the broader context, early DNase X activity at the phagocytic cup could represent an additional safeguard against inflammatory signalling by limiting extracellular or surface-associated DNA before phagolysosomal degradation by DNase II. This mechanism may be particularly relevant in settings where DNA fragmentation before engulfment is incomplete, such as necroptosis or NET formation. Determining whether DNase X deficiency exacerbates inflammatory responses, alters DNA clearance efficiency in vivo, or contributes to immune pathology will be critical for establishing its physiological and disease relevance.

Overall, this is a compelling study that introduces a novel concept of pre-phagolysosomal DNA digestion. The conclusions are well supported within the in vitro system used, but further investigation using diverse DNA substrates and physiologically relevant models will be required to fully define the impact of this mechanism on immune regulation and disease.

Comments on revised version.

The authors have responded constructively to the points raised in my original review. Fig. S5 extends the substrate range beyond the 18-bp SNS construct by showing degradation of plasmid DNA immobilised on microbeads, and fig. S8 addresses my concern that the surface-immobilised platform prevents phagocytic cup closure, by demonstrating DNase activity at cups forming over free, internalisable beads. Together with the live-imaging data in Fig. 1I, these additions support the central claim that DNase activity begins at the nascent phagocytic cup, before closure and well before phagolysosomal maturation. I consider my principal concerns addressed and see no reason to alter the eLife Assessment.

One point remains for readers rather than as a request. The new data establish that DNA at the cup is degraded, but do not themselves demonstrate that DNaseX is responsible, since neither fig. S5 nor the biofilm experiments in Fig. 7 include the PI-PLC or siRNA controls used in Fig. 3; the authors appropriately acknowledge this limitation for the biofilm data. The physiological questions raised in my original review - the handling of chromatin-associated DNA and NETs, and whether DNaseX activity at the cup limits downstream sensing of extracellular DNA - remain open, and the manuscript is appropriately framed as establishing the phenomenon rather than resolving its physiological role.

Reviewer #2 (Public review):

Summary:

This manuscript presents an elegant and innovative imaging approach to visualize DNase activity at the interface between macrophages and extracellular substrates. The platform is technically strong and enables the study of localized DNA degradation with high spatial resolution. The work is of clear interest and provides a useful framework to investigate how immune cells process extracellular DNA. However, several aspects of the mechanistic interpretation and conceptual framing would benefit from clarification.

Strengths:

(1) The study introduces a creative and well-designed imaging platform that allows visualization of localized DNase activity at cell-substrate interfaces.

(2) The approach is technically robust and represents a valuable tool that could be broadly useful to the field.

(3) The experiments are thoughtfully designed and address an important question regarding how immune cells interact with extracellular DNA.

(4) The work opens interesting avenues for studying DNA processing in contexts such as infection and inflammation.

Comments on revised version.

The authors have carefully addressed the points raised in my previous review, and the manuscript has been revised accordingly. I am satisfied with the revisions and have no further comments at this stage.

Author response:

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

eLife Assessment

This work by Pyne and Pandey et al. addresses DNase X (DNase1L1) activity at the macrophage phagocytic cup, using an innovative imaging approach that couples visualization of cup formation to spatially resolve DNA degradation. The methodology is technically sound, and the central finding that DNA digestion begins prior to phagolysosomal maturation is considered well supported, though some mechanistic claims may benefit from further evidence and more cautious framing. Overall, the study is solid and provides a valuable framework for investigating early events at the phagocytic cup that may shape responses to pathogens and inflammatory disease.

Thanks for the overall review.

Public Reviews:

Reviewer #1 (Public review):

Pyne and Pandey et al. report the observation of early DNA degradation at the phagocytic cup during macrophage engulfment. Using an elegant experimental system that combines actin staining to visualise cup formation with direct monitoring of DNA degradation, the authors identify rapid recruitment of the membrane-bound nuclease DNase X (DNase1L1) to nascent phagocytic cups. This recruitment occurs within minutes of cup formation, is independent of DNA presence at the substrate, and appears to originate from intracellular membrane structures rather than from the extracellular environment. The results support the conclusion that DNase X activity is present at the phagocytic cup and that DNA digestion can begin prior to phagolysosomal maturation.

The study is technically strong. The experimental system is clean, specific, and allows precise spatial and temporal detection of DNA degradation. The imaging-based approaches are carefully executed and enable convincing visualisation of DNase X recruitment and activity. The use of an alternative substrate beyond the primary SNS system strengthens the core observation, and the data broadly support the authors' central claim.

Thanks for the positive comments.

However, several limitations temper the physiological interpretation. The system relies largely on short, free DNA substrates, leaving open how efficiently DNase X processes more complex or physiologically relevant DNA structures, such as nucleosome-bound DNA or neutrophil extracellular traps (NETs). It remains unclear whether DNase X deficiency would alter macrophage responses to larger nucleic acid structures, influence engulfment efficiency, or modify downstream inflammatory signalling pathways such as TLR9 or STING activation.

Thanks for the comments. It is a constructive suggestion to test how DNase X in phagocytic cups (PCs) may degrade long DNA or chromosomal DNA, as the current experiment setup mainly used 18bp DNA as the DNase sensor construct. Now we included new data of DNase response of macrophage to plasmid DNA (extracted from E. coli) immobilized on microbead surface. The result show that macrophage also degrade the plasmid DNA, suggesting that the DNase in PC is rather versatile and can degrade long DNA. This result is included as fig. S5.

It would be certainly interesting to explore the role of DNase X in phagocytic efficiency and downstream inflammatory pathways. While these could be the future research directions, the goal of current manuscript is to demonstrate a new DNase activity in PCs and calibrate its several core features (temporal and spatial dynamics and response to biofilm, etc.). Expanding the topic further will require substantial resource and effort, and may make the current manuscript bloated with data. Therefore, we didn’t pursue the study of DNase of PCs in the context of phagocytosis efficiency and downstream pathways.

Moreover, the experimental setup prevents full phagocytic cup closure, potentially prolonging DNase activity compared with physiological phagocytosis, which typically proceeds rapidly to cargo internalisation. For example, the peak signal observed in Figure 5 occurs approximately 90 minutes after phagocytic cup formation, a time point at which many phagocytic cups would be expected to have already closed under physiological conditions.

Additional work using fully engulfed cargo in more physiological contexts would clarify whether early DNase X activity meaningfully contributes to overall DNA clearance kinetics.

Real-time imaging of SNS signal and F-actin in macrophages, shown in Fig. 1I in the manuscript, suggests that DNase activity in PCs appear within one minute after PC formation. We do resonate with the reviewer’s concern about the experiment platform used in this work, namely, the microbeads are immobilized on the glass surface, preventing the natural closure of PCs. This artificial design may create biased observation of DNase activity during phagocytosis.

In light of this concern, we developed a new assay by using free SNS-coated microbeads (non-immobilized). These beads can be completely internalized by macrophages through phagocytosis. During experiments, we searched for the PCs just starting to form over free SNS-coated beads. Despite the rare occurrence of these events (phagocytosis process has a rather short time windows), we did find the evidence that PCs already exhibit DNase activity before their closures, as shown in fig. S8.

Accordingly, we added one paragraph to the main text, “Because surface-immobilized microbeads prevent PC closure, they likely prolong PC formation and may alter the observed temporal dynamics of DNase activity within the PC. To determine whether the PC exhibits DNase activity before closure, we prepared SNS-coated free microbeads and fed them to adherent macrophages. The results showed that PCs indeed exhibited DNase activity before cup closure in response to free microbeads (fig. S8), confirming that DNase activity is initiated rapidly during PC formation.”

Mechanistically, the signal that triggers DNase X recruitment remains unresolved. Although actin rearrangement was excluded as the primary driver, the upstream cues that direct DNase X-containing membrane structures to the forming cup are not yet defined.

We agree that the biochemical signals responsible for initiating DNase X recruitment remain unresolved. Despite our efforts to identify such signals by exposing macrophages to a variety of immunogenic stimuli, we consistently observed DNase X recruitment to phagocytic cups under all conditions tested. Based on these findings, we conclude that DNase X recruitment to phagocytic cups is constitutive rather than stimulus-dependent.

Nevertheless, even if this recruitment is constitutive, it is likely that an intracellular biochemical cue initiates or regulates the recruitment process. Unfortunately, we have not yet been able to identify this upstream signal. We have acknowledged this limitation and included this consideration in the Discussion as an important direction for future investigation.

In previous studies, we also examined DNase activity at phagocytic cups following inhibition of the cGAS-STING and TLR9 pathways, two major cellular DNA-sensing mechanisms. However, inhibition of either pathway did not alter DNase activity at phagocytic cups. As these experiments yielded negative results and did not provide mechanistic insight into DNase X recruitment, we did not include them in the current manuscript.

In the broader context, early DNase X activity at the phagocytic cup could represent an additional safeguard against inflammatory signalling by limiting extracellular or surface-associated DNA before phagolysosomal degradation by DNase II. This mechanism may be particularly relevant in settings where DNA fragmentation before engulfment is incomplete, such as necroptosis or NET formation. Determining whether DNase X deficiency exacerbates inflammatory responses, alters DNA clearance efficiency in vivo, or contributes to immune pathology will be critical for establishing its physiological and disease relevance.

Thanks for the constructive comments which provide valuable suggestion for the future work.

Overall, this is a compelling study that introduces a novel concept of pre-phagolysosomal DNA digestion. The conclusions are well supported within the in vitro system used, but further investigation using diverse DNA substrates and physiologically relevant models will be required to fully define the impact of this mechanism on immune regulation and disease.

We appreciate the insightful comments and agree that in vivo experiments would provide valuable information for future studies. The current manuscript focuses on a cellular function identified using an in vitro experimental system. The results suggest that DNaseX may play important physiological roles in immune defense and extracellular DNA clearance, highlighting the need for further investigation using animal models or other in vivo systems. These are exciting directions that we are actively planning to pursue in our future research.

We are not able to incorporate in vivo studies into the current revision, as such experiments would require the establishment of fundamentally different experimental models, the acquisition of new technical expertise, and substantial additional financial support. These efforts are beyond the scope and timeline of the present manuscript revision. We believe that the current in vitro findings provide a foundation for these future in vivo investigations.

Reviewer #2 (Public review):

Summary:

This manuscript presents an elegant and innovative imaging approach to visualize DNase activity at the interface between macrophages and extracellular substrates. The platform is technically strong and enables the study of localized DNA degradation with high spatial resolution. The work is of clear interest and provides a useful framework to investigate how immune cells process extracellular DNA. However, several aspects of the mechanistic interpretation and conceptual framing would benefit from clarification.

Strengths:

(1) The study introduces a creative and well-designed imaging platform that allows visualization of localized DNase activity at cell-substrate interfaces.

(2) The approach is technically robust and represents a valuable tool that could be broadly useful to the field.

(3) The experiments are thoughtfully designed and address an important question regarding how immune cells interact with extracellular DNA.

(4) The work opens interesting avenues for studying DNA processing in contexts such as infection and inflammation.

Thanks for the positive comments.

Weaknesses:

While the experimental approach is strong, several key conclusions rely on interpretations that would benefit from further clarification:

(1) First, the conclusion that DNaseX is recruited to phagocytic cups from the "cytoplasm" appears conceptually imprecise. Given that DNaseX is a membrane-anchored protein, it is unlikely to exist as a freely soluble cytoplasmic pool. A more plausible interpretation is that DNaseX is supplied from intracellular membrane compartments. This interpretation would also be more consistent with the data showing dependence on a membrane anchor.

We thank the reviewer for this insightful comment. We agree that, as a GPI-anchored protein, DNaseX is likely transported from the endoplasmic reticulum and Golgi apparatus to the plasma membrane via secretory vesicles carrying the protein. The original subtitle “DNaseX in PCs is recruited from the cytoplasm, not from the plasma membrane” in the manuscript, is misleading, as it may imply that DNaseX is a cytoplasmic soluble protein. Our original point is to state that DNaseX is recruited intracellularly, not from the adjacent cell membrane domain. To avoid the ambiguity, we have revised that subtitle to “DNaseX in PCs is recruited intracellularly, not from the plasma membrane”. In the revision, we also emphasized the that DNaseX is likely delivered by the vesicles, which were observed in Fig. S13.

(2) Second, the interpretation that actin polymerization is not required for DNaseX recruitment raises concerns. Phagocytic cup formation is known to depend strongly on actin dynamics, and it is therefore unclear whether the structures observed under actin inhibition represent fully formed functional cups or partial cell-substrate contacts. This distinction is important for interpreting recruitment versus activity, particularly since enzymatic activity is reduced under these conditions.

We appreciate the reviewer’s comment and agree with this point. The data presented in the manuscript demonstrate the relationships, but not causalities, among F-actin signal, DNaseX signal (detected by immunostaining), and SNS signal (reflecting DNase activity) in PCs. Specifically, our results show a strong correlation between the F-actin and SNS signals, whereas the correlation between the F-actin and DNaseX signals is weak. These correlations do not warrant the causality relations.

Experimentally establishing causality is challenging because F-actin is an essential structural component of the phagocytic cup and cannot be completely eliminated from PCs. To ensure rigorous interpretation of the data, we have revised the manuscript to present the findings more cautiously and avoid implying a causal relationship where it has not been experimentally demonstrated.

We revised the section title: “Actin polymerization is dispensable for DNaseX recruitment but essential for DNase activity in the PCs” to “Actin polymerization is correlated with DNase activity in PCs, but not with DNaseX recruitment”. We also revise the abstract and the main text correspondingly to reflect this point.

(3) Third, the identification of DNaseX as the main nuclease responsible for the observed activity is not fully resolved. The conclusions rely primarily on gene silencing and staining approaches, but the specificity of these strategies relative to other nucleases is not addressed. It therefore remains possible that additional enzymes contribute to the observed activity.

The reviewer raises an important point. We acknowledge that gene silencing and immunostaining may not be entirely specific, as they could potentially affect or detect other members of the DNase family that share structural similarities with DNaseX. Indeed, the official name of DNaseX, Deoxyribonuclease-1-like 1, suggests a close structural relationship with DNase I.

Our conclusion regarding the specificity of DNaseX is based on the combined evidence from multiple independent approaches rather than any single experiment. In addition to gene silencing and immunostaining, we demonstrated that cleavage of the GPI anchor with PI-PLC markedly reduced or completely abolished DNase activity in PCs. This experiment provides particularly strong evidence because DNaseX is the only known membrane-bound DNase that is anchored to the plasma membrane via a GPI linker.

Taken together, the complementary results from GPI-anchor cleavage, gene silencing, and immunostaining provide convergent evidence supporting the conclusion that DNaseX is the primary enzyme responsible for the DNase activity observed in PCs. We therefore believe that the combined data provide sufficient specificity to support our conclusion.

(4) Finally, the interpretation of the biofilm experiments may be overstated. While the data clearly show localized DNA degradation in contact with macrophages, it is not fully established that this process depends specifically on phagocytic cup structures. An alternative explanation is that membrane-associated DNase activity more generally mediates this effect. In addition, the physiological relevance of this mechanism would benefit from further discussion.

We agree with the reviewer on this point. Although our data demonstrate that macrophages degrade eDNA in biofilms through direct physical contact, they do not establish that phagocytic cups are the primary structures responsible for mediating this process.

We attempted to obtain direct evidence by simultaneously imaging F-actin structures in macrophages and eDNA degradation within biofilms, with the goal of visualizing phagocytic cup formation around eDNA filaments. However, this proved to be technically challenging. Unlike a flat glass surface, biofilms possess substantial thickness and an uneven three-dimensional topology, making it difficult to achieve high-resolution imaging of F-actin structures with sufficient clarity. In addition, eDNA filaments, unlike microparticles, may not induce the formation of canonical F-actin-rich phagocytic cups. Consequently, we were unable to obtain convincing imaging data demonstrating that phagocytic cups are the structures responsible for eDNA degradation within biofilms.

To ensure a rigorous interpretation of our findings, we have revised the manuscript to avoid overstating this conclusion and to more accurately reflect the limitations of the current data. We added this paragraph in the revision: “While the results showed that macrophages degrade eDNA in biofilms through physical contact, it has not been confirmed that this degradation is mediated by DNaseX within the PC, as eDNA filaments are not expected to induce the formation of a typical PC structure. An alternative possibility is that DNaseX localized on the cell membrane, including within the plaque-like clusters (fig. S13), comes into direct contact with eDNA and mediates its degradation through physical contact.”

Overall, the study is technically strong and introduces a valuable methodology, but several central conclusions are only partially supported by the current data and would benefit from more cautious interpretation and clearer conceptual framing.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

The data are clearly presented, and the manuscript is well written, making it a pleasure to read.

Reviewer #2 (Recommendations for the authors):

(1) Clarify the description of DNaseX origin by replacing "cytoplasmic" with more precise terminology (e.g., intracellular membrane compartments), and revise the interpretation accordingly.

Revision has been made according to the reviewer’s suggestion. Please see the point-to-point response.

(2) Provide a clearer definition of what constitutes a phagocytic cup under actin inhibition conditions, and discuss whether the observed structures represent fully functional compartments.

Revision has been made according to the reviewer’s suggestion. Please see the point-to-point response.

(3) Consider including additional controls or discussion addressing the specificity of the gene silencing approach, particularly in relation to other nucleases. If feasible, assess the potential contribution or localization of other nucleases (immunofluoresce for DNase I, for example), or explicitly acknowledge this as a limitation.

Revision has been made according to the reviewer’s suggestion. Please see the point-to-point response.

(4) Refine the interpretation of the biofilm experiments to distinguish between phagocytic cup-specific mechanisms and more general membrane-associated activity. Expand the discussion on the physiological relevance of the findings, particularly in the context of complex environments such as biofilms or extracellular DNA structures in vivo.

Revision has been made according to the reviewer’s suggestion. Please see the point-to-point response.

(5) Minor: Improve consistency in terminology throughout the manuscript (e.g., membrane-associated vs cytoplasmic localization) to avoid conceptual confusion.

We revised all “membrane-associated” to “membrane-bound” for the consistency of terminology. The confusion due to the use of “cytoplasmic” has been rectified. Please see the point-to-point response.

eLife Assessment

This valuable study provides a very large data set of motor unit recordings in people with multiple sclerosis and, as such, could impact both foundational and clinical studies of the neuromuscular system. The study reports increased variability in firing rate profiles across participants and suggests that this observation reflects differences in the excitatory, inhibitory, and neuromodulatory drive reaching motor neurons. The descriptive finding is solid, but the link to specific neural mechanisms is incomplete as it depends heavily on assumptions related to its model-based framework.

Reviewer #1 (Public review):

This study applies a recently developed reverse-engineering framework for motor unit discharge to a large dataset from individuals with multiple sclerosis (MS) and neurologically intact controls. The authors aim to determine whether abnormalities in voluntary motor control in MS can be attributed to different patterns of excitatory, inhibitory, and neuromodulatory input to spinal motoneurons. A major conceptual emphasis of the study is on heterogeneity: rather than asking only whether people with MS differ from controls on average, the authors examine whether the distributions of motor unit discharge features and derived physiological variables are broader and more diverse across affected individuals.

A major strength of the work is the size and richness of the dataset. The study includes 89 participants with MS and 34 controls, with high-density surface electromyography (HDsEMG) used to obtain large populations of motor unit discharge patterns from the tibialis anterior (TA) and soleus (SOL) muscles. The resulting dataset contains many thousands of motor unit recordings and allows the authors to examine both group means and the shapes and variances of participant-level distributions. The finding that several motor unit discharge characteristics are more broadly distributed in MS than in controls is convincing and potentially important. In particular, the observation that affected individuals can occupy both high and low extremes of these distributions provides a useful empirical description of the diversity of motor unit behavior in this disease.

The principal limitation concerns the physiological interpretation assigned to these discharge patterns. The excitation, inhibition, and neuromodulation variables are not directly measured physiological inputs. They are composite variables derived from several features of motor unit discharge, with each feature weighted according to relationships identified in simulations reported previously by Chardon and colleagues (reference 1). Thus, the step from observed discharge behavior to specific underlying synaptic mechanisms is necessarily model-dependent. The distinction between these two levels of inference is important for interpreting the main conclusions of the study.

This issue is especially relevant because several different physiological processes can plausibly influence the same discharge features. Motor unit firing patterns reflect not only excitatory and inhibitory synaptic inputs and neuromodulation, but also intrinsic motoneuron properties, persistent inward currents (PICs), after-hyperpolarization (AHP) properties, tonic inhibition, the time course of synaptic excitation, and afferent input. Some of these factors are acknowledged as limitations of the modeling framework in the manuscript. The current data therefore provide strong evidence for heterogeneous motor unit discharge phenotypes in MS, but more indirect evidence that this heterogeneity can be uniquely attributed to distinct patterns of excitatory, inhibitory, and neuromodulatory input.

The interpretation of inhibition is a particularly clear example of this general inverse problem. In the underlying modeling framework, inhibition is represented along a continuum from proportional or balanced inhibition to reciprocal or push-pull inhibition. These different patterns influence PICs and consequently alter firing-rate nonlinearity and rate modulation. This provides a plausible forward-model relationship between inhibitory organization and motor unit discharge. However, observing a particular firing pattern in vivo does not necessarily identify the organization of inhibitory input uniquely, because similar changes in firing-rate modulation or hysteresis could arise from altered neuromodulation, intrinsic motoneuron properties, or other changes in synaptic drive. The inhibition composite is therefore best interpreted as a discharge phenotype that is consistent with a particular inhibitory organization under the assumptions of the model, rather than as a direct measure of inhibitory synaptic input.

A related methodological issue concerns the construction of the composite variables. The authors use mutual-information (MI) values from the previous simulation study as weights in signed linear combinations of normalized discharge features. Mutual information quantifies how informative a feature is about a modeled input parameter, but it is not itself a regression coefficient or a measure of the magnitude of a physiological effect. In addition, different discharge features may contain overlapping information about the same underlying process. The resulting composites are therefore useful summary measures of patterns associated with the modeled physiological variables, but their quantitative interpretation as direct estimates of those variables is less certain. This distinction is particularly relevant because the manuscript sometimes moves from describing the composite variables to describing the corresponding physiological inputs themselves.

The study's emphasis on variability also raises an important measurement issue. Because increased between-participant variance is itself one of the central biological findings, differences in measurement precision between the MS and control groups are more consequential here than in a conventional comparison of group means. Participants with MS sometimes had greater difficulty producing smooth triangular contractions, and motor unit yield and decomposition quality may plausibly vary more across affected participants. If measurement or decomposition uncertainty were more heterogeneous in the MS group, this could broaden participant-level distributions and thereby amplify the appearance of biological heterogeneity. The manuscript uses established decomposition and quality-control procedures, so this is not a general challenge to the validity of HDsEMG. Rather, it is a consideration that is particularly important when increased distributional spread is itself the primary result.

The manuscript also makes a stronger interpretive step from broad group distributions to patient-specific pathophysiology. The data convincingly show that motor unit discharge-derived measures are more heterogeneous among people with MS. They do not yet establish whether this variation represents distinct mechanisms in individual patients, continuous variation in a common mechanism, identifiable pathophysiological subgroups, differences in disease severity or lesion distribution, or some combination of these factors. The manuscript itself recognizes this distinction when it identifies the separation of individual-level variation from potential subgroups as an important goal for future work.

Overall, this is a valuable study with an unusually large motor unit dataset and a compelling demonstration that motor unit discharge behavior is markedly heterogeneous in MS. The work also provides an informative application of a model-based reverse-engineering framework to a clinically diverse human population. The evidence is strongest for the descriptive conclusion that motor unit discharge phenotypes are heterogeneous and altered in MS. The more specific attribution of these phenotypes to excitatory, inhibitory, and monoaminergic inputs is plausible and potentially useful, but remains contingent on the assumptions and identifiability of the underlying model. With this distinction in mind, the dataset and analytical approach should be useful to researchers interested in motor unit physiology, disease-related variability in motor control, and the possibilities and limitations of inferring latent physiological mechanisms from human motor unit discharge.

Reference 1: Chardon, M. K. et al. Supercomputer framework for reverse engineering firing patterns of neuron populations to identify their synaptic inputs. eLife 12, RP90624 (2024).

Reviewer #2 (Public review):

The Beauchamp and Chardon model, like any computational model, is only as informative as the physiological parameters it includes. By varying only three synaptic related dimensions, namely neuromodulatory (i.e., PIC strength from 5HT inputs, etc.), the pattern of inhibition relative to excitation, and the distribution of excitatory input across smaller versus larger motoneurons, the model necessarily holds constant many other properties that may differ substantially between individuals and between controls and patients with MS. These properties include intrinsic membrane conductance properties (with some conductances like NaV that may differ in MS not included in the Chardon model), afterhyperpolarization properties, tonic inhibition, dendritic and axonal structure, synaptic kinetics, glial changes, etc., etc. The present paper then moves one step further away from direct physiological estimation because it does not fit these three model parameters to each participant's data, but instead combines control-normalized firing pattern features using weights derived from the original simulations. As a result, the resulting "excitation," "inhibition," and "neuromodulation" scores should be regarded as indirect similarity scores within a restricted model space, with a substantial risk that changes caused by unmodeled physiology are misattributed to one of the three modeled components. An additional limitation is that the underlying motoneuron models were originally tuned to intracellular recordings from medial gastrocnemius motoneurons in decerebrate cats and then manually modified to generate more human-like firing rates and hysteresis, rather than being formally fitted to human motor unit recordings.

The authors should justify their conclusions using the Chardon reverse engineering method. In addition, more of the raw firing rate profiles should be presented to give a better sense of the data and the quality of the motor unit identification.

Reviewer #3 (Public review):

Summary:

The authors set out to describe how the three basic ingredients of a voluntary motor command, that is excitation, inhibition and neuromodulation, are altered in people with multiple sclerosis. They recorded high-density surface electromyograms from tibialis anterior and soleus during slow triangular contractions in 89 patients and 34 control participants, decomposed the signals into the discharge times of individual motor units, and extracted seven features of the resulting discharge patterns. These features were then combined into three composite scores intended to represent the three ingredients, and the authors asked whether the scores in patients differ from those in controls in their average value, in their spread, and in the shape of their distribution. The working hypothesis was that the motor command is disturbed in different ways in different patients rather than in one consistent direction, and the authors report findings that they consider consistent with that hypothesis.

Strengths:

The dataset is the main strength of this work, and it is quite a considerable one. Studies of motor unit behaviour in neurological populations are usually built on ten or twenty participants, whereas here there are 89 patients spanning the full range of disability together with 34 controls, two muscles per person, and more than 12,000 unique motor units. The clinical characterisation is thorough and includes disease subtype, symptom duration, disability score, walking tests and lesion locations from clinical imaging. The experimental protocol is appropriate and was clearly demanding to deliver in a population of this kind. The treatment of the motor unit data in the statistical models is also more careful than is common in this literature, in particular the recognition that motor unit labels are arbitrary and the correct nesting of units within muscle and within participant. The question itself is well worth asking, and the central observation, that average discharge rates and rate modulation are reduced while a minority of patients sit far above anything seen in controls, is interesting.

Weaknesses:

Three issues limit how far the conclusions can be taken as the paper stands.

First, the framework that gives the paper its title is not the framework that was used. The published approach searches a large database of simulations to find the combination of synaptic inputs that reproduces an observed discharge pattern. What is applied here is a weighted average of standardised discharge features, in which the weights are mutual information values taken from that earlier work and the signs of the weights are supplied by the authors on the basis of the literature. Mutual information measures how much a feature tells you about a parameter. It does not carry the direction of that association, and it does not become a regression coefficient by having a sign attached to it. Since the direction of every conclusion in the paper depends on those assigned signs, the reader has no way of judging how faithfully the composite scores track the physiological quantities they are named after. I do not understand why this was not done with simulations. The authors should expand or clarify this point.

The simulations have known inputs, so the composites could be computed on the simulated discharge patterns and their accuracy reported directly. Two of the features were also modified relative to those simulations, being computed against joint torque rather than against the synaptic drive, and using a normalised version of the hysteresis measure, while the weights derived for the original features were retained. The authors should justify this or specify why this was not done and how it affects the underlying physiology.

A related difficulty is that the three composite scores are not independent of one another. The hysteresis measure contributes to all three of them and several other features contribute to two. Finding abnormality in all three components of the motor command may therefore reflect a single underlying signal expressed three times over. The correlations between the composites are not reported, and without them the reader cannot tell which of these two readings is correct.

Second, the comparison with stroke and with spinal cord injury, which carries much of the novelty of the paper, is asserted rather than demonstrated. A good deal of recent motor unit work in spinal cord injury and in stroke is also omitted, which is an important weakness, and I would encourage the authors to engage with it directly rather than treat those populations as a settled contrast. The abstract and the discussion state that the variability seen here is fundamentally different from the consistency seen in those populations, but no stroke or spinal cord injury data are presented, and no quantitative comparison with published values is offered. This matters because the individual patterns illustrated in the paper, that is, reduced peak discharge rate, compressed rate modulation, a narrowed recruitment range, synchronisation between units, and continued firing after the end of the task, are all well-described features of spastic paresis of other causes. This is commonly observed in people with spinal cord injury as well. One of the three illustrated patients is described as having mild hemiparesis and spasticity. Most of this cohort also carries cervical and thoracic cord lesions in addition to lesions above the spinal cord, so a simpler reading of the spread in these data is a mixture of the mechanisms already known from stroke and from spinal cord injury, present in varying proportions in different patients. That would be a different and considerably less novel conclusion, and it deserves to be considered explicitly. There is a statistical asymmetry in the comparison as well, since the populations described as consistent have been studied in samples of ten to twenty people, and small samples cannot reveal the tails of a distribution.

Third, the central finding of greater variability between patients has plausible alternative explanations that have not been excluded. The value for each participant is the median across the motor units identified in that person, and the number of units varies widely between participants and between the two groups. The precision of a median depends on how many units contribute to it, so precision that differs systematically between groups will inflate the spread of the participant values on its own. In addition, the leg studied was the more affected leg in the patients but the dominant leg in the controls. Choosing the more affected of two limbs is a selection on an extreme value, and it will both shift the patient average and widen the patient distribution for reasons that are purely statistical. Since the Methods state that both legs were recorded, perhaps the authors could consider adding this as an additional control. Finally, the hypothesis is framed in terms of multiple peaks and possible subgroups of patients, yet no test of multimodality is performed, and the density estimates shown do not obviously support it.

The feature carrying by far the largest weight in the neuromodulation score shows no group difference at all, and depending on which version of the hysteresis measure entered the composite, either one or none of its six constituent features differs between groups. The group difference in neuromodulation is also absent in the primary model and emerges only when maximal strength is added as a covariate, a change that the text reports but describes as leaving the results generally unchanged.

Moreover, medication may be assessed more clearly considering the large cohort. Thirty-four of the 89 patients take antispastic drugs, and both tizanidine and baclofen act on the system that the neuromodulation score measures, while others take reuptake inhibitors that act in the opposite direction. A quantitative comparison, with the obvious caveat of confounding by indication, would be needed before the neuromodulation findings can be read as disease-related. The authors should consider this to strengthen the manuscript or justify why this was not done.

Appraisal:

The authors achieve their descriptive aim. They show convincingly that motor unit discharge is altered in multiple sclerosis, that the average change is towards lower discharge rates and reduced rate modulation, and that the patient group is more dispersed than the control group on most measures. I do not think they establish the two claims that give the work its stated significance, that is, that the abnormalities can be attributed specifically to excitatory, inhibitory, and neuromodulatory drive, and that the resulting pattern distinguishes multiple sclerosis from other conditions affecting the same pathways.

Impact and utility:

The findings and dataset are very novel and, once shared, will be a resource for the field, and I would encourage the authors to release the analysis code and the exact weights alongside it. Moreover, it would have been valuable to see more relationships reported between the clinical measures and motor unit behaviour, in particular disability, walking speed, lesion location and medication.

Author response:

We sincerely thank the reviewers and editors for their thorough and constructive comments. Our goal was to study characteristics of excitatory, inhibitory, and neuromodulatory components of voluntary motor commands in participants with multiple sclerosis. To do so, we drew on a novel reverse-engineering framework designed to estimate characteristics of these components from a selection of features extracted from motoneuron discharge patterns (Chardon et al., 2024).

We appreciate the reviewers’ recognition of the considerable size and richness of our dataset of motor unit firing patterns recorded from the tibialis anterior and soleus of 89 participants with MS and 34 neurologically intact controls, the rigor of our statistical analyses, the novelty and importance of our descriptive findings, and the potential value of our dataset and analytical approach to other researchers.

We address reviewer comments and discuss our planned revisions to the paper below.

Concerns about physiological inference of findings from the reverse engineering framework. Reviewer 1, for example:

The principal limitation concerns the physiological interpretation assigned to these discharge patterns. The excitation, inhibition, and neuromodulation variables are not directly measured physiological inputs. They are composite variables derived from several features of motor unit discharge, with each feature weighted according to relationships identified in simulations reported previously by Chardon and colleagues. Thus, the step from observed discharge behavior to specific underlying synaptic mechanisms is necessarily model-dependent. The distinction between these two levels of inference is important for interpreting the main conclusions of the study.

The issue is especially relevant because several different physiological processes can plausibly influence the same discharge features. … The interpretation of inhibition is a particularly clear example of this general inverse problem.

We acknowledge all three reviewers’ concerns that, as with any computational model, assumptions of the Chardon reverse engineering framework place limits on the physiological interpretation of our findings.

However, two key strengths of the reverse engineering framework warrant further emphasis when considering the extent to which model assumptions limit physiological inference: (1) the extensive experimental foundation and physiological realism of the motoneuron model, and (2) the demonstrated ability to address the problem of non-uniqueness when predicting characteristics of synaptic and neuromodulatory inputs from discharge patterns of model motoneurons. We will discuss these strengths and clarify the physiological properties that are modeled.

(1) Physiological foundation of the motoneuron model

We first highlight the extensive experimental foundation underlying the motoneuron model. The Heckman laboratory has devoted considerable effort to the development of a physiologically realistic motoneuron model. The motoneuron model is grounded in over 20 years of in situ voltage-clamp experiments that characterized motoneuron intrinsic properties and their responses to excitatory, inhibitory, and neuromodulatory inputs (e.g., Lee & Heckman, 1998a, 1998b, 1999; Kuo et al., 2003; Hyngstrom et al., 2008). The resulting model motoneurons used for simulations were described by the authors as “designed to closely recreate behaviors documented in our extensive database of current and voltage clamp studies in motoneurons within animal preparations” (Chardon et al., 2024).

Moreover, the motoneuron model did not originate with the reverse-engineering framework used in the present study. Earlier versions were refined and evaluated across multiple prior publications (Kim et al., 2009; Kim & Jones, 2012; Powers et al., 2012; Powers & Heckman, 2015, 2017; Beauchamp et al., 2023) and optimized to balance realism with computational speed.

While the motoneuron model’s extensive experimental foundation does not eliminate the inherent limitations of model-based inference, it constrains the model substantially and reduces the extent to which model predictions depend on arbitrary or purely theoretical assumptions.

(2) Addressing non-uniqueness in the inverse problem

We next discuss whether characteristics of motoneuronal inputs can be inferred with reasonable accuracy from motoneuron discharge patterns. This is the inverse problem for which non-uniqueness is a fundamental challenge.

Non-uniqueness must be considered when making physiological inferences from motor unit discharge. Many motoneuron discharge features, when examined on their own, are sensitive to changes in more than one type of motoneuronal input. For example, discharge hysteresis is affected by both the level of neuromodulation and the pattern of inhibition relative to excitation. Reviewers were concerned that the reverse engineering framework also suffered from non-unique solutions when predicting motoneuronal inputs from discharge patterns, suggesting as an example that changes to motoneuron firing patterns resulting from a change in the pattern of inhibition could also result from changes to the level of neuromodulation.

However, we must distinguish the general non-uniqueness problem from the performance of the reverse engineering framework specifically. Non-uniqueness is not an unaddressed weakness of the reverse-engineering framework. Rather, it is the central problem that the framework was explicitly designed to address.

The ability of the reverse-engineering framework to address and substantially reduce non-uniqueness was explicitly evaluated in Chardon et al., (2024) using simulated, known inputs to a population of model motoneurons. Characteristics of motoneuronal inputs (i.e., the distribution of excitatory inputs across motoneurons of different size, the pattern of inhibition relative to excitation, and the level of neuromodulation) were systematically varied in all combinations, generating a “library” of simulated spike trains from each model motoneuron. Seven reverse engineering features were calculated on the simulated discharge patterns and used to train linear and non-linear machine learning models. When applied to test data, the machine learning models predicted the known inputs with substantial accuracy. 

The key finding of Chardon et al. is that the discriminatory power of the reverse engineering framework comes from the collection of discharge features used in the inverse problem. Because individual features can be sensitive to more than one type of motoneuronal input, considering the features as an ensemble drastically reduces the inverse problem’s solution space. This allows the known inputs to the simulated model motoneurons to be predicted with substantial accuracy.

(3) Modeled and unmodeled parameters

Having discussed the unique strengths of the reverse engineering framework, we also acknowledge its limitations, as do the authors of the approach. There are aspects of motoneuronal inputs and intrinsic properties that were held constant or not explicitly represented in the Chardon et al. (2024) simulations. Therefore, application of the current framework to human motor unit data introduces uncertainty regarding the extent to which variation in unmodeled parameters may influence discharge patterns and, by extension, the inferred characteristics of excitatory, inhibitory, and neuromodulatory inputs.

However, we consider this uncertainty in the context of the physiological information gained by applying the framework to human data. Specifically, the framework allows us to interpret motor unit discharge patterns recorded in vivo using relationships established through systematic manipulation of physiologically grounded model parameters.

Thus, despite its limitations, the framework provides substantially greater physiological insight than can be obtained by interpreting individual motor unit discharge features in isolation and represents an important step toward identifying the combinations of motoneuronal inputs that contribute to typical and pathological motor unit discharge in humans. Further variation in these parameters could be incorporated in future iterations of the framework, as is possible with more complex models like that of Mousa and Elbasiouny (2026).

With regard to the inclusion of specific parameters, we discuss two points:

(A) It is necessary to clarify that some physiological properties identified in the reviewer comments as unmodeled were, in fact, represented in the motoneuron model and simulations used in Chardon et al. (2024).

Perhaps the most crucial is the presence of persistent inward currents (PICs), whose realistic representation is a defining feature of the motoneuron model due to its four dendritic compartments with varied Ca²⁺ channel densities (Powers & Heckman, 2017). In fact, in Chardon et al. (2024), the PIC-induced non-linearities in simulated firing patterns were not merely present in the model, but were a key factor contributing to the success of the reverse engineering framework.

Contributions from afferent inputs were also considered within the synaptic inputs to the model. The simulated distributions of excitatory input and patterns of inhibitory input both encompass potential contributions from afferent sources in addition to descending and spinal sources (Binder et al., 2002; Johnson et al., 2017; Chardon et al., 2024). Further, the opposing influence of inhibitory input from any source on facilitation of PICs is incorporated in the simulations.

(B) We agree with the reviewers that it is important to consider potential between-group differences in intrinsic membrane properties that affect PIC behavior in addition to neuromodulatory input. This is a point that we did not adequately discuss in the initial version of the paper. However, we see this primarily as a lack of precision in our discussion of the physiology rather than a deficit related to unmodeled parameters in the Chardon et al. model.

In the Chardon et al. (2024) simulations, the level of neuromodulatory input was varied with the intent of scaling PIC amplitude accordingly. This was done by changing the density of dendritic PIC channels to simulate how the number of channels activated scales with the amount of neuromodulatory input (i.e., serotonin and norepinephrine) present. Thus, the reverse-engineering features that predicted the simulated level of neuromodulatory input in the Chardon study more directly predicted the resulting variation in simulated PIC amplitude. Because other intrinsic membrane properties (e.g., NaV conductance) were held constant, changes in PIC amplitude could be attributed specifically to the simulated level of neuromodulatory input.

In contrast, variation in PIC amplitude estimated from our experimental data cannot be attributed to differences in neuromodulatory input alone. We must also consider other factors that affect PIC amplitude that are unknown in our participants, including differences in intrinsic membrane properties. Importantly, several discharge features incorporated into the reverse engineering framework, including braceheight and delta-F, are commonly used to estimate PIC amplitude in human motor unit recordings, independently of the Chardon et al. (2024) framework (e.g., Mesquita et al., 2024).

Thus, to the extent that brace height, delta-F, and the “neuromodulation” composite variable reflect PIC amplitude in our human data, our findings reflect variation in the physiological factors that determine PIC amplitude. These include neuromodulatory input and intrinsic membrane properties, as well as, for delta-F in isolation, the pattern of inhibition. In our revised version of the manuscript, we will incorporate discussion of these additional factors in addition to our current discussion of neuromodulatory input and the evidence for its alteration in MS.

Finally, while identifying the specific motoneuron inputs and properties that contribute to altered PIC amplitude among participants with MS is an important goal for future work, uncertainty regarding those contributors does not preclude the potential functional significance of our finding that estimated PIC amplitude can be abnormally high or low among participants with MS. Because PIC amplitude directly influences motoneuron excitability, abnormally high or low PIC amplitudes could contribute substantially to motor deficits that emerge in MS and their variation across patients.

Calculation of composite variables using mutual information scores. Reviewer 3, for example:

Mutual information measures how much a feature tells you about a parameter. It does not carry the direction of that association, and it does not become a regression coefficient by having a sign attached to it. Since the direction of every conclusion in the paper depends on those associated signs, the reader has no way of judging how faithfully the composite scores track the physiological quantities they are named after. I do not understand why this was not done with simulations. … The simulations have known inputs, so the composites could be computed on the simulated discharge patterns and their accuracy reported directly.

A related difficulty is that the three composite scores are not independent of one another. The hysteresis measure contributes to all three of them and several other features contribute to two. Finding abnormality in all three components of the motor command may therefore reflect a single underlying signal expressed three times over. The correlations between the composites are not reported, and without them the reader cannot tell which of these two readings is correct.

We agree that mutual information scores are non-directional and that they are not regression coefficients. We did not use them as regression coefficients; rather, we used them as an informed way to create a weighted average of the reverse engineering features that, individually, were most informative about each type of input. Signs were assigned to each feature to reflect the direction of the relationship between the feature and each type of input. We took great care to base the assigned signs on published literature, and we will update the paper to more explicitly link each assigned sign to the supporting literature.

Thank you for the useful suggestion to calculate our composite scores from the simulated features from Chardon et al. and compare them with the known input values. We obtained these data from the authors and found that two of the signs in the excitation composite (for torque at recruitment and duration) needed to be updated based on the Chardon data. After correcting these signs, the neuromodulation and excitation composite variables were both highly correlated with their respective known inputs (r > 0.85). Correction of the two signs in the excitation composite did not change the general pattern of our results. The inhibition composite variable was moderately correlated with the known input (r = 0.50), consistent with the original Chardon et al. study showing that machine learning predictions of inhibition using non-linear regression greatly outperformed those using linear regression. We will include these validation results in the revised manuscript, as they provide a direct assessment of how well the composite variables reflect the known inputs.

We agree that it is important to demonstrate the independence of the composite variables, and it was an oversight on our part not to include that information. The correlations among the composite variables were very low, ranging from r = 0.006 to r = 0.18, with none reaching statistical significance. As discussed, many of the discharge features are sensitive to more than one type of input, which is why it is difficult to interpret them physiologically in isolation. However, the different types of input affect the features in different ways. For example, push-pull/reciprocal inhibition increases both delta-F and rate attenuation slope compared with uniform inhibition. In contrast, an increase in neuromodulation increases delta-F but decreases rate attenuation slope. The signs within the composite variable calculations reflect these differences in the relationships between each feature and the inputs. Therefore, the shared constituent features do not necessarily result in strongly correlated composite variables, as confirmed by the very low correlations observed in our data.

We are now collaborating with the authors of Chardon et al. (2024) to explore application of their reverse engineering framework (and/or its subsequent updates currently under development) to our MS and control data to supplement our mutual information-based composite variable analyses.

Comparison with stroke and spinal cord injury populations, Reviewer 3.

Second, the comparison with stroke and with spinal cord injury, which carries much of the novelty of the paper, is asserted rather than demonstrated. A good deal of recent motor unit work in spinal cord injury and in stroke is also omitted, which is an important weakness, and I would encourage the authors to engage with it directly rather than treat those populations as a settled contrast. The abstract and the discussion state that the variability seen here is fundamentally different from the consistency seen in those populations, but no stroke or spinal cord injury data are presented, and no quantitative comparison with published values is offered. This matters because the individual patterns illustrated in the paper, that is, reduced peak discharge rate, compressed rate modulation, a narrowed recruitment range, synchronisation between units, and continued firing after the end of the task, are all well-described features of spastic paresis of other causes.

Although our discussion of the literature closely reflects the quantitative results from those populations, we agree that directly presenting those values would better support the comparison. We therefore will include quantitative comparisons with published values in these populations, to the extent that they are available, in the revised paper.

We are uncertain which additional studies the reviewer has in mind when stating that a “good deal of recent motor unit work in stroke and spinal cord injury is omitted, which is an important weakness.” We cited the published papers most directly relevant to our specific line of inquiry. Nonetheless, we will conduct an additional review of recent literature and update our references and discussion to address any relevant omissions in the paper revision.

Finally, we agree that the motor unit firing characteristics mentioned above can be found in spinal cord injury and stroke populations, although their prevalence and expression differ between the populations. However, the presence of those characteristics in some of our participants with MS does not undermine the primary distinction we intended to make between our findings and those reported in stroke and spinal cord injury. One novel aspect of our findings is the marked between-participant heterogeneity in motor unit firing patterns within the MS group, as shown in Figure 3. In particular, among participants with MS, alterations in multiple discharge features were not just more variable than controls, but in some cases, they deviated from controls in opposite directions. We will revise the manuscript to make clear that it is this heterogeneity, rather than the presence of any individual discharge characteristic, that distinguishes the patterns observed in our MS sample from the predominantly group-mean difference patterns typically reported in stroke and spinal cord injury. We will support this comparison quantitatively where published data permit.

Further, the motor unit characteristics listed by the reviewer were not our only finding. Another novel aspect of our findings is that a substantial number of participants with MS demonstrated decreases in estimated PIC amplitude. In contrast, recent work in spinal cord injury and stroke reported increases in estimated PIC amplitude during voluntary drive in both groups (Hassan, 2021; Benedetto et al., 2026).

Alternative explanations to increased variability in MS. Reviewers 3, for example:

… the central finding of greater variability between patients has plausible alternative explanations that have not been excluded.

In the revised manuscript, we will address the comments from Reviewers 1 and 3 regarding potential alternative contributors to the increased between-participant variability observed in MS, including motor unit yield, decomposition quality, task performance, anti-spastic medications, and other factors raised by reviewers.

Group differences in the constituent features of the composite variables. Reviewer 3:

The feature carrying by far the largest weight in the neuromodulation score shows no group difference at all, and depending on which version of the hysteresis measure entered the composite, either one or none of its six constituent features differs between groups.

There are two main points to clarify. First, a statistically significant group-mean difference in a composite variable does not require the individual constituent features to also have statistically significant group mean differences. As discussed, many of the features are sensitive to more than one type of motoneuronal input. Further, Chardon et al. (2024) demonstrated that prediction of motoneuronal inputs improved as more features were included.

Second, the absence of a group-mean difference in an individual discharge feature or composite variable does not imply that the feature or composite variable is unchanged among participants with MS. This is particularly important given that many of our MS distributions included values that deviated in opposite directions from controls, which can result in little or no difference in the group mean despite substantial differences at the individual level. For this reason, our analyses compared MS and control distributions not only in terms of central tendency but also their spread and shape.

Relationships between motor unit discharge features, composite variables, and clinical measures. Reviewer 3, for example:

Moreover, it would have been valuable to see more relationships reported between the clinical measures and motor unit behaviour, in particular disability, walking speed, lesion location and medication.

We agree that exploring these relationships is a logical next step and will be especially valuable for understanding the heterogeneity of discharge features and composite variables among participants with MS. The primary goal of the present study was to characterize this heterogeneity, whereas determining how specific clinical characteristics relate to that heterogeneity represents a substantial additional question. We are currently preparing a follow-up paper focused specifically on these relationships so that we have sufficient space to present and discuss them thoroughly. As mentioned above, however, in the paper revision, we will include more information about data from participants who were and were not taking anti-spastic medications.

In our revision, we will also use more cautious terminology when discussing physiological inference vs. motor unit phenotypes, where appropriate, revise the denominator used in calculating the weighted averages, add a formal test of multimodality, add more raw motor unit firing traces, and address the reviewers’ other outstanding minor suggestions.

Beauchamp JA, Pearcey GEP, Khurram OU, Chardon M, Wang YC, Powers RK, Dewald JPA & Heckman C (2023). A geometric approach to quantifying the neuromodulatory effects of persistent inward currents on individual motor unit discharge patterns. J Neural Eng 20, 016034.

Benedetto A, Jenz S, Farley M, Heit B, Sangari S, Beauchamp JA, McPherson L, Heckman C, Perez M & Pearcey G (2026). Muscle-specific motor unit firing characteristics in elbow flexors and extensors after cervical spinal cord injury. ; DOI: 10.64898/2026.06.03.729825. Available at: http://biorxiv.org/lookup/doi/10.64898/2026.06.03.729825 [Accessed June 9, 2026].

Binder MD, Heckman CJ & Powers RK (2002). Relative Strengths and Distributions of Different Sources of Synaptic Input to the Motoneurone Pool. In Sensorimotor Control of Movement and Posture, ed. Gandevia SC, Proske U & Stuart DG, pp. 207–212. Springer US, Boston, MA. Available at: https://doi.org/10.1007/978-1-4615-0713-0_25.

Chardon MK, Wang YC, Garcia M, Besler E, Beauchamp JA, D’Mello M, Powers RK & Heckman CJ (2024). Supercomputer framework for reverse engineering firing patterns of neuron populations to identify their synaptic inputs. eLife. Available at: https://elifesciences.org/articles/90624 [Accessed November 2, 2024].

Hassan AS (2021). Changes in Motor Unit Firing Patterns as a Function of Age, Muscle, and Following a Unilateral Brain Injury: Ionotropic and Metabotropic Effects (Ph.D. thesis). Northwestern University, United States -- Illinois. Available at: http://libproxy.wustl.edu/login?url=https://www.proquest.com/dissertations-theses/changes-motor-unit-firing-patterns-as-function/docview/2572564921/se-2?accountid=15159.

Hyngstrom AS, Johnson MD & Heckman CJ (2008). Summation of Excitatory and Inhibitory Synaptic Inputs by Motoneurons With Highly Active Dendrites. J Neurophysiol 99, 1643–1652.

Johnson MD, Thompson CK, Tysseling VM, Powers RK & Heckman CJ (2017). The potential for understanding the synaptic organization of human motor commands via the firing patterns of motoneurons. J Neurophysiol 118, 520–531.

Kim H & Jones KE (2012). The retrograde frequency response of passive dendritic trees constrains the nonlinear firing behaviour of a reduced neuron model. PloS One 7, e43654.

Kim H, Major LA & Jones KE (2009). Derivation of cable parameters for a reduced model that retains asymmetric voltage attenuation of reconstructed spinal motor neuron dendrites. J Comput Neurosci 27, 321–336.

Kuo JJ, Lee RH, Johnson MD, Heckman HM & Heckman CJ (2003). Active Dendritic Integration of Inhibitory Synaptic Inputs In Vivo. J Neurophysiol 90, 3617–3624.

Lee RH & Heckman CJ (1998a). Bistability in Spinal Motoneurons In Vivo: Systematic Variations in Persistent Inward Currents. J Neurophysiol 80, 583–593.

Lee RH & Heckman CJ (1998b). Bistability in Spinal Motoneurons In Vivo: Systematic Variations in Rhythmic Firing Patterns. J Neurophysiol 80, 572–582.

Lee RH & Heckman CJ (1999). Enhancement of Bistability in Spinal Motoneurons In Vivo by the Noradrenergic α₁ Agonist Methoxamine. J Neurophysiol 81, 2164–2174.

Mesquita RNO, Taylor JL, Heckman CJ, Trajano GS & Blazevich AJ (2024). Persistent inward currents in human motoneurons: emerging evidence and future directions. J Neurophysiol 132, 1278–1301.

Mousa MH & Elbasiouny SM (2026). A Multi-Scale, High-Fidelity Computational Model of the Mouse Triceps Surae Motor Pool.

Powers RK, ElBasiouny SM, Rymer WZ & Heckman CJ (2012). Contribution of intrinsic properties and synaptic inputs to motoneuron discharge patterns: a simulation study. J Neurophysiol 107, 808–823.

Powers RK & Heckman CJ (2015). Contribution of intrinsic motoneuron properties to discharge hysteresis and its estimation based on paired motor unit recordings: a simulation study. J Neurophysiol 114, 184–198.

Powers RK & Heckman CJ (2017). Synaptic control of the shape of the motoneuron pool input-output function. J Neurophysiol 117, 1171–1184.

eLife Assessment

This study presents a useful finding on the interplay of CCL5 and miR-324-5p during ischemic stroke injury. The authors have addressed some of the concerns in the revision. The methodological information has been updated, and additional experiments, such as FISH, have been included. After the revision, the evidence supporting the claims of the authors is solid. The work will be of interest to neuroscientists working on brain injury such as stroke.

Reviewer #1 (Public review):

Summary:

Here, the authors attempt to show that CCL5 is increased after stroke, possibly due to decreased miR-324, and that this is a modifiable system to decrease stroke damage. By bidirectionally manipulating CCL5 levels through direct injection of CCL5; a CCL5 blocking antibody; miR324; miR324 antagomir; or CCR5-blocking Maraviroc, they broadly show improvement with lower CCL5 levels. This includes infarct size, behavioral analysis, and immunohistochemical analysis of astrocytes, microglia, and neurons. They further try to mechanistically tie miR324 and CCL5 in astrocytes specifically to stroke-induced changes using a neuronal/astrocytic coculture system. They argue that decreasing CCL5 leads to increased ERK and CREB phosphorylation as a potential neuroprotective mechanism. CCL5 is one potential ligand for CCR5, and recent work identified CCR5 as a targetable mechanism by clinically-approved drug Maraviroc to enhance stroke recovery. Particularly given the high level of interest in CCR5 in stroke recovery, the focus on CCL5 - one of CCR5's potential ligands - and its miR regulation is an exciting expansion of this area of stroke biology.

Strengths:

The authors' findings that decreasing CCL5 acutely after stroke shows behavioral improvement appear robust. This broadly replicates work from other groups, although the finding that miR324 manipulation can phenocopy direct CCL5 manipulation is novel and intriguing. However, many of their other claims are difficult to evaluate based on a combination of missing methodological information, inappropriate statistical testing, and a flawed culture system.

Weaknesses:

Broadly speaking, the manuscript takes a zoomed-out view of what is fundamentally highly localized biology.

(1) miRNA-based regulation, by definition, has to include miR and mRNA in the same cell type; as the authors note, CCL5 is expressed in many cells. It is therefore impossible to propose any interaction on the basis of the tissue-level changes described; any evidence of in vivo cell-type specificity would dramatically improve the claims.

(2) The authors treat an extensive area of ipsilesional cortex uniformly as "IP". Astrocytic and microglial responses to localized injuries such as stroke are highly location-dependent and undoubtedly change dramatically within this area. The presented data cannot be interpreted without confirmation that these were taken at identical distances from the injury, and what that distance was. These do not appear to be adjacent to the injury, where the responses would presumably be the most informative. Similarly, it is difficult to interpret the neuronal Sholl and spine data without more information on where within the large IP region these neurons were found.

The authors attempt to narrow in on cell-type specificity via culture. However, astrocytes are notoriously prone to a dramatic change in culture and require careful methods (immunopanning; see eg doi: 10.1016/j.neuron.2011.07.022) to maintain much resemblance to their in vivo counterpart. It is difficult to conclude much about the role of astrocytes in the CCL5 pathway based on the use of this shaking-based culture system, particularly in the absence of cell-type specific validation in vivo.

There is missing methodological information, including infarct size measurements, TUNEL staining, and statistical testing. The TTC figures look very odd, like a collection of overlapping stars have been placed on the images rather than the natural relatively smooth infarct edges one would expect. It is unclear if the infarct volume measurements accounted for edema, as is standard; there is no description of the protocol used for quantification. It is also unclear if the infarct volume measurement comparisons were also done with t-tests vs ANOVA, as the statistical test used is not listed in the figure legends. In numerous cases where statistical testing is listed, repeated t-tests between subgroups are used vs the more appropriate ANOVA (assuming normality; nonparametric testing as appropriate), making it difficult to have confidence in the results.

Reviewer #2 (Public review):

The authors presented evidence from various in vivo and in vitro experiments demonstrating the mutual interaction between CCL5 and astrocytic miR-342-5p in the ipsilateral core of cerebral ischemia. However, miR-342-5p was downregulated only late after MCAO (D3-7). Additionally, this downregulation was observed not only in the ipsilateral core but also in the ipsilateral penumbra and contralateral sides. Therefore, it is not convincing that the upregulation of CCL5 in the ipsilateral core at later time points (D3 and D7) is attributable to the decreased expression of miR-342-5p. In particular, infarct injury was already evident within a short time period (say 24 h) following MCAO.

(1) The temporal and spatial expression patterns of miR-324-5p do not match those of CCL-5, especially at D1 and D3 (see Figure 1C, 1D). Despite the inverse relationship between miR-324-5p and CCL-5 expression at D7 after MCAO, what was the purpose of administering miR-324-5p agomir (or antagomir) at D1 post-MCAO? If the connection cannot be clearly established, the conclusion reached at the end will be difficult to accept.

(2) Would administering miR-342-5p or anti-CCL5 at later time points (e.g., after D3) reduce infarct size or improve functional recovery? If this is not the case, the effect of CCL5 on neuronal cell damage (infarct size formation) must occur within a very short time after MCAO. Additionally, if the increased CCL5 expression is due to the downregulation of miR-342-5p, its impact would likely be less significant.

(3) While the study offers valuable insights into the roles of CCL5 and its connection with the regulation of miR-342-5p (though this connection is somewhat weak), it is recommended that the authors explore potential translational applications of these findings.

Overall, given the experimental designs and results, it is difficult to support the conclusions drawn in the manuscript.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

(1) miRNA-based regulation, by definition, requires the miRNA and its target mRNA to be present in the same cell type. CCL5 is expressed in many cell types, making it impossible to propose any miRNA–mRNA interaction based solely on tissue-level expression changes. In vivo cell-type specificity data would substantially strengthen the claims.

We fully agree with this critique, and we consider this the most critical point to address. To directly examine cell-type-specific expression in vivo, we performed FISH combined with immunofluorescence for GFAP (astrocytes) or IBA1 (microglia) in the peri-infarct region of MCAO mice at D3 (Figure 1E–G). Our results show that Ccl5 mRNA is expressed in both GFAP-positive astrocytes and IBA1-positive microglia, with comparable co-localization rates across both cell populations. In contrast, miR-324-5p showed a significantly higher positive rate in astrocytes than in microglia (Figure 1G). Given that miR-324-5p is more abundantly expressed in astrocytes, the regulatory capacity of the miR-324-5p/CCL5 axis is predicted to be more pronounced in this cell type. These findings provide direct in vivo evidence supporting an astrocyte-predominant miR-324-5p/CCL5 regulatory interaction in the peri-infarct region. We acknowledge that the current data do not constitute cell-type-specific in vivo manipulation, and we discuss this limitation and propose future directions (including viral or transgenic approaches) in the revised Discussion.

(2) The authors treat an extensive area of ipsilesional cortex uniformly as "IP." Astrocytic and microglial responses to localized injuries such as stroke are highly location-dependent and undoubtedly change dramatically within this area. Data cannot be interpreted without confirmation that samples were collected at identical, defined distances from the injury. Similarly, it is difficult to interpret the Sholl and spine data without knowing where within the large IP region these neurons were found.

We thank the reviewer for identifying this important concern. Upon review, we recognized that part of the tissue samples used for qPCR, ELISA and Western blot in the original submission were collected from an excessively broad region of the ipsilateral cortex, which likely introduced heterogeneity into the data. We have re-collected these samples specifically from the peri-infarct zone, defined as the 1–2 mm cortical rim immediately surrounding the visibly pale infarct core, beginning from the second and third coronal slices from the most rostral aspect of the cerebral cortex. The qPCR, ELISA and Western blot data have been updated accordingly (Figures 1C–D, 2A, 3A, 6A), and the sampling definition has been specified in the Methods. We also confirmed that all immunofluorescence and Golgi staining analyses were performed within this same peri-infarct zone, where cells retain intact morphology and show the most informative between-group differences. This sampling region has now been defined consistently across all in vivo analyses in the revised manuscript.

(3) Astrocytes are notoriously prone to dramatic change in serum-containing culture. The shaking-based culture system makes it difficult to conclude much about the role of astrocytes in the CCL5 pathway, particularly without cell-type-specific validation in vivo.

We acknowledge this limitation. We attempted to implement the immunopanning protocol described by Barres et al. (doi: 10.1016/j.neuron.2011.07.022) to obtain a more purified astrocyte culture. However, several essential reagents — including sodium selenite, putrescine, and N-acetyl-L-cysteine — could not be procured due to import and purchasing restrictions in our region. The use of a commercially available O4 antibody substitute (clone O4, R&D, MAB1326) in place of O4 hybridoma supernatant, combined with the absence of these chemicals, likely contributed to the very low astrocyte yields and poor cell viability observed across multiple independent attempts. We therefore retained the shaking-based isolation and purification method for the present study.

To characterize the composition of our primary cortical astrocyte cultures, we have included immunofluorescence data from co-labeling of GFAP with Tuj1, Olig2, and IBA1 at P0 and P1 in Supplementary Figure S4, confirming that GFAP-positive cells comprised approximately 88% of total cells at P1. We have also added a paragraph to the Discussion acknowledging that more refined culture systems, as well as cell-type-specific in vivo manipulation of CCL5 and miR-324-5p via viral or transgenic approaches, would further consolidate the conclusions of the present study.

(4) Missing methodological information, including infarct size measurements, TUNEL staining, and statistical testing.

We apologize for these omissions. Detailed descriptions of infarct volume quantification (including the edema-correction formula), TUNEL staining procedures, NeuN/TUNEL co-labeling, and all statistical tests have been added to the Methods section.

(5) The TTC figures appear unusual, with infarct edges resembling overlapping stars rather than natural smooth boundaries. It is unclear whether infarct volume measurements accounted for edema, and no quantification protocol is described.

We apologize for the confusion. The unusual appearance of the infarct edges in the original TTC figures resulted from dotted-line annotations we had added to highlight the infarct boundaries; these have now been removed to present the unmodified TTC images (Figure 2B, 3B). Infarct volume was corrected for edema-induced hemispheric swelling using the formula: [(contralateral hemisphere volume − ipsilateral non-infarcted volume) / contralateral hemisphere volume] × 100%. This formula and the complete quantification protocol have been added to the Methods section.

(6) Repeated t-tests between subgroups are used instead of the more appropriate ANOVA, making it difficult to have confidence in the results.

We agree. We identified that t-tests had been applied inappropriately in the original qPCR and ELISA analyses. All qPCR and ELISA data have been re-analyzed using two-way ANOVA with Tukey's post-hoc test, as appropriate for datasets with multiple groups and time points. We have also reviewed all other figures and corrected any inappropriate use of t-tests.

Reviewer #2 (Public review):

(1) The temporal and spatial expression patterns of miR-324-5p do not match those of CCL5, especially at D1 and D3. Despite the inverse relationship between miR-324-5p and CCL5 being apparent only at D7 after MCAO, what was the purpose of administering miR-324-5p agomir (or antagomir) at D1 post-MCAO? If the connection cannot be clearly established, the conclusion reached at the end will be difficult to accept.

We thank the reviewer for identifying this critical issue. Upon re-examination, we recognized that the original qPCR and ELISA samples had been collected from an excessively broad region of the ipsilateral cortex, which likely introduced heterogeneity into the expression data and obscured the true temporal dynamics in the viable peri-infarct tissue. We have re-collected samples specifically from the peri-infarct zone — defined as the 1–2 mm cortical rim immediately surrounding the infarct core — and updated the data accordingly (Figures 1C–D, 2A, 3A).

The updated data reveal a clearer and more consistent temporal pattern. Ccl5 mRNA levels in the IP region are significantly elevated as early as D1 compared with sham controls, and continue to increase progressively through D7. Regarding miR-324-5p, although IP region levels at D1 do not yet differ significantly from sham controls, they are already significantly lower than in the contralateral CP region at this early time point. This ipsilateral-versus-contralateral difference at D1 indicates that miR-324-5p downregulation begins in the acute phase following stroke, even before it reaches statistical significance relative to the sham baseline. From D3 onwards, miR-324-5p levels in the IP region are significantly reduced relative to both sham and CP groups, coinciding with the period of sustained and progressive CCL5 upregulation. Taken together, these updated findings support an early and progressive inverse relationship between miR-324-5p and CCL5 in the peri-infarct cortex following MCAO, consistent with our previously published finding that miR-324-5p suppresses astrocytic CCL5 expression (Sun et al., Cell Death Dis., 2019), and functionally validated by the ELISA data showing that miR-324-5p agomir injection significantly reduces CCL5 protein concentrations in the IP region at D3 and D7 (Figure 3A).

Regarding the rationale for administering miR-324-5p agomir at D1: this timing was chosen to model a clinically realistic therapeutic scenario targeting the early post-stroke period. MicroRNA agomir/antagomir interventions typically require 2–10 days to achieve peak target gene modulation in the mouse brain; administration at D1 therefore ensures that meaningful miR-324-5p-mediated suppression of CCL5 is achieved during the critical acute-to-subacute transition period, as confirmed by the ELISA results at D3 (Figure 3A).

(2) Would administering miR-342-5p or anti-CCL5 at later time points (e.g., after D3) reduce infarct size or improve functional recovery? If this is not the case, the effect of CCL5 on neuronal cell damage must occur within a very short time after MCAO. Additionally, if the increased CCL5 expression is due to the downregulation of miR-342-5p, its impact would likely be less significant.

We acknowledge that the current study did not include experimental groups with delayed administration, and we recognize this as a limitation.

However, several points inform our interpretation. First, as described in our response to Comment 1 above, the updated data demonstrate that miR-324-5p downregulation in the IP region is already detectable relative to the contralateral CP region at D1 and progresses further through D3 and D7, indicating that the miR-324-5p/CCL5 regulatory axis is engaged from the acute phase of stroke, providing a biological basis for early intervention. Second, in experimental stroke models, the infarct core is largely established within the first 24–72 h following vessel occlusion, with the majority of ischemic neuronal death occurring during this window. CCL5, as a pro-inflammatory mediator, is expected to amplify immune cell recruitment and inflammatory cascades most consequentially during this early period, making D1 administration mechanistically rational for limiting neuronal loss. Third, the superior early behavioral outcomes in the CCL5 antibody group relative to the miR-324-5p agomir group (Figures 2D–E, 3D–E) support the value of early CCL5 suppression. As the antibody acts immediately while the agomir requires time for post-transcriptional regulation, this difference highlights that timely agomir delivery is essential to achieve effective CCL5 suppression during the critical early window.

(3) The study would benefit from the exploration of potential translational applications.

We thank the reviewer for this constructive suggestion. We are planning to investigate whether astrocyte-derived extracellular vesicles engineered to overexpress miR-324-5p can enhance neurological recovery after stroke. Extracellular vesicles offer several translational advantages: they can traverse the blood-brain barrier, provide a stable and biocompatible vehicle for miRNA delivery, and may be less immunogenic than viral approaches. This would leverage the neuroprotective regulatory mechanism identified in the present study while offering a clinically viable delivery strategy.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) In vivo cell-type specificity data for CCL5/miR-324-5p would substantially strengthen the manuscript.

We have performed FISH combined with GFAP and IBA1 immunofluorescence in the peri-infarct region of MCAO mice at D3 to characterize the cell-type-specific in situ expression of Ccl5 mRNA and miR-324-5p (Figure 1E–G).

(2) Cell-type-specific in vivo manipulation of CCL5/miR-324-5p (e.g., virally or transgenically) would further substantiate the conclusions. A more compelling astrocytic culture model is needed.

We fully agree that cell-type-specific in vivo manipulation would represent a major advance. As described in our response to Comment 3 above, we were unable to successfully implement immunopanning in the current study. The FISH data provide in vivo evidence supporting the astrocyte-enriched expression of miR-324-5p in the peri-infarct region. We have added a Discussion paragraph explicitly identifying viral or transgenic astrocyte-specific manipulation of CCL5 and miR-324-5p in vivo as a critical next step to validate and extend the conclusions of this study.

(3) The measurement shown in Figure 4D is unclear. A more informative measure might be TUNEL/DAPI, with additional cell-type-specific markers to identify what cells are dying in what proportions.

We agree with this suggestion and have revised the quantification accordingly. In the updated manuscript, apoptotic cell death is reported as the proportion of TUNEL-positive cells among total DAPI-positive nuclei (TUNEL/DAPI). As a complementary cell-type-specific measure, we quantified the proportion of NeuN-positive neurons among total DAPI-positive nuclei (NeuN/DAPI) to specifically assess neuronal survival within the co-culture system. These two measures together provide a clear and interpretable readout of both overall cell death and neuronal viability under each experimental condition. The revised quantification is presented in Figure 4C–E and Figure 5B–D.

(4) qPCR and ELISA data should be normalized to internal controls, sham values should be presented, and ANOVA (or appropriate non-parametric tests) should be used.

All qPCR data are now normalized to Gapdh (for mRNA) or U6 snRNA (for miRNA). Sham group values are presented in all relevant figures. Two-way ANOVA with Tukey's post-hoc test is now used for all qPCR and ELISA comparisons. The updated statistical approach is summarized in the Statistical Analysis section of the Methods.

(5) The description "within 24 hrs" for the timing of CCL5 in vivo manipulation is ambiguous.

We have revised the description to "at 24 h post-MCAO" throughout the Methods and Results sections to specify the precise time point of intervention.

(6) Only some statistical comparisons are shown in Figure 2E, which inaccurately implies that the other groups are not different.

We have updated Figure 2E and Figure 3E to include all statistically significant pairwise comparisons, ensuring that the significance markers accurately represent the complete set of statistical relationships among all groups.

(7) "Activation" is not the appropriate term for astrocytes in pathological contexts; A1/A2 terminology should be removed.

We thank the reviewer for this important correction. In line with the consensus recommendations by Escartin et al. (Nat Neurosci, 2021), we have replaced all instances of "astrocyte activation" in pathological contexts with "astrocyte reactivity" or "reactive astrogliosis" throughout the manuscript, including the Abstract, Results, and Discussion. All references to A1 and A2 subtypes have been removed from the Discussion.

(8) There are typographical errors, and the repeated use of "Besides" is awkward.

We have carefully proofread the entire manuscript to correct typographical errors. All instances of "Besides" used as a sentence-opening connector have been replaced with contextually appropriate alternatives, such as "Furthermore," "Moreover," "In addition," or "Additionally."

eLife Assessment

This study presents a valuable finding on the condition dependence of autophagy-mediated lifespan regulation in C. elegans. The evidence is solid, as the data broadly support the main claims, although variability between biological replicates in several of the performed experiments and mechanistic exploration with only a limited set of experimental techniques leave some conclusions less firmly established. The work will be of interest to researchers studying autophagy, ageing, and intracellular trafficking.

Reviewer #1 (Public review):

Summary:

Hsiung et al. investigated whether the effects of autophagy gene knockdown on the lifespan of long-lived C. elegans mutants depend on experimental conditions. The authors first compiled published data on autophagy-dependent lifespan regulation in daf-2 and wild-type backgrounds, highlighting that prior results are notably inconsistent and likely context-dependent. They then systematically tested the lifespan effects of RNAi knockdown of six autophagy genes (atg-2, atg-4.1, atg-9, atg-13, atg-18, and bec-1) in wild-type (N2), daf-2 (reduced insulin/IGF-1 signalling), and glp-1 (germlineless) animals, while varying temperature, daf-2 allele, FUDR concentration, and bacterial infection status.

The key findings are as follows. In wild-type animals, lifespan suppression by most autophagy gene knockdowns was more pronounced at 20{degree sign}C than at 25{degree sign}C, where little or no effect was observed. In daf-2 mutants, stronger lifespan suppression was seen in the weaker daf-2(e1368) allele at 20{degree sign}C, but not in the stronger daf-2(e1370) allele, and effects were largely absent at 25{degree sign}C. In glp-1 mutants, four of six gene knockdowns suppressed lifespan to a greater extent than in N2, though again in a temperature-dependent manner. FUDR at a high concentration (800 µM) abolished the life-shortening effects of most knockdowns and, in the case of atg-9 and atg-13, led to lifespan extension. Kanamycin treatment to eliminate bacterial proliferation did not fully account for the lifespan effects, suggesting that increased susceptibility to infection is not the primary mechanism. The authors also tested the programmed aging hypothesis that autophagy promotes lifespan reduction through biomass repurposing, but found no changes in vitellogenin levels upon knockdown of any of the six genes.

Altogether, among all genes tested, atg-18 knockdown produced the strongest and most consistent lifespan suppression across nearly all conditions, including both daf-2 and glp-1 backgrounds. The authors probed whether atg-18 acts through the FOXO transcription factor DAF-16 by examining dauer formation and ftn-1 expression, but found no evidence for this, suggesting a DAF-16-independent mechanism.

Strengths:

The primary strength of this work lies in its systematic and comprehensive approach to dissecting how experimental variables influence the outcome of autophagy-lifespan epistasis tests. The compilation of prior data alongside the authors' own multi-condition dataset is a genuinely useful resource for the field. The study raises a timely and important point about condition selection bias in interpreting autophagy lifespan relationships, with broader relevance to C. elegans studies. The finding that atg-18 behaves distinctly from other autophagy genes across a range of experimental conditions is particularly noteworthy and provides an interesting direction for future mechanistic investigation.

Comments on revised version.

The authors have carefully addressed the concerns raised in the previous review and have incorporated most of the suggested revisions. In particular, the revised manuscript provides greater clarity regarding the effects of glp-1 and FUDR and improves the presentation and interpretation of the experimental findings. The authors have also provided useful clarification regarding the variability between lifespan experiments, the number of biological replicates, and the interpretation of RNAi efficacy.

I have only two minor suggestions concerning the wording and organization of the manuscript, which I have communicated separately to the authors. These points are intended to further improve clarity and accuracy and do not affect my overall assessment of the study.

Overall, I am satisfied with the revisions and consider the manuscript to be a valuable contribution to our understanding of the context dependence of autophagy-mediated lifespan regulation.

Reviewer #2 (Public review):

Summary:

This study examines how genes involved in cellular recycling (autophagy) influence lifespan under different experimental conditions. The findings help clarify why previous studies have reported conflicting results about whether blocking autophagy shortens or extends lifespan. The work will be of interest to researchers studying aging and cellular stress responses, particularly those using model organisms.

Strengths:

The findings are valuable, as they help resolve inconsistencies within a specific subfield of aging research. The evidence presented is solid, as the data broadly support the primary claims of the study. In addition, the discussion is thorough and thoughtfully integrates the findings within the broader context of the field.

Weaknesses:

Additional functional validation would further strengthen the conclusions.

Comments on revised version.

I have reviewed the revised manuscript. Overall, the authors have addressed most of my concerns, and the revised manuscript has been improved. The study provides a comprehensive examination of the context-dependent effects of autophagy gene knockdown on lifespan in C. elegans. The data are solid and provide valuable information for understanding the conflicting results currently reported in the literature regarding the role of autophagy in longevity.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

Hsiung et al. investigated whether the effects of autophagy gene knockdown on the lifespan of long-lived C. elegans mutants depend on experimental conditions. The authors first compiled published data on autophagy-dependent lifespan regulation in daf-2 and wild-type backgrounds, highlighting that prior results are notably inconsistent and likely context-dependent. They then systematically tested the lifespan effects of RNAi knockdown of six autophagy genes (atg-2, atg4.1, atg-9, atg-13, atg-18, and bec-1) in wild-type (N2), daf-2 (reduced insulin/IGF-1 signalling), and glp-1 (germlineless) animals, while varying temperature, daf-2 allele, FUDR concentration, and bacterial infection status.

The key findings are as follows. In wild-type animals, lifespan suppression by most autophagy gene knockdowns was more pronounced at 20˚C than at 25˚C, where little or no effect was observed. In daf-2 mutants, stronger lifespan suppression was seen in the weaker daf-2(e1368) allele at 20˚C, but not in the stronger daf-2(e1370) allele, and effects were largely absent at 25˚C. In glp-1 mutants, four of six gene knockdowns suppressed lifespan to a greater extent than in N2, though again in a temperature-dependent manner. FUDR at a high concentration (800 µM) abolished the life-shortening effects of most knockdowns and, in the case of atg-9 and atg-13, led to lifespan extension. Kanamycin treatment to eliminate bacterial proliferation did not fully account for the lifespan effects, suggesting that increased susceptibility to infection is not the primary mechanism. The authors also tested the programmed aging hypothesis that autophagy promotes lifespan reduction through biomass repurposing, but found no changes in vitellogenin levels upon knockdown of any of the six genes.

Altogether, among all genes tested, atg-18 knockdown produced the strongest and most consistent lifespan suppression across nearly all conditions, including both daf-2 and glp-1 backgrounds. The authors probed whether atg-18 acts through the FOXO transcription factor DAF-16 by examining dauer formation and ftn-1 expression, but found no evidence for this, suggesting a DAF-16-independent mechanism.

We thank the reviewer for their meticulous review of the manuscript. Responding to their comments (particularly those not in the public review) has improved it substantially.

Strengths:

The primary strength of this work lies in its systematic and comprehensive approach to dissecting how experimental variables influence the outcome of autophagy-lifespan epistasis tests. The compilation of prior data alongside the authors' own multi-condition dataset is a genuinely useful resource for the field. The study raises a timely and important point about condition selection bias, which is relevant not only to autophagy research but to C. elegans aging studies more broadly. The finding that atg-18 behaves distinctly from other autophagy genes across all conditions is noteworthy and opens avenues for future mechanistic work.

Weaknesses:

Despite its breadth, the study has several weaknesses that limit the strength of some conclusions.

(1) Variability in control lifespan data. The N2 lifespan values under ostensibly identical conditions (e.g., GFP RNAi at 20˚C) differ substantially across experiments (compare Tables S2, S5, S6, S7, and S9). Since N2 serves as the baseline for calculating whether the effect is greater in long-lived mutants via Cox proportional hazard (CPH) analysis, this variability in controls directly affects the reliability of those comparisons.

Such inter-trial variability in N2 lifespan is not unusual in studies of C. elegans aging, ostensibly identical conditions notwithstanding, and its causes are unknown. A careful C. elegans lifespan study from 2017 compared results of tests performed independently at 3 sites under similar conditions. This showed that inter-trial variation occurred mainly at each site over time, rather than between sites (Lucanic et al., 2017). We note that the data for our study was gathered over a 5-year period. One possibility is that this long time frame may have contributed to the variability seen. 

Regarding statistical tests (whether the log-rank test to assess differences between pairs of populations, or the CPH test to assess differences between effects of treatments under two conditions): comparisons were made either of data within individual trials, or for pooled data. Thus, valid comparisons were made. We did not compare controls from one trial with treatments from another, which would have yielded misleading results. 

(2) Limited biological replication. Most experiments were performed with only two biological replicates. In several cases, the two replicates yield contradictory outcomes: one showing significant lifespan suppression and the other showing no effect or even extension. The authors combine these into cumulative datasets for analysis, which, while not incorrect in principle, may obscure genuine irreproducibility. Given that the central message of the paper concerns variability and condition dependence, additional replication would have substantially strengthened confidence in the reported results.

This is a natural issue to raise. In tests of an effect of a given treatment on C. elegans lifespan, a minimum of 3 trials is standard, and in this lab as a rule we follow this convention. However, after careful consideration during the design stages we opted not to do so for this particular study. Our reasons are set out in the manuscript as follows.

“A methodological note: for tests of effects of a given intervention on C. elegans lifespan an often-applied standard is to include 3 biological replicates. This is true of several recent studies where the effect of knockdown of a single atg gene on daf-2 longevity was studied (Minnerly et al., 2017; Wilhelm et al., 2017; Yang et al., 2024). However, given that the present condition dependence study effectively performs this test in 18 different ways, involving RNAi of 6 atg genes, 2 daf-2 mutants and 2 temperatures, N = 2 biological replicates were judged to be sufficient to draw robust conclusions; similarly, an earlier study of RNAi 14 atg genes under two conditions used 2-3 biological replicates (Hashimoto et al., 2009); for an overview of N sizes in previous studies, see Table S1.”

While this approach has yielded robust broad conclusions (e.g. that atg gene RNAi generally does not suppress daf-2(e1370) Age), it is true that for any one given treatment (say, effects of atg-9 RNAi on daf-2(e1370) Age) one may not draw conclusions with a high degree of confidence, and we do not do so. We therefore argue that in a study of this nature, as for instance in a whole genome RNAi screen for lifespan effects, it is reasonable and expedient to drop below the 3 replicates minimum standard; here we agree with the Nishida lab’s similar judgement, and from that study too robust conclusions may be drawn.

(3) Low sample sizes in individual trials. A number of lifespan assays were conducted with only 40-50 worms per replicate, and in some cases, as few as 30. Such sample sizes are below the standard commonly used in the C. elegans aging field and are likely to contribute to the variability observed.

Please see our response to point 2, which in essence responds to this concern.

(4) RNAi efficacy measured only in N2 at 20˚C. The authors demonstrated that atg-2 and atg-4.1 RNAi did not significantly reduce target mRNA levels, which may explain their weaker lifespan effects. However, these same RNAi treatments significantly affected lifespan in several other conditions (e.g., daf-2(e1368) at 20˚C, glp-1 at 20˚C and 25˚C, and N2 with 15 µM FUDR). Measuring RNAi efficacy across different genetic backgrounds and conditions would be needed to properly interpret these variable results.

The study would indeed be strengthened by inclusion of target atg mRNA measurements under all of the various conditions tested. However, this would have required a very large number of qPCR tests to be run; we note that in previous assessments of atg RNAi effects on lifespan (listed in Table S1 and Table S6), such tests were rarely performed. Regarding RNAi effects on daf-2 mutants, we note in the text the following: “While mRNA levels after RNAi under the various other conditions tested were not assayed, reduced IIS (including daf2(e1370)) has been shown to intensify the RNAi response (Wang and Ruvkun, 2004), thus lack of effect on lifespan in daf-2(e1370) is unlikely to reflect suppression of mRNA knockdown.” Here we have at least assessed, using N2, the most important issue relating to RNAi efficacy: the differential effects of different RNAi feeding clones on atg mRNA levels.

(5) Incomplete mechanistic exploration. The investigation of why atg-18 knockdown has uniquely strong effects was limited to DAF-16. Given published evidence that atg-18 may regulate HLH-30/TFEB, a master transcriptional regulator of autophagy and lysosomal biogenesis, testing whether atg-18 specifically affects HLH-30 nuclear localisation or activity could have provided valuable mechanistic insight and would distinguish atg-18 from the other genes tested.

We would have readily investigated this. However we learned of the interactions between atg-18 and hlh-30 only in Nov 2025, when one of us (David Gems) bumped into a member of Evandro Fan’s research group at an aging meeting at the Crick Institute in London. Their findings were very interesting for us, as they offered a possible explanation for the seeming idiosyncrasy of atg-18 RNAi effects on lifespan. This subject is currently under investigation by the Fan lab at the University of Oslo, who recently posted a preprint describing the work (Schmauck-Medina et al., 2026).

Reviewer #2 (Public review):

Summary:

This study examines how genes involved in cellular recycling (autophagy) influence lifespan under different experimental conditions. The findings help clarify why previous studies have reported conflicting results about whether blocking autophagy shortens or extends lifespan. The work will be of interest to researchers studying aging and cellular stress responses, particularly those using model organisms.

We thank the reviewer for their helpful remarks. Responding to their comments (particularly those not in the public review) has enabled us to improve it.

Strengths:

The findings are valuable, as they help resolve inconsistencies within a specific subfield of aging research. The evidence presented is solid, as the data broadly support the primary claims of the study. In addition, the discussion is thorough and thoughtfully integrates the findings within the broader context of the field.

Weaknesses:

Additional functional validation would further strengthen the conclusions.

We very much agree. Our original plan for this study was to include autophagic flux assays under different conditions. However, this line of investigation led us to a careful reassessment of reporter-based approaches to measuring autophagic flux in C. elegans, and attempts to improve them. This includes development of an automated, AI-based quantitative image analysis pipeline to improve reproducibility and data interpretation across studies. This investigation is still ongoing, and we are currently preparing a separate manuscript focused specifically on methodological clarity and quantitative assessment of autophagic flux.

Recommendations for the authors:

Reviewing Editor Comments:

To increase the evidence provided by the authors, they should at least address the comments from Reviewer 1 regarding the experimental inconsistencies. We acknowledge that the additional experiments suggested by Reviewer 2 regarding the use of C. elegans mutants and additional methods to assess autophagic flux would likely be a lot of additional work. However, adding results from such experiments would, of course, make the evidence more compelling.

Reviewer #1 (Recommendations for the authors):

Writing and presentation

(1) The abstract discusses results for daf-2 in detail but does not mention the glp-1 findings. Given that a substantial portion of the study addresses glp-1 longevity, including a summary of those results in the abstract would better represent the scope of the work.

Agreed. glp-1 is now referred to in the abstract.

(2) In the abstract, the sentence regarding FUDR effects is placed between statements about daf2, while it is referring to N2 lifespans, which may give the impression that the FUDR results were obtained in a daf-2 background. Consider restructuring this section for clarity.

Agreed. To improve clarity this now reads as follows. “In wild-type C. elegans, FUDR at a high concentration caused knockdown of several atg genes to increase lifespan”

(3) The definition of "robust" used in Figure 4E could be misleading to readers. For instance, bec-1 and atg-4.1 knockdowns in glp-1 at 20˚C are classified as robust, yet the actual percent suppression is modest (~8% and ~3%, respectively) and non-consistent in individual replicates. The "robust" designation arises because these knockdowns slightly increased N2 lifespan. Clarifying the definition in the figure legend or text would help readers interpret this correctly.

For glp-1 at 20˚C robust suppression is seen with atg-2, atg-18 and bec-1 RNAi, not atg-4.1 RNAi. But regarding bec-1: yes, the suppression is modest. bec-1 RNAi is an unusual case insofar as it meets the <30% definition partly because it caused an increase in N2 lifespan. Under the circumstances, arguably, it makes little sense to view it as an example of robust suppression, and Figure 4E has been altered accordingly, with a note added to the legend as follows. “Note that the bec-1 RNAi effect on glp-1 at 20˚C is not classified as robust here even though it reduces lifespan to within <30% of the mean lifespan of N2 under bec-1 RNAi, since the fact that it does so partly reflects an increase in N2 lifespan, rather than a robust life-shortening effect on glp-1.”

To try to improve clarity we have altered the definition of “robust” to read as follows. “R, robust suppression, i.e. knockdown reduces the extended lifespan of daf-2 or glp-1 to within <30% of the mean lifespan of N2 under the same RNAi. This designation (“robust”) indicates a high degree of suppression of the mutant longevity phenotype (see Figure 2, Figure 5 and Fig. S2).”

(4) On page 12, the text should read "9/30 suppresses robustly" (currently appears to contain a numerical error).

Fixed. This now reads “In 8/30 the RNAi effect was robust, i.e. the mutant longevity was largely suppressed.” (Now 8/30 since bec-1/glp-1/20˚C is no longer viewed as robust suppression).

(5) In Supplementary Sheet 8, the lifespan data from Hashimoto et al. are presented in a different format than the data from other studies. Standardizing the presentation would improve readability.

This is Supplementary Table 1. The inconsistencies have been ironed out.

Data and calculations:

(6) In Supplementary Table S3, the ΔΔCt values for atg-9 appear to be incorrect. Please verify and correct.

We thank the reviewer for highlighting this error. The ΔΔCt values for atg-9 in Supplementary Table S3 were incorrectly entered; the Fold Change values had been mistakenly placed in that column. This error has now been corrected. Please note that analyses and conclusions reported in the manuscript were based on the correct values.

(7) In Supplementary Table S4, the standard deviation values do not match my independent calculations. Please double-check these values.

This is correct: there was an error in the standard deviation (SD) values. We thank the reviewer for their diligence. We have updated both the SD and SEM (standard error of mean) columns in Table S4. The mean ΔΔCt values remain unchanged, as do the conclusions from analyses and statistical tests.

(8) In Supplementary Table S7 (kanamycin experiment), there appear to be several errors in the percent change calculations. Additionally, the statement "In the absence of Kan, atg-13 RNAi caused a slight reduction in lifespan" is not supported by the combined data, which actually shows a slight increase. Given that the reported changes are subtle but statistically significant, it would be prudent to re-verify the p-value calculations as well before drawing conclusions from this experiment.

The calculation for trial 1 atg-13 (-Kan) as a percentage of control (L4440 Kan) has been corrected so that the mean lifespan of the knockdown is divided by the mean lifespan of the control. In the previous version, the ratio was inadvertently calculated in the opposite orientation. All other values remain unchanged.

Responding further to this point, to strengthen the data here we have also conducted an additional trial, and Figure 3D and Table S7 have been updated accordingly. The more robust data still supports the conclusion that E. coli infection does not mask a life extending effect of atg-13 RNAi. However, in the new, summed data, atg-13 RNAi on no Kan does not shorten lifespan at all, in contrast to our previous trials (conducted several years earlier), but consistent with several other instances of variability in the study. Moreover, the modest life-shortening effect of atg-13 RNAi on Kan (-6.7%) is now statistically significant. The manuscript has been updated accordingly.

Experimental interpretation

(9) On page 9, the authors report testing N2 and daf-2(e1370) lifespan at 15˚C and 20˚C, but only the daf-2 results are discussed in the text. The N2 results at 15˚C appear in Table S5 but are never addressed. Notably, bec-1 knockdown significantly suppressed N2 lifespan at 20˚C in Table S2 but appears to significantly extend it at both 15˚C and 20˚C in Table S5. These discrepancies should be discussed.

The issue of inter-trial variability is discussed in our response to point 1 in the public review. More specifically: here it may be significant that the trials listed in Table S5 were performed several years after those in Table S2. The discrepancy is now noted and discussed as follows. “In these trials bec-1 RNAi also modestly increased N2 lifespan at both temperatures (Table S5), surprisingly given that in previous trials (performed several years earlier) bec-1 RNAi shortened N2 lifespan (Table S2). The reason for this discrepancy is unknown.”

(10) Regarding the CPH analysis of bec-1 in daf-2(e1368) (Figure 1), the authors state that bec-1 knockdown does not have a significantly greater effect in daf-2(e1368) relative to N2, and then note that this is consistent with the earlier observation by Hansen et al. that bec-1 shortens daf-2 lifespan without affecting N2. However, in the cumulative dataset, bec-1 does significantly suppress N2 lifespan. A more precise statement here would prevent readers from drawing an incorrect conclusion.

The point here is that our study and the Hansen et al study both point to atg RNAi suppression of daf-2 Age being limited to class 1 mutants, not that there are no effects on N2. To try to improve clarity we have rephrased as follows. “These findings are broadly consistent with the earlier observation that bec-1 and vps-34 RNAi shortened the lifespan of the daf-2(mu150) class 1 mutant but not of N2 at 20˚C (CPH analysis not performed) (Hansen et al., 2008).”

(11) The reference to Hashimoto et al. (2009) on page 10 states that atg-9 and atg-13 RNAi increased lifespan, but that study does not include data for atg-13. The lifespan extension reported by Hashimoto et al. was for atg-7, atg-9, bec-1, and unc-51. Please correct this citation.

Done. It now reads “where atg-9 (and also atg-7, bec-1 and unc-51) RNAi increased lifespan”.

(12) In a previous publication from this group, atg-2 and atg-13 knockdown with 15 µM FUDR led to significant lifespan extension, whereas in the current study, the same treatments significantly suppressed lifespan. Although this discrepancy is briefly mentioned in the Discussion, a more thorough discussion of possible explanations would strengthen the manuscript's value as a reference dataset for future studies.

A more detailed discussion has been added, as follows. “Regarding the causes of variability between results of ostensibly identical tests performed under ostensibly identical conditions: one clue is provided by a study comparing results of lifespan assays performed across three sites under similar conditions. This revealed that inter-trial variation occurred mainly at each site over time, rather than between sites (Lucanic et al., 2017). One possibility is that this reflects batch variation in media components, such as the BactoPeptone constituent of nematode growth medium (Petrascheck, 2014).”

(13) The N2 mean lifespan on GFP RNAi with 0 µM FUDR at 20˚C is approximately 15 days, whereas the N2 lifespan at 20˚C in Table S2 is approximately 20 days. While inter-experiment variability is expected, a difference of this magnitude warrants acknowledgement, as it could influence the interpretation of subsequent comparisons.

We have now acknowledged this in the legend to Figure 3, as follows. “We note that in (A) the lifespan of the gfp RNAi control is somewhat lower than in other experiments (mean 14.96 days, Table S6); see Discussion for consideration of possible reasons for inter-trial variability.”

(14) Regarding the glp-1 experiments (page 11 and Table S9), the N2 lifespan values in these experiments differ from earlier N2 results at 20˚C for several knockdowns (e.g., bec-1 knockdown appears to increase N2 lifespan in Figure 4). Additionally, the glp-1 lifespan results are not consistent between the two replicates for most genes except atg-2 and atg-18. The developmental shift (raised at 25˚C, then moved to 20˚C to obtain the glp-1 phenotype) could plausibly account for some of this variation compared to animals raised continuously at 20˚C. If so, this should be explicitly discussed.

Agreed. The following has been added to the Figure 4 legend. “That bec-1 RNAi increases N2 lifespan in (A) (+16.7%, p < 0.0001) but not (B) could imply an interaction with temperature during development, or merely variability of atg RNAi effects (see Discussion).”

(15) At 25˚C, N2 lifespan shows significant suppression upon atg-13 and atg-18 knockdown in Table S9, while these same effects were non-significant in Table S2. These and other interexperiment discrepancies should be noted and, if there are identifiable experimental differences, those should be specified.

This discrepancy has now been noted on page 9, immediately after the description of the data in Table S2, as follows. “(although in later tests at 25˚C, life-shortening effects of atg-13 and atg-18 RNAi were seen; Table S9)”

(16) If autophagy is already regulated by heat stress at 25˚C, this could explain the diminished effects of autophagy gene knockdown at higher temperatures. Measuring autophagy gene expression by qPCR across different temperatures and genetic backgrounds could provide useful mechanistic insight.

Agreed. However, more informative will be to measure effects of temperature and genotype on autophagy more directly, using fluorescent reporters of autophagic flux. We are addressing this as part of an ongoing study using improved and fully validated autophagic flux measurement methodologies (please see our response to reviewer 2, public review). 

(17) The authors report that atg-18 knockdown upregulates other autophagy genes (supplementary data). This is intriguing given that atg-18 shows the strongest phenotype. Whether this reflects a compensatory mechanism and why it does not rescue the lifespan suppression deserves further discussion.

On reflection we decided to remove from the manuscript the data relating to effects of atg-18 RNAi on mRNA levels of other atg genes due to concerns about data quality. 

(18) Regarding the FUDR and infection hypothesis: the logic that reduced bacterial infection upon FUDR treatment explains the loss of lifespan suppression is reasonable, but it does not account for why atg-13 knockdown actively extends lifespan in the presence of FUDR. This point could benefit from further discussion. In the kanamycin experiment, they further see that life-shortening effects of atg RNAi are not solely attributable to infection, but the question of lifespan extension remains unanswered.

Good point. We have addressed this as follows. “We also conclude the increase in lifespan upon atg-13 RNAi in the presence of 800 μM FUDR (Figure 3C) is not attributable to suppression of E. coli infection, but rather to some other, unidentified mechanism.”

(19) The timing of when lifespan assays are initiated relative to other experimental treatments is another potential source of variability (as seen from earlier reported data) that could be acknowledged as a consideration for future studies.

Good point. We have added the following to the section of the discussion about tackling condition dependency issues. “Another factor to take into account is the apparent tendency of results of C. elegans lifespan assays to vary over time (Lucanic et al., 2017).”

Reviewer #2 (Recommendations for the authors):

Major Comments:

(1) Assessment of autophagic activity

The authors demonstrate by qPCR that feeding RNAi reduces mRNA levels of autophagy-related genes. However, reduced transcript levels do not necessarily confirm functional inhibition of autophagy. Incorporating established assays of autophagic flux, such as Western blot analysis of lipidated ATG-8 (LGG-1/ATG-8-II) or validated fluorescence-based reporters, would substantially strengthen the mechanistic conclusions.

We agree with the reviewer that direct assessment of autophagic flux would provide additional functional insight. We are currently performing complementary assays to directly assess autophagic flux using reporter-based approaches, with a focus on standardizing reporter-based measurements and developing a quantitative image analysis pipeline to improve reproducibility and interpretation across studies. These analyses will be presented in a separate manuscript focused specifically on methodological clarity and quantitative assessment of autophagic flux.

(2) Use of genetic mutants

Several C. elegans loss-of-function mutants for autophagy genes are available. Validation of key findings using selected genetic mutants, where feasible, would provide complementary evidence and enhance confidence in the RNAi-based results.

In principle is this a good idea. In practice many loss-of-function mutants in core autophagy genes exhibit developmental defects or impaired viability, which can confound interpretation in aging studies. As our aim was to examine the effects of autophagy gene perturbation specifically during adulthood, RNAi provided a practical approach that allowed post-developmental knockdown while minimising disruption of normal development.

Minor Comments:

(1) Context-dependent effects of autophagy

The findings are conceptually consistent with prior work demonstrating dual roles of autophagy in C. elegans survival during starvation, where physiological levels promote survival but insufficient or excessive autophagy contributes to mortality (Kang et al., Genes & Development, 2007). Including a discussion of this study would help frame the present results within a broader biological context.

Good idea to cite this study, and we have now done so in the introduction, as follows. “It is by now clear that autophagy can enhance as well as inhibit the development of pathologies in C. elegans, including senescent ones (Kang et al., 2007).”

(2) Relevance beyond C. elegans

It would be helpful for the authors to clarify whether similar context-dependent effects of autophagy on lifespan have been reported in other organisms. Briefly referencing comparable findings in additional model systems would broaden the relevance and impact of the study.

This is a good idea, but our search for similar cases in other model organisms failed to identify clear examples. Perhaps more to the point here is that context dependent effects and, perhaps, condition selection bias, are a serious issue in scientific research in general. To emphasize this, the following has been added as the last line of the discussion. “More widely, condition dependency and conditional selection bias risk diminishing the reliability of research findings in many scientific disciplines.”

Hashimoto, Y., Ookuma, S. and Nishida, E., 2009. Lifespan extension by suppression of autophagy genes in Caenorhabditis elegans. Genes Cells. 14, 717-726.

Lucanic, M., Plummer, W., Chen, E., Harke, J., Foulger, A., Onken, B., Coleman-Hulbert, A., Dumas, K., Guo, S., Johnson, E., Bhaumik, D., Xue, J., Crist, A., Presley, M., Harinath, G., Sedore, C., Chamoli, M., Kamat, S., Chen, M., Angeli, S., Chang, C., Willis, J., Edgar, D., Royal, M., Chao, E., Patel, S., Garrett, T., Ibanez-Ventoso, C., Hope, J., Kish, J., Guo, M., Lithgow, G., Driscoll, M. and Phillips, P., 2017. Impact of genetic background and experimental reproducibility on identifying chemical compounds with robust longevity effects. Nat Commun. 8, 14256.

Minnerly, J., Zhang, J., Parker, T., Kaul, T. and Jia, K., 2017. The cell non-autonomous function of ATG-18 is essential for neuroendocrine regulation of Caenorhabditis elegans lifespan. PLoS Genet. 13, e1006764.

Schmauck-Medina, T., Anisimov, A., Meyer, D.H., Hu, Y., Huang, Z., Wu, Y., Taylor, S., Takla, M., MacArthur, M.R., Mitchell, S.J., Ai, R., Simonsen, A., Jensen, V., Labbadia, J., Shen, H.-M., Hansen, M., Schumacher, B., Rubinsztein, D., Lautrup, S., Lu, G. and Fang, E.F., 2026. ATG-18/WIPI2 drives longevity in an HLH-30/TFEB-dependent manner. bioRxiv.

Wang, D. and Ruvkun, G., 2004. Regulation of Caenorhabditis elegans RNA interference by the daf-2 insulin stress and longevity signaling pathway. Cold Spring Harb Symp Quant Biol. 69, 429-31.

Wilhelm, T., Byrne, J., Medina, R., Geisinger, J., Hajduskova, M., Tursun, B. and Richly, H., 2017. Neuronal inhibition of the autophagy nucleation complex extends life span in postreproductive C. elegans Genes and Development. 31, 1561–1572.

Yang, Y., Arnold, M.L., Lange, C.M., Sun, L.H., Broussalian, M., Doroodian, S., Ebata, H., Choy, E.H., Poon, K., Moreno, T.M., Singh, A., Driscoll, M., Kumsta, C. and Hansen, M., 2024. Autophagy protein ATG-16.2 and its WD40 domain mediate the beneficial effects of inhibiting early-acting autophagy genes in C. elegans neurons. Nat Aging. 4, 198-212.

eLife Assessment

This is an important study utilizing innovative CRISPR based approaches demonstrating the role of the KLF family of transcription factors in the post-natal maturation of cortical projection neurons. The strength of evidence overall is compelling, and the study is well executed. The screen data presented provides a number of interesting candidates for future analyses into the mechanism of action of KLF family members in neuronal maturation.

Reviewer #1 (Public review):

[Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. The original reviews were generally positive. In their revision, the authors partially address reviewers concerns, but mention that a number of requested experiments are for future studies.]

Summary:

This is an interesting manuscript by Kirk and colleagues describing a highly valuable knock-down system that leverages CRISPRi in order to further elucidate the role of the Kruppel-Like Factor (KLF) transcription factor family in regulating the maturation of postnatal cortical projection neurons. The authors firstly use RNA-Seq and ATAC-Seq data in order to identify the KLF TF family as a potential regulator of cortical neuron maturation in the postnatal brain and subsequently knock down four KLF family members; KLF9, KL13, KLF6 and KLF7, in order to ascertain the functions of specific KLF genes in the developing cortex. The described CRISPRi knock down strategy is highly robust and penetrant as evidenced by a KD efficiency > 95% (assessed by both qPCR and single molecule FISH) and demonstrates that KLF6 and KLF7 play an activating role in driving the expression of target genes relating to axonal growth whereas KLF9 and 13 play a repressive role that inhibits the expression of overlapping gene targets. Together, the authors propose a model where the KLF TF family acts as a regulatory "switch" from activation to repression in the postnatal cortex as a mechanism to control a shift in projection neuron function from axonal growth to circuit refinement. The findings and conclusions of the manuscript offer a valuable contribution to the field of postnatal cortical development and further our understanding of the regulatory mechanisms that govern neuron maturation.

Reviewer #2 (Public review):

Summary:

Kirk et al. use RNA-Seq and CRISPRi to provide evidence that KLF family transcription factors regulate postnatal neuronal maturation of pyramidal neurons. The genetic programs regulating postnatal neuronal maturation are not well understood. The authors first analyzed chromatin accessibility and gene expression data from layer 4 and 6 pyramidal neurons and found that KLF TFs are predicted regulators of postnatal neuronal maturation. They then use CRISPRi knockdown and find that KLF activators first activate genes and then this is followed by KLF repressors repressing genes. Interestingly, some genes, such as those with cytoskeletal functions, are shared targets of KLF activators and repressors.

Strengths:

The study is well-executed and the paper is well-written. A major strength of this study is the application of state-of-the-art transgenic approaches. The CRISPRi approach used to knock down multiple KLFs is compelling. The genomic data generated appears to be high quality and is carefully analyzed. The presented findings provide important insights into the genetic programs that regulate postnatal maturation in cortical pyramidal neurons. The discovery that KLF family activators/repressors regulate gene expression changes during this critical step of neuronal development fills an important gap in the field.

Weaknesses:

While beyond the scope of the current study, future studies should investigate the contributions of KLFs on postnatal morphological and physiological changes.

Reviewer #3 (Public review):

Summary:

In their manuscript "Multiplexed CRISPRi Reveals a Transcriptional Switch Between KLF Activators and Repressors in the Maturing Neocortex", Kirk and colleagues seek to dissect the developmentally regulated pan-neuronal gene programs that control the postnatal maturation of cortical neurons. For this, the authors analyzed newly generated and existing RNA-seq and ATAC-seq of Layer 4 and Layer 6 cortical pyramidal neurons at postnatal day 2 (P2) and day 30 (P30) and identified thousands of shared developmentally regulated genes and genomic (promoter) regions, including genes involved in axon growth (tend to be downregulated) and synaptic function (tend to be upregulated). Motif enrichment analysis of promoters of differentially regulated genes revealed a strong presence of KLF/Sp family binding motifs, pointing to Krüppel-Like Factors (KLFs) as key transcriptional regulators of cortical maturation. Expression profiling showed a developmental switch from activating KLFs (Klf6, Klf7) expressed neonatally to repressive KLFs (Klf9, Klf13) upregulated during maturation. Using an elegant in vivo multiplexed CRISPR interference (CRISPRi) system, the authors achieved efficient, cell-type-specific knockdown of these TFs and showed that Klf9 and Klf13 repress a set of genes that includes cytoskeletal regulators such as Tubb2b, Dpysl3, and Rac3. Conversely, Klf6 and Klf7 promoted the expression of these same genes in the early postnatal period, and their knockdown led to reduced expression of these genes, particularly at P10 when their activating influence is strongest. Since promoters of shared KLF targets were enriched for KLF/Sp motifs but showed little change in chromatin accessibility, the authors propose a model in which distinct KLF family members function either as transcriptional repressors and activators that compete at constitutively accessible promoters and thereby act as a developmental transcriptional switch that coordinates the downregulation of axon growth programs and upregulation of synaptic maturation genes during cortical development.

Strengths:

The study addresses an interesting question and advances our understanding of the transcriptional regulation underlying postnatal cortical development. A major strength of the study lies in the innovative use of in vivo multiplexed CRISPR interference (CRISPRi), which allows for cell-type-specific, combinatorial knockdown of redundant TFs - this an elegant solution to a long-standing challenge in transcription factor research and should be useful also for other neuroscience studies that require local and cell-type-specific gene loss-of-function. Also, the integration of RNA-seq and ATAC-seq across developmental time points provides a robust foundation for identifying direct targets of the KLF family, and the findings are reinforced by cross-species conservation and the identification of targets with clear neurodevelopmental relevance.

Weaknesses:

The major weakness of the study lies in its relatively narrow scope: the study focuses primarily on transcriptional mechanisms and largely lacks functional validation of the neuronal phenotypes that are predicted by the gene expression data (e.g. axonal morphology). Despite these limitations, the paper offers an interesting model for a transcriptional switch during neuronal maturation in the cortex and establishes a powerful methodological framework for dissecting redundant gene networks in vivo.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

This is an interesting manuscript by Kirk and colleagues describing a highly valuable knock-down system that leverages CRISPRi in order to further elucidate the role of the Kruppel-Like Factor (KLF) transcription factor family in regulating the maturation of postnatal cortical projection neurons. The authors firstly use RNA-Seq and ATAC-Seq data in order to identify the KLF TF family as a potential regulator of cortical neuron maturation in the postnatal brain and subsequently knock down four KLF family members; KLF9, KL13, KLF6 and KLF7, in order to ascertain the functions of specific KLF genes in the developing cortex. The described CRISPRi knock down strategy is highly robust and penetrant as evidenced by a KD efficiency > 95% (assessed by both qPCR and single molecule FISH) and demonstrates that KLF6 and KLF7 play an activating role in driving the expression of target genes relating to axonal growth whereas KLF9 and 13 play a repressive role that inhibits the expression of overlapping gene targets. Together, the authors propose a model where the KLF TF family acts as a regulatory "switch" from activation to repression in the postnatal cortex as a mechanism to control a shift in projection neuron function from axonal growth to circuit refinement. The findings and conclusions of the manuscript offer a valuable contribution to the field of postnatal cortical development and further our understanding of the regulatory mechanisms that govern neuron maturation.

The conclusions of this manuscript are generally supported by the data, but some aspects of the data collection and analysis require some further clarification. Specifically:

(1) The authors comprehensively assess the molecular effects of KLF TF knock-down, however, the authors do not deeply address the cellular effects of these knock-downs. The authors conclude that knockdown of KLF6/7 and KLF9/13 cause downregulation and upregulation, respectively, of a common set of genes involved in cytoskeletal or axon regulation such as Tubb2 and Dpysl3. How is the morphology of the cells affected by these knockdowns? For example, does KLF9/13 knockdown cause neurite/axonal outgrowth? The authors should perform some basic experiments to assess changes in cell morphology following KLF TF KD. This is the one key point that needs addressing, in my opinion.

We appreciate this comment and agree that the cellular effects of KLF activator and/or repressor KD are not addressed by the experiments in this manuscript. However, the effects of KLF9, KLF13, and KLF9/13 knockdown on neurite outgrowth have been previously examined in vitro (Avci et al., 2012; Avila-Mendoza et al., 2020) and in vivo (Apara et al., 2017). The collective findings of these papers (enhanced neurite outgrowth and axon regeneration following KLF repressor KD) align with the predicted outcome of upregulating the set of cytoskeletal remodeling genes identified as putative KLF family targets in this manuscript. Furthermore, the in vitro results demonstrate that the partial redundancy between KLF9 and KLF13 we identified at the transcriptional level is relevant for their roles in repressing neurite outgrowth. Our results suggest that these previous findings are likely to hold true in cortical neurons in vivo while offering a molecular explanation for these effects at the level of gene expression. Other groups have examined the effect of KLF6 or KLF7 overexpression on corticospinal axon regeneration in vivo and found that these transcription factors can individually promote axon regrowth after injury (Blackmore et al., 2012; Wang et al., 2018). Similarly, this earlier work did not identify transcriptional mechanisms underlying the observed effect so our results also offer a plausible set of targets that could mediate the link between KLF activator overexpression and axon regeneration.

(2) The authors identify 374 DEGs in P10 Klf6/7 KD neurons and 115 DEGs at P20 (figure 6B). Have the authors looked to see what proportion of these DEGs are upregulated in the KLF9/13 KDs in order to get a more global understanding of the degree of overlap in the genes regulated by the KLF family members? [MOU2] Along similar lines, the authors later indicate that there are 144 shared targets between the KLF activator and repressor pairs (Figure 7C). What percentage does this represent of the total number of DEGs between the KLF pairs. This could further illustrate the degree to which the KLF pairs regulate the same set of genes. If it is already indicated in the manuscript, it should be made a bit more clear to the reader.

We thank the reviewer for the suggestion to include more comprehensive quantitative measures of the degree of overlap between targets of KLF activator and repressor pairs. We have included the exact number and percentage of P10 and P20 KLF6/7 targets that are differentially expressed (adj. p-value <= 0.05) in KLF9/13 KD neurons and vice versa in the text in the section associated with Figure 7, which is devoted to describing this class of overlapping targets. Furthermore, figures 7D and S7.2B both show how the full set of KLF6/7 targets are affected by Klf9/13 KD and vice versa. Collectively, this demonstrates that both KLF activator and repressor targets trend towards opposite regulation by the opposing pair.

(3) Figures 5B and 6D2 are very interesting as they relate the changes in gene expression over time in neurons from P2 to P30 to the functions of KLF9/13 and KLF6/7, respectively. I would be curious to see how these two forms of analyses overlap with one another. For example, in Figure 6D2, where would the KLF9/13 upregulated genes fall on the plot shown in Figure 6D2? And would those overlapping genes fit a similar correlation?

We agree with the reviewer that this is a powerful way to visually demonstrate the overlap between KLF6/7 and KLF9/13 targets in a more unbiased way and within a developmental context. The suggested analysis has been included in the supplement to Figure 7 (Fig S7.3).

(4) Figure 7E shows expression levels of shared KLF TF targets in control or KD conditions. Interestingly, the expression of Tubb2b, shows higher expression in ScrGFP P10 when compared to KLF9/13 P20, suggesting that derepression of KLF9/13 does not fully restore the expression level of Tubb2b seen at P10. This may suggest that other repressive regulators may be involved in the downregulation of Tubb2b from P10 to P20[MOU4] . Can the authors further comment on this, perhaps in the discussion, and speculate if there are other regulatory factors at play that may be controlling some of the shared targets by KLF6/7 and KLF9/13?

We thank the reviewer for this keen observation. We have included a comment on this within the text associated with Figure 7. While other regulatory factors are plausible, this is easily explained by low expression of Klf6 and Klf7 in the P20 cortex, which cannot drive transcription of Tubb2b and other targets to the same level as what is observed at P10 when the KLF activators are still relatively abundant, even after repression by Klf9/13 is removed. Within this framework, overexpression of KLF activators on a KLF repressor background would be the only way to restore expression of Tubb2b to its P10 expression levels. This is wholly compatible with our model of ‘push-pull’ regulation of KLF targets outlined in Figure 9. The finding could also be affected by the addition of repressive histone marks by the KLF9/13-associated SID complex in early neonatal development that may persist following the knockdown of these repressors, preventing complete restoration of neonatal expression patterns.

Reviewer #2 (Public review):

Summary:

Kirk et al. use RNA-Seq and CRISPRi to provide evidence that KLF family transcription factors regulate postnatal neuronal maturation of pyramidal neurons. The genetic programs regulating postnatal neuronal maturation are not well understood. The authors first analyzed chromatin accessibility and gene expression data from layer 4 and 6 pyramidal neurons and found that KLF TFs are predicted regulators of postnatal neuronal maturation. They then use CRISPRi knockdown and find that KLF activators first activate genes and then this is followed by KLF repressors repressing genes. Interestingly, some genes, such as those with cytoskeletal functions, are shared targets of KLF activators and repressors.

Strengths:

The study is well-executed and the paper is well-written. A major strength of this study is the application of state-of-the-art transgenic approaches. The CRISPRi approach used to knock down multiple KLFs is compelling. The genomic data generated appears to be high quality and is carefully analyzed. The presented findings provide important insights into the genetic programs that regulate postnatal maturation in cortical pyramidal neurons. The discovery that KLF family activators/repressors regulate gene expression changes during this critical step of neuronal development fills an important gap in the field.

Weaknesses:

A limitation of the current study is that the functional importance of KLF for postnatal neuronal maturation is unclear. Although the authors find that KLFs regulate some of the gene expression changes during postnatal neuronal maturation, it is still unclear whether such gene expression changes mediate the postnatal changes in morphology and physiology. While beyond the scope of the current study, future studies should investigate the contributions of KLFs on postnatal morphological and physiological changes.

We thank the reviewer for their helpful comments on this manuscript. We agree that the effects of KLF knockdown identified in this study are primarily descriptive, but – as noted - a detailed analysis of the morphological and physiological consequences of KLF knockdown are beyond the scope of this paper. However, we believe that a more mechanistic model of KLF function during neuronal maturation can be obtained by considering our findings on the bidirectional regulation of core cytoskeletal genes by KLF activators and repressors alongside published in vivo and in vitro data on the opposing roles of KLF family members on axon outgrowth/regrowth (Apara et al., 2017; Avila-Mendoza et al., 2020; Blackmore et al., 2012; Moore et al., 2009; Wang et al., 2018). Thus, we offer a set of transcriptional targets that likely mediate these opposing effects and a developmental context within which they might operate.

Reviewer #3 (Public review):

Summary:

In their manuscript "Multiplexed CRISPRi Reveals a Transcriptional Switch Between KLF Activators and Repressors in the Maturing Neocortex", Kirk and colleagues seek to dissect the developmentally regulated pan-neuronal gene programs that control the postnatal maturation of cortical neurons. For this, the authors analyzed newly generated and existing RNA-seq and ATAC-seq of Layer 4 and Layer 6 cortical pyramidal neurons at postnatal day 2 (P2) and day 30 (P30), and identified thousands of shared developmentally regulated genes and genomic (promoter) regions, including genes involved in axon growth (tend to be downregulated) and synaptic function (tend to be upregulated). Motif enrichment analysis of promoters of differentially regulated genes revealed a strong presence of KLF/Sp family binding motifs, pointing to Krüppel-Like Factors (KLFs) as key transcriptional regulators of cortical maturation. Expression profiling showed a developmental switch from activating KLFs (Klf6, Klf7) expressed neonatally to repressive KLFs (Klf9, Klf13) upregulated during maturation. Using an elegant in vivo multiplexed CRISPR interference (CRISPRi) system, the authors achieved efficient, cell-type-specific knockdown of these TFs and showed that Klf9 and Klf13 repress a set of genes that includes cytoskeletal regulators such as Tubb2b, Dpysl3, and Rac3. Conversely, Klf6 and Klf7 promoted the expression of these same genes in the early postnatal period, and their knockdown led to reduced expression of these genes, particularly at P10 when their activating influence is strongest. Since promoters of shared KLF targets were enriched for KLF/Sp motifs but showed little change in chromatin accessibility, the authors propose a model in which distinct KLF family members function either as transcriptional repressors and activators that compete at constitutively accessible promoters and thereby act as a developmental transcriptional switch that coordinates the downregulation of axon growth programs and upregulation of synaptic maturation genes during cortical development.

Strengths:

The study addresses an interesting question and advances our understanding of the transcriptional regulation underlying postnatal cortical development. A major strength of the study lies in the innovative use of in vivo multiplexed CRISPR interference (CRISPRi), which allows for cell-type-specific, combinatorial knockdown of redundant TFs - this an elegant solution to a long-standing challenge in transcription factor research, and should be useful also for other neuroscience studies that require local and cell-type-specific gene loss-of-function. Also, the integration of RNA-seq and ATAC-seq across developmental time points provides a robust foundation for identifying direct targets of the KLF family, and the findings are reinforced by cross-species conservation and the identification of targets with clear neurodevelopmental relevance.

Weaknesses:

The major weakness of the study lies in its relatively narrow scope: the study focuses primarily on transcriptional mechanisms and largely lacks functional validation of the neuronal phenotypes that are predicted by the gene expression data (e.g. axonal morphology). For example, the authors analyzed the effects of KLF9/13 KD on the neurons' excitability and excitatory inputs, but did not assess the effects on inhibitory inputs and E/I-ratio or morphological parameters such as axonal length and axonal target fields - the manuscript would be strengthened considerably by such analyses (axonal projections could be analyzed e.g. via local injections of the gRNA AAVs and subsequent immunolabeling of brain sections). Similarly, the chromatin-based mechanisms underlying KLF activity remain relatively speculative, and the transcriptional mechanisms upstream of the KLFs remain unexplored (this could be addressed by analyzing existing datasets; see "Additional Point 1" below). Finally, the manuscript is too long (e.g., nearly five pages in the Discussion section are devoted to discussing various misregulated genes) and would benefit from presenting the Results and Discussion sections more concisely. However, despite these limitations, the paper offers an interesting model for a transcriptional switch during neuronal maturation in the cortex and establishes a powerful methodological framework for dissecting redundant gene networks in vivo.

Shorten discussion (possibly results also).

We appreciate the reviewer’s feedback and agree that linking transcriptional perturbations to cellular phenotypes of KLF activator and/or repressor KD would significantly strengthen this manuscript. However, we believe this is beyond the scope of this current manuscript and the effects of individual KLF activators and repressors on axon outgrowth/regrowth are known from prior in vitro and in vivo studies (Apara et al., 2017; Avila-Mendoza et al., 2020; Blackmore et al., 2012; Moore et al., 2009; Wang et al., 2018). Since there was no detectable effect of KLF repressor KD on excitatory synaptic transmission (Fig. S4.2), we did not believe it was likely that our excitatory neuron-specific knockdown would affect inhibitory synapses under our model where KLF targets have primarily axonal functions. In the absence of ATAC-seq from Klf9/13 KD neurons, the relationship between chromatin accessibility and KLF binding is limited to descriptions of chromatin accessibility around KLF targets in neonatal and mature neurons. However, we do offer two testable hypotheses in our Discussion of early developmental transitions driving the KLF switch (Thyroid Hormone and structured activity input). Finally, we recognize that the manuscript is too long and have edited or removed parts of the Discussion.

Recommendations for the authors:

Summary of Recommendations from Reviewing Editor: While the screen data is important and novel, there is consensus among the reviewers that the current study would greatly benefit by additional analyses as to the consequences of KLF knockdown on neuronal morphology and neurite outgrowth. It is recommended that this major open question be experimentally addressed. It is also recommended that additional comments/points below are addressed with changes to the figures/text as appropriate.

Reviewer #1 (Recommendations for the authors):

Minor points:

(1) In Figure 2C2, it can be assumed that green represents KLF7 and red represents KLF9 but a legend indicating this would be helpful here. The Y axis is labeled as "Binned D-V Axis". Can the authors further clarify what this means? Is it referring to the different layers of the cortex? It looks like the expression pattern of KLF7 and 19 is not consistent throughout the entire dorsal-ventral axis at P7 and there is a region where KLF19 expression is higher than KLF7. Could this suggest that KLF family members may act on different types of neurons in distinct layers at different timepoints? Could the authors expand on this at all?

We thank the reviewer for pointing this out and the appropriate legend has been added. A detailed description of our method for binning counts in the Dorsal-Ventral axis (y-axis) to normalize for differences in cortical thickness across ages is included in the methods section (see RNAScope Image Acquisition and Analysis), but text has been added to the results to clarify this. The non-uniform distribution of Klf7 across cortical layers at P2 and P7 likely indicates that that KLF switch, while conserved, occurs with variable kinetics across cortical cell types which could reflect their distinct rates of maturation in vivo (Gao et al., 2025). Minor edits have been made to the text associated with Figure 2 to call attention to this detail.

(2) Figure 6C1 describes three clusters of DEGs. If I understand correctly, cluster 1 represents downregulated DEGs by KLF6/7 that are also developmentally downregulated between P10 and P20 and cluster 2 represents downregulated DEGs by KLF6/7 that do not change between P10 and P20. It could be interesting to see how these gene sets compare with one another. For example, are the genes in cluster 1 that decrease from P10 to P20 more related to axon regulation/cytoskeleton when compared to cluster 2 which consist of genes that do not change from P10 to P20 and perhaps may reflect other regulatory functions of KLF6/7.

We appreciate this suggestion and we have updated Figure 6C1 to reflect this observation by highlighting genes that are part of significant enriched GO terms to allow the reader to see which cluster they belong to. Separate GO analyses performed for each downregulated gene cluster did not yield significant results and were therefore not included.

Reviewer #3 (Recommendations for the authors):

See above

Additional points:

(1) Upstream transcriptional regulation of KLFs:

The idea that KLFs are regulated by Thyroid hormone (T3) is interesting, and the manuscript would be strengthened by exploring this idea further. This could be done, e.g. by analyzing existing snRNA-seq on genes regulated in cortical neurons by T3 (see PMID: 39178853). Similarly, KLFs were suggested to act together with AP1, and many KLFs were previously found to be activity-regulated - hence, the manuscript would be strengthened by analyzing the connection between KLFs and AP1. This could be done e.g. by re-analyzing transcriptomic and proteomic data generated e.g. by the Greenberg laboratory.

We agree with the reviewer that the role of T3 in driving the KLF switch is intriguing and is an active area of investigation in our lab. However, these experiments are ongoing and are beyond the scope of this current manuscript. A preliminary analysis of the T3-treated snRNA-seq dataset present in Hochbaum et al., 2024 found that all major excitatory cortical cell types upregulate Klf9 upon T3 treatment while the expression of selected shared targets including Dpysl3 and Gap43 are downregulated in several cell types, supporting our hypothesis. These findings have been included in Figure S9.

It was not our intention to suggest that there may be cooperation between AP1 and the KLF family, as AP1 motif enrichment was detected in shared upregulated genes while the KLF/Sp motif was found in shared downregulated genes. Furthermore, DEGs from both KLF knockdown experiments had exclusive enrichment for promoter KLF/Sp motifs, making it unlikely that there is widespread co-regulation of these genes by the AP1 family.

(2) Figure 5D and Page 27, regarding Rac3:

The authors state that " no change in accessibility in motif-positive peaks upstream of Plppr1 and Rac3 (Figures 5D and S5C)." This statement seems incorrect for Rac3 in L6 neurons where the ATAC signal is strongly reduced at P30 relative to P2. The authors might want to explain this or choose a better example.

We acknowledge that Rac3 was a poor choice to represent a developmentally regulated Klf9/13 target with no change in promoter accessibility, and have updated Figure 5 with Atat1, a tubulin acetylase and shared KLF target, as our exemplar for this category of transcript.

(3) Supplemental Figure 2:

There seems to be a discrepancy between the expression levels in Panels A and C: the data in Panel A were generated from adult mouse cortex, and the levels of six KLFs are indicated as rather high - however, in Panel C, the adult levels of KLF6, 7 and 13 are rather low. How can this be explained?

We thank the reviewer for detecting this apparent discrepancy. Most of this can be attributed to the log scale used in Figure S2A, which flattens expression differences in the moderate to high expression range. We elected to use a log scale in Figure S2A to highlight KLF family members with higher cortical expression relative to those with low or no expression. In addition, the counts displayed in this figure were batch-corrected to account for significant batch effects between libraries from additional excitatory cortical cell types included in Figure S2A, which were prepared and sequenced at different locations (Brandeis University v. Janelia Research Campus, as in Sugino et al., 2019). The additional processing step has been included in our methods section.

(4) IGV plots in Figure 5D and Supplemental Figure 5B:

It might be worth highlighting/shading the promoter region to see the position of the peaks relative to the promoters and TSSs.

The TSS is indicated on all IGV plots by a dark arrow indicating the direction of transcription, so the promoter can be inferred to be immediately upstream.

(5) Middle of page 20:

Remove "saw"; this seems like a typo.

(6) Page 27, reference to the plot for Plppr1:

The plot for Plppr1 is in Supplemental Figure 5B2, not in S5C.

(7) Page 28:

There seems to be a typo/omission before "...general applicability of CRISPRi"

eLife Assessment

This study shows that partial cone photoreceptor loss induces pathway-specific circuit remodeling in the mouse retina, with alpha OFF-sustained and OFF-transient retinal ganglion cells adapting differently through changes in their pre- and postsynaptic circuits. The results are valuable because they provide a key understanding of the diversity of circuit remodeling in retinal degeneration. The data are convincing, although clearer reporting of the numbers of independent animals and retinas, a more rigorous distinction between compensation and circuit change, and discussion of the functional consequences would strengthen the mechanistic insights.

Reviewer #1 (Public review):

Summary:

Lee et al. investigate how parallel retinal pathways respond to a common loss of photoreceptor input. The authors induce partial cone loss in adult mice and compare the functional responses of sustained OFF alpha (sOFFa) and transient OFF alpha (tOFFa) ganglion cells, together with changes in their presynaptic circuits. Using targeted patch-clamp recordings, linear-nonlinear analyses, pharmacological dissection of inhibitory inputs, and quantitative synaptic imaging, they show that the two pathways do not respond uniformly to cone loss. tOFFa ganglion cells exhibit more extensive changes in spatiotemporal receptive fields than sOFFa ganglion cells, with contributions from excitatory transmission, presynaptic glycinergic inhibition, direct GABAergic and glycinergic inhibition, and intrinsic properties. At the same time, transformations between synaptic input and spike output partially preserve ganglion cell signaling despite the loss of cones.

Strengths:

This is a technically careful and high-quality study. The comparison of two well-defined ganglion cell types and their dominant bipolar-cell pathways provides an unusually detailed view of where circuit modifications arise following a shared perturbation. The combination of recordings at successive stages of signal processing, pharmacological manipulations, and synaptic imaging is a particular strength. The use of partial stimulation in control retina also helps distinguish the immediate consequence of reduced input from subsequent circuit changes. The resulting conclusion that common photoreceptor loss produces pathway-specific forms of remodeling rather than a uniform retinal response is interesting and well supported. The work adds to our understanding of the diversity and circuit specificity of responses to retinal degeneration.

Weaknesses:

The principal limitations concern the precision of some mechanistic interpretations rather than the central observation of pathway-specific remodeling. First, the framework used to classify effects as compensation or circuit change sometimes treats the absence of a statistically significant difference as evidence that two conditions are equivalent. Second, the numbers of animals and retinas contributing to the main physiological and anatomical comparisons are not consistently reported, making it difficult to evaluate the independence of measurements obtained from multiple cells, images, or synaptic puncta. Finally, the consequences of the observed remodeling for the visual signals carried by these pathways remain unclear. This is particularly relevant for tOFFa ganglion cells, which have been implicated in responses to looming or approaching dark objects. The altered temporal filtering, center-surround organization, and input-output transformation could preserve, degrade, or otherwise transform such signals. These issues qualify the mechanistic and functional interpretation but do not substantially weaken the main conclusion that the two pathways respond differently to partial cone loss.

Reviewer #2 (Public review):

Summary:

This is an elegant, rigorous, and thought-provoking study that examines how different neural circuits are altered in response to loss of a common sensory input. To study this question, the authors use the mouse retina as a model system to investigate how downstream retinal circuits undergo modifications following a well-controlled partial loss of cone photoreceptors.

Strengths:

The experiments were conducted with a high degree of rigor, and the authors carefully considered and implemented appropriate controls throughout the study. Multiple parameters were tested, including pharmacological approaches to assess responses from different ganglion cell types. In addition, the authors complemented their functional data with confocal imaging to further support their findings. Overall, this is a well-written paper that provides a thorough analysis demonstrating how two similar ganglion cell types undergo distinct adaptations (i.e., compensation versus remodeling) in response to the loss of the same sensory input.

Weaknesses:

No additional experiments are needed. However, the authors may wish to consider the following points:

(1) Do the differences in compensation versus remodeling observed in ganglion cells reflect changes in the OPL? Different bipolar types may remodel their dendrites and form aberrant contacts with rods in the absence of cones. However, this would be challenging to test because there are currently no good markers for different bipolar types.

(2) It would be interesting to determine whether these functional changes can be detected at the transcriptomic level or whether they are mediated primarily through post-translational modifications.

Author response:

Reviewer #1 (Recommendations for the authors):

(1) Please report effect sizes and confidence intervals for the key comparisons in Figure 1. Where equality between conditions is central to the interpretation, use an equivalence test or soften the equal signs and associated mechanistic language. Please also clarify the permutation analysis: the Methods state that the null distributions were centered on the observed difference of deltas, whereas a permutation null would ordinarily test relative to zero.

We thank the reviewer for these suggestions. We will report effect sizes and associated 95% confidence intervals for the key comparisons in Figure 1. We will also evaluate equivalence using defined margins based on control variability (e.g., standard deviation) and assess how conclusions depend on the margin definition. We will soften the associated mechanistic language and replace equal signs with approximate-equality symbols, clarifying that these indicate similarity without implying statistical equivalence. Finally, we will clarify the description of the permutation analysis in the Methods, including how the null distributions were constructed and centered.

(2) Please report, for each experimental group, the numbers of cells or images, retinas, and animals, and indicate when multiple observations came from the same animal. Where observations are nested, the analysis should account for this structure using an animal-level or hierarchical bootstrap, a mixed-effects analysis, or animal-level summaries. It would be reassuring to confirm that the main pathway-specific conclusions remain robust when the animal determines the independent sample size.

We thank the reviewer for highlighting the importance of accounting for nested observations. We will report the numbers of cells and images in the main figures. To supplementary table (Dataset S1), which already reports animal numbers, we will add the number of cells obtained from each animal. Finally, we will assess the robustness of our conclusions using animal-level summaries and mixed-effects analyses that account for observations nested within animals.

(3) Please expand the discussion of how the changes in tOFFa temporal filtering, spatial organization, and input-output transformation are expected to affect visual coding, including responses to looming or approaching dark objects. If readily feasible, direct electrophysiological recordings of responses to an expanding dark stimulus would be informative. I do not consider these recordings necessary, however; predictions from the existing linear-nonlinear models or a more developed discussion of the expected coding consequences would be sufficient.

We appreciate the suggestion and will use a linear-nonlinear model to predict how responses to a looming stimulus change following cone loss. We will use an expanding stimulus based on published looming protocols and apply the measured spatial and temporal filters and pass the resulting linear output through nonlinearities to predict the response. We will compare how the response changes over time in control and cone-DTR.

Reviewer #2:

Weaknesses:

(1) Do the differences in compensation versus remodeling observed in ganglion cells reflect changes in the OPL? Different bipolar types may remodel their dendrites and form aberrant contacts with rods in the absence of cones. However, this would be challenging to test because there are currently no good markers for different bipolar types.

We agree that bipolar cell remodeling could contribute to the differences in functional changes at ganglion cell level. Our functional measurements do not establish whether these differences originate in the OPL. We will address this possibility in the Discussion.

(2) It would be interesting to determine whether these functional changes can be detected at the transcriptomic level or whether they are mediated primarily through post-translational modifications.

We agree that determining the molecular mechanisms underlying these functional changes would be informative. As our current experiments do not distinguish between changes in gene expression and post-translational modifications, we will acknowledge the limitation and address these possibilities as directions for future work in the Discussion.

eLife Assessment

This Tools & Resources manuscript presents important findings that can help the implementation of research assessment reform in research-performing institutions. The authors employ a three-part Delphi method involving researchers, funders, journal editors, and academic leaders, ultimately reaching consensus on six standards. The manuscript's accompanying appendix provides helpful, modular guidance for research-performing organizations seeking to adopt these six parameters. Overall, the evidence presented is solid, and the manuscript makes a practical contribution that will add value. However, a significant limitation is the lack of evidence demonstrating whether and how implementation of the proposed guidance improves research evaluation in practice.

Reviewer #1 (Public review):

Summary:

This study addresses the gap between calls to incentivize Open Science practices through research assessment reform and its implementation by Research Performing Organizations (RPOs). To help RPOs get started and prioritize their assessment reforms, it makes a series of recommendations for action. It does so via an expert Delphi process, where consensus around six standards was reached. The purpose is to provide a resource for would-be reformers in Research Performing Organizations around the world, in the hope of advancing the Responsible Research Assessment and Open Science reform movements across global academia. The standards are not meant to be one-size-fits-all, but customizable to local needs and implemented selectively at the discretion of a given RPO.

Strengths:

Overall I was impressed with the contribution. Generally speaking, it clearly sets out its methods and arguments. I recognize there is indeed a growing need to support RPOs at varying levels of maturity, to get started and orient themselves around research assessment reform, so this seems to me like it will be a useful offering. The Delphi approach is well explained, and further information is provided in supplementary files.

Weaknesses:

Some further reflection on the disciplinary make-up of the Delphi participants would be helpful. Somewhat more hedging on the limitations of the manuscript would help improve the contribution, as would further details on how this contribution sits in relation to existing resources and studies on implementing open science-aware research assessment reforms.

Reviewer #2 (Public review):

Summary:

This paper describes the employment of a Delphi process, a common consensus-reaching practice based on consecutive rounds of discussion and voting. In this study, the authors engaged in the Delphi process with stakeholders to reach consensus on recommended guidelines for use in tenure and promotion, specific to reproducible and transparent research practices.

Strengths:

The paper provides very practical guidelines for research institutions willing to push for reform regarding reproducible and transparent practices. Guidelines like these are important for institutional leadership that wishes to drive change at their institutions but does not necessarily hold topic expertise on research and reproducibility. The study's authors represent longstanding expertise on the topic, and the use of the Delphi process ensures that stakeholders were engaged in the recommendations, something that helps ensure uptake and buy-in.

Weaknesses:

The biggest weakness of the paper is a lack of engagement with the literature of tenure and promotion reform. This is a field that has been heavily studied with regard to incorporating change in practice to promote equity and inclusion. Other specific subfields that have dealt in this space are those that have pushed for reform to increase the importance of teaching, service, and (inclusion of) mentorship in tenure and promotion packages. Because this paper deals heavily with implementing change in this specific space, and provides guidelines with very practical implications for change, some literature review in the introduction and discussion on where past efforts have succeeded or faced roadblocks would be welcome. This would add to the study's value in helping people implement these practices in a very real and tangible way. The other weakness is a lack of further expansion on the recommendations that did not reach consensus, many of which were added in Round 3. Are more rounds warranted for future studies? Changing landscapes require changing recommendations, which is probably why things like the use of LLMs emerged later and did not reach consensus. How should the readers engage with these results, and what future steps are needed to implement guidelines (if any) around these topics?

Reviewer #3 (Public review):

Summary:

This work provides a novel guide for Research Performing Organizations (RPOs) to evaluate researchers in ways that promote responsible research practices. Such a tool could help inform the development of policies that foster these practices within RPOs and potentially influence decisions related to the hiring and promotion of scientific staff. The guide was developed through a modified three-round Delphi process based on a preliminary set of recommendations provided by members of various organizations. Each round included a survey, and items that did not reach a consensus of >80% were reconsidered in the subsequent round. Round 3 consisted of an in-person meeting, during which the final guide was developed based on criteria that reached {greater than or equal to}80% consensus among participants. The guide includes six practices for evaluation (verification efforts, prospective study registration, data/code/materials sharing, reporting transparency, open-access publishing, and disclosure of funding interests) with an additional seventh component. Each practice includes a clearly defined aim and criteria describing what good practice looks like. The complete study methodology, statistical analyses, and supporting data are available through Open Science Framework (OSF).

Strengths:

A major strength of this work is the development of a practical tool that could help foster responsible research practices among candidates and provide promotion and hiring panel members with clearer criteria for assessing such practices. The three-round Delphi process, which included international participants, provides a structured approach to reaching consensus on practices that could contribute positively to the research ecosystem.

Weaknesses:

The recruitment strategy may have introduced selection bias and limited the representation of researchers and other interest holders from diverse geographical and institutional contexts, particularly from the Global South. Although the authors acknowledge that variability across regions and institutions should be considered when applying the guide, greater representation of these contexts would strengthen confidence in the broad applicability and generalizability of the proposed framework.

eLife Assessment

This study establishes the fast-developing tuxedo sea urchin, Mespilia globulus, as a genome-enabled experimental model. With the tools and methods described here, the authors provide a valuable platform with particular potential for developmental biology and physiology as well as comparative studies. The evidence supporting the principal claims of experimental tractability and quality of genomic resources is solid, although fuller documentation, e.g., of husbandry, aquarium-stock provenance and genetic diversity as well as gene-editing efficiency and imaging set up, would strengthen the work. This work will be of particular interest to developmental and evolutionary biologists.

Reviewer #1 (Public review):

Summary:

In this manuscript, Matar et al. introduce Mespilia globulus, the tuxedo sea urchin, as a new genomically-enabled model for echinoderm developmental biology. The authors establish a closed life-cycle culture system in a land-locked aquarium facility, demonstrate that key experimental techniques (hybridization chain reaction labelling and CRISPR/Cas9 gene knockout) are tractable in this species, and report chromosome-scale genome assemblies for two colour morphs and both sexes. Using these resources, they compare genome architecture and gene family evolution across sea urchin species, and investigate the genomic basis of sex determination.

Strengths:

The central motivation for this work is well justified: the long larval and juvenile periods of established sea urchin models such as Strongylocentrotus purpuratus have long limited the study of post-metamorphosis and adult biology, and M. globulus reaches metamorphosis in around two weeks and sexual maturity within four to six months, a substantial acceleration. The husbandry and life-cycle data are thorough, and the demonstration that hybridization chain reaction staining and CRISPR/Cas9 knockout both work as expected in this species convincingly establishes its experimental tractability. The two chromosome-scale genome assemblies are of high quality (BUSCO completeness above 99%, 21 chromosome-scale scaffolds consistent with other sea urchins), and the comparative synteny and gene family analyses are carefully constructed, drawing on a solid phylogenomic framework (CAFE-based gene family turnover analysis across six echinoderm species). The authors' finding that M. globulus has fewer duplicated genes in the gene repertoire relative to other camarodont urchins is a genuinely useful observation for researchers choosing a model system for functional genetics, since fewer paralogues should simplify interpretation of knockout phenotypes.

Weaknesses:

Some claims in the manuscript would benefit from additional supporting detail.

(1) The efficiency of the CRISPR/Cas9 knockout is illustrated qualitatively, but no sample size or penetrance value is reported, making it difficult for readers to judge how robust or reproducible this result is.

(2) The gene annotation is reported to have complete PFAM domain coverage for only 75% of predicted genes, but no independent completeness metric (such as BUSCO scored against the annotated gene set rather than the assembly) is provided, leaving open whether the remaining genes are genuinely novel, partial models, or annotation artefacts.

(3) Finally, at the time of review the NCBI BioProject accession cited for the genome and sequencing data (PRJNA1477966) could not be located, and it is not clear from the text whether this accession, once available, will include the gene annotation and RNA-seq datasets in addition to the raw genomic sequencing reads.

In summary, the authors achieve their stated aim of establishing M. globulus as a tractable, fast-developing echinoderm model, and the genomic and experimental resources presented support this conclusion. The comparative genomic conclusions - conservation of ancestral chromosome linkage groups, absence of a heteromorphic sex chromosome, and a comparatively low rate of gene family expansion - are well supported by the data shown, though some of the finer-grained claims (knockout efficiency, annotation completeness) require some clarification. Given the scarcity of tractable models for post-metamorphosis and adult echinoderm biology, this resource is likely to be of real value to the field, provided the genomic and transcriptomic data are made fully and clearly accessible to the community.

Reviewer #2 (Public review):

This manuscript introduces the tuxedo sea urchin, Mespilia globulus, as a new experimental model for developmental, reproductive and genomic biology. The authors establish culture methods that permit completion of the life cycle in a landlocked aquarium facility, demonstrate the applicability of developmental biology tools including HCR-FISH and CRISPR/Cas9-mediated gene disruption, generate chromosome-scale genome assemblies from two color morphs and both sexes, investigate potential sex determination mechanisms, and compare genome organization and gene family evolution with other sea urchin models. The authors conclude that M. globulus combines a relatively rapid life cycle with genomic tractability and therefore represents a valuable addition to the growing repertoire of genetically accessible sea urchin model systems.

Overall, I found this to be a strong and timely contribution that is well suited for the Tools and Resources category of eLife. The authors provide a comprehensive suite of resources, including husbandry protocols, genomic resources, developmental staging information, and proof-of-principle functional manipulations. The manuscript appropriately places M. globulus in the context of established sea urchin models, particularly Lytechinus pictus, and clearly argues that the new species is complementary rather than a replacement for existing systems. Given the increasing importance of genetically tractable echinoderm models, the development of an additional species that can be maintained and bred in closed aquarium systems is of considerable value to the field.

The evidence supporting the establishment of this model system is convincing, with multiple complementary datasets including life-cycle culture methods, genome assemblies, gene expression analyses, and gene-editing experiments.

My concerns primarily relate to two broader issues that should be addressed, followed by several specific comments.

(1) Genetic Background and Aquarium Trade Populations: A central argument of the manuscript is that M. globulus is attractive as a laboratory model because it is widely cultured in the aquarium trade and may exhibit reduced genetic variability due to captive propagation.

The manuscript states:

"M. globulus is a popular species in the aquarium industry and has excellent properties in tropical aquariums where it has been bred for many years in farming operations that reduce genomic heterogeneity..." (lines 96-98) and later:

"Captive breeding may also lower genetic variability compared to wild-caught individuals..." (lines 389-390).

However, the manuscript does not provide sufficient information to evaluate these claims. Several important questions remain unresolved:

(1) How genetically representative are the sequenced individuals relative to natural populations?

(2) What is known about the provenance and breeding history of the aquarium trade stocks used in this study?

(3) Are these animals derived from a small number of founder populations?

(4) Is there evidence for substantial inbreeding or genetic bottlenecks within commercial brood stocks?

(5) How similar are commercially available animals from different vendors and geographic sources?

These questions are important for both practical and biological reasons. From a practical perspective, researchers wishing to adopt M. globulus need to know whether animals purchased from aquarium suppliers are expected to resemble those analyzed here. From a biological perspective, reduced diversity or founder effects could influence genome assembly characteristics, heterozygosity estimates, gene family analyses, developmental traits, or responses to experimental manipulation.

Even if little information is currently available, the manuscript should explicitly discuss these uncertainties and provide available information regarding stock origin, aquaculture practices, and potential differences between captive and natural populations. A clear discussion of potential limitations would significantly strengthen the manuscript.

(2) Presentation and Interpretation of HCR and Phalloidin Data: Although the HCR and phalloidin experiments are not central to the major conclusions of the paper, they serve as important demonstrations of experimental tractability. At present, however, the presentation of these data is not fully convincing.

The HCR images show detectable signal, but the expression domains are only minimally documented. The manuscript states that expression patterns are consistent with known functions of Nodal and Notch signaling, yet the figures do not sufficiently guide readers to these conclusions (especially readers not familiar with sea urchin development). The signal is relatively diffuse and weak in some panels, and there is little annotation explaining exactly which embryonic territories are expressing the genes of interest. For readers without extensive sea urchin developmental biology expertise, it can be difficult to assess the validity and biological significance of the observed expression patterns.

Similarly, the phalloidin-labeled images provide limited anatomical information because the larvae are largely not labeled. I recommend:

(1) Adding labels identifying relevant embryonic regions and structures.

(2) Including arrows or overlays indicating key expression domains.

(3) Providing higher-magnification insets of relevant regions.

(4) Including selected optical sections rather than relying exclusively on 3D projections.

(5) Identifying known larval muscle groups in the phalloidin images.

(6) Improving image contrast and figure annotation where possible.

These changes would substantially strengthen the claim that established developmental biology methods are readily transferable to M. globulus.

Reviewer #3 (Public review):

Summary:

Omar Matar and colleagues describe how they successfully closed the life cycle of the tropical tuxedo urchin Mespilia globulus in a small, closed aquarium system. Its capacity to flourish under closed culturing conditions sets this sea urchin apart from most echinoderms and indeed most marine invertebrates, which require high-quality flow-through seawater (i.e. coastal access and a flow-through aquarium system). As M. globulus reaches sexual maturity in captivity in 4-6 months, which is shorter than other sea urchins, it opens the potential for transgenerational studies.

Matar et al. have developed a suite of molecular and genomic techniques and resources, including Cas9-mediated gene knockdown and chromosome-scale genome assemblies for both sexes and colour morphs. These strong aquaculture and genomics platforms suggest that M. globulus can provide insights into aspects of the sea urchin life cycle and history that would be difficult to study in other sea urchin models, which historically have been focused on embryogenesis.

Strengths:

This manuscript announces M. globulus as a novel, genome-enabled echinoderm model that can be cultured in a closed system with a few hundred litres of 23/24oC artificial seawater. This allows for experimental analysis of parts of the sea urchin life cycle that are not particularly tractable in other sea urchins, including the developmental biology of metamorphosis, symbiont interactions and immunity through the life cycle, post-settlement biomineralisation, and sex determination. The capacity to knock out embryonic/larval genes using CRISPR/Cas9 and visualise localised embryonic and larval gene expression using HCR are consistent with claims M. globulus can contribute novel insights into basic and applied (aquaculture) biology of sea urchins and echinoderms.

Weaknesses:

The general weakness of this manuscript is that the authors do not explain aspects of their study in enough detail (e.g. a single figure - Figure 2 - presents results of analysis of normal development, HCR and Cas9 knockdowns). This manuscript provides limited methodological detail about the culture system and the analysis of gene expression. Here are a few suggestions on how to improve the manuscript:

First, the authors should provide a thorough description of the methods used to cultivate and maintain M. globulus. This should include further details about the closed aquarium system; a schematic of the system would be insightful. Basic details about husbandry are needed, including (i) stocking densities of adults, embryos/larvae, postlarvae/juveniles; (ii) frequencies of level and water changes/top-ups; (iii) feeding regime at all phases of the life cycle (amount/sea urchin, post-feeding cleaning, etc.). These and other details are essential for the uptake of this model system. This documentation would provide the foundation for future improvements, which include shortening the time to acquire larval competence and sexual maturity, and improving and standardising larval settlement.

Second, the description of the procurement and analysis of mRNA is brief, unreferenced and reads as protocols used for an established model species (e.g. what is PFA in this case - the concentration of paraformaldehyde and the buffer can vary markedly between organisms and life stages). Even the RNA extraction protocols can vary between and within species. For instance, highly pigmented tissues tend to be more difficult to procure useable RNA. The HCR analysis, which is also scantily described, is restricted to embryonic and larval stages. Given the emphasis on the capacity of the M. globulus system to analyse all phases of the life cycle, it would be good to know if HCR can be performed on settled postlarvae, juveniles and adult tissues.

Third, the authors should consider dividing Figure 2, which consists of confocal images of normal development, HCR results and CRISPR/Cas9 knockout results, into three separate figures that explore these studies separately. A figure on normal development could, for instance, include documentation of metamorphosis, with a suite of images of postlarval stages. A figure documenting HCR could be expanded to include more stages, higher magnification images and other genes. A figure on the Cas9 knockdown of a PKS gene can provide details on the normal expression of this gene using HCR and possibly qRT-PCR.

Fourth, there should be consideration of providing more characterisation about the protein-coding genes comprising the chromosomal region (Chr. 4) that has marked differences between sexes. This could go beyond Supplementary Table 4 and Supplementary Figure 5B, and include analysis of expression in adult tissues (this would be enhanced by having matching male and female tissues), and KEGG pathways and GO enrichments.

Overall, this is an exciting development in the study of echinoids and echinoderms. This manuscript can be improved by providing the reader with more details. Given the exciting prospect of being able to study M. globulus settlement and metamorphosis in detail, the authors may consider providing more information about this part of the life cycle.

Author response:

Reviewer #1 (Public review):

(1) The efficiency of the CRISPR/Cas9 knockout is illustrated qualitatively, but no sample size or penetrance value is reported, making it difficult for readers to judge how robust or reproducible this result is.

We agree that quantitative documentation of the knockout experiments is needed. In the revised manuscript we will report the number of injected embryos, the number of independent experiments, and the penetrance of the albino phenotype obtained with the PKS guide RNAs, so that readers can judge the robustness and reproducibility of gene editing in M. globulus.

(2) The gene annotation is reported to have complete PFAM domain coverage for only 75% of predicted genes, but no independent completeness metric (such as BUSCO scored against the annotated gene set rather than the assembly) is provided, leaving open whether the remaining genes are genuinely novel, partial models, or annotation artefacts.

We have now scored BUSCO (metazoa_odb10, n = 954) directly against the annotated gene sets rather than the assemblies. The annotation of the blue male genome is 96.9% complete (S: 89.7%, D: 7.2%; F: 1.2%, M: 1.9%), and the annotation of the red female genome is 97.6% complete (S: 97.0%, D: 0.6%; F: 0.8%, M: 1.6%). These values indicate that the predicted gene sets show a high degree of completeness and that the genes lacking full PFAM domain coverage are not simply the product of fragmented or artefactual gene models. These metrics will be added to the revised manuscript.

(3) Finally, at the time of review the NCBI BioProject accession cited for the genome and sequencing data (PRJNA1477966) could not be located, and it is not clear from the text whether this accession, once available, will include the gene annotation and RNA-seq datasets in addition to the raw genomic sequencing reads.

We apologise for not commissioning the release of these records, the BioProject was still being processed by NCBI at the time of review. We confirm that the accession will include the gene annotations and the RNA-seq datasets in addition to the raw genomic sequencing reads, and we will verify that all records are publicly accessible before submitting our revision. We will also state explicitly in the Data Availability section which datasets are deposited under this accession.

Reviewer #2 (Public review):

(1) Genetic Background and Aquarium Trade Populations: A central argument of the manuscript is that M. globulus is attractive as a laboratory model because it is widely cultured in the aquarium trade and may exhibit reduced genetic variability due to captive propagation. [...] The manuscript does not provide sufficient information to evaluate these claims: how genetically representative are the sequenced individuals relative to natural populations; what is known about the provenance and breeding history of the aquarium trade stocks; are these animals derived from a small number of founder populations; is there evidence for substantial inbreeding or genetic bottlenecks within commercial brood stocks; and how similar are commercially available animals from different vendors and geographic sources?

We thank the reviewer for raising this important point, which we have investigated directly. The two sequenced colour morphs have distinct provenances: the red individual derives from a line bred in captivity in North America, whereas the blue individual is wild-caught from the Indo-Pacific. We therefore expected the captive-bred red animal to show reduced polymorphism relative to the wild blue animal. We will thoroughly check differences of polymorphism between and across individuals from either population and discuss the results.

(2) Presentation and Interpretation of HCR and Phalloidin Data: [...] The HCR images show detectable signal, but the expression domains are only minimally documented [...] Similarly, the phalloidin-labeled images provide limited anatomical information because the larvae are largely not labeled. I recommend: (1) adding labels identifying relevant embryonic regions and structures; (2) including arrows or overlays indicating key expression domains; (3) providing higher-magnification insets of relevant regions; (4) including selected optical sections rather than relying exclusively on 3D projections; (5) identifying known larval muscle groups in the phalloidin images; (6) improving image contrast and figure annotation where possible.

We agree that the HCR and phalloidin panels should stand on their own for readers who are not sea urchin specialists. The revised manuscript will implement the reviewer's suggestions and provide more detailed labelling of embryonic territories and high-magnificant insets.

Reviewer #3 (Public review):

First, the authors should provide a thorough description of the methods used to cultivate and maintain M. globulus. This should include further details about the closed aquarium system; a schematic of the system would be insightful. Basic details about husbandry are needed, including (i) stocking densities of adults, embryos/larvae, postlarvae/juveniles; (ii) frequencies of level and water changes/top-ups; (iii) feeding regime at all phases of the life cycle.

We agree that full documentation of the culture system is essential for uptake of the model, and we will expand the Methods accordingly. A construction schematic of the closed aquarium system, produced for us by Aquarium Connections (London, UK), is provided in Author response image 1 and will be included in the revised manuscript as a supplementary figure. The system is a three-tier rack (2000 x 1800 x 700 mm) comprising a brood-stock holding tank with egg-crate divisions, a row of settlement tanks, and a lower sump level with a UKS 200 protein skimmer, XHO algae lighting for live-feed culture, and a reverse-osmosis top-up reservoir, controlled via timer and switch boxes.

The revised Methods will include the following husbandry parameters. Adults are stocked at 20 M. globulus per 300 L tank and fed kombu seaweed once per week. Embryos are stocked at 1 embryo per 4 mL, which corresponds to the density required at the larval stage; approximately 20% of larvae are lost by the time competency is reached. Once feeding begins, larval cultures are cleaned every three days, with water topped up at the same time. The day-by-day larval feeding and rearing schedule, from fertilisation through metamorphosis and the transition to the juvenile diet, is summarised in Author response table 1 and will be included in the revised Methods. We agree that this documentation will provide the foundation for future improvements, including shortening time to competence and maturity and standardising settlement.

Author response table 1.

Larval feeding and rearing schedule for M. globulus.

Author response image 1.

Construction plan of the closed M. globulus culture system (Aquarium Connections, drawing GC-001, rev. P1; scale 1:10 at A3). The three-tier rack (2000 x 1800 x 700 mm) houses the brood-stock holding tank with egg-crate divisions (top), settlement tanks (middle), and sump level with UKS 200 protein skimmer, XHO algae lighting, and RO top-up reservoir (bottom).

Second, the description of the procurement and analysis of mRNA is brief, unreferenced and reads as protocols used for an established model species (e.g. what is PFA in this case - the concentration of paraformaldehyde and the buffer can vary markedly between organisms and life stages). Even the RNA extraction protocols can vary between and within species. [...] The HCR analysis, which is also scantily described, is restricted to embryonic and larval stages. Given the emphasis on the capacity of the M. globulus system to analyse all phases of the life cycle, it would be good to know if HCR can be performed on settled postlarvae, juveniles and adult tissues.

We will substantially expand the Methods to give a complete, referenced account of fixation (including the paraformaldehyde concentration and buffer used at each stage), RNA extraction, and the HCR protocol as applied to M. globulus, rather than assuming familiarity with protocols from established models. We will also address the applicability of HCR beyond embryonic and larval stages in the revision. We agree that demonstrating in situ methods in post-settlement stages would reinforce the central premise of the model, and we will report our experience with postlarval, juvenile and adult material in the revised manuscript.

Third, the authors should consider dividing Figure 2, which consists of confocal images of normal development, HCR results and CRISPR/Cas9 knockout results, into three separate figures that explore these studies separately. A figure on normal development could, for instance, include documentation of metamorphosis, with a suite of images of postlarval stages. A figure documenting HCR could be expanded to include more stages, higher magnification images and other genes. A figure on the Cas9 knockdown of a PKS gene can provide details on the normal expression of this gene.

We agree that Figure 2 is currently overloaded. In the revision we plan to split it into separate figures: one devoted to confocal documentation of normal development. We intend to further include one (or more) juvenile stage presenting the HCR expression data together with the CRISPR/Cas9 knockout results, with the improved annotation described in our response to Reviewer #2.

Fourth, there should be consideration of providing more characterisation about the protein-coding genes comprising the chromosomal region (Chr. 4) that has marked differences between sexes. This could go beyond Supplementary Table 4 and Supplementary Figure 5B, and include analysis of expression in adult tissues (this would be enhanced by having matching male and female tissues), and KEGG pathways and GO enrichments.

We agree that the sex-differentiated region on chromosome 4 deserves fuller characterisation. In the revised manuscript we will extend the analysis of the protein-coding genes in this region beyond Supplementary Table 4 and Supplementary Figure 5B, including functional characterisation (GO and KEGG enrichment) and, where material permits, analysis of expression in adult tissues.

eLife Assessment

This study demonstrates a critical role of the glycolipid membrane protein insertase (MPIase) in the twin-arginine translocation (TAT) pathway, a protein translocation system that is conserved across all domains of life. The successful reconstitution of the bacterial TAT system in both bacteria-derived and artificial liposomes provides solid experimental evidence that MPIase has a broader role in membrane protein translocation than previously recognized. By revealing the essential function of a nonproteinaceous membrane component in catalyzing a core cellular process, this work offers fundamental insights into the molecular mechanisms of protein translocation.

Reviewer #1 (Public review):

[Editors' note: Overall, the reviewing editor and reviewers agreed that authors rigorously addressed reviewers' major scientific concerns.]

Hanako and colleagues demonstrated that glycolipid MPIase is essential for the TAT system, and they successfully reconstituted the TAT system in vitro for the first time. This will facilitate the understanding of the mechanism of the TAT system.

Comments on revised version from Reviewer 1 and the Reviewing Editor:

In Reviewer 1's points 1 and 2 regarding statistical significance of immunoblotting data (in recommendation to authors), authors stated that they repeated the experiments three times in the rebuttal letter but not in the figure legends. We recommend authors to add a statement such as "errors denote +-SD (N = 3)" in the final version, which will make the data more solid and reliable.

Reviewer #2 (Public review):

Summary:

In this manuscript, the authors investigated the relationship between the Tat system and MPIase, a glycolipid that facilitates protein integration into the bacterial cell membrane. The TAT (twin-arginine translocation) system is a unique membrane transport machinery that exports fully folded proteins containing a twin-arginine signal peptide. Using both in vivo and in vitro approaches, the authors demonstrated that a sufficient amount of MPIase is required for Tat-dependent protein translocation. Furthermore, the authors successfully reconstituted the Tat transport system by combining recombinant TatA, TatB, TatC, MPIase, and FoF1-ATP synthase.

Strengths:

The reconstituted system clearly demonstrated the requirement for each component, as substrate translocation occurred only when all components were present. Based on these findings, the authors proposed a mechanistic role for MPIase in facilitating Tat-mediated membrane translocation. Previous studies have shown that MPIase is involved in Sec-dependent protein translocation and membrane protein integration, as well as YidC-dependent membrane insertion. The present study further demonstrated that MPIase also plays an essential role in the Tat translocation pathway. Overall, this work highlights the central importance of MPIase in bacterial membrane protein biogenesis and provides new insights into the molecular mechanism of Tat-dependent protein transport.

Comments on revised version from Reviewer 2 and Reviewing Editor:

(Reviewer 2) It is nice to find the mistake in Fig. 4B (RR, EK413, (not +MPIase)) and corrected it in the revised version. However, the data of Fig. 4B and the related modified text on P.9 L256-269 are not very easy to understand what the authors are trying to argue. This reviewer believes that the magnitudes of MPIase expression level are Ek413 > +MPIase > delta-MPIase. Therefore, the means of supernatant in the delta-MPIase and EK413 represents mostly cytosol and periplasm, respectively. On the other hand, the +MPIase shows less supernatant that means a large part of TorA-GFP are trapped nearby membrane but not aggressively transported. Similar results are found also in (KK) case, whereas the sup level in EK413 is lower than that of (RR) case due to the lack of twin-arginine motif in the substrate.

Although these results are convincing in explaining the model in Figure 5, authors might be able to attempt to refine them more understandable. Especially, the revised text on P.9 L256-269 is not enough for leading readers to the author's claim. So that, the authors should revise the text about Fig. 4B again and the related description in the manuscript. Another option is it would be possible to delete whole Fig. 4B, because Fig. 4A, C already well support the model in Fig. 5.

(Reviewing Editor) I agree with Reviewer 2. The first paragraph in the section "MPIase serves as the TAT signal receptor" runs very long and the clarity of the descriptions on Fig. 4B could be further improved. Fig. 4A, C would sufficiently support author's arguments. I recommend that authors consider these options for the final version.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Hanako and colleagues demonstrated that glycolipid MPIase is essential for the TAT system, and they successfully reconstituted the TAT system in vitro for the first time. This will facilitate the understanding of the mechanism of the TAT system.

My major points are listed below for the authors to consider:

(1) The authors successfully reconstituted the TAT system using the purified TatA/B/C, but the translocation efficiency was much lower than that of native INV. The authors partly attributed this to the reason that "MPIase recovery would be too low to detect the TAT activity" in the Discussion part. So, what would happen to the translocation efficiency if you added more MPIase to the reconstituted system? How about the abundance of MPIase from the INV and reconstituted proteoliposomes?

Fig. 3A shows that proteoliposomes reconstituted with active INV were inactive. However, TAT activity was observed when the proteoliposomes were fused with MPIase-containing liposomes. This indicates that MPIase was not sufficiently recovered during reconstitution. Typically, we added MPIase to proteoliposomes at 5% to phospholipids. However, increasing the amount of MPIase to exceed that of INV did not significantly increase the activity. The amounts of MPIase were stated in the text (P17, L3-4).

(2) Why were only TatC levels measured in Figure 2C, whereas the expression levels of TatA were not detected? Also, from my observation, the amount of TatC in the third lane is lower than that in the previous two lanes.

We also wanted to check the TatA levels, but the antibody was no longer available. Here, we confirmed that inactive INVs still contain Tat components.

(3) The authors should explain why the TatA/B/C ratios in Figure 3C (1:1:1) and Figure 3D (10:1:1) are inconsistent.

We examined various TatA/B/C ratios. We found that the ratio of 10:1:1 was better than 1:1:1. This was added to the text (P8, L11-16).

(4) ~30% of the fluorescence was recovered in the membrane fraction (Figure 4A) both in the functional TAT signal sequence (RR) and in the inactivating mutant signal sequence (KK), which suggests that MPIase acts as a relatively broad recognition factor. Given that MPIase does not discriminate between RR and KK, why do un-translocated substrates remain in the cytoplasm rather than non-specifically adhering to the membrane when MPIase is depleted in vivo?

Our results strongly suggest that MPIase acts as a signal sequence receptor. Therefore, in the absence of MPIase, the targeting of TAT precursors should be inhibited. Furthermore, because the TorA signal is less hydrophobic than the Sec signal, the level of nonspecific membrane binding would decrease.

Reviewer #2 (Public review):

Summary:

In this manuscript, the authors investigated the relationship between the Tat system and MPIase, a glycolipid that facilitates protein integration into the bacterial cell membrane. The TAT (twin-arginine translocation) system is a unique membrane transport machinery that exports fully folded proteins containing a twin-arginine signal peptide. Using both in vivo and in vitro approaches, the authors demonstrated that a sufficient amount of MPIase is required for Tat-dependent protein translocation. Furthermore, the authors successfully reconstituted the Tat transport system by combining recombinant TatA, TatB, TatC, MPIase, and FoF1-ATP synthase.

Strengths:

The reconstituted system clearly demonstrated the requirement for each component, as substrate translocation occurred only when all components were present. Based on these findings, the authors proposed a mechanistic role for MPIase in facilitating Tat-mediated membrane translocation. Previous studies have shown that MPIase is involved in Sec-dependent protein translocation and membrane protein integration, as well as YidC-dependent membrane insertion. The present study further demonstrated that MPIase also plays an essential role in the Tat translocation pathway. Overall, this work highlights the central importance of MPIase in bacterial membrane protein biogenesis and provides new insights into the molecular mechanism of Tat-dependent protein transport.

Weaknesses:

(1) To show the importance of the Tat system in bacterial cells, it would be good to describe in the introduction how many proteins are translocated via the Tat system.

The E. coli K12 cells possess 27 TAT substrates. This was added to the text (P2, L14).

(2) Figure 2B and D show that a sufficient amount of MPIase is important in SufI translocation. However, the reason why MPIase level was upregulated in the BL21 strain but not in the KS46 strain remains unexplained. The authors should address this point.

CdsA is a rate-limiting enzyme in MPIase biosynthesis. When MPIase production needs to increase, the cdsA promoters are activated. In KS46, however, cdsA is under the control of the arabinose promoter on a plasmid, so CdsA cannot be induced even when MPIase is necessary. This was added to the text (P6, L26-27).

(3) In Figures 4A and B, the authors explain that MPIase first works as a receptor of TorA-GFP without recognizing the RR motif. This conclusion is based on the results of the fractionation assays, where "sup" indicates the cytoplasmic and periplasmic fractions, and "ppt" indicates the membrane fraction. In Figure 4B, under the TatABC+++, (RR), +MPIase condition, the substrate is secreted most efficiently via the Tat pathway and should therefore be recovered in the periplasm fraction (sup). However, the authors point out that efficiently processed substrate was recovered in the ppt fraction rather than the sup fraction. The authors should explain why this occurred.

We found a mistake in the processing of the results for the sample of the TatABC+++, (RR), +MPIase condition, in Fig. 4B. Therefore, we remeasured the sample and corrected the figure. The relevant text was also modified (P9, L16-29). It was found that a large part of fluorescence was recovered in the supernatant fraction. We also uploaded the raw data for the fluorescence values.

Recommendations for the authors:

Reviewing Editor Comments:

Overall, I agree with the reviewers that the conclusions are reasonably well supported by the data. I have additional scientific/editorial concerns and recommendations:

(A) - Scientific:

(1) P.4 lines 10-15 and Fig. 1 top-left (the effect of overexpressed TatABC).

The authors state that the extent of Surf1 maturation increased upon overexpression of TatABC, but the band intensities for mature Surf1 in the TatABC-induced vs uninduced lanes are very similar.

The expression was weakened: ' The level of the mature form increased....' was changed to ' The level of the mature form slightly increased....'

(2) P.7 lines 23-24, Figure 3A. (the TAT system reconstituted in liposomes).

The authors state that a small portion of Surf1 is successfully translocated and protected by ProK. But unlike the assay results on IMVs, the size of the mature form (i.e., translocated form) appears to be the same as the untranslocated form. Some explanation seems to be needed.

In INV, the signal sequence is cleaved off by Lep, however, the cleavage is not coupled with translocation. In the reconstituted proteoliposomes, such cleavage is hardly observed since the membrane proteins were diluted by fusing with MPIase-containing liposomes. This was added to the text (P7, L26-29).

(3) Based on the model (Figure 5), the interaction between the RR motifs and MPI appears non-electrostatic as the substrate binds to the sugar moieties of the MPIase, not to the pyrophosphate part. Then what is the molecular nature of the substrate-MPI interaction? Some explanation/speculation seems to be needed.

MPIase has been identified as a factor that drives membrane protein insertion. Through the analysis of the mechanisms of insertion, we found that the glycan chain interacts directly with the transmembrane region of the membrane proteins through the numerous acetyl residues on the glycan. Moreover, we found that the positive charges of substrate membrane proteins interact with the pyrophosphate residue through the electrostatic interaction. Therefore, it is reasonable that the h region of the TAT signal binds to the MPIase glycan through the hydrophobic interactions, and the n region including the RR motif binds to the pyrophosphate residue. This was added to the text (P8, L25-27; P11, L14-17).

(B) - Editorial:

(1) P.2. lines 5-10. (Introduction)

The description of the TAT-targeting signal needs to be more clearly described for a broader readership. For example, what are "h" and "c"?

The signal sequence is composed of three regions, 'n', 'h', and 'c' from N-terminus. The n region contains positive charges including the RR motif. The next h region contains a hydrophobic stretch. The c region contains a cleavage site. This was added to the text (P2, L4-11).

(2) P.7 lines 23-24.

What is the rationale for using Pm-Fob-His as a control?

This is a control for a membrane protein unrelated to the TAT system to reveal that MPIase is specifically interacts with TatABC.

(3) P.8 lines 7-8.

CCCP is not defined.

CCCP (Carbonyl cyanide 3-chlorophenylhydrazone) is a protonophore. This was added to the text (P8, L18).

Reviewer #1 (Recommendations for the authors):

(1) The conclusions derived from Figures 1-3 heavily rely on representative immunoblotting images. Given that representative tracks can inherently introduce selection bias, how do the authors ensure the statistical robustness of these findings without quantitative bar graphs and rigorous significance analysis? To fully substantiate these interpretations, it is recommended that these immunoblotting assays be quantified across at least three independent biological replicates.

We did not quantify the results in sections where we presented qualitative discussions. However, we performed all experiments at least three times.

(2) Is it possible that the error values be added after the statistical values of translocation efficiency in Figures 2 and 3?

We added the SD values to some results in Fig. 3.

(3) "MPIase depletion of was then confirmed", an extra "of".

Corrected.

(4) The capitalization style of "TAT" in the whole text is suggested to be unified.

Corrected.

eLife Assessment

The study presents a fundamental structure interaction model describing how polymyxin interacts with hHepT2 the human oligopeptide transporter's 2 Aspartic acid residues to mediate nephrotoxicity via the D215 mutagenesis. These findings guided the rational design of potent polymyxin B analogues with reduced toxicity. The conclusions are supported by compelling evidence from computational and experimental data. The work will be of use to biochemists and medicinal chemists in developing the next generation of antibiotics for clinically relevant multidrug resistant bacteria.

Reviewer #1 (Public review):

Summary:

Polymyxins are the last line of drugs to treat gram-negative bacteria induced multi-drug resistance, however, they cause nephrotoxicity in 60% of patients. In this work, authors have studied the structure-interaction relationship (SIR) of polymyxins with hPepT2 using computational and experimental methods. Moreover, it is observed that the electrostatic interactions coordinate the hPepT2-Polymyxin interactions, hence, an alanine scanning strategy is used to understand the interactions and derive the polymyxin variants.

Computational methods such as molecular modeling, coarse grained and all atom MD simulations, and interaction studies are performed. While the results are validated in the mouse model which is a great strategy to prove the hypothesis.

Strengths:

A clear understanding on the hPepT2-Polymyxin interactions and role of electrostatic interactions is one of the very important strengths of the paper. In addition, this work proposes a great pipeline for using computational approaches and experimental validation methods to guide the development of newer antibiotics.

Overall, the study proposes novel polymyxin analogues with reduced or no nephrotoxicity, thereby providing a promising foundation for the rational development of safer lipopeptide antibiotics.

Comment on revised manuscript:

I appreciate the effort by Authors to address my comments. The manuscript now contains details of ACE inhibitors in the introduction. In addition, method section is updated for better reproducibility of the MD simulation and structure modeling methods. I congratulate authors for reporting a wonderful scientific study. I don't have any further recommendations. Thank you!

Reviewer #3 (Public review):

Summary:

Jiang et al. described findings aimed at interrogating the interactions of the antibiotic polymyxin B with human kidney proteins that mediate nephrotoxicity. Their findings using both computational molecular dynamics simulations and experimental approaches illustrate the importance of aspartic acid residues (D215) in mediating the antibiotic uptake into the cells, and upon mutagenesis with Alanine, the effects are less pronounced. Further, they could modify the antibiotic units interacting with proteins into less toxic peptides with retained antibacterial properties.

Strengths:

I was impressed by this text, which advances the knowledge of how the antibiotic causes human nephrotoxicity and how this could be exploited into less problematic antibiotic peptides.

Comments on revised version.

Majority of the raised issues were addressed satisfactorily.

Author response:

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

Reviewer #1 (Public review):

(1) The introduction is well articulated; however, including a paragraph on the known inhibitors might be helpful in understanding the current status. In addition, it might also help to introduce Dabs, FADDI variants, Gly-sar and MIPS.

We thank the reviewer for the suggestion. We have included a paragraph on the substrates of hPepT2 (Lines 104-112).

(2) The following details of modeling with AlphaFold2 should be included: how the final structure was selected, what the RMSD and structure alignment of the template are, and the final selected structure. A section on modeling with all the parameter details might be useful for reproducing the structure. In addition, specify how the alanine scanning was performed alongside the structure prediction of polymyxins.

We appreciate the reviewer's careful assessment of our computational methodology. In the Materials and Methods section of the revised manuscript, we have added a dedicated subsection describing the detailed methods: "Structural modelling of hPepT2 and polymyxins" as shown below (Lines 435-447).

(1) hPepT2 structure prediction: The inward-open conformation of hPepT2 was predicted using AlphaFold2 via the ColabFold implementation with default parameters, employing the MMseqs2 multiple sequence alignment pipeline. To obtain the physiologically relevant outward-open conformation of hPepT2 for substrate binding, homology modeling was performed using MODELLER with the cryo-EM structure of rabbit PepT2 in the outward-open state (PDB: 7NQK) as the template. The sequence alignment between hPepT2 and rabbit PepT2 was conducted using Clustal Omega. One hundred models were generated, and the final model was selected based on the lowest Discrete Optimized Protein Energy score and verified by Ramachandran plot analysis (with >95% of residues in favored regions).

(2) Polymyxin structure and alanine scanning: The structure of polymyxin B1 was constructed and energy-minimized using the CHARMM36 force field. Computational alanine scanning of polymyxin B1 was performed by individually replacing each Dab (2,4-diaminobutyric acid) residue at positions 1, 3, 5, 8, and 9 with alanine using the mutagenesis wizard in PyMOL.

(3) In the all-atom MD simulation method, detailing several parameters might help in reproducing the results: simulation time for each system, water model, system composition, protonation state, box type and dimensions, salt ions and concentration, membrane parameters and ligand parameterization methods. Also, the following details on energy minimization might be useful: minimization algorithm, number of steps for minimization and structure restraints in place.

We thank the reviewer for the suggestion. In this study, all-atom molecular dynamics simulations were performed using the TIP3P water model for solvation. The system was placed in a rectangular simulation box with dimensions of 11 × 11 × 14 nm³. Neutralization was achieved by adding 0.15 M NaCl, resulting in a total of approximately 158,000 atoms. The membrane lipid composition consisted of 50% phosphatidylcholine, 25% phosphatidylethanolamine, and 25% phosphatidylserine, consistent with the coarse-grained molecular dynamics simulations. Force field parameters for polymyxin B1 were generated using CGenFF. Energy minimization was carried out using the steepest descent algorithm with default parameters. For production simulations, 500-ns simulations were run for sampling the outward‑open conformation, and 100-ns simulations were run for analyzing the polymyxin‑hPepT2 interaction. We have provided these methodological details in the revised manuscript (Lines 477-492).

(4) On page 6, line 210, the MIC is used for the first time; although MIC is given in the abbreviation list, the first occurrence should have a complete name. A one-line explanation of MIC in the introduction or wherever suitable might be better but is not mandatory.

We thank the reviewer for the suggestion. We have included the full term of MIC and its definition in the revised manuscript (Lines 236-238).

(5) Similarly, Gly-sar is first mentioned on page 8, line 301, but its complete name is only mentioned later on page 10, line 368. This can be addressed if a short description is included in the introduction section.

We apologize for the overlook. We have now referred to the full term of Gly-Sar and a short description on Page 3 (Lines 104-106), when it was first mentioned.

(6) For coarse-grained MD simulation, why were 2 replicates performed? Most studies perform 3 replicates, which are also good in terms of statistics and error bar calculations. In addition, the authors should specify whether an independent minimization is done for each of the two replicates or whether the minimization step is common for both.

In our study, coarse-grained MD simulations served as a preliminary exploration to identify the potential binding trajectory of polymyxin B to hPepT2, rather than the primary source of quantitative interaction data. The two independent CG-MD replicates yielded highly consistent final binding poses of polymyxin B at the lateral opening gate of hPepT2, serving this exploratory purpose. Furthermore, more detailed interaction energetics were subsequently characterized using all-atom MD simulations with four replicates (Fig. 4). We have now clarified in the revised Methods that independent energy minimization was performed for each of the two CG-MD replicates (Lines 465-469).

(7) For MD simulation results, giving simulation movies in supplementary results might be a better way to show how the trajectories behaved.

We thank the reviewer for the suggestion. We have prepared an animation of the coarse-grained MD trajectory showing the binding trajectory of polymyxin B to hPepT2, which has been submitted as a Supplementary Movie (.mp4 format). This movie visualizes how polymyxin B molecules spontaneously approach and bind to the lateral opening gate of hPepT2 over the 3-μs simulation. The Supplementary Movie is referenced in the revised Methods section (Lines 471-474).

(8) The description of visualisation software such as VMD or PyMol is missing. The authors should specify if any visualization tool is used.

We apologize for this omission. In the revised manuscript, we have specified in the Methods section that molecular visualizations, structural figures, and trajectory animations were prepared using PyMOL (Schrödinger) and ChimeraX (Lines 471-474).

(9) For the mouse model study, the authors claim that FADDI-795 has no observable nephrotoxicity; however, the n=3 shows that a very small number of mouse models were used to make the assumption. In addition, the number of mice used in each experiment is not explicitly mentioned in the methods section.

We have now added “n=3 each group” in the Methods section (Line 603). In this proof-of-concept study, we employed acute kidney damage analysis in a small number of mice to rapidly screen these analogues for nephrotoxicity. Three biological replicates were used because our mouse nephrotoxicity model is robust as demonstrated by a large number of animals in our drug discovery program (e.g. Nature Communications 2022, 13, 1625). Moreover, minimising animal use is a fundamental principle of the 3Rs (Replacement, Reduction, and Refinement).

(10) In Table 2, the column 8 header is not visible.

We apologize for the format error and have now fixed it in Table 2.

Reviewer #1 (Recommendations for the authors):

(1) The MD simulation trajectories or movies are not provided.

As mentioned in our response to Comment 7 above, a Supplementary Movie (mp4 format) showing the CG-MD binding pathway has been provided with the revised manuscript and referenced in the Methods section.

(2) Mouse model experiments are performed over n=3, which might not be statistically significant in case of an experiment with a large n.

Please see our response to Comment 9 above.

Reviewer #2 (Public review):

(1) Several conclusions would benefit from a more cautious interpretation. A major limitation is that several transporter mutations substantially altered total or membrane protein expression, making it difficult to distinguish effects on substrate binding from indirect effects caused by impaired transporter stability or trafficking. The authors acknowledge this limitation in the Discussion, but some mechanistic conclusions remain stronger than the available evidence supports.

We agree with the reviewer that it is difficult to draw definite conclusions when both K_m and V_max are altered in some mutants, especially when expression of the mutant transporter is impaired. However, we consider kinetic analysis to be more indicative of altered substrate binding, as changes in K_m are generally more closely associated with substrate recognition and binding, whereas reduced transporter expression predominantly affects V_max. Nonetheless, to avoid overinterpretation, we have moderated our conclusions by replacing “was/is” with “may be” (such as Lines 318, 319, 323, and 347).

(2) Similarly, while the proposed binding model is biologically plausible and supported by mutagenesis, it remains an inferred model derived from molecular simulations rather than a direct structural determination. Statements describing the model as "validated" should therefore be moderated to indicate that the experimental data provide support rather than definitive structural confirmation.

We thank the reviewer for the suggestion and have replaced “validated” with “explored” in the main text (Lines 298).

(3) The translational implications are promising but remain preliminary. Although FADDI-795 demonstrated reduced nephrotoxicity in the mouse model while maintaining antibacterial activity, no pharmacokinetic studies were presented to demonstrate reduced renal accumulation or altered tissue distribution, and additional efficacy studies in infection models would further strengthen the therapeutic claims.

We thank the reviewer for the suggestion. The current study is a proof-of-concept investigation to determine whether the structure-interaction relationship (SIR) model of hPepT2 and polymyxin can be used to design new lipopeptides with reduced nephrotoxicity. Following screening of the newly designed lipopeptides, we will select the most promising candidate for comprehensive pharmacological evaluations, including pharmacokinetics, toxicology, and efficacy.

Reviewer #2 (Recommendations for the authors):

(1) Strengthen the interpretation of the mutagenesis data. For transporter mutants that exhibit reduced total or cell surface expression, consider discussing more explicitly the extent to which reduced polymyxin uptake may result from altered protein stability or trafficking rather than direct effects on substrate recognition. Where possible, clarify this distinction throughout the Results and Discussion.

We thank the reviewer for the recommendation. As suggested, we have expanded the discussion on the mechanisms underlying the altered protein expression of the hPepT2 mutants (Lines 325-333). 

(2) Moderate statements regarding structural validation. The molecular dynamics simulations, mutagenesis, and functional studies provide strong support for the proposed structure-interaction model; however, wording such as "validated the first SIR model" could be softened to reflect that the binding model remains computationally inferred rather than directly resolved by structural methods.

We thank the reviewer for the recommendation. We have revised the statement as described in our response to Comment 2 above.

(3) Expand the discussion of study limitations. A more explicit discussion of the limitations associated with molecular dynamics predictions, the influence of altered transporter expression on functional interpretation, and the lack of direct pharmacokinetic measurements of renal accumulation would improve the balance of the manuscript.

We thank the reviewer for the recommendation. We have expanded the discussion by elaborating the limitations of the molecular dynamics predictions (Lines 289-291), the effect of altered transporter expression on functional interpretation (Lines 324-333), and the lack of direct pharmacokinetic and renal accumulation measurements (Lines 393-401).

(4) Provide additional methodological detail where appropriate. Please clarify the number of biological replicates used for each experimental approach, report exact P values where practical, and indicate whether assumptions for the statistical tests were assessed.

We thank the reviewer for the recommendation. We have provided more information on experimental replicates in the figure legends. P values have been added (Lines 208-217), and the statistical tests used for every analysis are now specified in the corresponding table or figure legends.

(5) Future validation of the lead compound. Although this may fall outside the scope of the present manuscript, it would be helpful to briefly discuss future studies aimed at evaluating the pharmacokinetics, renal exposure, and efficacy of FADDI-795 in relevant infection models, as these will be important for establishing its translational potential.

We thank the reviewer for the recommendation. We have now incorporated this discussion into the revised manuscript (Lines 396-399).

(6) Improve figure presentation. Several figure legends could include additional methodological information to make the figures more self-contained. In particular, indicating sample sizes, statistical tests used, and definitions of error bars would improve readability.

We thank the reviewer for the recommendation. We have added this information to the figure legends as suggested.

(7) Language and style. The manuscript would benefit from careful language editing to improve grammar, sentence structure, and readability. Several long sentences in the Discussion could be shortened, and minor typographical errors should be corrected throughout the manuscript.

We thank the reviewer for the recommendation. We have revised the long sentences where possible and carefully proofread the manuscript to improve grammar, sentence structure, and readability. The revised manuscript has also been reviewed by a native English speaker.

(8) Terminology. Please ensure that abbreviations such as "SIR" are defined at first use and are used consistently throughout the manuscript.

We have confirmed that SIR is defined when first introduced in both the Abstract and main text (Lines 47 and 120).

(9) Minor corrections.

- Check for consistency in the naming of hPepT2/hPEPT2 throughout the manuscript.

- Verify that all figures, supplementary figures, and tables are cited sequentially in the text.

- Consider reporting confidence intervals alongside kinetic parameters (K_m and V_max) where appropriate to facilitate interpretation.

We thank the reviewer for the recommendation. We have revised the manuscript accordingly. “hPepT2” is now used consistently throughout the text. We have revised the order of the supplementary figures to ensure that they are cited sequentially in the text; and included the 95% confidence interval of the K_m and V_max values of the D215A mutant compared to the wild type (Lines 208-212).

Reviewer #3 (Public review):

(1) Interactions of Polymyxin B with kidney proteins were not demonstrable in vivo, and with reliable technologies such as X-ray or NMR.

We thank the reviewer for the comment. To the best of our knowledge, polymyxin-induced nephrotoxicity is primarily driven by renal tubular accumulation, followed by mitochondrial dysfunction, oxidative stress, inflammatory activation, and apoptosis, ultimately resulting in proximal tubular injury. Little is known about the molecular interactions of polymyxins with kidney proteins.

Reviewer #3 (Recommendations for the authors):

(1) Title: I suggest that you bring out the actual findings to convey the message.

We thank the reviewer for the comment. We have changed the title to “Lateral opening site of human oligopeptide transporter 2 plays a key role in the interaction with polymyxins”.

(2) Line 133: Provide data for these 10 molecules (possibly by supplementary file/table).

We thank the reviewer for the suggestion. As described in the Methods, ten coarse-grained polymyxin molecules were randomly inserted into the upper water layer of the simulation system (Lines 452-456). Because these molecules are structurally identical and share the same force field parameters, they differ only in their initial random positions. Thus, we believe that presenting the coordinates or labelling the individual molecules would not provide additional information and can be misleading.

(3) Line 135: Figure S3 mentioned prior Figure S2.

We apologize for the error. We have now revised the sequence of the supplementary figures.

(4) Lines 159 - 171: Cite Figure 5 a - b appropriately. Describe the obtained results fully.

We thank the reviewer for the suggestion. We have expanded the relevant section to include a more detailed description of the results, with reference to Fig. 5a and 5b (Lines 187-193).

(5) Line 208: Was cytotoxicity performed to validate this claim?

We did not specifically evaluate the cytotoxicity of the lipopeptides in cultured cells because the cellular uptake data already indicated their potential toxicity. Importantly, histopathological assessment in animal models provides the most reliable evaluation of their in vivo toxicity (Nature Communications 2022, 13, 1625).

(6) Line 509: Provide the approval number.

The animal ethics approval number is AEC37419, which has been added to the “Ethics approval and consent to participate” section.

(7) Table 2: Add the Standard deviation and the number of replicates performed.

We thank the reviewer for the suggestion. We have added “n = 3 mice per group” to the table legend. As described in the Methods (Lines 590-597), the MIC of each lipopeptide was determined. MIC values are reported as absolute concentrations rather than continuous variables; therefore, reporting the mean ± standard deviation is not applicable. Likewise, the maximum kidney Semi-Quantitative Score (SQS) was determined by histological examination and is presented as the representative maximum score for each treatment group. Accordingly, the calculation of a mean and standard deviation is not appropriate for either MIC or SQS values presented in Table 2.

(8) The authors need to indicate the number of replicates performed in the methods section.

We thank the reviewer for the suggestion. As mentioned in our response to Recommendation 6 from Reviewer 2 above, we have added this information to the figure legends as suggested.

eLife Assessment

This study presents important findings on how cardiac regenerative capacity diverges across species by examining heart repair in two species of livebearers, platyfish and swordtails. In contrast to zebrafish, the livebearer species show persistent scarring after cryo-injury, and the work highlights how lineage-specific anatomical and immunological traits may constrain regenerative competence. The study is compelling, the data are convincing, and the results contribute to our understanding of the mechanisms underlying heart regeneration across vertebrates.

Reviewer #1 (Public review):

Summary:

How the regenerative capacity of the heart varies among different species has been a long-standing question. Within teleosts, zebrafish can regenerate their hearts, while medaka and cavefish cannot. The authors examined heart regeneration in two livebearers, platyfish and swordtails. Interestingly, they found that these two fish species lack the compact myocardium layer that contains coronary vessels. Furthermore, these fish form a "pseudoaneurysm" after cryoinjury without initial deposition of fibrotic tissues. However, delayed leukocyte infiltration and prolonged inflammation lead to permanent scar tissue in the injured heart. Although their cardiomyocytes can also proliferate, platyfish and swordtails can only regenerate partially. The authors argue that the restorative mechanism of platyfish and swordtails likely reflects "evolutionary innovations in the ventricle type and the immune system".

Strengths:

The authors took advantage of the annotated genome of platyfish to perform transcriptomic analyses. The histological analyses and immunostaining are beautifully done.

Comments on revised version:

The authors have addressed all my previous concerns. I don't have further questions.

Reviewer #2 (Public review):

This manuscript by Hisler, Rees, and colleagues examines the cardiac regenerative ability of two livebearer species, the platyfish and swordtail. Unlike zebrafish, these species lack cortical myocardium and coronary vasculature. Cryoinjury to their hearts caused persistent scarring at 60 and 90 days post-injury and prevented most of the myocardium from regenerating. Although the wound size progressively shrinks and fibronectin content decreases, the myocardial wall does not recover. Transcriptomic profiling at 7 dpi revealed significant differences between zebrafish and platyfish, including alterations in ECM deposition, immune regulation, and signaling pathways involved in regeneration, such as TGFβ, mTOR, and Erbb2. Platyfish exhibit a delayed but chronic immune response, and although some cardiomyocyte proliferation is observed, it does not appear to contribute to myocardial recovery significantly.

Overall, this is an excellent manuscript that tackles a crucial question: do different fish lineages have the ability to regenerate hearts, or is this capability limited to a few groups? Therefore, this work is relevant to the fields of cardiac regeneration and comparative regenerative biology for a broad audience. I am very enthusiastic about expanding the list of species tested for their heart regeneration abilities, and this study is detailed and rigorous, providing a solid foundation for future comparative research. However, there are several aspects where additional work could significantly strengthen the manuscript.

Comments on revised version:

The authors have done a fantastic job addressing all my comments, covering both the experimental and the more conceptual aspects of my critique. I am fully satisfied with their response.

Author response:

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

We sincerely thank the editors and reviewers for their thorough evaluation of our manuscript. We are grateful for their enthusiastic support of our comparative approach to understanding cardiac regeneration across fish lineages, and we appreciate the care with which they captured and summarized the key findings of our study. Their constructive suggestions have been invaluable in strengthening the work, and we are pleased to submit the revised manuscript enclosed herewith.

The revised manuscript contains 7 main figures and 8 supplementary figures, representing the addition of two figures relative to the previous version. Specifically, we have included additional Podocalyxin immunostaining to clarify our characterization of ventricular vascularization, and Picrosirius Red staining to improve the characterization of fibrillar collagen in uninjured Xiphophorus hearts relative to zebrafish. We have also refined our RNA-seq analysis by manually curating well-defined markers of cardiomyocyte cell fate change and immune system activation, such that the manuscript no longer relies on a single-cell RNA-seq dataset. All remaining revisions have been made in direct response to the reviewers' comments, as detailed below.

Reviewer #1 (Public review):

Minor Weaknesses:

Transcriptomic analysis was only done for one time point. Different time points could be included to validate whether some processes occur at different time points. But this can be done in the future for more detailed studies.

We agree with this valuable suggestion. While RNA-seq at multiple time points would undoubtedly enrich our understanding of the temporal dynamics of Xiphophorus heart regeneration, such an expansion would constitute a substantial independent study extending well beyond the scope of this initial characterization. We agree that this represents a compelling avenue for future investigation, and we have acknowledged this limitation explicitly in the Discussion section of the revised manuscript (lines 357-363).

Reviewer #1 (Recommendations for the authors):

(1) Figure 5B and quantification in Figure 5D seem inconsistent. The numbers of Mpx neutrophils at Day 7 seem the same as Day 14 and 30, but overall, there are very few Mpx+ cells. The authors should examine earlier time points, such as Day 1 and 3, to determine if neutrophil and macrophage infiltration is different in zebrafish.

We thank the reviewer for this careful observation. We agree that the apparent similarity in Mpx⁺ cell numbers across Day 7, 14, and 30 in the original figures warranted closer examination of earlier time points. We have therefore performed new Mpx and L-plastin immunostaining at 1 and 3 dpci in Xiphophorus hearts. These analyses confirm low immune cell infiltration at these early time points, with Mpx⁺ cell numbers peaking at 14 dpci, which is a pattern strikingly different from the rapid and robust neutrophil infiltration observed in zebrafish within the first few days post-injury. These results reinforce our conclusion that the inflammatory response in Xiphophorus is both delayed and attenuated relative to zebrafish, and the updated data are now presented in Figure 6.

(2) Their transcriptomic data at 7 days post injury suggest that TGFß signaling was not activated after injury, and tenascin C was not expressed in platyfish. The authors might check whether TGFB signaling is activated and tenascin is expressed at later time points, since platyfish show persistent scarring. This may help determine whether the molecular mechanisms of scarring in platyfish are the same as in zebrafish, and it just happens late.

We think it is a good suggestion. Unfortunately, our anti-TnC antibodies do not produce a clear or specific signal in platyfish tissue, precluding immunofluorescence-based analysis of TnC expression at later regenerative stages. To partially address this limitation, we have included transcript abundance data for several genes associated with TGF-β signaling in the new Figure 5, which provides an initial view of this pathway's activity during Xiphophorus heart regeneration. Nevertheless, a thorough characterization of TGF-β activity and TnC expression, especially including protein-level validation and temporal profiling, would require dedicated methodological development and constitutes a new study. We therefore consider this to be an important avenue for future investigation rather than a component of the present initial characterization.

(3) The PCNA staining BrdU labeling experiments suggest that proliferating cardiomyocytes are not maintained even though they re-enter the cell cycle. Since the platyfish hearts lack the proliferative compact cardiomyocytes that account for myocardial regeneration in zebrafish, do the authors suggest that their trabecular cardiomyocytes proliferate or that they only undergo DNA synthesis?

This question touches on a central unresolved aspect of our findings. Our results indicate that a subset of cardiomyocytes re-enter the cell cycle and undergo DNA synthesis at 7 and 14 dpci, as evidenced by PCNA staining (new Figure 7D). However, BrdU incorporation combined with Tropomyosin immunostaining reveals an absence of BrdU⁺/Tropomyosin⁺ cardiomyocytes within the border zone myocardium (new Figure 7F), indicating that newly synthesized DNA does not translate into efficient cardiomyocyte repopulation of the injured area. Whether the detected S-phase entry is followed by mitosis, or whether these cardiomyocytes undergo DNA synthesis without completing cell division, a phenomenon known as endoreplication, which has been described in other cardiac contexts, remains to be determined. Resolving this question would require live imaging or mitotic marker analyses beyond the scope of the present study, and we have highlighted this as an important open question in the revised Discussion.

(4) Line 136-137. The authors might clarify what they meant by "N2.261 antibody recognizes different myosin types in zebrafish versus platyfish". Are these N2.261+ cardiomyocytes in the atria of platyfish also immature cardiomyocytes, but are there more immature cardiomyocytes in platyfish than in zebrafish?

We thank the reviewer for this good question. In zebrafish, N2.261 has been established as a marker of immature cardiomyocytes in larvae and regenerating myocardium; however, we cannot currently conclude that N2.261-positive atrial cardiomyocytes in platyfish represent an analogous immature population. Rather, the broad atrial immunoreactivity in platyfish most likely reflects the presence of the N2.261 epitope within the dominant atrial myosin heavy chain isoform of this species. Thus, platyfish atrial myosin shares greater sequence similarity with the antibody's target epitope than the corresponding zebrafish atrial myosin does. In other words, the differential labeling pattern between species is more likely attributable to evolutionary divergence in myosin heavy chain sequences than to differences in cardiomyocyte maturation state. To directly identify which amino acid residues are essential for N2.261 immunoreactivity and to resolve these evolutionary differences, we have initiated a dedicated epitope-mapping study.

To clarify this interpretation in the manuscript, we have replaced the previous concluding statement with the following: "Together, these findings indicate that N2.261 recognizes distinct myosin heavy chain isoforms in zebrafish and Xiphophorus: an embryonic ventricular isoform in the former and an atrial isoform in the latter. The differential labeling pattern between species is more likely attributable to evolutionary divergence in myosin heavy chain sequences than to differences in cardiomyocyte maturation state, reflecting the substantial lineage-specific reshaping of cardiac myosin repertoires that has occurred between cyprinids and poeciliids."

(5) Line 212-"Interspecies comparison revealed that upon cryoinjury, 199 and 268 orthologous gene transcripts were more abundant in zebrafish than in platyfish, respectively". Does the author mean that "more abundant in zebrafish than in platyfish and vice versa"?

We thank the reviewer for flagging this ambiguity. The original sentence was indeed unclear, and we have revised it to read: "Interspecies comparison revealed that upon cryoinjury, 199 orthologous gene transcripts showed a higher log₂FC in zebrafish than in platyfish, while 268 showed the opposite pattern, indicating that the two species mount distinct transcriptional responses to cardiac injury." We believe this phrasing now unambiguously conveys that the comparison is bidirectional.

Reviewer #2 (Public review):

Major comments

(1) Title selection

The title the authors chose suggests that platyfish and swordtails "partially regenerate," but I do wonder how much these animals truly regenerate. This may be a semantic discussion and a matter of personal preference. Still, based on other significant work on regenerative capacity (see, for example, the landmark cavefish regeneration paper PMID: 30462998 or work on medaka PMID: 24947076), the persistence of such a prominent fibrotic scar would be considered a minimal regenerative capacity. Measuring this "partial regeneration" more precisely by comparing zebrafish with platyfish and swordtails would also greatly strengthen the comparisons made here - see below.

The same can be said about line 152-153 - do these hearts "regenerate" with deformation and partial scarring, or would it be more fair to say that they are "healed" or "repaired" with a process that involves fibrosis?

We thank the reviewer for raising this conceptual point, and we appreciate the references to the cavefish and medaka literature. We acknowledge that the term "partially regenerate" requires careful justification given the persistence of a substantial collagenous scar at the injury site. We have retained this terminology for the following reasons:

(1) The bulging wound undergoes resorption over time, suggesting that the initial structural deformation is transient rather than permanent.

(2) Wound size is significantly reduced and the proportion of hearts retaining visible injury decreases over the course of the experiment.

(3) Cardiomyocyte proliferation is detectably elevated in the myocardium at 7 and 14 dpci, indicating that some regenerative machinery is engaged.

We would also note that partial heart regeneration accompanied by residual scarring has been described in newts, a classically regenerative vertebrate. This suggests that the boundary between regeneration and fibrotic repair is not always clear-cut. Rather than a binary distinction, the cardiac injury response may exist on a continuum, where hallmarks of regeneration, such as wound resorption, reduced scar size, and cardiomyocyte proliferation, can coexist with persistent fibrosis. We therefore believe that "partial regeneration with persistent scarring" accurately and honestly reflects our findings, and we have refined the relevant passages in the manuscript, including the indicated lines, to make this nuance explicit.

(2) Cross-species comparisons

Having two species of livebearers strengthens the findings of this paper, but the presentation of results from both species is inconsistent. For example, the reader should not be asked to assume that the architecture of the swordtail ventricle is similar to that of the platyfish (line 125). The same applies to the presence or absence of coronary vessels (Figure 1), the reduction in wound area over time (Figure 3), and the immune system's response (Figure 5). Most importantly, the authors miss an opportunity to move from qualitative observations to quantifying the "partial regeneration" phenotype they observe. Specifically, providing a side-by-side comparison between these new species and zebrafish would help define the extent of differences in regeneration potential. For instance, in Figure 6, while the authors provide excellent quantification of PCNA staining in platyfish, these data are less meaningful without a direct comparison with zebrafish results. The same applies to Figures 6E and 6F - although differences are noted, quantifying these results would enable a more rigorous assessment of the process.

We thank the reviewer for this constructive critique. We agree that greater consistency in the cross-species presentation strengthens the comparative framework of the paper, and we have made several additions to address this.

To document the cardiac architecture of swordtails explicitly, rather than asking the reader to assume similarity with platyfish, we have added the following data:

(1) New supplementary figure (Figure S1) dedicated to the swordtail ventricle, incorporating 1) AFOG, 2) Picrosirius Red, 3) Alkaline Phosphatase Assay for the vasculature, 4) and immunostaining against Fibronectin, N2.261, and F-actin.

(2) The dynamics of swordtail heart regeneration across 7, 14, 30, 60, and 90 dpci are presented in a separate supplementary figure (Figure S5), which also includes quantification of wound size over time. These data show that the wound, representing approximately 20% of ventricular area at 7 dpci, is reduced to approximately 2.5% by 60–90 dpci, providing quantitative support for our characterization of partial regeneration in this species.

We appreciate the reviewer's point regarding side-by-side quantitative comparisons with zebrafish for markers such as PCNA. We respectfully note, however, that cardiomyocyte proliferation in zebrafish following cryoinjury is extensively documented across multiple independent studies, and we consider this body of evidence sufficient to contextualize our platyfish findings without requiring full parallel quantification. Nevertheless, in response to this comment, we have included side-by-side zebrafish and platyfish data for BrdU staining in the new Figure 7, with zebrafish quantification drawn from our previously published dataset (Sallin et al., 2015). We believe this addition meaningfully strengthens the cross-species comparison at this key figure while remaining within the scope of the present study.

(3) Lack of coronary vasculature

There is a growing body of evidence highlighting the importance of the coronary vessels during zebrafish heart regeneration (PMIDs: 27647901, 31743664). Surprisingly, this finding has not been integrated or discussed in the context of this literature.

The results of the alkaline phosphatase assay and anti-podocalyxin-2 staining appear inconsistent. Specifically, in Supplementary Figure 1L-M, we can see some vessels covering the bulbus arteriosus and also what appears to be a signal in the ventricle. However, in Figures 1 K and 1L, we cannot see any vessels, even in the bulbus. The authors should also be more rigorous and add a description of how many animals were analyzed, their ages, and sizes. In zebrafish, the formation of the coronary arteries appears to depend on animal size and age. With the data provided, we cannot say whether this is a one-time observation or a consistent finding across many animals at different ages and across both species.

We thank the reviewer for raising this point and for directing us to the relevant literature. We agree that the role of coronary vasculature in zebrafish heart regeneration is an important and growing area of research, and we have now integrated a discussion of this evidence into the manuscript, highlighting the contrast with the avascular Xiphophorus ventricle and its potential implications for regenerative capacity.

Regarding the apparent inconsistency between the alkaline phosphatase assay and the anti-Podocalyxin (anti-Podxl) staining, we thank the reviewer for this careful observation. We have performed additional staining to resolve this discrepancy and conclude that the two approaches, rather than being inconsistent, reflect distinct and complementary aspects of ventricular organization.

To better characterize the anti-Podxl signal, we performed immunofluorescence on thick (50 µm) sections of both platyfish and zebrafish hearts. In contrast to zebrafish, anti-Podxl immunoreactivity in platyfish is confined to a very thin outer layer of the myocardium. This subtle signal accounts for the weak staining previously observed in whole-ventricle preparations and was already visible, though not highlighted, in the original figures (see white arrow in new Supplementary Figure S3C, H, O–R). A direct comparison of anti-Podxl staining across species (Supplementary Figure S3A, D, N, R) clearly demonstrates the absence of coronary vascularization in the platyfish ventricle. Consistent with this, no vascular signal is detected in the platyfish bulbus arteriosus (Supplementary Figure S3I).

To further clarify the nature of the outer myocardial layer in Xiphophorus, we performed Picrosirius Red (PSR) staining, which selectively labels fibrillar collagen, across all three species. Whereas the compact outer myocardium of zebrafish is PSR-negative, a thin but distinct PSR-positive layer is present at the ventricular surface of both platyfish and swordtails, closely mirroring the anti-Podx1 staining pattern.

Taken together, these findings indicate that the outer ventricular layer of Xiphophorus fish is composed of a thin collagen- and Podxl-immunoreactive matrix, which is structurally distinct from the vascularized compact myocardium of zebrafish and is consistent with the absence of coronary vessels in these species.

Finally, we acknowledge the reviewer's request for greater rigor regarding sample sizes, animal ages, and body sizes. These details have now been added to the Methods section. We note that, in line with the reviewer's observation regarding zebrafish coronary development, we have ensured that animals of comparable size and age ranges are described for each species to allow meaningful cross-species interpretation.

(4) The link between livebearers' responses and pseudoaneurysms is overstated. This work is already extremely relevant without trying to make it medically oriented.

We agree that the clinical parallel with pseudoaneurysms was overstated in the original manuscript, and that the comparative and evolutionary relevance of our findings stands on its own merits without requiring a medical framing. We have accordingly removed the term from the Results section and now invoke it only briefly at the close of the Discussion, where a concise note on broader medical relevance is appropriate without overshaping the narrative of the paper.

Reviewer #2 (Recommendations for the authors):

(1) The description of the N2.261 staining (Figure 1 and Supplementary Figure 1) is entirely irrelevant for the rest of the manuscript. One wonders why the authors have not used this antibody to characterize the presence or absence of "dedifferentiated" muscle after injury. Given that they make a point later about potential differences in dedifferentiation in livebearers, this is a missed opportunity to address it using tools the authors have characterized extensively in zebrafish.

This is a valuable suggestion that we have now addressed. We performed N2.261 immunostaining on injured platyfish and swordtail hearts at 7 dpci, and included the results as an additional supplementary figure (Suppl. Fig. S4C). Importantly, N2.261 immunoreactivity was not detected in the peri-injury zone of the myocardium, suggesting that, unlike in regenerating zebrafish hearts, the embryonic cardiac myosin heavy chain isoform is not upregulated at the injury site in platyfish. This finding is particularly informative given the broad atrial N2.261 reactivity observed in intact platyfish hearts: the absence of enhanced staining in the injury zone argues against a dedifferentiation-associated upregulation of this isoform and instead suggests that platyfish cardiomyocytes may not undergo the same embryonic gene re-expression program that characterizes zebrafish cardiac regeneration. This result therefore directly informs our later discussion of potential differences in dedifferentiation between zebrafish and livebearers, and we have updated the relevant section of the manuscript accordingly.

(2) Many of the markers highlighted here as part of the differential gene expression analysis are not the most canonical ones, and it is unclear how the authors selected them. For example, in Figure 5, anxa2a, hlx1, and nup153 are presented as macrophage markers. However, anxa2a appears to be expressed predominantly in the endocardium in response to injury (see PMID: 32341028), and other markers (mpeg, mfap4, etc) would have been more consistent as macrophage markers according to other literature. The same is true for cardiomyocyte proliferation markers and dedifferentiation markers in Figure 6. In this last case, N2.261 could have been used as reported by the authors before. L-plastin is a pan-leukocyte marker, not a macrophage marker. This should be corrected (Figure 5 and lines 258-279).

We agree with this critique that several of the markers used in the original analysis were neither sufficiently canonical nor specific for the cell populations we intended to characterize. We apologize for this oversight. In response, we have moved away from scRNA-seq-derived marker sets and now rely entirely on manually curated markers drawn from the established literature, as detailed below.

(1) Cardiomyocyte cell fate change (new Figure 7, Supplementary Figure S8): We now report the transcript abundance of genes encoding proteins with well-documented roles in the transcriptional reprogramming associated with cardiomyocyte dedifferentiation and redifferentiation. These include members of the Activator Protein-1 (AP-1) complex, the SWI/SNF chromatin remodeling complex, and the transcriptional coactivator cited4a, alongside established markers of cardiomyocyte proliferation (cx43) and sarcomere reassembly (the Rbfox family). We also include myh7 transcript abundance, which in zebrafish is upregulated in dedifferentiated cardiomyocytes.

(2) Immune response (new Figure 6, Supplementary Figure S7): New Figure 6 focuses on pan-leukocyte markers, while Supplementary Figure S7 presents genes involved in innate myeloid activation and inflammation, including components of the NF-κB/TNF-α pathway, TLR signaling, and inflammatory regulation, as well as specific markers of neutrophils (mpx) and macrophages (mpeg1, mfap4), in line with the markers recommended by the reviewer.

Regarding L-plastin, we thank the reviewer for this correction. The text has been amended throughout to describe L-plastin accurately as a pan-leukocyte marker rather than a macrophage/phagocyte marker.

(3) In several instances, the authors reference papers without citing the original source. In many other instances, there are some oversights regarding citations. I would advise revising many of these:

We thank the reviewer for drawing our attention to these citation oversights. We have carefully revised the reference list throughout the manuscript to ensure that original discovery papers are cited alongside, or in place of, review articles wherever appropriate.

(4) Lines 49-51 - several very interesting reviews of cardiac regeneration are listed here, but referencing the source of the discovery is always most rigorous.

For lines 49–51, we have replaced the previous review citations with three original research papers reporting the discovery of cardiac regeneration in zebrafish and three reporting it in axolotl. We have applied the same principle of prioritizing primary sources across all other instances flagged by the reviewer.

(5) Line 76. When discussing the recovery of the muscle after cryoinjury, Poss et al. 2002 shouldn't be referenced. Sánchez-Iranzo et al 2018 should be replaced by González-Rosa et al. 2011, which is the contemporary manuscript to those of Chablais 2011 and Schnabel 2011.

Done

(6) Line 284 - when discussing cardiomyocyte dedifferentiation, it would be fair to reference also Kikuchi et al 2010.

Done

eLife Assessment

Davies et al. present a valuable study proposing that the Drosophila homologue of MACF1, Shot, can act as a molecular linker between microtubules and actin during dendrite pruning, suggesting an intriguing role in non-centrosomal microtubule organization. The study provides robust evidence that Shot helps control microtubule organization and dendrite pruning, but the evidence that Shot coordinates these processes through its interaction with actin remains incomplete, particularly because the actin-disrupting experiment does not affect the distal dendrites where the authors propose Shot acts. Overall, the study contains several interesting findings, but additional experimental support is needed to establish how Shot, actin, and microtubules work together to control dendrite pruning.

Reviewer #1 (Public review):

The neuronal microtubule cytoskeleton is essential for long-range transport in axons and dendrites. Axon-specific plus-end-out microtubule organization versus dendrite-specific plus-end-in organization (in invertebrates) enables selective transport into each neurite, establishing neuronal polarity. Dendritic microtubule organization is also thought to be important for dendrite pruning in Drosophila during metamorphosis. However, the precise mechanisms that organize microtubules in neurons remain incompletely understood.

In this manuscript, the authors show that the spectraplakin Shot is important for developmental dendrite pruning. Shot mutants display dendritic microtubule polarity defects, which-based on the authors' rescue experiments and prior work-likely underlie the pruning defects. Since Shot is a known actin-microtubule crosslinker, the authors also examine the role of actin and find that it, too, is required for dendrite pruning.

Notably, Shot localizes transiently to dendrite tips during the 1st instar larval stage, a time when microtubule defects are already detectable. Given that Shot interacts with both plus- and minus-end microtubule-binding proteins, and that overexpression of Shot's microtubule-binding domain alone is sufficient to rescue the mutant phenotype, the authors propose that Shot's principal function is to stabilize microtubules at dendrite tips during the 1st instar stage (and possibly along the entire dendritic shaft at later stages).

In parallel, the authors investigate microtubule nucleation and find that, specifically at the 1st instar stage, microtubules can nucleate from RAB11-positive vesicles carrying γ-tubulin. Interestingly, this nucleation occurs only toward the cell body, suggesting additional layers of control. The authors do not observe nucleation from the dendrite tip, as previously reported in C. elegans. While this discrepancy could be technical, it more likely reflects a genuine mechanistic difference between the two systems.

Strengths:

Overall, this work was technically well performed, using advanced genetics and imaging. The authors convincingly show that Shot plays an important role in microtubule stabilization, and that its interaction with actin is likely important for localizing the protein.

The transient localization of Shot to dendrite tips and microtubule nucleation from RAB11 endosomes during the 1st instar stage is intriguing. This suggests that specific mechanisms may act during early neuronal development to establish microtubule organization, while different mechanisms maintain this organization later in development.

Weaknesses:

The link between nucleation (rab11/gamma-tubulin) and stabilization (Shot/Actin/EB/Patronin) at this stage is unclear. The authors are correct to discuss that we may be missing early events, e.g. maybe we miss early microtubule nucleation from the tip. Or alternatively could MTs frow with their minus-ends into the dendrite (see Feng et al)?

Distal enrichment of shot at 1st instar is a critical piece of data (it appears in the title). However, this was seen upon overexpression of GFP::Shot. However, does this correctly reflect the endogenous situation? Also, because its binding partner, EB1, can only be seen enriched to dendrite tips upon shot overexpression but not in controls.

Reviewer #2 (Public review):

Summary:

The authors' data reveal that the spectraplakin Shot, which can interact with both actin and microtubules, is essential for the proper pruning of dendrites in a Drosophila model. A molecular basis for the coordination of these two cytoskeletons during neuronal developmental has been elusive. The authors' data support a role for Shot in regulating microtubule polarity in young neurons, and the authors link this early function to a later role in dendrite pruning. The story has many interesting components, but lacks experimental depth to solidly (and sufficiently) support the authors' claims.

Strengths:

(1) A strength of the manuscript is the authors' data supporting the idea that Shot is needed for proper dendritic microtubule polarity in young neurons and that Shot regulates dendrite pruning.

(2) Another strength of the manuscript is the data in support of Rab11 functioning as a MTOC in young larvae but not older larvae; this is an important finding that may resolve some debates in the literature. The finding that Rab11 and Msps coimmunoprecipitate is nice evidence in support of the idea that Rab11(+) endosomes serve as MTOCs.

Weaknesses:

(1) A major concern is that many of the authors' main conclusions are weakly supported. The idea that Shot exerts an effect on microtubules via an interaction with actin relies predominantly on the findings that deleting Shot's actin-interacting domains (CH domains) does not rescue the dendritic microtubule polarity defect and also alters the dendrite-tip enrichment of Shot (the change in dendrite-tip enrichment, however, appears fairly modest as Shot appears to still be localized quite well near the tip of the dendrite, if not at the actual tip). The over-expression of Mical, which was intended the disrupt actin, does not affect actin in distal dendrites in 1st instar neurons; 1st instar distal dendrite tips is where (and when) the authors propose Shot functions. It does not seem possible to draw conclusions about the actin cytoskeleton from the overexpression of Mical since this manipulation only (mildly) affects actin in the proximal dendrites of 1st instar neurons.

Given the reliance on the Shot transgenes, it is also important to test, or at least comment on, whether the Shot transgenes are expressed at equivalent levels (the transgenes appear to be integrated at different positions in the genome). A difference in transgene expression should be ruled out as a potential contributing factor.

(2) The authors propose that Shot acts by (locally) stabilizing microtubules. This conclusion is not supported by the data. For example, the fluorescently tagged tubulin images show signal that is quite diffuse, and the images are not convincing (fluorescent tubulin reports both the free tubulin pool as well as microtubules). A high-resolution image in the supplementary materials is also not convincing. These images are of primary dendrites, but the authors' model proposes a local role for Shot in dendrite tips.

(3) A general weakness is the use of data derived from different parts of the dendritic arbor, and at varying developmental stages, to generate a model that proposes a spatially and temporally specific role for Shot.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

The Neuronal microtubule cytoskeleton is essential for long-range transport in axons and dendrites. The axon-specific plus-end out microtubule organization vs the dendritic-specific plus-end in organization allows for selective transport into each neurite, setting up neuronal polarity. In addition, the dendritic microtubule organization is thought to be important for dendritic pruning in Drosophila during metamorphosis. However, the precise mechanisms that organize microtubules in neurons are still incompletely understood.

In the current manuscript, the authors describe the spectraplakin protein Shot as important in developmental dendritic pruning. They find that Shot has dendritic microtubule polarity defects, which, based on their rescues and previous work, is likely the reason for the pruning defect.

Since Shot is a known actin-microtubule crosslinker, they also investigate the putative role of actin and find that actin is also important for dendritic pruning. Finally, they find that several factors that have been shown to function as a dendritic MTOC in C. elegans also show a defect in Drosophila upon depletion.

Strengths:

Overall, this work was technically well-performed, using advanced genetics and imaging. The author reports some interesting findings identifying new players for dendritic microtubule organization and pruning.

We thank reviewer 1 for this assessment.

Weaknesses:

(1) The evidence for Shot interacting with actin for its functioning is contradictory. The Shot lacking the actin interaction domain did not rescue the mutant; however, it also has a strong toxic effect upon overexpression in wildtype (Figure S3), so a potential rescue may be masked. Moreover, the C-terminus-only construct, which carries the GAS2-like domain, was sufficient to rescue the pruning. This actually suggests that MT bundling/stabilization is the main function of Shot (and no actin binding is needed).

Thank you for this comment. We agree that the rescue and overexpression experiments with UAS-Shot ΔCH1 (lacking only the first actin-binding domain) and UAS-Shot CTerm (containing only the MT binding domain) allow for other interpretations than ours that both features are necessary for Shot function in dendrites. To address this issue, we tested two more Shot domain deletion mutants, Shot ΔABD (lacking both actin-binding domains) and Shot ΔCTail, which lacks the MT binding domain (Figure 3). Neither of the latter two constructs can rescue the pruning and MT orientation phenotypes of shot mutants, confirming that actin binding is required for these functions in the context of full-length Shot. Importantly, in contrast to the previously used Shot ΔCH1, neither Shot ΔABD nor Shot ΔCTail inhibit pruning upon overexpression, ruling out other interpretations of overexpression toxicity (Fig. S2). The observation that Shot CTerm (containing only the MT binding domain) is sufficient to rescue indicates that actin binding is only required for Shot function in the context of full-length Shot. Interestingly, this is reminiscent of a proposed autoregulation mechanism where actin binding disinhibits Shot's MT binding domain (Applewhite 2013).

Figure 3, Figure S2 (Shot domain analysis): We added two new constructs (Shot ΔABD, Shot ΔCtail) and a Shot BAC as additional control in Figure 3.

(2) On the other hand, actin depolymerization leads to some microtubule defects and subtle changes in shot localization in young neurons (not old ones). More importantly, it did not enhance the microtubule or pruning defects of the Shot domain, suggesting these act in the same pathway. Interesting to note is that Mical expression led to microtubule defects but not to pruning defects. This argues that MT organization effects alone are not enough to cause pruning defects. This may be good to discuss.

Thank you for this comment. We think that microtubule orientation defects are tightly linked to dendrite pruning defects, as all known conditions causing orientation defects either cause pruning defects, or they act as genetic enhancers of pruning mutants. We now clarify this in the introduction citing the example of EB1 RNAi (causes orientation defects (Matties et al., 2010), enhances pruning mutants (Herzmann et al., 2018)). This is also the case for Mical overexpression, which causes orientation defects and enhances the pruning defects caused by Shot RNAi. To make this point clearer, we now also show representative images of pruning defects caused by either Shot RNAi alone or Shot RNAi combined with Mical overexpression.

Introduction - we explain better the genetic relationship between microtubule orientation and pruning. Figure 4 (phenotypic enhancement of shot knockdown by Mical overexpression): added representative images.

(3) For the actin depolymerization, the authors used overexpression of the actin-oxidizing Mical protein. However, Mical may have another target. It would be good to validate key findings with better characterized actin targeting tools.

Thank you for this comment. To our knowledge, there are no good tools to globally manipulate actin dynamics in vivo in a similar manner to, e. g., latrunculin in cultured cells. According to the literature, Micals are highly specific for actin. To enhance our analysis, we assessed the effect of Mical overexpression on actin distribution in c4da neurons (as assessed by LifeAct-GFP) and found significant effects at the first instar.

Added new analysis of effect of Mical overexpression on Lifeact::GFP distribution in c4da neurons (new Fig. S3), added references for Mical targets.

(4) In analogy to C. elegans, where RAB-11 functions as a ncMTOC to set up microtubules in dendrites, the authors investigated the role of these in Drosophila. Interestingly, they find that rab-11 also colocalizes to gamma tubulin and its depletion leads to some microtubule defects. Furthermore, they find a genetic interaction between these components and Shot; however, this does not prove that these components act together (if at all, it would be the opposite). This should be made more clear. What would be needed to connect these is to address RAB-11 localization + gamma-tubulin upon shot depletion.

All components studied in this manuscript lead to a partial reversal of microtubules in the dendrite. However, it is not clear from how the data is represented if the microtubule defect is subtle in all animals or whether it is partially penetrant stronger effect (a few animals/neurons have a strong phenotype). This is relevant as this may suggest that other mechanisms are also required for this organization, and it would make it markedly different from C. elegans. This should be discussed and potentially represented differently.

Thank you for this comment. We agree that the genetic interaction with pruning as readout does not prove that Rab11 and Shot act in the exact same pathway during establishment of microtubule organization. To address this question, we live-imaged Rab11::GFP vesicles in control and Shot knockdown neurons and found no significant difference in motile behavior (new Figure 7). As Rab11::GFP also is not visibly enriched at dendrite tips at the first instar stage, we cannot confidently state that a Rab11 tip MTOC exists in Drosophila neurons. Still, we only see EB1 comets originating from Rab11 puncta at the first instar, but not the third. We therefore rephrased and toned down our interpretation and state now that Rab11 may be a component of a developmentally transient MTOC but likely acts independently of Shot.

Unfortunately, we do not see many comets per neuron with our fluorescently tagged EB1 constructs, indicating that only a fraction of them are labeled. This makes it very difficult to judge severity between single neurons. The impression is that usually only some comets per neuron have reversed orientation, which would be consistent with several independent mechanisms for orientation.

Reviewer #2 (Public review):

Summary:

In their manuscript, the authors reveal that the spectraplakin Shot, which can bind both microtubules and actin, is essential for the proper pruning of dendrites in a developing Drosophila model. A molecular basis for the coordination of these two cytoskeletons during neuronal development has been elusive, and the authors' data point to the role of Shot in regulating microtubule polarity and growth through one of its actin-binding domains. The authors also propose an intriguing new activity for a spectraplakin: functioning as part of a microtubule-organizing center (MTOC).

Strengths:

(1) A strength of the manuscript is the authors' data supporting the idea that Shot regulates dendrite pruning via its actin-binding CH1 domain and that this domain is also implicated in Shot's ability to regulate microtubule polarity and growth (although see comments below); these data are consistent with the authors' model that Shot acts through both the actin and microtubule cytoskeletons to regulate neuronal development.

(2) Another strength of the manuscript is the data in support of Rab11 functioning as an MTOC in young larvae but not older larvae; this is an important finding that may resolve some debates in the literature. The finding that Rab11 and Msps coimmunoprecipitate is nice evidence in support of the idea that Rab11(+) endosomes serve as MTOCs.

Thank you for these comments.

Weaknesses:

(1) A significant, major concern is that most of the authors' main conclusions are not (well) supported, in particular, the model that Shot functions as part of an MTOC. The story has many interesting components, but lacks the experimental depth to support the authors' claims.

Thank you for this comment. We agree that we cannot conclusively state that Shot is part of a classical MTOC where microtubules are nucleated and therefore tone down our conclusions regarding this.

(2) One of the authors' central claims is that Shot functions as part of a non-centrosomal MTOC, presumably a MTOC anchored on Rab11(+) endosomes. For example, in the Introduction, last paragraph, the authors summarize their model: "Shot localizes to dendrite tips in an actin-dependent manner where it recruits factors cooperating with an early-acting, Rab11-dependent MTOC." This statement is not supported. The authors do not show any data that Shot localizes with Rab11 or that Rab11 localization or its MTOC activity is affected by the loss of Shot (or otherwise manipulating Shot). A genetic interaction between Shot and Rab11 is not sufficient to support this claim, which relies on the proteins functioning together at a certain place and time. On a related note, the claim that Shot localization to dendrite tips is actin-dependent is not well supported: the authors show that the CH1 domain is needed to enrich Shot at dendrite tips, but they do not directly manipulate actin (it would be helpful if the authors showed the overexpression of Mical disrupted actin, as they predict).

Thank you for these comments. In response, we tested whether Shot knockdown affects Rab11 vesicle motility in first instar dendrites and find that this is likely not the case (new Fig. 7 K, L). We therefore remove the proposal that Shot directly regulates Rab11. Regarding the actin dependence of Shot tip localization, we based this conclusion on the observations that Shot ΔCH1 does not localize to tips, and that Mical overexpression significantly broadens the Shot tip signal (Fig. 5 E, F, J - previously Fig. 5 C, D, G). To further deepen this conclusion, we now add a localization analysis of a Shot mutant lacking both actin-binding CH domains (Shot ΔABD) and find that this also abrogates tip enrichment.

(3) The authors show an image that Shot colocalizes with the EB1-mScarlet3 comet initiation sites and use this representative image to generate a model that Shot functions as part of an MTOC. However, this conclusion needs additional support: the authors should quantify the frequency of EB1 comets that originate from Shot-GFP aggregates, report the orientation of EB1 comets that originate from Shot-GFP aggregates (e.g., do the Shot-GFP aggregates correlate with anterogradely or retrogradely moving EB1 comets), and characterize the developmental timing of these events. The genetic interaction tests revealing ability of shot dsRNA to enhance the loss of microtubule-interacting proteins (Msps, Patronin, EB1) and Rab11 are consistent with the idea that Shot regulates microtubules, but it does not provide any spatial information on where Shot is interacting with these proteins, which is critical to the model that Shot is acting as part of a dendritic MTOC.

Thank you for these comments. Due in part to low overall EB1 comet density in c4da neurons, we could rarely see comets originating from tip-localized Shot::GFP at the first instar stage - one example is shown in new Fig. S5C (previously Fig. 7D). At later stages, Shot is relatively evenly localized along the dendritic plasma membrane. As we cannot be certain about the relatively vague MTOC conclusions, we decided to rephrase our conclusions and state - based on genetic and functional data - that Shot's ability to stabilize microtubules is required, likely at dendritic tips.

(4) It is unclear whether the authors are proposing that dendrite pruning defects are due to an early function of Shot in regulating microtubule polarity in young neurons (during 1st instar larval stages) or whether Shot is acting in another way to affect dendrite pruning. It would be helpful for the authors to present and discuss a specific model regarding Shot's regulation of dendrite pruning in the Discussion.

Thank you for these comments. It is indeed our hypothesis that Shot's early role in setting up microtubule organization is also crucial for pruning, this is the most parsimonious explanation as pruning and orientation defects always go hand in hand. We added the following sentence in the Discussion: "As we do not have evidence for Shot regulation at the onset of the pupal stage (Fig. S4), the function of Shot crucial for pruning is most arguably its early role in setting up uniform microtubule orientation."

(5) The authors argue that a change in microtubule polarity contributes to dendrite pruning defects. For example, in the Introduction, last paragraph, the authors state: "Loss of Shot causes pruning defects caused by mixed orientation of dendritic microtubules." The authors show a correlative relationship, not a causal one. In Figure 4, C and E, the authors show that overexpression of Mical disrupts microtubule polarity but not dendrite pruning, raising the question of whether disrupting microtubule polarity is sufficient to cause dendrite pruning defects. The lack of an association between a disruption in microtubule polarity and dendrite pruning in neurons overexpressing Mical is an important finding.

Thank you for this comment. Microtubule orientation defects are tightly linked to dendrite pruning defects, as all known conditions causing orientation defects either cause pruning defects, or they act as genetic enhancers of pruning mutants. We now clarify this in the introduction citing the example of EB1 RNAi (which causes orientation defects (Matties et al., 2010), and does not cause pruning defects by itself, but enhances the effects of other pruning mutants (Herzmann et al., 2018)). This is also the case for Mical overexpression, which causes orientation defects and enhances the pruning defects caused by Shot RNAi. To make this point clearer, we now also show representative images of pruning defects caused by either Shot RNAi alone or Shot RNAi combined with Mical overexpression.

Introduction - we explain better the genetic relationship between microtubule orientation and pruning. Figure 4 (phenotypic enhancement of shot knockdown by Mical overexpression): added representative images

(6) The authors show that a truncated Shot construct with the microtubule-binding domain, but no actin-binding domain (Shot-C-term), can rescue dendrite pruning defects and Khc-lacZ localization, whereas the longer Shot construct that lacks just one actin-binding domain ("delta-CH1") cannot. Have the authors confirmed that both proteins are expressed at equivalent levels? Based on these results and their finding that over-expression of Shot-delta-CH1 disrupts dendrite pruning, it seems possible that Shot-delta-CH1 may function as a dominant-negative rather than a loss-of-function. Regardless, the authors should develop a model that takes into account their findings that Shot, without any actin-binding domains and only a microtubule-binding domain, shows robust rescue.

Thank you for this constructive criticism. We agree that the rescue and overexpression experiments with UAS-Shot ΔCH1 (lacking only the first actin-binding domain) and UAS-Shot CTerm (containing only the MT binding domain) allow for other interpretations than ours that both features are necessary for Shot function in dendrites. To address this issue, we tested two more Shot domain deletion mutants, Shot ΔABD, lacking both actin-binding domains, and Shot ΔCTail, which lacks the MT binding domain (Figure 3). Neither of these two constructs can rescue the pruning and MT orientation phenotypes of shot mutants, confirming that both actin and microtubule binding are required for these functions in the context of full-length Shot. Importantly, in contrast to the previously tested Shot ΔCH1, neither Shot ΔABD nor Shot ΔCTail inhibit pruning upon overexpression, ruling out other interpretations of overexpression toxicity. The observation that Shot CTerm is sufficient to rescue indicates that actin binding is only required for Shot function in the context of full-length Shot. Interestingly, this is reminiscent of a proposed autoregulation mechanism where actin binding disinhibits Shot's MT binding domain (Applewhite 2013).

Figure 3, Figure S2 (Shot domain analysis): We added two new constructs (Shot ΔABD, Shot ΔCTail) and a Shot BAC as additional control in Figure 3.

(7) The authors state that: "The fact that Shot variants lacking the CH1 domain cannot rescue the pruning defects of shot[3] mutants suggested that dendrite tip localization of Shot was important for its function." (pages 10-11). This statement is not accurate: the Shot C-term construct, which lacks the CH1 domain (as well as other domains), is able to rescue dendrite pruning defects.

Thank you for this comment. Shot ΔCH1 indeed does show a partial rescue of the pruning defects (but not of the polarity defects). To clarify whether actin binding is required for Shot function in dendrites, we tested the additional Shot construct Shot ΔABD and found that this construct could not rescue the defects at all (Fig. 3).

Fig. 3, added Shot ΔABD.

(8) The authors state that: "In further support of non-functionality, overexpression of Shot[deltaCH1] caused strong pruning defects (Fig. S3)." (page 8). Presumably, these results indicate that Shot-delta-CH1 is functioning as a dominant-negative since a loss-of-function protein would have no effect. The authors should revise how they interpret these results. This comment is related to another comment about the ability of Shot constructs to rescue the shot [3] mutant.

Thank you for this comment. We agree that the rescue and overexpression experiments with UAS-Shot ΔCH1 (lacking only the first actin-binding domain) allow for other interpretations than ours - that actin binding is necessary for Shot function in dendrites. To address this issue, we tested another Shot domain deletion mutant, Shot ΔABD (lacking both actin-binding domains) (Figure 3). This construct did not rescue the pruning and MT orientation phenotypes of shot mutants, and also does not cause pruning defects upon overexpression, confirming that actin binding is required in the context of full-length Shot. As our domain analysis is reminiscent of a proposed autoregulation mechanism where actin binding disinhibits Shot's MT binding domain (Applewhite 2013), it is interesting to speculate that the overexpression toxicity of Shot ΔCH1 may stem from a residual ability of the second CH domain to relieve autoinhibition.

Fig. 3, Fig. S2, added Shot ΔABD and Shot ΔCTail to domain analyses.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) Not enough info was given about how the statistics were performed in e.g., Figures 5E, 5H and 7C. Many data are highly significant, but this is not always directly clear to me from the images, e.g., 7C 6-10um. Or comparing 7I to 7L, how come the significance is so much different

Thank you for this comment. For quantification of the localization data, we pooled the data points from individual intensity profiles for each dendrite segment and compared them statistically. Since this generates a relatively high number of data points, samples are quite likely to be significantly different. Acknowledging this, we now only compare the most distal segments (0-2 μm, 2-4 μm). We describe our approach now in more detail in the methods section.

Better description of fluorescence quantifications in Figs. 5, 7 in Methods section.

(2) Abstract "while plus-en out MT just grow from the soma" - I think this is probably only partially correct. There are various players important for organizing/nucleating these microtubules (e.g., Augmin for local MT nucleation).

Thank you for this comment. This was oversimplified and we rephrased the abstract.

(3) "We and others have previously shown that c4da neuron dendrite pruning depends on local microtubule disassembly in proximal dendrites (Herzmann et al, 2017)." Nice to add other references.

Thank you for this comment. We added additional references.

(4) Figure 2 EB imaging. Where was this performed along the dendrite? Close to the CB, this may reflect CB MT growth entering the dendrite.

Thank you for this comment. We usually perform EB1 imaging in primary or secondary dendrites -- now mentioned in the text (p. 7). The idea that anterograde comets in this region could reflect microtubules entering from the soma is interesting, especially as the fast anterograde comets upon combined shot knockdown and Mical overexpression behave so differently. We take up this idea in the Discussion - thanks again!

(5) Figure 3H-L I cannot see the magenta in the overlays. Separate channels seem to be needed. And where is the axon? Moreover, in the text it is stated that the I 3H fusion protein is exclusively in the axon. And control is not in 3M.

Thank you for this comment. Our lacZ antibodies are relatively weak, so the axons are not always visible. We therefore changed the text to only mention the soma now. We also now include all shown genotypes in the quantification.

(6) Figure 4A I'm not sure what is seen in the image. Is this a single dendrite? And if so, why is Shot only seen on one side?

Thank you for this comment. In this panel (now moved to Fig. 5A) we had to take a fairly thin projection of a confocal stack because of strong surrounding tissue signal. Because of this, the Shot::GFP signal on one edge of the dendrite is not as well visible. We try to improve this by indicating the edges of the dendrite in the panel.

Fig. 5A - D, indicated dendrite outline in confocal/SIM images.

(7) Figure S2C description is not the same as in the graph (neurons / dendrites).

Thank you for pointing this out. We corrected the legend (now Fig. 2C).

(8) Figure S4 - is this a single dendrite or a bundle?

The images (now Fig. S5) correspond to the microtubule signal from single primary dendrites, i. e., mostly microtubule bundles. We clarify now in the legend.

Fig. S5, specified nature of samples in legend.

Reviewer #2 (Recommendations for the authors):

(1) In the Discussion, page 13, the authors state that: "On the one hand, we provide evidence that Shot anchors microtubules via actin in mature neurons." This is not supported by the authors' evidence; indeed, the authors themselves state on page 10 that "... the CH1 domain is unlikely to be required for Shot cortical localization in dendritic shafts." If the authors mean to say that the cortical localization of Shot is restricted to dendrite tips, then superresolution microscopy should be done on dendrite tips.

Thank you for pointing this out. As the CH1 domain is clearly not required for Shot third instar localization, we agree with this notion and rephrase our interpretation accordingly: "Our data also do not rule out that Shot could anchor dendritic microtubules in mature third instar neurons, likely in cooperation with actin." (p. 17)

(2) The authors characterize the localization of Shot during larval stages but not during pupal stages. The authors should characterize where Shot localizes during dendrite pruning, since it is possible that Shot may localize differently at this stage. For example, the authors mention that Shot localizes to dendrite tips during early larval stages but not late larval stages; it seems that the localization of Shot is dynamic.

Thank you for this comment. In response, we analyzed Shot localization in a time-course experiment including the second instar larval stage and at 5 h APF in the early pupal stage (new Figure S4). This showed that Shot localizes almost exclusively to dendrite tips at the first instar, but is more broadly distributed in the soma and along dendrites shafts at all later developmental stages.

New Fig. S4, time course of Shot::GFP localization including first to third instars and early pupal stage.

(3) Figure 3: Khc-lacZ is not an ideal read-out of microtubule polarity per se. Better, read-out microtubule polarity using EB1-GFP.

Thank you for this comment. Unfortunately, this experiment is not feasible in this situation. In our MARCM system, we use a red fluorescent tdTomato to label c4da neurons, and all Shot transgenes carry GFP tags, prohibiting the use of both fluorescent EB1 transgenes that we have, EB1::GFP and EB1::mScarlet3.

(4) Figure 4 A-A' (also Figure S4): Superresolution microscopy data are presented but not analyzed. Analysis should be included before drawing a conclusion from these data.

Thank you for this comment. In response, we quantified the width of the dendritic microtubule bundles in the STED experiments (now Fig. S5) and found that these bundles are significantly thinner upon shot knockdown.

New Fig. S5 B, quantification of STED data.

(5) Figure 5, A and B: These images are extremely difficult to interpret on their own. It is not clear what the authors are interpreting as a positive signal. Analysis and quantification of Shot-FL and Shot-deltaCH1 should be included, not just representative images. This seems like a missed opportunity.

Thank you for this comment. In response, we show higher magnification images of Shot(endo)GFP, UAS-Shot::GFP, and the two actin-binding mutants (UAS-Shot ΔCH1::GFP and UAS-Shot ΔABD::GFP. For better interpretation, we indicate the boundaries of the dendrites as determined by tdTomato expression in the neurons (new Fig. 5 A - D).

(6) Figure 6: If gamma-tubulin is knocked down, does this decrease the number of EB1-GFP comets that originate from Rab11(+)-endosomes?

Thank you for this comment. This experiment is technically challenging because of the number and location of UAS transgenes involved. In addition, gamma-tubulin knockdown only has minor effects on pruning (no defects) or microtubule behavior (no orientation defects) in c4da neurons (Wang et al., 2019, eLife (Fengwei Yu lab).

(7) Figure 6: Rab11(+) endosomes correlate with the initiation of EB1-GFP comets in first instar larvae, but the orientation of microtubules is analyzed in third instar larvae -is there an effect on microtubule polarity in younger larvae? What about second instar larvae? On a related note, what is the orientation of microtubules that originate from Rab11(+) endosomes? These should be quantified data.

Thank you for this comment. We analyzed the effect of Rab11 knockdown at the first instar stage and found that this led to a non-significant increase in anterograde comets. These data are now included as new Fig. S6. We also checked the orientation and found that all comets from Rab11-cherry puncta were retrograde, this is now mentioned in the text.