Reviewer #1 (Public review):

Summary:

Despite setting global conservation goals for the conservation of genetic diversity, gathering enough genetic data to assess its current status or monitor change over time for all species would require substantial time and resources. Finding reliable proxies or predictors of genetic diversity change would be a valuable alternative. Selmoni and Schuman test for patterns and trends in coral reef animals' genetic diversity across space and time and evaluate how well environmental variables predict their genetic diversity.

Strengths:

The authors have compiled a large genomic dataset from 19 studies, 18 species, and 173 reefs, and use a standardized k-mer-based pipeline to process all datasets. The authors propose that using k-mers may be especially suitable for macrogenomic analyses because they are less computationally intense and do not require a reference genome. This is interesting because there are currently few examples of macrogenetics studies using genomic data, likely due in part to the difficulty associated with processing multiple disparate datasets in an efficient and standardized way.

Weaknesses:

I have concerns that the models don't fit the data structure, the data are not suitable to test for temporal change, and that the values being predicted do not reflect meaningful genetic diversity. The manuscript would also benefit from a more cohesive narrative structure, clearly stated research goals, and better engagement with previous literature on this topic.

Effects of time: The data are not suitable to test for temporal changes across all species. 11 of 19 datasets were only sampled over 1-2 years, and only 5 were sampled across 5 or more years. I don't know generation times for these species offhand, but this is generally too short to draw any conclusions about genetic diversity change over time. While it makes sense to include year in the models as a control variable, in my opinion these patterns should not be interpreted and removed from the discussion, or at least the conclusions should be tempered (for example the section "Are coral reef populations rapidly losing their genetic diversity?").

Environmental effects models: I'm not convinced that the geographic random intercept term (response variable) represents a local genetic distance that should be related to environments. How many species were sampled at each reef? If only 1 species is sampled, then the reef's average diversity reflects that species, not necessarily environmental effects. If multiple species are sampled at a reef, then I suspect this value is the average diversity of all the species sampled there. Knowing that species have different levels of diversity (e.g., Toczydlowski et al 2025 cited here), what does this metric mean and how is it affected by species composition vs environments? I do not see the value in predicting this variable without considering species identity.

Introduction: The macrogenetics literature is not well described in the introduction. Most of the cited studies indeed use raw data rather than summary statistics, and most do not aim to predict genetic diversity.

I wouldn't say that explaining more than 20% of variation in data is a major challenge in macrogenetics. The choice of how many, and which, predictors to use depends on the research question (e.g., on the spectrum of understanding to predicting, see Shmueli 2010 "To Explain or To Predict?"; Arif & MacNeil 2022 "Predictive models aren't for causal inference"). It's proposed here that explained variance can be increased by adding more, or more relevant, environmental predictors-but particularly for macrogenetics studies reusing data from different sources with different study and sampling designs and different species, much of the variation in the data will be due to these factors. That is why conditional R2s are typically quite high (>80%) in mixed models with species/study as random effects, see e.g. Clark & Pinsky 2024 or Karachaliou 2025. Having strong effects of particular variables would also require that all species respond to the predictor in the same way, which is not necessarily expected. Low explanatory capacity alone is not an issue if the goal is not prediction.

Reviewer #2 (Public review):

Summary:

The authors measured whether a phylogenetically wide sample of marine species was experiencing declining genetic diversity, as one might expect from widespread habitat threat, over a recent 2-decade time span. They next identified key environmental variables that are impacting genetic diversity within and between reefs. A key insight was to apply k-mer-based genetic distances to massively speed up the reanalysis of genetic data into a common pipeline, which is otherwise onerous. The manuscript ends by highlighting key seascape variables that were associated with increases or decreases with genetic diversity through time. The authors achieved their overall aims, though I remain unsure of how well the identified seascape variable-genetic diversity predictions can be generalized, as implied in the abstract.

Strengths:

A key strength of the study was its rigor in vetting the k-mer based distance metrics, checking whether they give population structure patterns (Figure S2) and correlated with nucleotide diversity. This surpasses previous studies that used a similar method. The modeling procedure of genetic diversity predictions from environmental variables is also rigorous and presented with some appropriate nuance.

Weaknesses:

I noted five weaknesses, listed below in order of potentially more severe at the top to more minor at the bottom.

First, only one k-mer distance metric was tested. There are many k-mer distance metrics that will potentially give different weight to different frequencies of polymorphism, like how Watterson's theta and nucleotide diversity weight polymorphisms, depending on their frequency. It would be interesting to see if the seascape variable predictions hold with a Jaccard or cosine distance metric, or if the observed results are purely restricted to the choice of Bray-Curtis distance. Another option would be to use mash, skmer, or (very recent development) re-skmer distances, which attempt to more directly approximate the average nucleotide identity between two sequence sets based on the k-mer sets of their reads while also being faster than traditional alignment. This would potentially give cleaner trends, seeing as the goal is to have a proxy for nucleotide diversity, with the downside of not including the impacts of non-SNP variation.

Second, it is unclear to me whether the number of species analyzed, 18, can accurately identify important environmental variables in early warning systems, as claimed in the abstract. While this likely represents the best available balance of evidence, it is worth highlighting the manuscript's note that the environment-diversity predictions did not scale across marine realms. This is likely a limitation of data availability, rather than a study design flaw, but is nonetheless important for readers to keep in mind. Would recommend that the abstract acknowledge this limitation.

Third, it is unclear to me how the included datasets compare in terms of genome-wide coverage. Figure S1 gives sequencing depths in terms of read number, but what is the range normalized for genome size - are the datasets 1x, 5x, 10x, on average, etc? This is probably most key to how the k-mer distance metrics will perform, because low coverage will make two samples appear artificially distant due to rarity of sampling the same k-mer multiple times. However, the correlation between k-mer distance and regular alignment-based SNP distance (Figure 2C) gives some confidence that this effect could be small.

Fourth, though k-mer distances may in some sense better capture the breadth of DNA sequence diversity, they lack a concrete interpretation of what loci may/may not be under selection as marine environments are increasingly threatened over time. This is a different question than what the manuscript tries to address, but is of interest to the field and is something that would be seemingly difficult to do with k-mers.

Finally, I had a more minor concern: the k-mer-based distances use only k-mers that are mapped to a reference genome. This is done to thoroughly remove contaminant sequences, which is important, but is a double-edged sword because there is potentially additional pangenomic variation that is real but does not map to a single reference. The authors state in the first paragraph of the discussion that this choice did not affect the results, but I did not find an associated analysis in the supplement or main text. It would be interesting to compare distance metrics based on screening against all known microbial+human genomes vs screening against the reference genomes.

eLife Assessment

This study presents a valuable analysis of the TARA oceans dataset, advancing genome- and ecosystem-scale metabolic modeling. The authors present convincing evidence for the factors that shape microbial ecology at the molecular scale. The work will be of broad interest to oceanographers and scientists in other areas where metabolic interactions in diverse communities are of importance.

Reviewer #1 (Public review):

The authors use metabolic reconstruction to build genome-scale models from TARA metagenomes.

This is not the first paper that attempts a reconstruction of metabolism on this scale. And like those that came before, its weakness is in the analysis of the resulting model.

I was initially quite excited about this manuscript as the reconstruction itself seems well done. However, the analysis does not deliver major or minor results.

It could be said that the modelling cuts some corners, for example in the identification of the flux modes. I was not quite able to follow why this is necessary, as numerical linear algebra methods seem to be able to handle problems of this size. But then again, I am quite confident that the sampling approach that the authors use produces a good enough approximation.

The weak point of the manuscript is the analysis that follows. The conclusions contain very few hard results, and the results that are presented talk mostly about concepts that are somewhat arbitrarily introduced and hard to relate back to nature. The authors construct a mix of their own concepts and others borrowed from cancer research, and neither of the metrics that are used is very convincing.

For example, the authors say they contribute to the complexity-stability debate, but that debate focuses on a particular notion of stability. What the authors call stability here is an entirely different concept that is completely unrelated to the stability of the complexity-stability debate; I would rather describe it as a resilience metric rather than stability.

The problem is that the metrics that are used lack an underlying firm grounding. Ecology was for a long time struggling with the same problem (and to some extent still is), but eventually much progress has been made by rigorously deriving metrics that are easy to relate back. The same would be possible here. Using Steuer's structural kinetic method, the reconstructed metabolism could be described dynamically, which would allow genuine stability analysis as well as other crucial metrics such as the observability and controllability, impact and sensitivity, which would make it far easier to relate results back.

Reviewer #2 (Public review):

This study used publicly available Tara Oceans and Tara Oceans Polar Circle metagenomic and metatranscriptomic datasets, including viromes, to construct sample-specific, gene-scale metabolic models. This reviewer understood that, for each sample, genes or transcripts associated with metabolic pathways were integrated into a single virtual "superorganism" or "community cell." These sample-level models were then compared across global ocean regions to investigate spatial patterns in heterotrophic prokaryotic metabolism, metabolic synergy, and the potential effects of virus-encoded auxiliary metabolic genes. However, it was not clear whether archaeal genes were also included in the heterotrophic prokaryotic fraction.

The study represents an ambitious and potentially valuable attempt to connect large-scale environmental omics data with constraint-based metabolic modeling. The authors handled a very large dataset and introduced quantitative approaches based on flux sampling, Reaction Cumulative Correlation, and synergy scores to describe community-level metabolic phenotypes using several mathematical formulations. The recovery of previously defined oceanic ecological zones from reaction-based models may provide preliminary support for the ecological relevance of the framework, although only approximately 26% of prokaryotic genes and 7% of viral genes were mapped to known metabolic reactions.

However, the framework should be understood as gene- or reaction-resolved, sample-level metabolic modeling rather than genome-resolved community modeling. By combining all detected genes within a sample into a single superorganism, the approach likely loses taxon-specific metabolic information and cannot directly distinguish intracellular metabolism from interspecies metabolic exchange. Consequently, several ecological interpretations, including cooperation, stability, reaction essentiality, and viral impacts, remain strongly dependent on the underlying model assumptions.

Overall, the manuscript presents a novel and scalable framework with considerable potential. However, greater methodological clarification and more cautious interpretation are needed before the ecological and biogeochemical conclusions can be fully supported.

eLife Assessment

This is a valuable manuscript that leverages information already being collected in mosquito surveillance, but that is currently discarded, building on ideas developed during the SARS-CoV-2 pandemic. The work is mostly rigorous, involving ample data from real-world surveillance, independent laboratory testing, and simulation modeling to validate the inference method. The evidence is solid for demonstrating feasibility and biological plausibility, although some conclusions would be strengthened by additional sensitivity analyses on key assumptions.

Reviewer #1 (Public review):

Summary:

This manuscript seeks to make use of information about Ct values from PCR testing of mosquito pools for West Nile virus infection to make inferences about mosquito prevalence and West Nile risk. It does so through analysis of empirical data and simulated data with a realistic agent-based model.

Strengths:

This work is conceptually innovative for mosquito-borne viruses, building on ideas developed primarily during work on SARS-CoV-2. Exploring this topic is worthwhile regardless of the outcome. The use of data, testing in multiple labs, and the complementarity of modeling and empirical data analysis are all strengths of the approach.

Weaknesses:

Some of the primary weaknesses include a dependence of the results on relatively narrow model assumptions, and a lack of compelling improvement over existing methods. None of these weaknesses are fatal flaws; they are modest weaknesses that limit the potential of or excitement about the method.

Reviewer #2 (Public review):

Summary:

The authors extend their previous population-based Ct-value framework for inferring community epidemic trajectories from human infections to vector infections, using mosquitoes as vectors for West Nile virus. They use agent-based modelling to distinguish virus-positive detections arising from non-active infection states from those reflecting active infections, and then apply this framework to mosquito surveillance data from Colorado and Texas.

Overall, this is a well-designed and carefully evaluated study. The manuscript proposes a feasible and potentially valuable framework for vector infection surveillance. The findings are supported by both mechanistic agent-based simulations and applications to real-world mosquito surveillance data, which strengthens the biological plausibility and practical relevance of the proposed approach.

Strengths:

A major strength of the study is its clear methodological extension from human infection surveillance to vector infection surveillance. The agent-based modelling framework provides a useful basis for distinguishing active infections from virus-positive detections that may reflect non-active infection states. The application to surveillance data from two different geographic settings further supports the feasibility of the framework. Overall, the study is carefully designed, and the model schematic and main analyses are generally clear.

Weaknesses:

(1) It would be helpful if the authors could provide plots showing variation across locations and over time. This would further support the claim made in the paragraph at lines 101-107.

(2) Figure 2: The model schematic is clear in terms of workflow, but it would benefit from more information on model parameterization. In particular, it would be helpful to clarify which parameters or migration rates were estimated from the data and which were assumed based on prior literature.

(3) Figure 4: I wonder whether the authors examined how changes in the proportion of mosquitoes with static viral-kinetics trajectories would affect the observed bimodal distribution. Relatedly, it would be useful to know whether there is a threshold proportion at which the method becomes less able to distinguish active from static viral-kinetics patterns.

Conclusion:

Overall, the evidence is reasonably strong for demonstrating the feasibility and biological plausibility of the proposed framework. Some conclusions would be further strengthened by additional sensitivity analyses on key assumptions, especially the proportion of static viral-kinetics trajectories and spatial-temporal heterogeneity across surveillance sites.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors characterize the phospholipid scramblase Xkr in Drosophila. They generate null mutants in both S2 cells and flies and find that phosphatidylserine (PS) exposure is reduced during apoptosis; they show reduced engulfment of apoptotic cells, and that the protein is localized partially within the cytoplasm, overlapping with the ER. They go on to identify Xkr binding partners and show that they overlap with plasma membrane-ER contact sites, suggesting that Xkr facilitates PS transfer from the ER to PM. Overall, this reveals a new role for Xkr and identifies new binding partners, which are valuable contributions to the field.

Strengths:

(1) The generation of new Xkr reagents in both S2 cells and flies to analyze its function. Tools are used to quantify both PS exposure and efferocytosis, and the effects of Xkr knockout are significant.

(2) The discovery of new binding partners of Xkr which also affect PS exposure and efferocytosis.

(3) The authors demonstrate that the binding partners are conserved in mammalian cells.

Weaknesses:

(1) Throughout the manuscript (e.g, lines 105, 165, 274 and discussion), the authors describe Xkr as being activated in a caspase-independent manner, and use this as the rationale for identifying binding partners. However, this is never shown in the manuscript or clearly referenced. Interestingly, there is a TEVDA sequence in the fly ortholog at the same location as the caspase cleavage site in C. elegans Ced-8 (Figure S1), suggesting the caspase cleavage site is conserved. This should be further investigated, or the statements regarding caspase independence should be modified. I don't think the N- and C-terminal GFP fusions indicate caspase independence, especially since apoptosis was not induced in Figure 1A, B. If cleavage occurred at the TEVDA site in Figure S1A, it would not lead to a noticeable change on the Western blot, although the size does look a bit smaller in Figure S2B at the 8 h time point.

We thank the reviewer for pointing out that, as E/DXXD has been considered a conserved caspase-3 cleavage site, TEVDA has also been validated as a caspase-6 cleavage site, which we have missed. We will further confirm this using site-mutated expression vectors in S2 cells.

(2) The authors examine overlap between tagged Xkr and cellular compartment markers and find substantial overlap with Lamp (and other vesicle markers to a lesser extent) (Figure S2). This is not addressed in the paper and could indicate engulfment of other cells since S2 cells are macrophages. To test this, the staining could be tested on the mixed cells (vesicle-GFP tagged S2 + apoptotic xkr-mcherry). Similarly, calreticulin is an eatme signal that gets translocated to the PM of apoptotic cells. This could affect interpretation of colocalization (Figure 2J), and ideally another ER marker should be used.

We thank the reviewer for the suggestion. We will attempt to label Xkr-mCherry under apoptosis with other vesicles and change the ER marker to Cnx99A (Calnexin ortholog in Drosophila).

(3) There are some places where there is over- or incorrect interpretation, and these instances should be corrected.

We thank the reviewer for their careful reading, and we will correct the mistakes in the revised manuscript.

Specific examples:

a) Line 342 "Relative expression analysis by RT-qPCR showed that all three mutants were likely null alleles." This does not make sense since there is still mRNA present. In Figure S7A, the tm9sf4 allele is expressed at 75% of the control. The others show a greater reduction, but this is not proof of a null allele.

We agree with the reviewer’s opinion. These mutants from the BDSC are not completely deleted but partially deleted; therefore, the RT-qPCR assay may not be very accurate. We will detect the mRNA levels of tm9sf4, dorp9, and sac1 using RT primers from different cDNA regions to make the results more convincing.

b) Figure S3I - It looks like mCherry-Lact:C2 does get localized to the PM with AcD treatment in the xkr[ko], although the authors conclude "this disrupted PS localization to the PM could not be restored by apoptosis induction". However, the PM localization does look disrupted in the tm9sf4 and sac1 knockdowns.

We thank you for raising this intriguing hypothesis. Indeed, PM localization of Lact:C2 was reduced in xkr^ko cells, and the distribution could not be rescued after apoptosis. Unlike xkr^ko, tm9sf4, and sac1 RNAi-treated cells displayed weak PS disorder, which may be due to the efficiency of knockdown. However, the statistical results indicated that the ratio of PM/Cyto was reduced in tm9sf4 and sac1 RNAi-treated cells.

c) Figure 3I. The control Lact:C2 staining looks very different from the staining in Figure 2J, with abundant Lact:C2 outside the cell. Given the variability in the staining, were the contact sites quantified? On lines 287-288, it is stated that "fewer ER-PM MCSs were detected in xkrko cells than in WT", but no quantification is provided.

We thank for the reviewer’s suggestion. We will add the statistical results of Fig. 3I in the revised version.

d) Line 299-300 - "the interaction between Xkr and dORP9 was enhanced after apoptosis induction". The interaction does not look enhanced in Figure S5F, so this statement should be removed or data supporting the statement should be provided. The interaction between Xkr and dORP2 looks enhanced upon apoptosis induction, but also paradoxically looks even more enhanced when apoptosis is blocked.

We thank you for raising this intriguing hypothesis. We will delete the relevant statement to eliminate unnecessary misunderstandings.

e) The data in Figure S6 are highlighted in the abstract. If this is a major conclusion, it would be best to move it to the main text and provide quantification.

We thank for the reviewer’s suggestion. We will move this to the main text and provide quantification in the revised version.

f) Lines 392-4. The concluding statement seems overstated given that there was only a modest inhibition of PS exposure in the osbpl5 knockdown (Figure 6A) and no defects in efferocytosis (Figure 6C). The osbpl8 showed a stronger effect on PS exposure but still a very modest effect on efferocytosis.

We thank for the reviewer’s suggestion. We will weaken the statement in the Results section of Figure 6 and perform osbpl9 knockdown to observe efferocytosis in Raw264.7 cells, as OSBPL9 interacts with Xkr8 strongly.

Reviewer #2 (Public review):

In this study, the authors investigate the mechanisms underlying phosphatidylserine (PS) exposure during efferocytosis in Drosophila. They first show that Xkr promotes PS exposure and apoptotic cell clearance in both S2 cells and Drosophila embryos. As Drosophila Xkr lacks the canonical caspase cleavage site found in mammalian XKR proteins, the authors further explore the underlying mechanism by which Xkr regulates PS externalization. Through protein interaction studies, they identify TM9SF4 as an interacting partner of Xkr that regulates PS distribution and show that non-vesicular PS transport contributes to apoptotic PS exposure and efferocytosis. Using protein interaction studies, they further demonstrate that Xkr interacts with the lipid transfer protein dORP9 at ER-PM contact sites to facilitate non-vesicular PS transport to the plasma membrane. Loss of these proteins affects PS externalization and efferocytosis in Drosophila. Finally, using human cells, they demonstrate that human OSBPL8 interacts with XKR8 to regulate apoptotic PS exposure. Overall, the study supports a model in which Xkr promotes efferocytosis by facilitating lipid transport in addition to its role as a phospholipid scramblase.

Thank you for your comprehensive and generous assessment of our work and for the time and expertise you have devoted to reviewing our manuscript. We will revise the manuscript accordingly and provide a point-by-point response in the revised version.

Reviewer #3 (Public review):

Summary:

The manuscript investigates the function of the Drosophila Xkr protein, a homolog of mammalian Xkr8 that lacks the canonical caspase-cleavage motif. The authors show that apoptotic stimuli increase Xkr protein abundance through a post-transcriptional mechanism and that Xkr promotes phosphatidylserine (PS) exposure during apoptosis. Using immunoprecipitation coupled with mass spectrometry, they identify TM9SF4 as an Xkr-interacting protein and further implicate TM9SF4, Sac1, dORP2, dORP9, and Vap33 in regulating apoptotic PS exposure and efferocytosis. Based on these findings, the authors propose that Xkr regulates PS transport at ER-PM contact sites. Similar observations are also presented in human cells.

Strengths:

Overall, this is an interesting study. The authors provide convincing evidence that Drosophila Xkr participates in apoptotic PS exposure and employ multiple complementary approaches to support the involvement of several proteins in this pathway. The identification of TM9SF4 as a potential regulator of Xkr-mediated PS exposure is likely to be of broad interest.

Weaknesses:

I am less convinced by the evidence supporting the proposed role of ER-PM contact sites, and several mechanistic conclusions appear to extend beyond the data presented. Addressing the following points would substantially strengthen the manuscript.

We sincerely thank you for your careful reading and accurate summary of our manuscript. We appreciate the time, effort, and expertise you have dedicated to evaluating our work, and we will try our best to improve our manuscript according to your suggestions.

Major concerns:

(1) In Figure 2A and related text, it is unclear whether the mass spectrometry analysis was performed using untreated cells or AcD-treated cells. If the objective was to identify apoptosis-associated Xkr interactors, it would be helpful to clarify the experimental condition and explain whether apoptosis-specific interactors were analyzed separately.

We thank for the reviewer’s suggestion. We used AcD-treated S2 cells and untreated S2 cells to perform mass spectrometry. To clarify this, we will add a detailed method description in the method section.

(2) In Figure 2B, 2E, and several other co-IP results, a negative control of Flag tag only is required to exclude experimental errors like insufficient washing, etc.

We thank for the reviewer’s suggestion. We used anti-HA magnetic beads to perform immunoprecipitation, and single HA-TM9SF4 was used as a negative control.

(3) In Figure S3B, S3F, and several other BiFC results, an mVC-only negative control would be important to exclude nonspecific fluorescence complementation.

We thank for the reviewer’s suggestion, we will add the negative control for BiFC results in the revised version.

(4) In Figure 2G, the quantitative values appear inconsistent with the flow cytometry histograms. The peak shift following Sac1 knockdown appears smaller than that of TM9SF4 knockdown, whereas the quantified values suggest the opposite. Please clarify this apparent discrepancy.

We sincerely thank you for the careful consideration of our statistical results, which were obtained from 3 repeats. We will choose another flow cytometry histogram of tm9sf4 and sac1 to make the data and images more consistent.

(5) I find the interpretation in Lines 223-227 difficult to reconcile with the data. Knockdown of both tm9sf4 and sac1 impaired apoptotic PS exposure to a similar extent as xkr knockout. However, while xkr deficiency significantly reduced efferocytosis, sac1 knockdown produced only a modest, statistically insignificant effect. These observations suggest that impaired PS exposure alone may not fully account for the efferocytosis phenotype observed in xkr-deficient cells. These results appear difficult to reconcile with the proposed model, which needs careful discussion.

We sincerely thank the reviewer for their careful and thoughtful observations. Given the results we have observed, we will add this to the discussion section in the revised version.

(6) In Lines 274-275, the authors state that 'increased Xkr may accelerate non-vesicular PS transport for efficient apoptotic PS exposure'. However, Xkr protein levels increase only ~8 h after AcD treatment, whereas PS exposure occurs much earlier. Thus, alternative explanations like Xkr relocalization (Figure S5C), rather than increased abundance, may also explain how Xkr mediates PS transport. An Xkr overexpression experiment could be helpful to support this statement.

We thank for the reviewer’s suggestion. We will overexpress Xkr with or without AcD treatment to observe whether the localization or amount of Lact:C2 changes and to re-evaluate the role of Xkr in PS exposure.

(7) The interpretation of the MAPPER experiments requires further clarification. In Line 283, the authors refer to "the intracellular proportion of the signal for each protein overlapping with MAPPER." Since MAPPER is designed to label ER-PM contact sites, which are located on the plasma membrane, intracellular MAPPER fluorescence likely represents the ER network rather than bona fide ER-PM contacts. Throughout the manuscript (including Figure S6, etc.), intracellular MAPPER puncta appear to be interpreted as ER-PM contacts, which may not be appropriate. In contrast, the peripheral MAPPER puncta observed along the cell cortex (e.g., Figure S5C after AcD treatment) are more consistent with authentic ER-PM contact sites. It is also not obvious that these cortical MAPPER signals colocalize with Xkr(Figure S5C). Thus, while the data support a role for the ER, they do not yet convincingly demonstrate Xkr clustering at ER-PM contact sites.

We thank the reviewer for the suggestion, and we believe that the TIRF technique can help us demonstrate the ER-PM signal. Since our college has no TIRF microscope, we will try our best to seek cooperation from other colleges to achieve this experiment.

(8) In the Xkr knockout cells, all fluorescence signals appear substantially low in intensity. Differences in protein distribution are difficult to interpret when overall probe expression also appears altered. It would be helpful to demonstrate that probe expression levels are comparable between conditions. Furthermore, as noted above, intracellular MAPPER signal may primarily represent ER rather than ER-PM contacts. Finally, despite the reduced signal intensity, the remaining MAPPER and PS signals still appear well colocalized in the knockout cells, similar to the observations in Figure 2J. The interpretation in Lines 285-288 should therefore be reconsidered.

We sincerely thank the reviewer for this careful and thoughtful observation, and we agree that the interpretation in Line 285-288 is overstated. To explain this, we plan to detect the Lact:C2 and MAPPER signals in S2 and xkr^ko cells with or without AcD to confirm how Xkr regulates PS via ER-PM under apoptotic conditions.

eLife Assessment

This study presents valuable findings implicating Xkr and its newly identified binding partners in regulating the exposure of apoptotic signal for phagocytosis, suggesting that Xkr might facilitate phosphatidylserine transfer at the ER-plasma membrane contact sites. However, several conclusions, including caspase-independent activity of Xkr and its role at the ER-PM contact sites, are not well supported by experimental data. Furthermore, some key controls are missing, and in numerous places the authors' discussion has extended far beyond what the data support. The findings presented are currently incomplete and do not fully support key mechanistic claims of the study.

Reviewer #1 (Public review):

Summary:

The authors characterize the phospholipid scramblase Xkr in Drosophila. They generate null mutants in both S2 cells and flies and find that phosphatidylserine (PS) exposure is reduced during apoptosis; they show reduced engulfment of apoptotic cells, and that the protein is localized partially within the cytoplasm, overlapping with the ER. They go on to identify Xkr binding partners and show that they overlap with plasma membrane-ER contact sites, suggesting that Xkr facilitates PS transfer from the ER to PM. Overall, this reveals a new role for Xkr and identifies new binding partners, which are valuable contributions to the field.

Strengths:

(1) The generation of new Xkr reagents in both S2 cells and flies to analyze its function. Tools are used to quantify both PS exposure and efferocytosis, and the effects of Xkr knockout are significant.

(2) The discovery of new binding partners of Xkr which also affect PS exposure and efferocytosis.

(3) The authors demonstrate that the binding partners are conserved in mammalian cells.

Weaknesses:

(1) Throughout the manuscript (e.g, lines 105, 165, 274 and discussion), the authors describe Xkr as being activated in a caspase-independent manner, and use this as the rationale for identifying binding partners. However, this is never shown in the manuscript or clearly referenced. Interestingly, there is a TEVDA sequence in the fly ortholog at the same location as the caspase cleavage site in C. elegans Ced-8 (Figure S1), suggesting the caspase cleavage site is conserved. This should be further investigated, or the statements regarding caspase independence should be modified. I don't think the N- and C-terminal GFP fusions indicate caspase independence, especially since apoptosis was not induced in Figure 1A, B. If cleavage occurred at the TEVDA site in Figure S1A, it would not lead to a noticeable change on the Western blot, although the size does look a bit smaller in Figure S2B at the 8 h time point.

(2) The authors examine overlap between tagged Xkr and cellular compartment markers and find substantial overlap with Lamp (and other vesicle markers to a lesser extent) (Figure S2). This is not addressed in the paper and could indicate engulfment of other cells since S2 cells are macrophages. To test this, the staining could be tested on the mixed cells (vesicle-GFP tagged S2 + apoptotic xkr-mcherry). Similarly, calreticulin is an eatme signal that gets translocated to the PM of apoptotic cells. This could affect interpretation of colocalization (Figure 2J), and ideally another ER marker should be used.

(3) There are some places where there is over- or incorrect interpretation, and these instances should be corrected.

Specific examples:

a) Line 342 "Relative expression analysis by RT-qPCR showed that all three mutants were likely null alleles." This does not make sense since there is still mRNA present. In Figure S7A, the tm9sf4 allele is expressed at 75% of the control. The others show a greater reduction, but this is not proof of a null allele.

b) Figure S3I - It looks like mCherry-Lact:C2 does get localized to the PM with AcD treatment in the xkr[ko], although the authors conclude "this disrupted PS localization to the PM could not be restored by apoptosis induction". However, the PM localization does look disrupted in the tm9sf4 and sac1 knockdowns.

c) Figure 3I. The control Lact:C2 staining looks very different from the staining in Figure 2J, with abundant Lact:C2 outside the cell. Given the variability in the staining, were the contact sites quantified? On lines 287-288, it is stated that "fewer ER-PM MCSs were detected in xkrko cells than in WT", but no quantification is provided.

d) Line 299-300 - "the interaction between Xkr and dORP9 was enhanced after apoptosis induction". The interaction does not look enhanced in Figure S5F, so this statement should be removed or data supporting the statement should be provided. The interaction between Xkr and dORP2 looks enhanced upon apoptosis induction, but also paradoxically looks even more enhanced when apoptosis is blocked.

e) The data in Figure S6 are highlighted in the abstract. If this is a major conclusion, it would be best to move it to the main text and provide quantification.

f) Lines 392-4. The concluding statement seems overstated given that there was only a modest inhibition of PS exposure in the osbpl5 knockdown (Figure 6A) and no defects in efferocytosis (Figure 6C). The osbpl8 showed a stronger effect on PS exposure but still a very modest effect on efferocytosis.

Reviewer #2 (Public review):

In this study, the authors investigate the mechanisms underlying phosphatidylserine (PS) exposure during efferocytosis in Drosophila. They first show that Xkr promotes PS exposure and apoptotic cell clearance in both S2 cells and Drosophila embryos. As Drosophila Xkr lacks the canonical caspase cleavage site found in mammalian XKR proteins, the authors further explore the underlying mechanism by which Xkr regulates PS externalization. Through protein interaction studies, they identify TM9SF4 as an interacting partner of Xkr that regulates PS distribution and show that non-vesicular PS transport contributes to apoptotic PS exposure and efferocytosis. Using protein interaction studies, they further demonstrate that Xkr interacts with the lipid transfer protein dORP9 at ER-PM contact sites to facilitate non-vesicular PS transport to the plasma membrane. Loss of these proteins affects PS externalization and efferocytosis in Drosophila. Finally, using human cells, they demonstrate that human OSBPL8 interacts with XKR8 to regulate apoptotic PS exposure. Overall, the study supports a model in which Xkr promotes efferocytosis by facilitating lipid transport in addition to its role as a phospholipid scramblase.

Reviewer #3 (Public review):

Summary:

The manuscript investigates the function of the Drosophila Xkr protein, a homolog of mammalian Xkr8 that lacks the canonical caspase-cleavage motif. The authors show that apoptotic stimuli increase Xkr protein abundance through a post-transcriptional mechanism and that Xkr promotes phosphatidylserine (PS) exposure during apoptosis. Using immunoprecipitation coupled with mass spectrometry, they identify TM9SF4 as an Xkr-interacting protein and further implicate TM9SF4, Sac1, dORP2, dORP9, and Vap33 in regulating apoptotic PS exposure and efferocytosis. Based on these findings, the authors propose that Xkr regulates PS transport at ER-PM contact sites. Similar observations are also presented in human cells.

Strengths:

Overall, this is an interesting study. The authors provide convincing evidence that Drosophila Xkr participates in apoptotic PS exposure and employ multiple complementary approaches to support the involvement of several proteins in this pathway. The identification of TM9SF4 as a potential regulator of Xkr-mediated PS exposure is likely to be of broad interest.

Weaknesses:

I am less convinced by the evidence supporting the proposed role of ER-PM contact sites, and several mechanistic conclusions appear to extend beyond the data presented. Addressing the following points would substantially strengthen the manuscript.

Major concerns:

(1) In Figure 2A and related text, it is unclear whether the mass spectrometry analysis was performed using untreated cells or AcD-treated cells. If the objective was to identify apoptosis-associated Xkr interactors, it would be helpful to clarify the experimental condition and explain whether apoptosis-specific interactors were analyzed separately.

(2) In Figure 2B, 2E, and several other co-IP results, a negative control of Flag tag only is required to exclude experimental errors like insufficient washing, etc.

(3) In Figure S3B, S3F, and several other BiFC results, an mVC-only negative control would be important to exclude nonspecific fluorescence complementation.

(4) In Figure 2G, the quantitative values appear inconsistent with the flow cytometry histograms. The peak shift following Sac1 knockdown appears smaller than that of TM9SF4 knockdown, whereas the quantified values suggest the opposite. Please clarify this apparent discrepancy.

(5) I find the interpretation in Lines 223-227 difficult to reconcile with the data. Knockdown of both tm9sf4 and sac1 impaired apoptotic PS exposure to a similar extent as xkr knockout. However, while xkr deficiency significantly reduced efferocytosis, sac1 knockdown produced only a modest, statistically insignificant effect. These observations suggest that impaired PS exposure alone may not fully account for the efferocytosis phenotype observed in xkr-deficient cells. These results appear difficult to reconcile with the proposed model, which needs careful discussion.

(6) In Lines 274-275, the authors state that 'increased Xkr may accelerate non-vesicular PS transport for efficient apoptotic PS exposure'. However, Xkr protein levels increase only ~8 h after AcD treatment, whereas PS exposure occurs much earlier. Thus, alternative explanations like Xkr relocalization (Figure S5C), rather than increased abundance, may also explain how Xkr mediates PS transport. An Xkr overexpression experiment could be helpful to support this statement.

(7) The interpretation of the MAPPER experiments requires further clarification. In Line 283, the authors refer to "the intracellular proportion of the signal for each protein overlapping with MAPPER." Since MAPPER is designed to label ER-PM contact sites, which are located on the plasma membrane, intracellular MAPPER fluorescence likely represents the ER network rather than bona fide ER-PM contacts. Throughout the manuscript (including Figure S6, etc.), intracellular MAPPER puncta appear to be interpreted as ER-PM contacts, which may not be appropriate. In contrast, the peripheral MAPPER puncta observed along the cell cortex (e.g., Figure S5C after AcD treatment) are more consistent with authentic ER-PM contact sites. It is also not obvious that these cortical MAPPER signals colocalize with Xkr(Figure S5C). Thus, while the data support a role for the ER, they do not yet convincingly demonstrate Xkr clustering at ER-PM contact sites.

(8) In the Xkr knockout cells, all fluorescence signals appear substantially low in intensity. Differences in protein distribution are difficult to interpret when overall probe expression also appears altered. It would be helpful to demonstrate that probe expression levels are comparable between conditions. Furthermore, as noted above, intracellular MAPPER signal may primarily represent ER rather than ER-PM contacts. Finally, despite the reduced signal intensity, the remaining MAPPER and PS signals still appear well colocalized in the knockout cells, similar to the observations in Figure 2J. The interpretation in Lines 285-288 should therefore be reconsidered.

eLife Assessment

An intriguing set of electrophysiological and analytical results suggests that adult rat neurons in the medulla oblongata crucial for mediating descending pain modulation are embedded in a network that generates a specific pattern of oscillatory activity. These valuable findings may broaden our understanding of how descending pain modulation is achieved. The supporting evidence is incomplete, and the authors need to provide several methodological clarifications and additional data interpretations to strengthen the manuscript.

Reviewer #1 (Public review):

Summary:

The authors hypothesized that "RVM neurons operate across multiple temporal scales, integrating fast responses associated with reflex-linked control with slower fluctuations reflecting ongoing network or state-dependent modulation". The hypothesis was tested with the established ON/OFF-cell model and probabilistic modeling. The study is conceptually interesting and methodologically sophisticated. The findings build toward the conclusion that pain-control circuits operate across multiple timescales.

Strengths:

The use of Bayesian regression and Gaussian process modeling to quantify and characterize recovery dynamics and ongoing oscillatory activity.

The authors show that slow rhythmic activity appears preferentially in ON- and OFF-cells but not in NEUTRAL-cells, suggesting that the oscillations are related to pain-modulatory circuitry rather than being a generic feature of all recorded neurons.

The observation that some oscillatory activity is coherent with autonomic measures aligns with broader views of the RVM as a hub integrating nociceptive and homeostatic regulation.

Some pitfalls are appreciated and discussed by the authors, including the functional significance of slow fluctuations, the influence of anesthetics on global brain-state dynamics, the molecular profiles of the studied ON- and OFF-cells, and the heart rate as a covarying signal of RVM neuronal activity.

Weaknesses:

A general weakness is that the work is mostly descriptive and relies on anesthetized preparations. Whether the observed rhythms occur in awake animals and are linked to fluctuations in pain behavior needs to be confirmed in future studies.

The study measures limited autonomic variables. The causal relationship between "slow fluctuations" and "ongoing physiological state" is unclear and overstated, since the data presented appear correlational.

ON- and OFF-cells in the RVM are identified by their responses correlated with reflexive activity. The significance of the observed oscillations in spontaneous pain conditions is unclear.

It is uncertain whether the observed rhythms truly reflect intrinsic RVM organization rather than anesthesia-dependent phenomena; the authors appreciated this pitfall, though.

The Gaussian process analysis suggests predictability and quasi-periodicity, but predictability alone does not necessarily imply a true biological oscillator.

Conclusion:

The results support the authors' hypothesis. The findings provide a compelling conceptual message about the multiscale organization and dynamics of descending pain-control circuits and encourage further studies on the topic.

Reviewer #2 (Public review):

Using electrophysiological recordings in a well-characterized animal model of acute pain, and analytical and modeling methods, the authors show that descending pain-modulatory neurons in the rostral ventromedial medulla (RVM) operate across various timescales. They have both rapid multi-phase responses to noxious stimuli that unfold over tens of seconds, with distinct fast and slow recovery dynamics. Additionally, they generate slow quasi-periodic oscillations with approximately 5-minute periods during ongoing activity. These oscillations are statistically predictable and cell-type specific, demonstrating that descending pain control is organized through structured temporal dynamics that encompass immediate stimulus-evoked responses and slower fluctuations associated with physiological state.

A novel discovery is a ~5-minute quasi-periodic oscillation in ongoing ON- and OFF-cell activity. This oscillation, along with its coherence with heart rate, forms the basis for the claim that descending pain circuits exhibit intrinsic multi-timescale organization. However, it's crucial to demonstrate that this periodicity is independent of external experimental cycles such as methohexital infusion pharmacokinetics, servo-controlled temperature regulation, or slow autonomic feedback loops, all of which operate on similar timescales. For instance, the 300-second period closely matches typical drug infusion cycling and thermoregulatory feedback intervals. Therefore, heart-rate coherence peaks at multiples of this period could equally reflect a shared external driver rather than intrinsic RVM organization. Although the absence of this cyclic structure in Neutral cells argues against this possibility, the authors might want to explicitly discuss this potential confound.

The findings are important and novel in that they characterize an intriguing structure in the activity of ON and OFF neurons in the RVM. However, in the absence of a causal manipulation causality can only be inferred. That there is no phase-dependence of withdrawal latency argues against a causal role. The author are encouraged to qualify their conclusions (and their title) accordingly.

Because anesthesia can affect global dynamics, this might affect the oscillations reported. Without awake validation, it remains uncertain whether these rhythms reflect an intrinsic property or an anesthesia-induced regime. Again, the absence of oscillations in Neutral cells argues against this possibility, but it is still possible that ON/OFF cells are embedded in different circuits that are affected differently by anesthesia.

Analyses of many of the ON-cells had longer training windows (>1 sec) compared to those for the NEUTRAL cells. Could this have reduced the ability to fit and validate periodicity for the latter cell type?

Reviewer #3 (Public review):

Summary:

In this manuscript by Ashworth and colleagues, the authors investigate the temporal dynamics of the rostral ventromedial medulla (RVM), a key output node in a major descending pain-modulation circuit. Using data from extrasellar single-unit recordings of RVM ON, OFF, and NEUTRAL cells in lightly anesthetized rats, the authors' computational modeling yielded two major findings: (1) heat-evoked ON burst and OFF pause, followed by exponential recovery components in10s of seconds; and (2) ON and OFF cells exhibit periodic fluctuations in ~5-minute cycles that are statistically predictable.

Strengths:

The manuscript's concept is innovative, offering the first quantitative analysis of multi-timescale dynamics in physiologically characterized RVM pain-modulating neurons. This advances a field that has mostly depended on qualitative or single-timescale descriptions. The authors use contemporary Gaussian process and probabilistic models to capture statistically predictable slow dynamics. The study is further strengthened by identifying ON-, OFF-, and NEUTRAL-type cells using well-established criteria grounded in decades of RVM research. The combination of rapid reflex-related responses and slower ongoing rhythms supports a dual-timescale framework, providing a more integrated understanding of how these neurons may regulate reflex activity and state-dependent processes.

Weaknesses:

Several limitations are noted. Incomplete characterization of light anesthesia during recording sessions, such as methohexital stability and clear criteria for identifying "lightly anesthetized" states. While the NEUTRAL cell control is helpful, it does not fully address concerns about circuit specificity or systemic confounds. The findings are male-dominant, which may limit their generalizability. The synchrony between ON and OFF cells was suggested but not directly tested. The heart rate coherence with ON, OFF, and NEUTRAL cell activity results is intriguing but does not fully clarify how these neurons influence heart rate, particularly within the "lightly anesthetized" model.

Author response:

We are pleased that the reviewers found the study conceptually novel and the analytical framework rigorous. In response we have substantially revised the manuscript to clarify methodological details, temper several interpretations, expand discussion of alternative explanations, and include additional analyses using the existing dataset. We have deliberately revised the manuscript so that our conclusions are limited to those directly supported by the data, namely that physiologically identified RVM pain-modulatory neurons exhibit structured dynamics spanning multiple temporal scales. We do not interpret the slow fluctuations as evidence for a specific intrinsic oscillator or for a causal role in physiological state regulation. We have also expanded the rationale for the lightly anaesthetized preparation, emphasizing that it provides both the recording stability required for prolonged single-unit recordings from sparse neurons in the deep RVM and a controlled physiological setting in which the baseline temporal organization of the circuit can be characterized while minimizing ongoing sensory, motor, and behavioral influences.

Regarding the rationale for the lightly anaesthetized preparation, these experiments take advantage of the well-validated lightly anaesthetized Sprague-Dawley rat in which much of the foundational data concerning physiology and function of RVM neurons was obtained. This “middle-out” strategy [1] has allowed direct connections between the activity and pharmacology of identified RVM neurons and altered nociceptive behavior. This protocol demonstrably spares the essential links between brainstem pain-modulating neurons and nociceptive transmission pathways. Although the focus here was on ongoing activity, precluding the repeated nociceptive testing needed to link neuronal activity to nociceptive threshold, previous work has demonstrated that ongoing activity of OFF and ON-cells is correlated with nociceptive sensitivity [2] and that alterations in OFF- and ON cell firing in response to pharmacological manipulation and in models of persistent pain states, stress, and sickness have behavioral relevance [3–6,6–24]. Further, conclusions from work in lightly anaesthetized rats have repeatedly been found to be congruent with behavioral observations by other groups in awake rats and mice [19,25–36] and with functional imaging evidence in humans [37–40]. The lightly anaesthetized model has thus established a circuit-level explanatory framework for behavioral findings obtained in several species in multiple laboratories.

A further consideration for the present study is that the lightly anaesthetized preparation allows us to examine the underlying temporal organization of the RVM under controlled conditions, without the additional factors that would necessarily come into play in an awake animal. Dynamics would inevitably be influenced by ongoing sensory input, behavioral priorities, arousal and other internal state changes. These factors would make it difficult to distinguish the intrinsic dynamics of the descending pain-modulatory system from the effects of the animal’s constantly changing experience.

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors hypothesized that “RVM neurons operate across multiple temporal scales, integrating fast responses associated with reflex-linked control with slower fluctuations reflecting ongoing network or state-dependent modulation”. The hypothesis was tested with the established ON/OFF-cell model and probabilistic modeling. The study is conceptually interesting and methodologically sophisticated. The findings build toward the conclusion that pain-control circuits operate across multiple timescales.

Strengths:

The use of Bayesian regression and Gaussian process modeling to quantify and characterize recovery dynamics and ongoing oscillatory activity.

The authors show that slow rhythmic activity appears preferentially in ON- and OFF-cells but not in NEUTRAL-cells, suggesting that the oscillations are related to pain-modulatory circuitry rather than being a generic feature of all recorded neurons.

The observation that some oscillatory activity is coherent with autonomic measures aligns with broader views of the RVM as a hub integrating nociceptive and homeostatic regulation.

Some pitfalls are appreciated and discussed by the authors, including the functional significance of slow fluctuations, the influence of anesthetics on global brain-state dynamics, the molecular profiles of the studied ON- and OFF-cells, and the heart rate as a covarying signal of RVM neuronal activity.

Weaknesses:

A general weakness is that the work is mostly descriptive and relies on anesthetized preparations. Whether the observed rhythms occur in awake animals and are linked to fluctuations in pain behavior needs to be confirmed in future studies.

The study measures limited autonomic variables. The causal relationship between “slow fluctuations” and “ongoing physiological state” is unclear and overstated, since the data presented appear correlational.

ON- and OFF-cells in the RVM are identified by their responses correlated with reflexive activity. The significance of the observed oscillations in spontaneous pain conditions is unclear.

It is uncertain whether the observed rhythms truly reflect intrinsic RVM organization rather than anesthesia-dependent phenomena; the authors appreciated this pitfall, though.

The Gaussian process analysis suggests predictability and quasi-periodicity, but predictability alone does not necessarily imply a true biological oscillator.

Conclusion:

The results support the authors’ hypothesis. The findings provide a compelling conceptual message about the multiscale organization and dynamics of descending pain-control circuits and encourage further studies on the topic.

We thank Reviewer R1 for their thoughtful and balanced assessment of our work. We are grateful for the reviewer’s positive evaluation of the conceptual framework, the analytical methodology, and the conclusion that the results support the hypothesis that RVM pain-modulatory neurons operate across multiple temporal scales. We agree with the reviewer’s central assessment that the present study is primarily descriptive and that several important questions regarding the origin and functional significance of the slow dynamics remain unresolved. We also appreciate the reviewer’s emphasis on clearly distinguishing observations directly supported by the data from their mechanistic interpretation. Although the manuscript already acknowledged that the present findings do not establish the mechanistic origin of the slow dynamics, we agree that this distinction could be made more explicit. We have therefore revised the manuscript to clarify that the approximately 5-minute fluctuations represent structured, quasi-periodic activity whose underlying origin cannot be determined from the present experiments. Throughout the manuscript we now explicitly acknowledge that these dynamics may arise from interactions between RVM circuitry and broader physiological or network processes, including anaesthesia-related state modulation, autonomic regulation, or other slow network influences. We also emphasise that the relationship between RVM activity and heart rate is correlational and does not establish a causal interaction. Finally, we now discuss more explicitly the complementary roles of controlled lightly anaesthetized and awake preparations. The present preparation was chosen to characterize baseline RVM dynamics under controlled sensory and behavioral conditions, whereas future awake studies will be important for determining how these dynamics are expressed and modulated during ongoing behavior, sensory experience, and chronic pain.

(1) In the abstract, the ”timescales” are vaguely stated as ”rapid activation”, ”fast recovery dynamics,” and ”slow dynamics”. Quantifying these expressions with approximate ranges (milliseconds, seconds, tens of seconds, minutes, etc.) whenever possible would benefit readers.

We thank the reviewer for this helpful suggestion. We have revised the Abstract to provide approximate timescales for the different phases of neuronal activity, distinguishing the rapid stimulus-evoked response (sub-second), recovery dynamics (seconds to hundreds of seconds), and ongoing quasi-periodic fluctuations (approximately 5 minutes). We believe these revisions improve the clarity of the Abstract and better convey the central findings of the study.

(2) The abstract states that ”Effective pain therapies increasingly target neural circuits...” The connection to therapy is not clarified in the manuscript. A brief statement about how temporal dynamics might influence neuromodulation, analgesic interventions, or chronic pain could strengthen translational impact.

We appreciate this suggestion. We have revised both the Abstract and Discussion to better explain the potential translational relevance of our findings. Rather than making a broad statement regarding pain therapies, we now briefly discuss how understanding the temporal organisation of descending pain-modulatory circuits may ultimately inform the design and timing of neuromodulatory interventions. We also emphasise that these implications remain speculative and require future investigation.

(3) To address inter-animal and inter-neuron variability. Are the effects consistent across animals? Are all neurons oscillatory? Are the reported timescales driven by a subset of cells?

We thank the reviewer for raising this important point. We have expanded the Results and Discussion to clarify the degree of variability observed across neurons and animals. In particular, we now emphasise that the slow quasi-periodic dynamics are not uniformly expressed across all neurons, but rather represent a structured population-level phenomenon with variability in predictability and modulation strength between cells, particularly within the OFF-cell population. We also clarify the consistency of the observed timescales across animals and discuss this variability as an important feature of the underlying circuitry rather than evidence for a single homogeneous oscillatory process.

(4) Discuss what circuit mechanisms generate the oscillations. Are they driven by inputs from the PAG or intrinsic to RVM?

We agree that the mechanisms underlying the slow temporal dynamics are an important question. We have expanded the Discussion to consider several possible sources of these dynamics, including intrinsic RVM circuitry, descending inputs from higher-order structures, and broader physiological or brain-state fluctuations. We emphasise that the present experiments cannot distinguish between these possibilities and have revised the manuscript to make this limitation more explicit while highlighting it as an important direction for future work.

Reviewer #2 (Public review):

Using electrophysiological recordings in a well-characterized animal model of acute pain, and analytical and modeling methods, the authors show that descending pain-modulatory neurons in the rostral ventromedial medulla (RVM) operate across various timescales. They have both rapid multi-phase responses to noxious stimuli that unfold over tens of seconds, with distinct fast and slow recovery dynamics. Additionally, they generate slow quasi-periodic oscillations with approximately 5-minute periods during ongoing activity. These oscillations are statistically predictable and cell-type specific, demonstrating that descending pain control is organized through structured temporal dynamics that encompass immediate stimulus-evoked responses and slower fluctuations associated with physiological state.

A novel discovery is a 5-minute quasi-periodic oscillation in ongoing ON- and OFF-cell activity. This oscillation, along with its coherence with heart rate, forms the basis for the claim that descending pain circuits exhibit intrinsic multi-timescale organization. However, it’s crucial to demonstrate that this periodicity is independent of external experimental cycles such as methohexital infusion pharmacokinetics, servo-controlled temperature regulation, or slow autonomic feedback loops, all of which operate on similar timescales. For instance, the 300-second period closely matches typical drug infusion cycling and thermoregulatory feedback intervals. Therefore, heart-rate coherence peaks at multiples of this period could equally reflect a shared external driver rather than intrinsic RVM organization. Although the absence of this cyclic structure in Neutral cells argues against this possibility, the authors might want to explicitly discuss this potential confound.

The findings are important and novel in that they characterize an intriguing structure in the activity of ON and OFF neurons in the RVM. However, in the absence of a causal manipulation causality can only be inferred. That there is no phase-dependence of withdrawal latency argues against a causal role. The author are encouraged to qualify their conclusions (and their title) accordingly. Because anesthesia can affect global dynamics, this might affect the oscillations reported. Without awake validation, it remains uncertain whether these rhythms reflect an intrinsic property or an anesthesia-induced regime. Again, the absence of oscillations in Neutral cells argues against this possibility, but it is still possible that ON/OFF cells are embedded in different circuits that are affected differently by anesthesia.

We thank Reviewer R2 for their careful and constructive assessment of our work and for recognising the novelty of identifying structured multi-timescale dynamics in physiologically characterised RVM neurons. We particularly appreciate the reviewer’s thoughtful consideration of alternative explanations for the observed low-frequency temporal structure.

The reviewer raises an important question regarding the extent to which the approximately 5-minute quasi-periodic dynamics reflect processes generated within descending pain-modulatory circuitry versus broader physiological or experimental influences. As discussed in the original manuscript, the present experiments cannot determine the precise mechanistic origin of these dynamics, and we have revised the Discussion to make this distinction more explicit. We now consider possible contributions from autonomic regulation, thermoregulatory processes, anaesthesia-related state modulation, and other slow physiological influences. We also clarify that the NEUTRAL-cell population argues against a uniform global effect acting similarly across all RVM neurons, but cannot exclude systemic influences that preferentially engage ON- and OFF-cell circuitry.

We have additionally expanded the rationale for the lightly anaesthetized preparation. This preparation was not used solely for technical convenience. Stable single-unit recordings from physiologically identified ON- and OFF-cells are technically challenging because the RVM is a deep brainstem structure and these functional cell classes are relatively sparse; suppression of spontaneous movement therefore permits substantially greater recording stability over the prolonged epochs required here. Importantly, the preparation also provides a controlled physiological setting in which the underlying temporal organization of RVM activity can be examined while reducing the continuously changing sensory, motor, arousal, and behavioral influences that would necessarily contribute to RVM activity in an awake animal. Awake preparations are essential for determining how RVM neurons respond during ongoing behavior and natural sensory experience, but that is a complementary question to the one addressed here: whether physiologically identified RVM neurons exhibit structured temporal dynamics under controlled conditions.

We have therefore revised the manuscript to present the lightly anaesthetized preparation as both a methodological choice and an important boundary condition on interpretation. We continue to acknowledge that anaesthesia may influence slow network dynamics, and that future awake recordings will be required to determine how the temporal structure identified here is expressed in the behaving animal. We have also revised the title and several sections of the manuscript to ensure that our conclusions consistently reflect the correlational nature of the data and do not imply mechanistic or causal interpretations beyond those directly supported by the experiments.

(1) Analyses of many of the ON-cells had longer training windows (> 1 sec) compared to those for the NEUTRAL cells. Could this have reduced the ability to fit and validate periodicity for the latter cell type?

We thank the reviewer for raising this important point. The difference between ON/OFFand NEUTRAL-cell analyses reflects the available recording durations rather than differences in the Gaussian process fitting procedure. All cell classes were fitted using the same GP model and training strategy; however, some NEUTRAL-cell recordings were shorter (960 s versus 1500 s for ON- and OFF-cells), resulting in correspondingly shorter training segments. Whilst the minimum frequency recoverable from a 960 s training segment is 0.00104 Hz, meaning that the 0.0033 Hz frequency observed in ON- and OFF-cells would be recoverable if present in a 960 s recording. We agree that shorter recordings could, in principle, reduce the ability to estimate slow periodic structure. We have therefore clarified this point in the Methods and Discussion. Importantly, the absence of predictable low-frequency dynamics in NEUTRAL-cells is supported not only by GP prediction performance but also by the independent power spectral analysis, the low latent GP variance, and the near-flat phase-normalised reconstructions, suggesting that the difference between cell classes is not solely attributable to recording duration. To address this concern, we will revise the manuscript to repeat the GP analysis after truncating the ON- and OFF-cell recordings to match the duration of the NEUTRAL-cell recordings. We will also include a 960-second-long simulated NEUTRAL-cell recording with periodic structure, to demonstrate that this would be located by our method if present.

Reviewer #3 (Public review):

Summary:

In this manuscript by Ashworth and colleagues, the authors investigate the temporal dynamics of the rostral ventromedial medulla (RVM), a key output node in a major descending pain-modulation circuit. Using data from extrasellar single-unit recordings of RVM ON, OFF, and NEUTRAL cells in lightly anesthetized rats, the authors’ computational modeling yielded two major findings: (1) heat-evoked ON burst and OFF pause, followed by exponential recovery components in10s of seconds; and (2) ON and OFF cells exhibit periodic fluctuations in 5-minute cycles that are statistically predictable.

Strengths:

The manuscript’s concept is innovative, offering the first quantitative analysis of multitimescale dynamics in physiologically characterized RVM pain-modulating neurons. This advances a field that has mostly depended on qualitative or single-timescale descriptions. The authors use contemporary Gaussian process and probabilistic models to capture statistically predictable slow dynamics. The study is further strengthened by identifying ON-, OFF-, and NEUTRAL-type cells using well-established criteria grounded in decades of RVM research. The combination of rapid reflex-related responses and slower ongoing rhythms supports a dual-timescale framework, providing a more integrated understanding of how these neurons may regulate reflex activity and state-dependent processes.

Weaknesses:

Several limitations are noted. Incomplete characterization of light anesthesia during recording sessions, such as methohexital stability and clear criteria for identifying “lightly anesthetized” states. While the NEUTRAL cell control is helpful, it does not fully address concerns about circuit specificity or systemic confounds. The findings are male-dominant, which may limit their generalizability. The synchrony between ON and OFF cells was suggested but not directly tested. The heart rate coherence with ON, OFF, and NEUTRAL cell activity results is intriguing but does not fully clarify how these neurons influence heart rate, particularly within the “lightly anesthetized” model.

We thank Reviewer R3 for their thoughtful and constructive assessment of our work. We appreciate the reviewer’s emphasis on providing additional methodological detail and placing the findings within the context and limitations of the experimental preparation. In response, we have substantially expanded the Methods to provide a more complete description of the lightly anaesthetized preparation, the methohexital infusion protocol, physiological monitoring, and the rationale for the ongoing recording paradigm.

We have also clarified why this preparation was appropriate for the question addressed here. In addition to enabling stable long-duration single-unit recordings from sparse, physiologically identified neurons in the deep RVM, the lightly anesthetized preparation provides a controlled physiological setting in which baseline temporal dynamics can be characterized while minimizing ongoing sensory, motor, and behavioral influences. We nevertheless acknowledge that anaesthesia may alter slow brain-state dynamics, and we now make this limitation more explicit throughout the manuscript. We have also revised the Discussion to more clearly acknowledge the predominantly male sample, the interpretation of the NEUTRAL-cell population as a comparison group rather than a definitive control for systemic effects, the limitations of inferring synchrony from pseudo-population data, and the correlational nature of the heart-rate coherence analysis.

(1) How does the lightly anesthetized preparation affect evoked and oscillation activity modeling? Given that cell activities can be highly influenced by the state of sedation and the pharmacology of methohexital, detailing how light anesthesia was achieved and determined can help interpret the limitations of the current model. For example, did the methohexital rate adjustments occur during the ongoing activity period used for GP modeling? What specific criteria defined “lightly anesthetized” beyond stable paw withdrawal latency, such as stable respiratory rate, EMG (reflex vigor?), and core temperature? Given RVM activity coupled to autonomic/thermoregulatory circuits, data on these variables should be reported, or their absence should be acknowledged.

We thank the reviewer for this important comment. We have substantially expanded the Methods and Discussion to describe both the rationale for the lightly anaesthetized preparation and the criteria used to maintain it.

The preparation offers both technical and conceptual advantages for the present question. Technically, the RVM is a deep brainstem structure and physiologically identified ON- and OFF-cells are relatively sparse. Prolonged extracellular recordings therefore depend on maintaining stable electrode–neuron contact, which is readily disrupted by spontaneous movement. Light methohexital anaesthesia suppresses spontaneous movement while preserving nocifensive withdrawal responses and the canonical physiological response patterns used to identify ON-, OFF-, and NEUTRAL-cells. Conceptually, the aim of the present study was to characterize the baseline temporal organization of identified RVM neurons rather than to determine which sensory, cognitive, or behavioral events drive their activity in an awake animal. An awake preparation would necessarily introduce continuously changing sensory input, motor activity, arousal, behavioral priorities, and other internal-state variables, all of which are known to influence RVM activity. These are important influences in their own right, but for the present question they would make it more difficult to distinguish underlying temporal structure from activity driven by ongoing experience. We therefore view controlled lightly anaesthetized and awake preparations as complementary: the former is useful for identifying foundational circuit dynamics under controlled conditions, whereas the latter will be essential for determining how those dynamics are modified and expressed during natural behavior.

This preparation has also been extensively used to establish the canonical relationship between ON-/OFF-cell activity and nocifensive responses, pharmacological modulation of the RVM, and top-down control from structures including the hypothalamus and amygdala, with many of these functional relationships subsequently confirmed in awake behavioral experiments. We have added this context to the revised manuscript.

With respect to physiological monitoring, core temperature was continuously monitored and maintained at 36–37 °C, heart rate was monitored by EKG, and EMG was recorded to monitor withdrawal responses. Light anesthesia was defined functionally by preservation of a stable nocifensive withdrawal response in the absence of spontaneous movement. Respiratory variables were not recorded, and we now acknowledge this explicitly as a limitation. We have also clarified in Methods that the Methohexital rates were not adjusted during the recording windows used for the gaussian process analysis.

We therefore agree that the findings must be interpreted within the context of the lightly anaesthetized preparation, but we do not view awake recordings as a direct substitute for the present experiment. Rather, awake studies provide the important next step of determining how the structured dynamics identified under controlled conditions are modulated by sensory experience, behavioral state, and ongoing cognition.

(2) It is unclear how the absence of slow oscillations in NEUTRAL cells can be used as an internal control for anesthesia and systemic drift. It is unlikely that NEUTRAL cells are identified in every single-cell recording session for them to be used as a consistent internal control. Also, as the authors suggested that the shared modulatory inputs to ON/OFF cells explain the coordinating mechanism for ON/OFF rhythmicity, the lack of rhythmicity or coherence in majority of the NEUTRAL cells may indicate that they do not receive the same modulatory inputs as ON/OFF cells. Would this make NEUTRAL cells insensitive to systemic changes throughout the recording sessions? Do rhythmic vs. non-rhythmic cells differ in location within the RVM?

We appreciate the reviewer’s important distinction. We agree that NEUTRAL-cells should not be considered a definitive internal control for anaesthesia or systemic physiological drift. NEUTRAL-, ON-, and OFF-cells were not necessarily recorded simultaneously within the same session, and the functional classes may differ in the systemic or modulatory inputs they receive. Our intended inference is therefore narrower: the absence of comparable low-frequency temporal structure in most NEUTRAL-cells argues against a uniform global process that imposes the same temporal pattern on all RVM neurons. It does not exclude anaesthesia-related, autonomic, thermoregulatory, or other systemic processes that preferentially influence ON- and OFF-cell circuitry. We have revised the manuscript throughout to make this distinction explicit and now refer to NEUTRAL-cells as an informative comparison population rather than as a definitive control for systemic influences.

Indeed, as the reviewer suggests, differential sensitivity to common modulatory inputs could itself contribute to the distinction between ON/OFF- and NEUTRAL-cell dynamics. This interpretation is also compatible with the observation that a subset of NEUTRAL-cells shows low-frequency coherence with heart rate despite lacking the structured approximately 5-minute temporal dynamics observed in the ON/OFF populations.

We additionally examined the reconstructed recording locations and found no obvious anatomical segregation between neurons showing stronger versus weaker low-frequency structure within the sampled RVM region. We now state this in the revised manuscript. We appreciate the reviewer’s point that there may be locational differences between RVM rhythmic and non-rhythmic cells, which should be addressed in future work; however, determining this would require a substantially larger sample size, for example with multichannel probe recording, for a valid analysis.

(3) It is important to acknowledge that findings are effectively male-only (77 M and 6 F). Although a recent publication demonstrated that RVM ON and OFF cell activities do not differ substantially on an individual level between male and female rats, sex differences in RVM population dynamics remain unexplored. The current finding may not be generalizable to females.

We thank the reviewer for highlighting this important limitation. We now explicitly acknowledge in the Discussion that the present dataset is predominantly male (77 males, 6 females) and therefore does not permit meaningful assessment of sex differences in population dynamics. Although previous studies suggest that individual ON- and OFF-cell responses are broadly comparable between sexes, the generalisability of the present findings to female animals remains unknown and should be addressed in future work.

(4) It was suggested that strong synchrony exists within each functional population (e.g., ON and OFF cells). However, phase-relationship or coherence analyses were lacking. Since there were recoding sessions with > 2 cells/animals, were there enough recordings that contain simultaneous ON/OFF pairs to allow for these analyses?

We thank the reviewer for this helpful suggestion. We agree that direct analyses of synchrony between simultaneously recorded neurons would provide valuable additional information. However, the number of simultaneous recordings containing identifiable ON/OFF-cell pairs was insufficient to support a robust phase or coherence analysis. We have therefore revised the Discussion to avoid implying that synchrony has been directly demonstrated and instead describe the results as evidence for consistent low-frequency temporal structure across recordings. We also identify direct analysis of synchrony in larger simultaneously recorded neuronal populations as an important direction for future work.

(5) It was intriguing that the ON-cell population’s ongoing activity shows a predictive structure, while the OFF-cell population does not (Figure 5). However, this interesting asymmetry in ongoing activity between two cell classes was not adequately explained in the discussion. For example, since shared modulatory inputs were proposed as the coordinating mechanism for ON/OFF rhythmicity, how may this difference in ON and OFF rhythm predictivity occur?

We appreciate the reviewer drawing attention to this interesting observation. We have expanded the Discussion to consider possible explanations for the greater predictability observed in ON-cells relative to OFF-cells. Although both populations exhibited similar dominant timescales, ON-cells displayed larger latent GP variance and more consistent predictive performance, whereas OFF-cells exhibited greater heterogeneity across recordings. We now discuss several possible explanations for this asymmetry, including differences in intrinsic cellular properties, network coupling, or modulation amplitude, while emphasising that the present data do not allow these possibilities to be distinguished.

(6) The relationship between RVM activity oscillations and cardiac rhythms appears to be covariate but may not support the ”physiologically meaningful” claim with the current analysis. Additional discussion could help clarify the findings of a) how the RVM oscillation period of 300s relates to the heart rate peak/oscillation period of 600s (Figure 6d) and b) how NEUTRAL cells show heart rate coherence but lack rhythmicity.

We thank the reviewer for this thoughtful comment. We have revised the Discussion to more carefully interpret the heart-rate coherence analysis. In particular, we now emphasise that the observed coherence demonstrates shared low-frequency temporal structure but does not establish a causal relationship between RVM activity and cardiac dynamics. We also discuss the relationship between the approximately 300-s RVM timescale and the broader low-frequency components observed in the heart-rate spectrum, noting the limited frequency resolution available at these timescales. Finally, we expand our discussion of the NEUTRAL-cell results, clarifying that significant coherence in NEUTRAL-cells despite the absence of comparable structured low-frequency firing dynamics is consistent with shared physiological influences acting on multiple cell classes without implying that the slow temporal structure originates within NEUTRAL-cells.

References

(1) Noble, D. The Music of Life: Biology beyond the Genome (Oxford University Press, 2006).

(2) Heinricher, M. M., Barbaro, N. M. & Fields, H. L. Putative Nociceptive Modulating Neurons in the Rostral Ventromedial Medulla of the Rat: Firing of On- and Off-Cells Is Related to Nociceptive Responsiveness. Somatosensory & motor research 6, 427–39 (1989).

(3) Barbaro, N. M., Heinricher, M. M. & Fields, H. L. Putative Nociceptive Modulatory Neurons in the Rostral Ventromedial Medulla of the Rat Display Highly Correlated Firing Patterns. Somatosensory & Motor Research 6, 413–425 (1989).

(4) Heinricher, M. M., Haws, C. M. & Fields, H. L. Evidence for GABA-mediated control of putative nociceptive modulating neurons in the rostral ventromedial medulla: Iontophoresis of bicuculline eliminates the off-cell pause. Somatosensory & Motor Research 8, 215–225 (1991).

(5) Heinricher, M. M. & Kaplan, H. J. GABA-mediated inhibition in rostral ventromedial medulla: Role in nociceptive modulation in the lightly anesthetized rat. Pain 47, 105–113 (1991).

(6) Heinricher, M. M. & Tortorici, V. Interference with GABA transmission in the rostral ventromedial medulla: Disinhibition of off-cells as a central mechanism in nociceptive modulation. Neuroscience 63, 533–546 (1994).

(7) Heinricher, M. M., McGaraughty, S. & Grandy, D. K. Circuitry Underlying AntiOpioid Actions of Orphanin FQ in the Rostral Ventromedial Medulla. Journal of Neurophysiology 78, 3351– 3358 (1997).

(8) Heinricher, M. M., McGaraughty, S. & Farr, D. A. The role of excitatory amino acid transmission within the rostral ventromedial medulla in the antinociceptive actions of systemically administered morphine. Pain 81, 57–65 (1999).

(9) Heinricher, M. M., McGaraughty, S. & Tortorici, V. Circuitry Underlying Antiopioid Actions of Cholecystokinin Within the Rostral Ventromedial Medulla. Journal of Neurophysiology 85, 280–286 (2001).

(10) Heinricher, M. M., Schouten, J. C. & Jobst, E. E. Activation of brainstem N-methyl-daspartate receptors is required for the analgesic actions of morphine given systemically. Pain 92, 129–138 (2001).

(11) McGaraughty, S. & Heinricher, M. M. Microinjection of morphine into various amygdaloid nuclei differentially affects nociceptive responsiveness and RVM neuronal activity. Pain 96, 153–162 (2002).

(12) Heinricher, M. M. & Neubert, M. J. Neural Basis for the Hyperalgesic Action of Cholecystokinin in the Rostral Ventromedial Medulla. Journal of Neurophysiology 92, 1982–1989 (2004).

(13) Heinricher, M. M., Martenson, M. E. & Neubert, M. J. Prostaglandin E2 in the midbrain periaqueductal gray produces hyperalgesia and activates pain-modulating circuitry in the rostral ventromedial medulla. Pain 110, 419–426 (2004).

(14) Heinricher, M. M., Neubert, M. J., Martenson, M. E. & Gonc¸alves, L. Prostaglandin E2 in the medial preoptic area produces hyperalgesia and activates pain-modulating circuitry in the rostral ventromedial medulla. Neuroscience 128, 389–398 (2004).

(15) Kincaid, W., Neubert, M. J., Xu, M., Kim, C. J. & Heinricher, M. M. Role for Medullary Pain Facilitating Neurons in Secondary Thermal Hyperalgesia. Journal of Neurophysiology 95, 33–41 (2006).

(16) Ortiz, J., Heinricher, M. & Selden, N. Noradrenergic agonist administration into the central nucleus of the amygdala increases the tail-flick latency in lightly anesthetized rats. Neuroscience 148, 737–743 (2007).

(17) Xu, M., Kim, C. J., Neubert, M. J. & Heinricher, M. M. NMDA receptor-mediated activation of medullary pro-nociceptive neurons is required for secondary thermal hyperalgesia. PAIN 127, 253 (2007).

(18) Ortiz, J. P., Close, L. N., Heinricher, M. M. & Selden, N. R. α2-Noradrenergic antagonist administration into the central nucleus of the amygdala blocks stress-induced hypoalgesia in awake behaving rats. Neuroscience 157, 223–228 (2008).

(19) Edelmayer, R. M. et al. Medullary pain facilitating neurons mediate allodynia in headache-related pain. Annals of Neurology 65, 184–193 (2009).

(20) Martenson, M. E., Cetas, J. S. & Heinricher, M. M. A possible neural basis for stress-induced hyperalgesia. Pain 142, 236–244 (2009).

(21) Heinricher, M. M., Maire, J. J., Lee, D., Nalwalk, J. W. & Hough, L. B. Physiological Basis for Inhibition of Morphine and Improgan Antinociception by CC12, a P450 Epoxygenase Inhibitor. Journal of Neurophysiology 104, 3222–3230 (2010).

(22) Heinricher, M. M., Martenson, M. E., Nalwalk, J. W. & Hough, L. B. Neural basis for improgan antinociception. Neuroscience 169, 1414–1420 (2010).

(23) McGaraughty, S., Farr, D. A. & Heinricher, M. M. Lesions of the periaqueductal gray disrupt input to the rostral ventromedial medulla following microinjections of morphine into the medial or basolateral nuclei of the amygdala. Brain Research 1009, 223–227 (2004).

(24) Rogness, V. M. et al. Descending Facilitation of Nociceptive Transmission From the Rostral Ventromedial Medulla Contributes to Hyperalgesia in Mice with Sickle Cell Disease. Neuroscience 526, 1–12 (2023).

(25) Smith, D. J. et al. Dose-Dependent Pain-Facilitatory and -Inhibitory Actions of Neurotensin Are Revealed by SR 48692, a Nonpeptide Neurotensin Antagonist: Influence on the Antinociceptive Effect of Morphine1,2. The Journal of Pharmacology and Experimental Therapeutics 282, 899–908 (1997).

(26) Hurley, R. W. & Hammond, D. L. The Analgesic Effects of Supraspinal µ and δ Opioid Receptor Agonists Are Potentiated during Persistent Inflammation. Journal of Neuroscience 20, 1249–1259 (2000).

(27) Kovelowski, C. J. et al. Supraspinal cholecystokinin may drive tonic descending facilitation mechanisms to maintain neuropathic pain in the rat. Pain 87, 265–273 (2000).

(28) Porreca, F. et al. Inhibition of Neuropathic Pain by Selective Ablation of Brainstem Medullary Cells Expressing the µ-Opioid Receptor. Journal of Neuroscience 21, 5281–5288 (2001).

(29) Zhang, Y. et al. Identifying local and descending inputs for primary sensory neurons. Journal of Clinical Investigation 125, 3782–3795 (2015).

(30) Franc¸ois, A. et al. A Brainstem-Spinal Cord Inhibitory Circuit for Mechanical Pain Modulation by GABA and Enkephalins. Neuron 93, 822–839.e6 (2017). URL https://www.ncbi.nlm. nih.gov/pmc/articles/PMC7354674/.

(31) Kim, J.-H. et al. Yin-and-yang bifurcation of opioidergic circuits for descending analgesia at the midbrain of the mouse. Proceedings of the National Academy of Sciences 115, 11078–11083 (2018).

(32) Nguyen, E. et al. Medullary kappa-opioid receptor neurons inhibit pain and itch through a descending circuit. Brain 145, 2586–2601 (2022).

(33) Jiao, Y. et al. Molecular identification of bulbospinal ON neurons by GPER, which drives pain and morphine tolerance. The Journal of Clinical Investigation 133, e154588 (2023).

(34) Nguyen, E., Grajales-Reyes, J. G., Gereau, R. W. & Ross, S. E. Cell type-specific dissection of sensory pathways involved in descending modulation. Trends in Neurosciences 46, 539–550 (2023).

(35) Fatt, M. P. et al. Morphine-responsive neurons that regulate mechanical antinociception. Science 385, eado6593 (2024).

(36) Wang, Q. et al. Deconstruction of a spino-brain–spinal cord circuit that drives chronic pain. Nature 1–10 (2026).

(37) Brooks, J. C., Davies, W.-E. & Pickering, A. E. Resolving the Brainstem Contributions to Attentional Analgesia. The Journal of Neuroscience 37, 2279–2291 (2017).

(38) Mills, E. P. et al. Brainstem Pain-Control Circuitry Connectivity in Chronic Neuropathic Pain. Journal of Neuroscience 38, 465–473 (2018).

(39) Mills, E. P., Keay, K. A. & Henderson, L. A. Brainstem Pain-Modulation Circuitry and Its Plasticity in Neuropathic Pain: Insights From Human Brain Imaging Investigations. Frontiers in Pain Research (Lausanne, Switzerland) 2, 705345 (2021).

(40) Oliva, V., Hartley-Davies, R., Moran, R., Pickering, A. E. & Brooks, J. C. Simultaneous brain, brainstem, and spinal cord pharmacological-fMRI reveals involvement of an endogenous opioid network in attentional analgesia. eLife 11, e71877 (2022).

eLife Assessment

This important technical study introduces SCOPE, an optics-free spatial reconstruction method based on bidirectional sender and receiver oligonucleotides on barcoded hydrogel beads. By sequencing proximity-encoded chimeric molecules, the authors computationally reconstruct 2D and 3D spatial information at an impressive scale. The technical demonstrations in synthetic bead systems are convincing and establish proof-of-principle that large spatial domains can be reconstructed without microscopy. The methodological advance is clear and the scale is impressive. This work will be of interest to those working on spatial mapping.

Reviewer #1 (Public review):

Summary:

Liao et al. present SCOPE (Spatial reConstruction via Oligonucleotide Proximity Encoding), a method for reconstructing spatial organization from diffusion-defined DNA barcode interactions without the use of optical imaging. In SCOPE, hydrogel beads bearing unique DNA barcodes contain both "sender" and "receiver" oligonucleotides. Upon enzymatic release, sender oligos diffuse locally and hybridize to receiver oligos on neighboring beads, forming chimeric molecules that encode spatial proximity. Sequencing these products yields an interaction matrix, which is then used to reconstruct a spatial coordinate map.

The authors demonstrate reconstruction of synthetic two-dimensional shapes, a large multicolor Snellen eye chart, and the interior surface of three-dimensional molds. The work expands the conceptual and experimental landscape of optics-free spatial sequencing.

Strengths:

SCOPE employs bidirectional sender and receiver oligonucleotides on every bead, rather than using asymmetric transmitter-receiver architectures found in other diffusion-based methods. The symmetric design may improve detection sensitivity and reconstruction strategies, and represents a meaningful variation on optics-free spatial encoding.

A notable strength of this study is the physical scale achieved. The authors reconstruct a Snellen chart spanning approximately 704 mm² and demonstrate molded 3D structures on the order of 75-100 mm³. Although some larger-scale warping is evident, and is discussed as potentially due to non-uniform diffusion, the relative local positioning across these large areas appears impressively accurate.

The authors extend reconstruction beyond two-dimensional arrays to three-dimensional molded surfaces. This demonstrates that the assay and the computational methods for interpreting proximity graphs can support non-planar spatial relationships, expanding the scope of optics-free spatial inference.

The revised manuscript also strengthens the interpretation of these three-dimensional experiments by providing additional evidence that the limited recovery of interior regions primarily reflects technical constraints associated with molecular recovery from the hydrogel beads on the interior rather than an inherent limitation of the SCOPE framework.

Weaknesses:

Although the method is discussed in the context of spatial genomics and potential tissue applications, it is currently demonstrated only on engineered two-dimensional bead arrays and three-dimensional shapes fabricated in molds. The authors appropriately acknowledge that additional work will be required to establish performance in heterogeneous biological tissues, where diffusion and molecular recovery are likely to be more complex.

The revised manuscript substantially clarifies the limitations of the current three-dimensional implementation by providing additional discussion and control experiments supporting the interpretation that reduced recovery of interior beads is primarily a technical limitation of the present protocol. Although limited volumetric sampling remains a current constraint of the method, the authors appropriately discuss applications in which surface-resolved reconstruction may still be informative.

The computational reconstruction workflow is now described in greater detail, including automated parameter selection, fixed versus optimized hyperparameters, and explicit discussion of the manual flattening step required for the largest Snellen reconstruction. The revised manuscript also explains that many anticipated tissue-section applications present a more constrained reconstruction problem than the intentionally challenging proof-of-concept demonstrations presented here, providing additional context for the expected performance of the method in future biological applications.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

Liao et al. present SCOPE (Spatial reConstruction via Oligonucleotide Proximity Encoding), a method for reconstructing spatial organization from diffusion-defined DNA barcode interactions without the use of optical imaging. In SCOPE, hydrogel beads bearing unique DNA barcodes contain both "sender" and "receiver" oligonucleotides. Upon enzymatic release, sender oligos diffuse locally and hybridize to receiver oligos on neighboring beads, forming chimeric molecules that encode spatial proximity. Sequencing these products yields an interaction matrix, which is then used to reconstruct a spatial coordinate map.

The authors demonstrate reconstruction of synthetic two-dimensional shapes, a large multicolor Snellen eye chart, and the interior surface of three-dimensional molds. The work expands the conceptual and experimental landscape of optics-free spatial sequencing.

Thank you for this accurate summary of the work.

Strengths:

SCOPE employs bidirectional sender and receiver oligonucleotides on every bead, rather than using asymmetric transmitter-receiver architectures found in other diffusion-based methods. The symmetric design may improve detection sensitivity and reconstruction strategies, and represents a meaningful variation on optics-free spatial encoding.

A notable strength of this study is the physical scale achieved. The authors reconstruct a Snellen chart spanning approximately 704 mm² and demonstrate molded 3D structures on the order of 75-100 mm³. Although some larger-scale warping is evident, and is discussed as potentially due to non-uniform diffusion, the relative local positioning across these large areas appears impressively accurate.

The authors extend reconstruction beyond two-dimensional arrays to three-dimensional molded surfaces. This demonstrates that the assay and the computational methods for interpreting proximity graphs can support nonplanar spatial relationships, expanding the scope of optics-free spatial inference.

Thank you for highlighting these strengths of SCOPE.

Weaknesses:

Although the method is discussed in the context of spatial genomics and potential tissue applications, it is currently demonstrated only on engineered two-dimensional bead arrays and three-dimensional shapes fabricated in molds. It remains unclear how SCOPE would perform in heterogeneous biological environments, where diffusion may exhibit additional non-uniformities. A biological proof-of-concept, even limited in scope, would help define the method's strengths and limitations more clearly.

We concur with the reviewer that a biological proof-of-concept is a key next step, and that diffusion will be more heterogeneous in this more complex environment. To this end, we are actively working to further develop SCOPE for use in tissue sections, with the goal of capturing transcriptomes, accessible chromatin, and genomes. As part of this work, we also hope to systematically explore a range of tissue permeabilization and tissue clearing approaches to mitigate the impact of heterogeneity on performance.

The reconstruction of three-dimensional structures lacks strong sampling from volume interiors. This is speculated to be due to several possible factors; however, this limitation constrains the method to reconstruction of volume surfaces rather than comprehensive three-dimensional profiling.

Thank you for highlighting this important limitation. The 3D reconstructions are indeed constrained by undersampling of volume interiors. We anticipate that this might be addressed via relatively minor adjustments to the protocol, e.g. using light- or base-labile linkers to trigger oligo release, with the expectation that this will improve reaction consistency throughout the volume. However, even if we are unable to resolve this issue, we note that surface-resolved reconstructions may be useful for some goals, e.g. embedding a bead-packed gel within a tissue lumen, such as the gut. This could enable surface beads to capture RNA transcripts from adjacent cells, while bead–bead associations serve to define the surface topology.

The reconstruction workflow involves multiple preprocessing steps and embedding choices. While these appear to work well for synthetic shapes with known geometry, it is less clear how parameter choices would be made in contexts where ground truth is unknown. Clarifying how reconstruction robustness is assessed without prior knowledge of spatial structure would help readers understand how the method could be practically deployed, particularly in more heterogeneous tissue contexts.

Thank you for the opportunity to clarify. The computational pipeline used for 2D SCOPE reconstruction is designed to operate on a standardized input format and can be applied to arbitrary datasets without prior knowledge of spatial structure. For example, as shown in Figure 3, both the circle and “swoosh” geometries were reconstructed using the same algorithm and identical initial parameters. While certain hyperparameters are pre-specified (e.g. the number of k-nearest neighbours used to compute the pairwise distance matrix for UMAP), these are fixed across datasets. Other parameters, such as UMAP’s “min_dist,” are selected via an automated heuristic grid search that proceeds without user intervention. The agreement with ground truth in these controlled settings, together with the reproducibility of stochastic reconstructions (see Figure 3E-F), supports the robustness of the approach.

Importantly, there was one exception. Reconstruction of the Snellen eye chart dataset required a manual step, involving an initial 3D UMAP embedding followed by a 2D projection to “flatten” the result. We suspect this reflects radial non-uniformities in sender/receiver oligo diffusion at larger spatial scales. Addressing such confounders algorithmically by explicitly modelling diffusion heterogeneity represents an important area for future work, with the goal of entirely eliminating the need for manual intervention.

Finally, we note that these benchmark shapes represent somewhat contrived examples, and the geometries encountered in practice may often be much less complex. For example, in conventional spatial genomics, the geometry consists of a bead monolayer forming a flat, regular surface on a rectangular slide of known dimensions. Regardless of the tissue architecture overlaid on this surface, the reconstruction problem is defined by the bead monolayer itself, inferred through sender-receiver interactions.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

It would be helpful to further clarify the limitations in interior sampling of three-dimensional structures by providing a more explicit comparison with Qian and Weinstein's volumetric DNA microscopy (UMI-UEI) approach. In particular, do the authors anticipate that the current limitation in SCOPE's interior sampling can be mitigated through experimental optimization, or might this represent an inherent challenge associated with hydrogel-based bead scaffolds relative to substrate-free approaches? A more detailed discussion of this point would help readers understand whether the observed volumetric constraint is technical and potentially solvable, or structural to the platform design.

We thank the reviewer for this suggestion and agree that a more explicit comparison to volumetric DNA microscopy helps clarify the origin of the current limitation in interior sampling. Based on our experiments to date, we view this constraint as primarily technical and, in principle, addressable.

A key distinction between SCOPE and volumetric DNA microscopy(Qian and Weinstein, 2025), using the UMI– UEI framework, lies in the recovery of recorded molecules from the interior of the sample. In volumetric DNA microscopy, the hydrogel-embedded specimen can be fully digested and treated with Proteinase K, enabling efficient liberation and recovery of molecules throughout the volume for downstream sequencing(Qian et al., 2026). In contrast, in the current implementation of SCOPE, we do not dissolve the polyacrylamide hydrogel, and recovery therefore relies largely on diffusion of chimeric molecules out of the scaffold. This likely biases against molecules generated in the interior and leads to reduced sampling of internal regions. This interpretation is supported by a control experiment in which barcoded beads were allowed to settle in solution at the bottom of a tube in the absence of a polymerized hydrogel scaffold. In this setting, 3D UMAP reconstruction yielded a solid, non-hollow structure consistent with the expected conical geometry of the tube bottom, indicating that SCOPE is capable of recovering volumetric structure when recovery is not diffusion-limited. Taken together, these observations suggest that the apparent “hollowing” in current 3D reconstructions reflects a limitation in molecule recovery from hydrogel scaffolds, rather than an inherent constraint of the SCOPE framework itself.

We are currently exploring potential solutions, including the use of reducible crosslinkers to enable hydrogel dissolution and/or mechanical shearing of the gel. If these experiments are successful, we would plan to include the results in revisions to the manuscript, together with an appropriately edited version of the paragraph above. If they are unsuccessful and major experimental effort is going to be required to address this issue, we would likely move forward with textual changes only, incorporating the points made in the paragraph above into the discussion.

References

Qian N, Li J, Yasser R, Yu M, Weinstein JA. 2026. Volumetric DNA microscopy for mapping spatial transcriptomes in three dimensions. Nat Protoc. doi:10.1038/s41596-025-01329-3

Qian N, Weinstein JA. 2025. Spatial transcriptomic imaging of an intact organism using volumetric DNA microscopy. Nat Biotechnol 1–11.

eLife Assessment

This important study investigates how sleep loss and circadian disruption affect whole-organ metabolism in flies (Drosophila melanogaster) and reports that wild-type flies align metabolism in anticipation of diurnal rhythm, while mutant flies with impaired sleep or circadian function shift to reactive or misaligned metabolism. The integration of chamber-based flow-through respirometry with LC-MS metabolomics is innovative, and the significance of the findings is significant. The strength of evidence needed to support the conclusions is generally convincing, although the absence of direct measures of locomotor activity makes it difficult to separate intrinsic metabolic changes from potential behavioral differences in mutants.

Reviewer #1 (Public review):

Summary:

This study by Akhtar et al. aims to investigate the link between systemic metabolism and respiratory demands, and how sleep and circadian clock regulate metabolic states and respiratory dynamics. The authors leverage genetic mutants that are defective in sleep and circadian behavior in combination with indirect respirometry and steady-state LC-MS-based metabolomics to address this question in the Drosophila model.

First, the authors performed respirometry (on groups of 25 flies) to measure oxygen consumption (VO2) and carbon dioxide production (VCO2) to calculate the respiratory quotient (RQ) across the 24-hour day (12h:12h light-dark cycle) and assess metabolic fuel utilization. They observed that among all the genotypes tested, wild type (WT) flies and per0 flies in LD and WT flies in DD exhibit RQ >1. They concluded the >1 RQ is consistent with active lipogenesis. In contrast, the short-sleep mutants fumin (fmn) and sleepless (sss) showed significantly different RQ; the fmn exhibits a slight reduction in RQ values, suggesting increased reliance on carbohydrate metabolism, while sss exhibit even lower RQ (0.94), consistent with a shift toward lipid and protein catabolism.

The authors then proceeded to bin these measurements in 12-hour partitions, ZT0-12 and ZT12-24, to assess diurnal differences in average values of VO2, VCO2, and RQ. They observed significant day-night differences in metabolic rates in WT-LD flies, with higher rates during the day. The diurnal differences remain in the short-sleep mutants, but the overall metabolic rates are higher. WT-DD flies exhibit the lowest respiratory activity although the day-night differences remain in free-running conditions. Finally, per01 mutants exhibit no significant change in day-night respiratory rates, suggesting that a functional circadian clock is necessary for diurnal differences in metabolic rates.

They then performed finer resolution 24-hours rhythmic analysis (RAIN and JTK) to determine if VO2, VCO2, and RQ exhibit 24-hour rhythmic and if there are genotype-specific differences. Based on their criteria, VCO2 is rhythmic in all conditions tested while VO2 is rhythmic in all conditions except in fmn-LD. Finally, RQ is rhythmic in all 3 mutants but not in WT-LD and WT-DD. Peak phases for the rhythms were deduced using JTK lag values.

The authors proceeded to leverage a previously published steady-state metabolite datasets to investigate potential association of RQ with metabolite profiles. Spearman correlation was performed to identify metabolites that exhibit coupling to respiratory output. Positive and negative lag analysis were subsequently performed to further characterize these associations based on the timing of the metabolite peak changes relative to RQ fluctuations. The authors suggest that a positive lag indicates that metabolite changes occur after shifts in RQ, and a negative lag signifies that metabolite changes precede RQ changes. To visualize metabolic pathways that exhibit these temporal relationships, clustered heatmap and enrichment analysis were performed. Through these analyses, they concluded that both sleep and circadian systems are essential for aligning metabolic substrate selection with energy demands, and different metabolic pathways are misregulated in the different mutants with sleep and circadian defects.

Strength:

The research questions this study explore are significant given metabolism and respiratory demand are central to animal biology. The experimental methods used, including the well characterized fly genetic mutants, the newly developed method for indirect calorimetry measurements, and LC-MS based metabolomics, are all appropriate. This study provides insights into the impact of sleep and circadian rhythm disruption on metabolism and respiratory demand and serves as a foundation for future mechanistic investigations.

Comments on revised version.

The authors have thoughtfully revised the manuscript. They have now provided clarifications regarding perceived conceptual flaws in the original version. They have also performed additional data analysis to support their conclusions and provide clarifications on statistical methods when appropriate. Overall, the revised manuscript is much improved and the conclusions are generally well supported by their results.

Reviewer #2 (Public review):

This is an innovative and technically strong study that integrates dual-gas respirometry with LC-MS metabolomics to examine how sleep and circadian disruption shape metabolism in Drosophila. The combination of continuous O₂/CO₂ measurements with high-temporal-resolution metabolite profiling is novel and provides fresh insight into how wild-type flies maintain anticipatory fuel alignment, while mutants shift to reactive or misaligned metabolism. The use of lag-shift correlation analysis is particularly clever, as it highlights temporal coordination rather than static associations. Together, the findings advance our understanding of how circadian clocks and sleep contribute to metabolic efficiency and redox balance.

However, there are several areas where the manuscript could be strengthened. The authors should acknowledge that their findings may be gene-specific. Because sleep deprivation was not performed, it remains uncertain whether the observed metabolic shifts generalize to sleep loss broadly or are restricted to the fmn and sss mutants. This concern also connects to the finding of metabolic misalignment under constant darkness despite an intact clock. The conclusion that external entrainment is essential for maintaining energy homeostasis in flies may not translate to mammals. It would help to reference supporting data for the finding and discuss differences across species. Ideally, complementary circadian (light-dark cycle disruption) or sleep deprivation (for several hours) experiments, or citation of comparable studies, would strengthen the generality of the findings. Figures 1-4 are straightforward and clear, but when the manuscript transitions to the metabolite-respiration correlations, there is little description of the metabolomics methods or datasets, which should be clarified. The Discussion is at times repetitive and could be tightened, with the main message (i.e., wild-type flies align metabolism in advance, while mutants do not) kept front and center. Terms such as "anticipatory" and "reactive" should be defined early and used consistently throughout.

Overall, this is a strong and novel contribution. With clarification of scope, refinement of presentation, and a more focused Discussion, the paper will make a significant impact.

Comments on revised version.

The authors have satisfactorily addressed my concerns in the revised manuscript

Reviewer #3 (Public review):

Summary:

The authors investigate how sleep loss and circadian disruption affect whole-organism metabolism in Drosophila melanogaster. They used chamber-based flow-through respirometry to measure oxygen consumption, carbon dioxide production, in wild-type flies and in mutants with impaired sleep or circadian function. These measurements were then integrated with a previously published metabolomics dataset to explore how respiratory dynamics align with metabolic pathways. The central claim is that wild-type flies display anticipatory coordination of metabolic processes with circadian time, while mutants exhibit reactive shifts in substrate use, redox imbalance, and signs of mitochondrial stress.

Strengths:

The study has several strengths. Continuous high-resolution respirometry in flies is challenging, and its application across multiple genotypes provides good comparative insight. The conceptual framework distinguishing anticipatory from reactive metabolic regulation is interesting. The translational framing helps place the work in a broader context of sleep, circadian biology, and metabolic health.

Weaknesses:

At the same time, the evidence supporting the conclusions is somewhat limited. The metabolomics data were not newly generated but repurposed from prior work, reducing novelty. The biological replication in the respirometry assays is low, with only a small number of chambers per genotype. Importantly, respiratory parameters in flies are strongly influenced by locomotor activity, yet no direct measurements of activity were included, making it difficult to separate intrinsic metabolic changes from behavioral differences in mutants. In addition, repeated claims of "mitochondrial stress" are not directly substantiated by assays of mitochondrial function. The study also excluded female flies entirely, despite well-documented sex differences in metabolism, which narrows the generality of the findings.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

This study by Akhtar et al. aims to investigate the link between systemic metabolism and respiratory demands, and how sleep and the circadian clock regulate metabolic states and respiratory dynamics. The authors leverage genetic mutants that are defective in sleep and circadian behavior in combination with indirect respirometry and steady-state LC-MS-based metabolomics to address this question in the Drosophila model.

First, the authors performed respirometry (on groups of 25 flies) to measure oxygen consumption (VO2) and carbon dioxide production (VCO2) to calculate the respiratory quotient (RQ) across the 24-hour day (12h:12h light-dark cycle) and assess metabolic fuel utilization. They observed that among all the genotypes tested, wild type (WT) flies and per0 flies in LD and WT flies in DD exhibit RQ >1. They concluded the >1 RQ is consistent with active lipogenesis. In contrast, the short-sleep mutants fumin (fmn) and sleepless (sss) showed significantly different RQ; the fmn exhibits a slight reduction in RQ values, suggesting increased reliance on carbohydrate metabolism, while sss exhibits even lower RQ (0.94), consistent with a shift toward lipid and protein catabolism.

The authors then proceeded to bin these measurements in 12-hour partitions, ZT0-12 and ZT12-24, to assess diurnal differences in average values of VO2, VCO2, and RQ. They observed significant day-night differences in metabolic rates in WT-LD flies, with higher rates during the day. The diurnal differences remain in the short-sleep mutants, but the overall metabolic rates are higher. WT-DD flies exhibit the lowest respiratory activity, although the day-night differences remain in free-running conditions. Finally, per01 mutants exhibit no significant change in day-night respiratory rates, suggesting that a functional circadian clock is necessary for diurnal differences in metabolic rates.

They then performed finer-resolution 24-hour rhythmic analysis (RAIN and JTK) to determine if VO2, VCO2, and RQ exhibit 24-hour rhythmic and if there are genotypespecific differences. Based on their criteria, VCO2 is rhythmic in all conditions tested, while VO2 is rhythmic in all conditions except in fmn-LD. Finally, RQ is rhythmic in all 3 mutants but not in WT-LD and WT-DD. Peak phases for the rhythms were deduced using JTK lag values.

The authors proceeded to leverage a previously published steady-state metabolite dataset to investigate the potential association of RQ with metabolite profiles. Spearman correlation was performed to identify metabolites that exhibit coupling to respiratory output. Positive and negative lag analysis were subsequently performed to further characterize these associations based on the timing of the metabolite peak changes relative to RQ fluctuations. The authors suggest that a positive lag indicates that metabolite changes occur after shifts in RQ, and a negative lag signifies that metabolite changes precede RQ changes. To visualize metabolic pathways that exhibit these temporal relationships, a clustered heatmap and enrichment analysis were performed. Through these analyses, they concluded that both sleep and circadian systems are essential for aligning metabolic substrate selection with energy demands, and different metabolic pathways are mis regulated in the different mutants with sleep and circadian defects.

We thank the reviewer for summarizing the contributions made by this manuscript.

Strength:

The research questions this study explores are significant, given that metabolism and respiratory demand are central to animal biology. The experimental methods used, including the well-characterized fly genetic mutants, the newly developed method for indirect calorimetry measurements, and LC-MS-based metabolomics, are all appropriate. This study provides insights into the impact of sleep and circadian rhythm disruption on metabolism and respiratory demand and serves as a foundation for future mechanistic investigations.

We thank the reviewer for the positive comments.

Weaknesses:

There are some conceptual flaws that the authors need to address regarding circadian biology, and some of the conclusions can be better supported by additional analysis to provide a stronger foundation for future functional investigation.

At times, the methods, especially the statistical analysis, are not well articulated; they need to be better explained.

Thank you for this suggestion and have revised and we expanded the Methods and figure legends to improve transparency and reproducibility.

Specifically, we have:

(i) Strengthened the rhythmicity description by specifying that rhythmicity was assessed in Nitecap using the RAIN algorithm with FDR-adjusted p-values (significant p ≤ 0.05; trending 0.05–0.1), and that period and peak phase were estimated with JTK_CYCLE (JTK lag = peak phase), with per-genotype period, phase, and p-values reported in Table 1;

(ii) Clarified sample size and replication by adding these lines to the methods section and indicating sample sizes in figure legends. Figure legends now report n (chambers) and SEM.

“Each genotype measurement represents an average of ~300 flies (25 flies per chamber × 4 chambers per experiment × 3 experimental days). The chamber was treated as the experimental unit for all analyses.”

In addition, we have expanded the description of metabolomics-respirometry correlation analyses to include dataset structure, time-matching across ZT, normalization steps, the use of Spearman correlations, and interpretation of lagged associations.

(iii) We have added these lines to the methods section:

“To integrate respirometry with metabolomics, RQ was recorded continuously at 1-second resolution and averaged into 5-minute bins. Because steady-state metabolite measurements were acquired at 2-hour intervals, we extracted the RQ values corresponding to each 2-hour Zeitgeber Time (ZT) sampling point from the 5-minutebinned dataset to generate time-matched RQ-metabolite pairs. We additionally evaluated temporal relationships using a lag analysis by systematically shifting the RQ time series relative to the metabolite time points (−120, −60, −30, −15, −5, +5, +15, +30, +60, and +120 minutes). Metabolite abundances were normalized as described above, and associations between RQ and individual metabolites were quantified using Spearman rank correlations (ρ) at each lag. Metabolites showing strong associations (e.g., |ρ| > 0.7 with nominal p < 0.05) were carried forward for visualization and summary, and lag direction was interpreted as metabolites preceding (negative lag) or following (positive lag) changes in RQ.”

Reviewer #2 (Public review):

This is an innovative and technically strong study that integrates dual-gas respirometry with LC-MS metabolomics to examine how sleep and circadian disruption shape metabolism in Drosophila. The combination of continuous O₂/CO₂ measurements with high-temporal-resolution metabolite profiling is novel and provides fresh insight into how wild-type flies maintain anticipatory fuel alignment, while mutants shift to reactive or misaligned metabolism. The use of lag-shift correlation analysis is particularly clever, as it highlights temporal coordination rather than static associations. Together, the findings advance our understanding of how circadian clocks and sleep contribute to metabolic efficiency and redox balance.

We thank the reviewer for the positive comments.

However, there are several areas where the manuscript could be strengthened.

The authors should acknowledge that their findings may be gene specific. Because sleep deprivation was not performed, it remains uncertain whether the observed metabolic shifts generalize to sleep loss broadly or are restricted to the fmn and sss mutants. This concern also connects to the finding of metabolic misalignment under constant darkness despite an intact clock.

We agree that our findings should be framed as genotype- and condition-specific. The phenotypes we report arise from chronic, genetically encoded sleep loss (fmn, sss) and clock loss (per01); because acute sleep deprivation was not performed, we do not claim these effects generalize to sleep loss broadly. This also bears on the reviewer's point about constant darkness: the metabolic misalignment we observe in WT-DD occurs despite an intact clock, and we therefore interpret it as a consequence of removing external light-dark cues under our conditions. We have scoped the claims accordingly in the Abstract, Results, and Discussion (subsection “Metabolic Desynchrony and Redox Imbalance in Wild-Type Flies Under Constant Darkness, DD”).

The text now reads as follows:

“We restrict our conclusions to the genotypes and conditions tested (fmn, sss, and per⁰¹), and we do not generalize these effects to acute sleep deprivation because sleep deprivation was not performed in this study. Accordingly, the ‘metabolic misalignment’ observed in constant darkness (DD) likely results from a decrease in synchrony due to the removal of external light:dark cues.”

The conclusion that external entrainment is essential for maintaining energy homeostasis in flies may not translate to mammals. It would help to reference supporting data for the finding and discuss differences across species. Ideally, complementary circadian (lightdark cycle disruption) or sleep deprivation (for several hours) experiments, or citation of comparable studies, would strengthen the generality of the findings.

Thank you. We have tempered the interpretation and expanded both the discussion and its citations. We now (i) avoid stating that external entrainment is universally “essential” for energy homeostasis, (ii) explicitly discuss fly-mammal differences (sleep architecture, thermoregulation, feeding control, and entrainment mechanisms), and (iii) anchor the translational comparison to the mammalian circadian-misalignment and sleep-loss literature already integrated in our Discussion (refs [3, 43-46]), noting that establishing cross-species generality will require additional paradigms (constant light, acute sleep deprivation).

The text now reads as follows:

“These phenotypes parallel mammalian systems, where sleep loss and circadian misalignment are linked to elevated basal metabolic rate, a shift toward carbohydrate oxidation and lipid/protein catabolism, and blunted, phase-shifted respiratory oscillations [3, 43-46]; physiological differences between flies and mammals nonetheless caution against direct mechanistic extrapolation. In constant darkness, our DD data show that endogenous free-running regulation persists but that removing external light-dark cues degrades temporal coordination between respiration and metabolism; however, we acknowledge that the coupling may be different in mammals.”

Figures 1-4 are straightforward and clear, but when the manuscript transitions to the metabolite-respiration correlations, there is little description of the metabolomics methods or datasets, which should be clarified.

Thank you for noting this. We agree that the transition to the metabolite–respiration correlation analyses required clearer description of the metabolomics datasets and processing. We have revised the Methods and the corresponding Results text to briefly summarize the metabolomics dataset parameters and workflow, including how metabolomics and respirometry measurements were time-matched across ZT, the normalization procedures applied prior to analysis, the use of Spearman rank correlations, and how we interpret lagged relationships between metabolite abundance and respiratory outputs.

The text now reads as follows:

Methods:

“Metabolomics-respirometry integration and lag analysis

To integrate respirometry with metabolomics, RQ was recorded continuously at 1-second resolution and averaged into 5-minute bins. Because steady-state metabolite measurements were acquired at 2-hour intervals, we extracted the RQ values corresponding to each 2-hour Zeitgeber Time (ZT) sampling point from the 5-minutebinned dataset to generate time-matched RQ-metabolite pairs. We additionally evaluated temporal relationships using a lag analysis by systematically shifting the RQ time series relative to the metabolite timepoints (−120, −60, −30, −15, −5, +5, +15, +30, +60, and +120 minutes). Metabolite abundances were normalized as described above, and associations between RQ and individual metabolites were quantified using Spearman rank correlations (ρ) at each lag. Metabolites showing strong associations (e.g., |ρ| > 0.7 with nominal p < 0.05) were carried forward for visualization and summary, and lag direction was interpreted as metabolites preceding (negative lag) or following (positive lag) changes in RQ.”

Results:

“Temporal Profiling of Respiratory Quotient in Wild-Type Flies Under Light-Dark Conditions

RQ values corresponding to each 2-hour Zeitgeber Time (ZT) point were extracted from the 5-minute-binned dataset. Building on this alignment, we explored temporal relationships by systematically shifting the RQ time series by −120, −60, −30, −15, −5, +5, +15, +30, +60, and +120 minutes relative to the metabolite dataset. The continuous respirometry time series showed an oscillatory day-night pattern in RQ; metabolomics was then used to relate time-matched and lagged metabolite dynamics to RQ patterns (Figure 4).”

The Discussion is at times repetitive and could be tightened, with the main message (i.e., wild-type flies align metabolism in advance, while mutants do not) kept front and center.

Thank you for this helpful suggestion. We have revised the Discussion to reduce repetition and improve focus by keeping the central takeaway explicit throughout, and by consolidating overlapping paragraphs into a more streamlined narrative.

We added this revision at the start of the Discussion, in the opening subsection “Temporal Misalignment Alters Fuel Utilization and Respiratory Rhythms.”

The Discussion now reads as follows:

“Across the manuscript, the central takeaway is that wild-type flies under LD exhibit anticipatory alignment of fuel selection with time of day, whereas short-sleep mutants (fmn, sss) and clock-disrupted flies (per01) show reactive or misaligned metabolism under our conditions. We therefore focus the Discussion on loss of temporal coordination between respiratory output and pathway-level metabolism, rather than reiterating rate changes alone.”

Terms such as "anticipatory" and "reactive" should be defined early and used consistently throughout.

Thank you for this suggestion. We agree and have revised the manuscript to define these terms early (at first use) and apply them consistently throughout. We added this definition in two places:

(i) In the Results, at the start of the metabolomics-respirometry integration section where we first introduce the lag analysis, and

(ii) In the Methods, within the paragraph describing the lag analysis workflow, using identical wording.

The text now reads as follows:

“We define ‘anticipatory’ as metabolite changes that precede the associated respiratory shift (negative lag) and ‘reactive’ as changes that follow or coincide with the respiratory shift (positive lag), and we use these terms consistently throughout.”

Overall, this is a strong and novel contribution. With clarification of scope, refinement of presentation, and a more focused Discussion, the paper will make a significant impact.

We again thank the reviewer for the positive comments.

Reviewer #3 (Public review):

Summary:

The authors investigate how sleep loss and circadian disruption affect whole-organism metabolism in Drosophila melanogaster. They used chamber-based flow-through respirometry to measure oxygen consumption and carbon dioxide production in wild-type flies and in mutants with impaired sleep or circadian function. These measurements were then integrated with a previously published metabolomics dataset to explore how respiratory dynamics align with metabolic pathways. The central claim is that wild-type flies display anticipatory coordination of metabolic processes with circadian time, while mutants exhibit reactive shifts in substrate use, redox imbalance, and signs of mitochondrial stress.

We thank the reviewer for summarizing the contributions made by this manuscript.

Strengths:

The study has several strengths. Continuous high-resolution respirometry in flies is challenging, and its application across multiple genotypes provides good comparative insight. The conceptual framework distinguishing anticipatory from reactive metabolic regulation is interesting. The translational framing helps place the work in a broader context of sleep, circadian biology, and metabolic health.

We thank the reviewer for the positive comments.

Weaknesses:

At the same time, the evidence supporting the conclusions is somewhat limited. The metabolomics data were not newly generated but repurposed from prior work, reducing novelty.

Thank you for raising this point. We now make the provenance of the metabolomics dataset explicit in the manuscript. Importantly, the current study uses this dataset in a new analytical context by integrating it with continuous VCO₂/VO₂ respirometry through timematched and lag-aware analyses. This approach allows us to evaluate dynamic relationships between respiratory output and metabolite profiles across circadian time, which was not addressed in the original metabolomics study. We have clarified this point in the Introduction and Methods.

The text now reads as follows:

In the Introduction:

“To provide a more comprehensive perspective on metabolic regulation, we complemented newly generated respiratory measurements with steady-state metabolomic profiling using liquid chromatography-mass spectrometry (LC-MS) data previously published from our group [27]. This integrative framework enabled timematched and lag-aware analysis of respiratory output and metabolite profiles across Zeitgeber time in the LD cycle….”

In the Methods:

“The metabolomics dataset analyzed in this study was previously published and is publicly available, as described in detail in [27, 31]. In the present study, these data were integrated with respirometry measurements to assess temporal relationships between metabolite abundance and respiratory output.”

The biological replication in the respirometry assays is low, with only a small number of chambers per genotype.

Thank you for highlighting this concern. We suggest that this is a lack of clarity in our initial description of the design and that the replication structure should be stated more explicitly. We have revised the Methods and all relevant figure legends to clearly report biological replication using the chamber as the experimental unit, including n (number of chambers) per genotype and the associated error structure. We also clarify sampling depth by stating that each genotype measurement reflects an average of ~300 flies (25 flies per chamber × 4 chambers per experiment × 3 experimental days). This information is now reported consistently to make the unit of analysis transparent.

We added this clarification in the Methods under “Respirometry Setup” (where chamber loading and experimental design are described) and ensured that each relevant figure legend explicitly reports n (chambers) and SEM.

The text now reads as follows:

“Each genotype measurement represents an average of ~300 flies (25 flies/chamber × 4 chambers/experiment × 3 experimental days), with the chamber as the experimental unit; n (chambers) and SEM are reported in each figure legend.”

Importantly, respiratory parameters in flies are strongly influenced by locomotor activity, yet no direct measurements of activity were included, making it difficult to separate intrinsic metabolic changes from behavioral differences in mutants.

A detailed timing comparison between behavior (feeding and locomotion) compared to respirometry is given in our master response to Reviewer 1, Major comment 4.

In addition, repeated claims of "mitochondrial stress" are not directly substantiated by assays of mitochondrial function.

Thank you. We agree that “mitochondrial stress” requires direct functional evidence. We therefore directly assayed mitochondrial respiration (baseline gut-tissue OCR in fmn and per01 versus iso31 controls), added a new Methods subsection and Figure 9, and reframed our wording from “mitochondrial stress/impairment” to altered (elevated) baseline mitochondrial respiration. The experimental details are now in the Methods and the result in the Results (both quoted below). sss was not assayed, so we removed the functional mitochondrial-stress claim for sss; we retain Vaccaro et al. (2020) as prior support for fmn.

Methods — new subsection “Gut Tissue Respirometry” now reads: “Oxygen consumption rate (OCR) was measured in dissected gut tissue from iso31, fmn, and per01 flies using the Resipher System (Lucid Scientific, GA, USA). Baseline OCR (fmol/mm²/s) was averaged over a 24-hour window following a 12-hour acclimation; values from 3 independent runs were pooled, median-normalized to iso31 within each run, and log2(x+10)-transformed. A single iso31 outlier (third per01 run) was excluded; no other values were removed. Each mutant was compared with iso31 using the MannWhitney test (GraphPad Prism 10), with significance at p < 0.05 (fmn vs iso31, n = 12 vs 12; per01 vs iso31, n = 12 vs 11 after excluding one iso31 outlier).”

The text has been added to Results:

“Because pathway-level metabolomics implicated mitochondrial pathways in the sleep and circadian mutants, we directly assayed mitochondrial respiration by measuring baseline oxygen consumption rate (OCR) in dissected gut tissue from fmn and per01 relative to iso31 controls. Both fmn and per01 guts showed significantly elevated baseline OCR (fmn vs iso31, n = 12 vs 12; per01 vs iso31, n = 12 vs 11 across 3 runs; Mann-Whitney test, p<0.05; Figure 9A,B). Because only baseline OCR was measured, we interpret this as altered (elevated) baseline mitochondrial respiration rather than reduced capacity or a specific coupling defect (Figure 9).”

Discussion- fmn:

“These interpretations are further supported by gut-tissue respirometry showing elevated baseline mitochondrial respiration in fmn relative to iso31 controls. Together with prior evidence of ROS accumulation (oxidative stress) in fmn (Vaccaro et al., Cell, 2020), these functional data indicate that chronic sleep loss in fmn is associated with altered mitochondrial respiration.”

Discussion- per01:

“Gut-tissue respirometry in per01 likewise showed elevated baseline mitochondrial respiration, providing functional evidence consistent with the metabolomic signatures of disrupted redox balance and mitochondrial metabolism.”

The study also excluded female flies entirely, despite well-documented sex differences in metabolism, which narrows the generality of the findings.

Thank you for raising this point. We agree that sex is an important biological variable in metabolic regulation. While our Methods state that male flies were collected, we have now made this explicit and unambiguous by stating that only males were used for respirometry and metabolomics integration, and we have added this as a limitation in the Discussion, noting that sex-specific physiology could influence the magnitude and/or timing of the effects we report. We also highlight inclusion of females as an important future direction.

The text now reads as follows (Methods):

“Only male flies were used for all respirometry experiments and for integration with the metabolomics dataset. This was done to reduce variability introduced by female reproductive status (e.g., mating/egg production) and associated metabolic differences, enabling a clearer comparison across genotypes and lighting conditions.”

The text now reads as follows (Discussion):

“Because only males were analyzed, our conclusions may not generalize to females, which can show sex-specific metabolic physiology. Inclusion of female flies and direct sex comparisons across LD and DD conditions will be an important future direction. More specifically, females carry a higher reproductive and biosynthetic load (egg production) that typically raises metabolic rate and can shift RQ toward lipogenesis and alter the amplitude and phase of diurnal respiratory rhythms; females might therefore show larger or differently-timed effects than the males studied here.”

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Major comments:

(1) The authors appear to be using WT-DD as a condition to disrupt circadian rhythm (line 216). Although rhythmicity is often dampened in DD compared to in LD, circadian rhythm is defined as rhythm in a constant condition (e.g., DD) after entrainment. WT-DD is not a condition that the authors should use if they want to disrupt circadian rhythms in flies. WTLL would be a better condition to use since flies become arrhythmic in LL, not DD.

Thank you for this clarification. We agree that DD is not a circadian-disrupting condition; circadian rhythmicity is defined by rhythms that persist under constant conditions after entrainment. Our intent was not to treat WT-DD as arrhythmic, but to use DD to assess free-running circadian regulation in the absence of external light–dark cues. We have revised the manuscript to clearly distinguish diurnal rhythms under LD from free-running circadian rhythms under DD, and to avoid implying that DD abolishes rhythmicity.

The Results text now reads as follows:

“We used constant darkness (DD) to assess free-running circadian regulation in the absence of external light–dark cues.”

In the Discussion, in the subsection “Metabolic Desynchrony and Redox Imbalance in Wild-Type Flies Under Constant Darkness,” we revised the interpretation of WT-DD to clarify that the observed metabolic effects reflect removal of external light–dark cues under our experimental conditions, while avoiding overgeneralization.

The Discussion text now reads as follows:

Our DD data show that endogenous free-running circadian regulation persists but that removing external light–dark cues degrades the temporal coordination between respiration and metabolism, indicating that external entrainment normally strengthens this coordination under our conditions; however, we acknowledge that the coupling may be different in mammals.

(2) The authors need to be more precise in the use of the term "circadian" throughout the manuscript. When describing 24h rhythmicity in the LD condition in flies, they can use the diurnal rhythm. Circadian rhythm is an endogenous rhythm without external time cues (e.g., DD rhythm).

Thank you for this point. We agree and have revised the manuscript to use terminology consistently: rhythms measured under LD are now referred to as diurnal (LD) rhythms/patterns, and the term circadian is reserved for endogenous free-running rhythms in DD. We updated the Methods, Results, Discussion and figure legends throughout to correct instances where LD rhythmicity was previously labeled as “circadian.”

We added this clarification in the Methods (“Drosophila Strains, Entrainment and Collection”) and in the Results/figure legends where LD time courses are described.

The text now reads as follows (Methods):

“Male flies were collected shortly after eclosion and entrained in light-dark (LD) incubators for a minimum of three days before diurnal (LD) time-course collection across Zeitgeber time (ZT).”

We revised the Results text where WT-DD and LD time courses are described, replacing imprecise references to ‘circadian disruption’ or ‘circadian cycle’ with ‘free-running conditions in DD,’ ‘24-hour cycle,’ or ‘diurnal pattern under LD,’ as appropriate. We also revised the Discussion to avoid describing LD patterns as circadian and to avoid implying that DD disrupts circadian rhythmicity.

(3) Lines 253-256: The authors' interpretation of this data is not accurate. The authors observed significant day-night differences in VO2 and VCO2 in WT-DD. This suggests there is circadian control over metabolism rhythms. The authors noted there is "limited circadian control".

Thank you for pointing this out. We agree that the significant day-night differences in VO₂ and VCO₂ in WT-DD support persistent endogenous (circadian) control of respiratory rhythms under constant darkness. We have revised the Results text (lines 253-256) to remove the statement implying “limited circadian control” and instead describe the WTDD effect as maintained rhythmicity with altered amplitude and/or phase relative to LD, rather than loss of rhythmic regulation.

We added this revision in the Results section under “Diurnal Variation in CO₂ Production, O₂ Consumption, and Respiratory Quotient Across Genotypes” (WT-DD description; lines 253-256).

The text now reads as follows:

WT-DD flies, maintained in constant darkness, exhibited the lowest overall respiratory activity. Despite the absence of environmental light cues, VCO₂ and VO₂ retained significant day-night differences (Figure S4), consistent with persistent free-running circadian control in constant darkness (with altered amplitude and/or phase relative to LD).

(4) Besides sleep, metabolic rates are known to be affected by food consumption. Measuring food consumption of the sleep and circadian mutants might provide insights into whether the metabolic rates are more affected by changes in sleep profile or food consumption. This might also be important given fumin displays impaired dopamine transport function and defective dopamine reuptake, and dopamine is known to affect eating behavior. This was not considered and/or discussed.

It is technically challenging to measure feeding during respirometry, so we acknowledge it as a limitation. To test whether behavioral timing could explain our results, we compared our respiratory rhythms with the feeding and activity rhythms reported in Malik et al. (2026); the comparison and its interpretation are now in the Discussion (quoted below). This is our master response to the activity/feeding concern and is crossreferenced from Reviewer 3’s public review and Recommendation 1; feeding and activity were not measured for sss.

We added this clarification in the Discussion (Limitations/confounds) where we address potential behavioral contributors (activity/feeding) to respirometry outcomes.

The text now reads as follows:

“Locomotor activity and feeding could not be measured during the respirometry recordings, so genotype differences in respiratory parameters should be interpreted with caution. To assess whether behavioral timing could account for these differences, we compared our respiratory rhythms with the feeding and activity rhythms reported for these genotypes in Malik et al. (2026): wild-type feeding peaked at ZT ~3.25 and fmn feeding was phase-delayed to ZT ~4.5, while fmn also showed elevated locomotor activity, particularly during the dark period. In our data the fmn VCO2 rhythm peaks at ZT ~3.25 (RQ at ZT ~4.25), so the respiratory peak slightly precedes the feeding peak; the phase of the fmn respiratory rhythm is therefore not driven by feeding, although the elevated activity of fmn may contribute to its higher overall metabolic rate. Feeding and activity were not measured for sss.”

(5) It is not clear whether the authors are simply analyzing the SAME dataset in Figure 1, Figure S4, and Figure 2-3, but just with different resolutions. They need to better articulate this point.

We thank the reviewer for pointing this out. While the source data for Figures 1, 2-3, and Figure S4 use the same underlying respirometry dataset, they present different analyses to address specific questions. Figure 1 shows the full time-course traces (fine time bins), Figures 2-3 extract rhythmicity metrics (e.g., period/phase) from those same time-series, and Figure S4 collapses the same data into simple day (ZT0-12) vs night (ZT12-24) averages.

We added this clarification in the Figure legends for Figure 1, Figures 2-3, and Figure S4, and also noted it in the Methods where the respirometry analysis outputs (time-series binning, rhythmicity analysis, and day/night averaging) are described.

(6) Figure 3 and page 14: What are their criteria for differentiating "conserved" vs "distinct" phase alignment? It is not clear whether this conclusion is supported by any statistical analysis.

We now specify that the “conserved” vs “distinct” phase descriptions refer to early- vs late-peaking rhythms, and we state the criterion explicitly: a phase was called “conserved” when it fell within ±3 h of the WT-LD peak. The revised text reads: “VO₂ peaked … suggesting conserved phase alignment, with all phases within ±3 h of WT-LD (Figure 3, Table 1)”; “RQ peaked … indicating distinct phase alignment, with the sleep-mutant phases ~8 h apart (Figure 3, Table 1).”

The text now reads as follows:

“VO₂ peaked …suggesting conserved phase alignment, with all phases within ±3 h of WT-LD (Figure 3, Table 1)”

“RQ peaked … indicating distinct phase alignment compared to respiratory output, with phases of the sleep mutants 8 h apart (Figure 3, Table 1).”

(7) Figure 4: The authors need to provide more details as to how the metabolite dataset was utilized to generate this figure and how they made the conclusion that their analysis "revealed a distinct circadian rhythmicity in RQ, characterized by oscillatory patterns indicative of coordinated substrate utilization across the day-night cycle".

We have clarified how the respirometry and metabolomics data are used for Figure 4 and corrected the overstated rhythmicity claim:

(1) The RQ patterning in Figure 4 is derived from the continuous respirometry time series, not from the metabolomics dataset, which is used only for the time-matched and lagged correlation analyses that relate metabolite dynamics to RQ. (2) We removed the statement that the analysis “revealed a distinct circadian rhythmicity in RQ”: RQ was not statistically rhythmic in WT-LD or WT-DD (Table 1), and Figure 4 instead shows the day-night RQ pattern that serves as the reference for the lag-based metabolite correlations. We revised the Methods, the Results paragraph introducing Figure 4, and the Figure 4 legend accordingly.

The text now reads as follows:

“Respiratory quotient (RQ) was recorded continuously and averaged into 5-minute bins. To integrate with metabolomics collected every 2 hours, we extracted the corresponding 2-hour ZT RQ values and performed a lag analysis (−120 to +120 min) to relate metabolite dynamics to RQ patterns; rhythmicity of RQ itself was assessed from the respirometry time series.”

(8) It is unclear why the examples of hydroxyhexadecenoylcarnitine and quinolinate were chosen to be presented in Figure 5a. The authors should clarify their choice of these two examples. Also, the authors should generate a supplemental table with the "several metabolites demonstrating strong correlations (line 294).

Thank you for this suggestion. Hydroxyhexadecenoylcarnitine and quinolinate were selected as representative examples, and we agree that the metabolites supporting the strong RQ-associated correlations should be provided more explicitly. We have now added a Supplementary Table listing the metabolites demonstrating strong correlations with RQ across WT-LD, fmn, sss, per⁰¹, and WT-DD conditions, using the same selection criterion applied in the heatmap analyses (|ρ| ≥ 0.7, p < 0.05).

We also revised the Results text near the statement describing strong metabolite-RQ correlations to direct readers to this new table.

The text now reads as follows:

Hydroxyhexadecenoylcarnitine and quinolinate are among the strongest positively- and negatively-lagged RQ-correlated metabolites in WT-LD (ρ = +0.78 at +120 min and ρ = −0.77 at −120 min; Supplementary Table 1), illustrating the two opposite lag directions of the workflow. The full set of metabolites showing strong RQ-associated correlations across WT-LD, fmn, sss, per⁰¹, and WT-DD conditions is provided in Supplementary Table 1.

(9) Although Spearman correlation analysis suggests some correlation between RQ and the two metabolites shown in Figure 5, the correlation shown in Figure 5b does not appear to be compelling. Results shown in Figure 5b do not provide confidence that conclusions based on clustered heatmap analysis shown in Figures 6 to 8 are meaningful. In addition to Spearman correlation, the authors might consider performing additional statistical methods to provide further support.

Thank you for this comment. We suggest that Fig. 5b was not explained clearly and have clarified. Figure 5b is a lag analysis, not a separate correlation result: it shows how the Spearman correlation changes when the RQ time series is shifted forward or backward in time relative to the metabolite timepoints. The goal is to illustrate lead-lag timing (which shift gives the strongest association), rather than to present a single “strong” correlation as standalone proof.

To address the concern about confidence in the heatmap-based results (Figs. 6-8), we have strengthened the reporting by providing effect sizes (Spearman ρ) and lag for the metabolite-respirometry associations (now included as Supplementary Table 1). This allows readers to evaluate the statistical support underlying the clustering, beyond the visual patterns in the heatmaps.

We additionally report multiple-testing–corrected significance for the metabolite–RQ correlations (Benjamini–Hochberg FDR) alongside nominal p in Supplementary Table 1, using the same correction already applied to the pathway enrichment in Supplementary Table 2.

Minor comments:

(1) Line 82: The authors should clarify what they mean by "circadian collection". Except for WT-DD, my interpretation is that they collected their samples in LD, so that would not be "circadian collection".

Thank you for catching this. We agree that “circadian collection” was imprecise. We have revised line 82 to clarify that samples collected under LD were collected across diurnal (LD) time (ZT), and we now reserve “circadian” specifically for collections under constant conditions (DD). We also updated the wording throughout the manuscript to maintain this distinction consistently. We added this clarification in the Methods section “Drosophila Strains, Entrainment and Collection” (line 82).

The text now reads as follows:

“Male flies were collected shortly after eclosion and entrained in light-dark (LD) incubators for a minimum of three days before diurnal (LD) time-course collection across Zeitgeber time (ZT).

(2) The authors cited Frayn 1983 to indicate how the RQ value can be used to reflect metabolic fuel utilization. Is this interpretation accepted for all animals, including flies?

RQ is widely used in indirect calorimetry as an index of relative substrate utilization, including in small model organisms, but we agree that it should be interpreted with appropriate caveats. We have revised the manuscript to clarify that we interpret RQ conservatively as reflecting relative shifts in substrate utilization over time and between genotypes, rather than as a precise quantitative measure of absolute carbohydrate versus lipid oxidation.

We made this change in two places: in the Introduction, where RQ is introduced and Frayn is cited, and in the Methods, under “Carbon Dioxide and Oxygen Analysis and Calculations,” where RQ is defined.

The text now reads as follows:

Introduction: “These measurements allow for the estimation of energy expenditure and respiratory quotient (RQ), which can provide an index of relative substrate utilization, with appropriate caveats[13, 14]. In this study, we interpret RQ conservatively as reflecting relative shifts in substrate utilization over time and between genotypes, rather than as a precise quantitative measure of absolute carbohydrate versus lipid oxidation.”

Methods: “RQ was used as an index of relative shifts in substrate utilization over time and between genotypes, interpreted conservatively rather than as a precise measure of absolute carbohydrate versus lipid oxidation as this has not been directly characterized in flies.

(3) Figure S4: Since the authors are comparing day-night differences, they should plot them in the same graph to make it easier to compare.

We have revised Figure S4 to plot day and night within the same graph/panel for each metric (VCO₂, VO₂, RQ), using side-by-side day vs night groupings per genotype to facilitate direct visual comparison, and we updated the legend accordingly.

(4) Line 252: When comparing diurnal differences, the authors should not use the word "rhythm". Pattern or profile might be a better word to use.

We appreciate the suggestion. Where we use “rhythm” we refer specifically to 24-hour oscillations established statistically by JTK_CYCLE and RAIN (Figures 2-3, Table 1); for the coarser day-versus-night comparisons we agree “pattern” or “profile” is preferable and have adopted it there. We have gone through the manuscript to apply this distinction consistently.

(5) Figure 6a: larger font labels are necessary for the metabolites.

Thank you. We have revised Figure 6a to increase the metabolite label font size for improved readability.

Reviewer #3 (Recommendations for the authors):

(1) Activity controls: To strengthen the paper, include or reference direct measures of locomotor activity (e.g., DAM system). This would allow the separation of metabolic changes from behavioral differences and would enable better analysis of circadian patterns.

This activity/feeding confound is addressed in full in our response to Reviewer 1, Major comment 4; we cross-reference it here to avoid repetition.

(2) Mitochondrial function: "mitochondrial stress" should be supported by additional assays such as mitochondrial enzyme activities, high-resolution respirometry, or reactive oxygen species measurements.

See our full response to the mitochondrial point in Reviewer #3's public review above, including new Figure 9 and the Gut Tissue Respirometry Methods.

(3) Sex differences: Provide a clear justification for excluding female flies. If feasible, incorporate female data or explicitly discuss how sex differences could alter metabolic outcomes.

This is addressed in the public review for reviewer 3.

(4) Statistical presentation: Increase n. Clarify in figure legends whether error bars represent SD or SEM and ensure consistency across all figures (Figure legends).

We have revised all figure legends to clearly state that error bars represent SEM and ensured this is applied consistently across all figures. We also explicitly report the n for each genotype (with chambers as the experimental unit) in the relevant legends.

The text now reads as follows:

“Error bars represent SEM, and n denotes the number of chambers (experimental units) per genotype.”

(5) Sample size and replication: Indicate more clearly that the chamber, not the individual fly, is the experimental unit. Discuss limitations of replication and statistical power in the text.

Thank you for this comment. We have clarified throughout the Methods and figure legends that the chamber (not the individual fly) is the experimental unit. We now state that each genotype measurement reflects an average of 300 flies (25 flies/chamber × 4 chambers/experiment × 3 experimental days), and we report n (number of chambers) and the error structure (SEM) for each genotype.

Added to Discussion:

“We also acknowledge the limits of this replication: with n = 12 chambers per genotype, statistical power to detect small-magnitude differences and subtle phase shifts is limited, and negative calls (e.g., arrhythmicity) should be interpreted with this caveat.”

(6) Writing and clarity: (a) Streamline the Discussion to focus on mechanistic themes (anticipatory vs reactive alignment, substrate shifts, redox imbalance).

(b) The opening phrase ("Precise temporal regulation of metabolism by sleep and circadian rhythms is essential for dynamic energy homeostasis") is dense, vague, and difficult to interpret. Consider rephrasing to something more concrete, for example: "Sleep and circadian rhythms tightly control when and how the body uses energy, but we do not yet know exactly how this timing connects to oxygen use and breathing needs."

Thank you for these helpful suggestions. We have revised the Discussion to reduce repetition and improve focus by organizing it around the key mechanistic themes raised by our data, including anticipatory vs reactive metabolic alignment, substrate-use shifts, and redox/mitochondrial imbalance. We revised the opening sentence of the Abstract to make the biological question more concrete and accessible, following the reviewer’s suggestion.

eLife Assessment

This study presents fundamental results on questions related to the presence/absence of the Entner-Doudoroff pathway in cyanobacteria. In contrast to an earlier study, compelling evidence is given that Synechocystis PCC 6803 lacks both an Entner-Doudoroff pathway and a related bypass but contains a promiscuous aldolase. This study successfully reconciles data from different studies and lessons learned from a previous misconception.

Reviewer #1 (Public review):

Strengths:

Thorough reanalysis of the experimental results obtained in previous studies, which led to the publication of the PNAS paper in 2016.

New experimental evidence to confirm that enzymes previously considered as participating in the ED, actually are not catalyzing the ED biochemical reactions, but are involved in other metabolic pathways. Also, the authors completely discarded the occurrence of the GDH/GK shunt in Synechocystis PCC 6803. Generally speaking, the manuscript is very clearly written, with a precise description of the previous findings, the mistakes which took place in the 2016 paper, and the strategies they have used to address those issues, in order to reach a thoroughly revised vision of the glucose metabolic pathways in Synechocystis PCC 6803. In this regard, the drawings shown in Figures 1 and 7 are very helpful for the reader to follow the story and understand the possible metabolic transformations depending on the working hypothesis.

Also, I commend the authors for openly describing previous mistakes. In this paper, they reassess past observations under the light of more recent findings, and to integrate the information in this manuscript. The scientific conclusions are solid and very interesting, and besides they use the opportunity to offer valuable advice to researchers. This is especially focused on the importance of careful biochemical characterization of enzymes, which should always be carried out when studying proteins which have been identified as a specific enzyme on the basis of sequence homology. In a similar way, they found that an insertional mutant was the cause for the absence of specific metabolites, which had been attributed to particularities of a metabolic pathway in that mutant, when it was actually due to a nucleotide insertion; given that currently, genome sequencing is an affordable technique, this kind of mistakes can now be easily prevented by confirming the correct generation of the mutant by DNA sequencing, as proposed by the authors in a recently published preprint (Theune et al, bioRxiv 10.64898/2026.04.08.717167).

Weaknesses:

The authors propose that EDA might be involved in the PEP-pyruvate-OAA node, or in the proline metabolism, but this requires further experimental work for clarification; what their results indicate clearly is that this enzyme is not actually catalyzing the transformation of KDPG to GAP, which is the second specific enzyme of the ED pathway. But the real physiological function in this cyanobacterium is still unconfirmed.

Another aspect which could be improved is that the recombinant expression of some genes was carried out in E. coli; even if this is a useful and valid research strategy, in studies like this (where there is a strong focus on the physiological function of enzymes in the original organism, Synechocystis PCC 6803), I think it would have been more appropriate to express the 6803 genes in another cyanobacterium easily amenable for genetic transformation and gene expression, which would produce the protein in a physiological environment more similar to another cyanobacterium (compared to E. coli, which is an heterotrophic bacterium). I am not sure this would change any of the obtained results, but certainly would confer additional robustness to the enzymatic results.

Comments on revised version.

The authors have provided satisfactory replies to all my suggestions and corrections, and I have no further changes to suggest.

Reviewer #2 (Public review):

Summary:

The study presents novel results on the presence of the Entner Doudoroff pathway in Synechocystis sp. PCC 6803. In contrast to an earlier study, compelling evidence is given that this strain lacks both an ED pathway and a glucose dehydrogenase/glucokinase bypass but contains a promiscuous aldolase, which also decarboxylates oxaloacetate and cleaves 2-keto-4-hydroxyglutarate (as it occurs in proline degradation). The study concludes with successfully reconciling data of different studies and with lessons learned from the previous misconception.

Strengths:

Solid biochemical data is presented to reconcile contradicting data of earlier studies and to serve as basis for disclosing possible functions of a promiscuous aldolase. Earlier misconceptions and lessons to be learned are well discussed.

Weaknesses:

The materials and methods section is rather lengthy, suffering from a lack of conciseness and repetitions, and nevertheless misses some specifications.

Comments on revised version.

The materials and methods section has been significantly improved. The revised manuscript is now recommended for publication as it is.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

Some of the authors proposed in a PNAS paper in 2016 the occurrence of the EntnerDoudoroff (ED) pathway in cyanobacteria and plants, on the basis of several lines of biochemical and genetic evidence. However, more recent results indicated that one of the two specific enzymes of the ED pathway (EDD) is missing in Synechocystis PCC 6803. The authors carried out additional experiments, which demonstrated that EDD is missing, and one of the enzymes (ED aldolase) is a promiscuous enzyme which seems to be involved in proline metabolism and is not actually participating in the ED pathway as initially believed. The results described in this paper are strong evidence that this new interpretation is appropriate, and therefore, it corrects the previous proposal, providing an honest description of the reasons why the authors had reached the wrong conclusion about the existence of the ED pathway in cyanobacteria and plants.

We thank Reviewer 1 for the summary and comments. Based on our finding that EDA is a promiscuous aldolase that, in addition to the cleavage of KDPG to GAP and pyruvate (a reaction of the ED pathway) catalyzes other reactions in vitro, we proposed potential in vivo functions of EDA, including its involvement in proline metabolism. However, these assumptions require further experimental testing. We do not yet have definitive findings regarding the function of the promiscuous aldolase EDA in Synechocystis in vivo, but respective studies are currently underway.

Strengths:

Thorough reanalysis of the experimental results obtained in previous studies, which led to the publication of the PNAS paper in 2016.

New experimental evidence to confirm that enzymes previously considered as participating in the ED actually are not catalyzing the ED biochemical reactions, but are involved in other metabolic pathways. Also, the authors completely discarded the occurrence of the GDH/GK shunt in Synechocystis PCC 6803. Generally speaking, the manuscript is very clearly written, with a precise description of the previous findings, the mistakes which took place in the 2016 paper, and the strategies they have used to address those issues, in order to reach a thoroughly revised vision of the glucose metabolic pathways in Synechocystis PCC 6803. In this regard, the drawings shown in Figures 1 and 7 are very helpful for the reader to follow the story and understand the possible metabolic transformations depending on the working hypothesis.

Also, I commend the authors for openly describing previous mistakes. In this paper, they reassess past observations in light of more recent findings and to integrate the information in this manuscript. The scientific conclusions are solid and very interesting, and besides, they use the opportunity to offer valuable advice to researchers. This is especially focused on the importance of careful biochemical characterization of enzymes, which should always be carried out when studying proteins which have been identified as a specific enzyme on the basis of sequence homology. In a similar way, they found that an insertional mutant was the cause of the absence of specific metabolites, which had been attributed to particularities of a metabolic pathway in that mutant, when it was actually due to a nucleotide insertion; this could have been easily prevented by confirming the correct generation of the mutant by DNA sequencing.

We thank the reviewer for this kind comment. We agree that biochemical characterization of enzymes as well as DNA sequencing to check deletion mutants, are important and valuable tools. As outlined in the manuscript and additionally in more detail in a recently submitted article, which is available at bioRxiv (Theune et al. 2026, doi: https://doi.org/10.64898/2026.04.08.717167) and is currently under review at PLOS One, we suggest that genome sequencing of deletion mutants in combination with complemented strains as controls are required to minimize the risk of misinterpretation based on secondary mutations (1). During the early stages of our research on the ED pathway, and later as well when we were already trying to resolve the conflicting results that had accumulated concerning the ED pathway, genome sequencing for Synechocystis mutants was not affordable as a routine procedure (2-4). Therefore, we could not have easily prevented this misconception based on this technique at that time. However, we strongly encourage genome sequencing of deletion mutants (in combination with complemented strains) as routine procedures these days (1).

Weaknesses:

The authors propose that EDA might be involved in the PEP-pyruvate-OAA node, or in the proline metabolism, but this requires further experimental work for clarification; what their results indicate clearly is that this enzyme is not actually catalyzing the transformation of KDPG to GAP, which is the second specific enzyme of the ED pathway. But the real physiological function in this cyanobacterium is still unconfirmed.

As stated above and in the manuscript, we agree that the in vivo role of EDA requires further experimental work which is in progress. However, our results demonstrate that EDA splits KDPG into GAP and pyruvate in vitro, but we assume that this reaction does not play a role in vivo due to the absence of its substrate.

Another aspect which could be improved is that the recombinant expression of some genes was carried out in E. coli; even if this is a useful and valid research strategy, in studies like this (where there is a strong focus on the physiological function of enzymes in the original organism, Synechocystis PCC 6803), I think it would have been more appropriate to express the 6803 genes in another cyanobacterium easily amenable for genetic transformation and gene expression, which would produce the protein in a physiological environment more similar to another cyanobacterium (compared to E. coli, which is an heterotrophic bacterium). I am not sure this would change any of the obtained results, but it certainly would confer additional robustness to the enzymatic results.

Synechocystis is easily amenable to genetic manipulation, and we agree that expression and purification of all enzymes from this host would have been ideal. However, the first characterization of Synechocystis EDA was performed with proteins that were purified from Synechocystis and showed activity on KDPG at comparable rates as proteins that were purified from E. coli in this study (2). Moreover, most biochemical characterizations of EDAs from archaea, bacteria and plants were performed after recombinant expression in E. coli and yielded highly active enzyme as in the case of Synechocystis is this study (5-7). Therefore, we currently have no reason to worry that the expression in E. coli might affect the enzymatic activity of EDA. The main reason for utilizing E. coli as an expression strain in this study was to gain higher yields of protein for in-depth analyses.

Bibliography:

I think the list of papers used in this manuscript is complete and up to date. However, I do miss recent papers which addressed one aspect that was proposed in the original 2016 PNAS paper: the authors wrote, "We therefore suggest that Prochlorococcus might oxidize glucose via the ED pathway under mixotrophic conditions, as shown for Synechocystis." Recent studies checked this hypothesis and have shown that the ED pathway seems to be also missing in Prochlorococcus and marine Synechococcus, and I think this manuscript is a good place to cite them, since these results are consistent with the findings of this paper.

We will include a references from Moreno-Cabezuelo et a. 2023 (DOI: 10.1128/spectrum.03275-22) in which the proteomes of three marine Prochlorococcus and three marine Synechococcus strains were investigated upon exposure to glucose (8). Protein levels of EDA were either downregulated or not affected while proteins involved in OPP pathway and CBB cycle were upregulated. The authors of this study conclude that this indicates that the latter processes rather than the ED pathway are involved in photomixotrophy in these strains. However, flux analyses are still missing.

Reviewer #2 (Public review):

Summary:

The study presents novel results on the presence of the Entner-Doudoroff pathway in Synechocystis sp. PCC 6803. In contrast to an earlier study, compelling evidence is given that this strain lacks both an ED pathway and a glucose dehydrogenase/glucokinase bypass but contains a promiscuous aldolase, which also decarboxylates oxaloacetate and cleaves 2-keto-4-hydroxyglutarate (as it occurs in proline degradation). The study concludes with successfully reconciling data from different studies and with lessons learned from the previous misconception.

Strengths:

Solid biochemical data are presented to reconcile contradicting data of earlier studies and to serve as a basis for disclosing possible functions of a promiscuous aldolase. Earlier misconceptions and lessons to be learned are well discussed.

Weaknesses:

The materials and methods section is rather lengthy, suffering from a lack of conciseness and repetition, and nevertheless misses some specifications.

We thank Reviewer 2 for the kind summary and comments and will improve the materials and methods part accordingly in a revised version.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Some additional aspects that could be improved:

(1) L182-184: Are there any known in vitro attempts to determine whether some DHADs can accept 6PG as substrate? If not, did the authors check this possibility in the lab?

To the best of our knowledge, as mentioned in the manuscript, some DHADs are tested for their activity towards gluconate and some other substrates but not 6PG. It was suggested that since gluconate is smaller it might fit in the catalytic site of DHADs normally occupied by DHIV (7). We discussed this point in the manuscript (Line 177-185).

In this study we tested the DHAD Slr0452 from Synechocystis and DHAD from Synechococcus with 6PG as substrate but the enzyme did not catalyze 6PG dehydration. The result for Slr0452 is shown in Figure S3.

(2) L234-241: This paragraph shows how important it is to avoid relying only on sequence alignments to assign functions to proteins (and enzymes in particular). The authors stress this idea elsewhere in the paper, but I think it should receive even more attention in a manuscript like this. Physiological characterization of the protein function is paramount, especially in the case of enzymes. Also, the GDH1 overexpression mutant was done in E. coli; this adds an additional layer of uncertainty, since the protein processing in E. coli might not be entirely identical to that carried out in cyanobacteria, as I mentioned above. This, in turn, could be one of the reasons for not finding the expected enzymatic activity. This comment is also relevant for the results shown in Table 1 (page 12).

As outlined following lines L234-241 we tested crude cell extracts from Synechocystis WT, a Synechocystis strain overexpressing putative GDH1 (Sll1709) and a Dzwf deletion mutant which can be assumed to upregulate a GDH/GK bypass if present in Synechocystis for GDH activity but did not find any. This strongly indicates that sll1709 does not code for an active GDH in Synechocystis. We thereafter also overexpressed GDH1 in E. coli and again could not detect any GDH activity. In case of GDH1, enzyme activity was therefore tested both in Synechocystis and E. coli and yielded similar results.

(3) L309-310: "we currently have no explanation for the gluconate that was detected in previous IC-ESI-MSMS measurements in Synechocystis". This is a serious issue, given how important this evidence was for the conclusions of the 2016 PNAS paper and the hypothesis of the ED pathway in cyanobacteria. I would suggest that the authors provide some possible explanations, even if it is based on studies from other teams.

It would be very speculative and therefore in our eyes not helpful to search for explanations in this case as indeed the measurements were done in another lab. One explanation might be that 6P gluconate got dephosphorylated and yielded gluconate as an artifact. However, as we have no experimental validation for this idea and furthermore cannot test it. We therefore prefer to not further comment on this aspect.

(4) L347: The presence of an insertion in the sequence of zwf in the ∆gnd mutant is a welcome explanation for the observed results: abolition of 6PG production in this mutant. This is another important message to stress in this manuscript: construction of mutants should always be double checked by DNA sequencing of the relevant genomic regions, to ensure that these kinds of side problems do not appear, leading to confusing results.

We agree with the reviewer and would get even one step further and rather suggest combining whole-genome sequencing with complemented mutants as it is difficult to know all relevant genomic regions that should be sequenced. We discuss this issue in even more detail in another manuscript that is currently available as online preprint in bioRxiv and is under review (9). This work was now added as a citation in line 566 and the end of the following statement (line 564-566): Routine complementation of deletion mutants and sequencing of selected genes or the entire genome are effective means of identifying secondary mutations that can lead to misleading phenotypes (9).

(5) L418-419 and Table 2: The results observed for the authors (i.e., that EDA could also catalyze reactions with OAA and KHG, albeit with substantially lower catalytic efficiency than with KDPG), is a matter for concern: if their hypothesis is correct (meaning that this EDA is fundamentally involved in the PEP-pyruvate-OAA node and/or proline metabolism, and not with the ED pathway), then why should it keep in the evolution of these organisms such a strong preference for KPDG, when it is not being used physiologically for the ED pathway)? Furthermore, the Km for KDPG is lower than for OAA or KHG, leading to a very big difference in the Kcat/Km values.

To solve these questions further respective studies are underway. As EDD is absent from Synechocystis no KDPG should be available in the cells so that catalytic activity on KDPG should be irrelevant in vivo. The in vivo role of Eda requires further clarification.

Hereafter, I will mention some aspects, following the instructions of eLife, which are related to suggestions for improved experiments/data/analyses, improvements of writing and presentation, and minor corrections to text/figures.

(1) L81: I would modify the text to "Accordingly, this raises further questions...".

Thanks for this hint. We modified the text accordingly.

(2) L189: Add "pages" after "following".

Thanks for pointing this out. We replaced “following” by “below”.

(3) Page 7: The whole beginning of the Results section is actually more discussion than description of results, and the first mention of figures appears in L207 of page 8. Given the content of the paper, I think the authors might reconsider using "Results and Discussion" rather than different, specific sections for Results and Discussion. This is one of the papers where I think the combined use of both makes sense and will allow an easier understanding of the message.

We thank the reviewer for this valuable suggestion and changed the heading to Results and Discussion.

(4) L181: Add reference regarding the llvD/EDD superfamily.

We added the following references in lines 172-176 and 185-189:

(1) Melse, O., Sutiono, S., Haslbeck, M., Schenk, G., Antes, I., & Sieber, V. (2022). Structure Guided Modulation of the Catalytic Properties of [2Fe− 2S]-Dependent Dehydratases. ChemBioChem, 23(10), e202200088.

(2) Ren, Y., Vettenranta, E., Penttinen, L., Jänis, J., Rouvinen, J., & Hakulinen, N. (2025). The engineered dimer of L-arabinonate dehydratase from Rhizobium leguminosarum bv. trifolii: The role of intersubunit interactions in IlvD/EDD family. Biochemical and Biophysical Research Communications, 757, 151610.

(3) Ahmed, H., Ettema, T. J., Tjaden, B., Geerling, A. C., Van Der Oost, J., & Siebers, B. (2005). The semi-phosphorylative Entner–Doudoroff pathway in hyperthermophilic archaea: a reevaluation. Biochemical Journal, 390(2), 529-540. ff

(4) Bräsen, C., Esser, D., Rauch, B., & Siebers, B. (2014). Carbohydrate metabolism in Archaea: current insights into unusual enzymes and pathways and their regulation. Microbiology and Molecular Biology Reviews, 78(1), 89-175.

(5) Figure 2: The data shown in column plots in Fig 2A, B and C, and 4B, could be presented in tables, which would save space while providing the same information: basically, very little/no activity in some cases vs high levels of activity in others.

We would like to keep the column plots as we still think that they visualize our data well.

(6) L255 "Unfortunately, we were not able to overexpress putative GDH2". It would be interesting to give more details about the possible reasons for this fact.

We tested different growth conditions for recombinant GDH2 expression. The expression culture was incubated at 37 °C for 3 hours as well as overnight at 18 °C for overnight after induction. Both experiments did not resolve the expression problem.

(7) L409-410: I think this sentence should include a brief part explaining the kind of essay used to test this activity.

We added now that the LDH-coupled continuous assay was used (see line 403).

(8) Figure 6E: Please give the specific activity in U/mg, as in Fig 6F, instead of percents.

100% is given in U/mg units in the figure legend as “control without effector (100 %; specific activity of 4.3 U/mg)”. For easy comparison of effectors, the relative activity (%) is often used. We would therefore prefer to keep the current data presentation.

(9) L576: Provide the origin of the utilized PCC 6803 strain, given there is a certain level of variability in this strain (glucose tolerance, etc). Also, even if there are some methods which are very widely used, I think the Materials and Methods section should either properly describe them or else cite the source. For instance, BG11 medium is mentioned, but no further information is given.

We included the information that the glucose-tolerant Synechocystis strain was utilized and added the receipt of and a citation for BG11 medium (10).

(10) L582 Generation of mutants: This section mentions the Gibson Assembly cloning method, but I miss further information to allow the reader to reproduce the methodology with as many details as possible, or at least cite papers which do so.

We added a reference in which Gibson Assembly is described (11). Together with the primers listed in Table S3 the generation of mutants is now reproducible.

(11) L609 Please give information in g, not rpm, for centrifugation. Also, mention the model and brand of the centrifuge and rotors used. Also, immunoblotting is very loosely described. This is also valid for other sections, for instance, L618, L636.

We now added the following information: Cells were harvested by centrifugation at an RCF (relative centrifugal force) of 3,992 x g in a Beckman Coulter with a JLA-8.1000 Rotor for 20 minutes at 4°C. We now added a reference (12) in which immunoblotting is described in more detail.

(12) L613: Describe the "small scale purification".

We now added the information that the small-scale purification was performed using a 50-ml aliquot of the large culture which was treated as described below for the remaining sample.

(13) L619-620: Describe the composition of the lysis buffer.

The composition of the lysis buffer is already described as follows: lysis buffer (50 mM NaPO₄ pH=7.0; 250 mM NaCl; 1 tablet complete protease inhibitor EDTA-free

(Roche) per 50 mL)

(14) L691: Specify which amounts of auxiliary enzymes in coupled enzymatic assays were used.

Thanks for pointing this out. We have now integrated the information that 1U of each of the auxiliary enzymes was utilized in the coupled enzymatic assays.

(15) L716: The authors mention several times using a "double beam spectrophotometer". Please provide the model and brand.

Model and brand were now added for the double-beam spectrophotometer (Uvikon 810, Kontron, Augsburg, Germany).

(16) L717 and 718: define "∆absorption".

In line 715, the information is given that absorption was measured at 340 nm, "∆absorption" is accordingly the ∆absorption at 340 nm. This information was added.

(17) L723: "Synechocystis" should be in italics.

Synechocystis is now written italics.

(18) L749-759: This section should be described in more detail: preparation of protein extracts, SDS, immunoblotting, etc, or cite references of the same team where these methods were properly described.

In this section the listed methods are already described in detail.

(19) 798-799: "frozen cell pellets". Please provide numbers to specify the amount of material used.

Thanks for pointing this out. We now added the information that frozen cell pellets with a wet weight of 2.4 g wet weight were resuspended.

(20) L871-872: "It was ensured that auxiliary enzymes were not rate-limiting. One unit (1 U) of enzyme activity is defined as 1 µmol substrate consumed or product formed per minute" is repeated several times in the manuscript (L 907-909, L936-938). I would advise using it the first time, and on other occasions, refer to the same conditions as described above.

We have accordingly circumvented the repetition of 1 U definition from the manuscript.

Reviewer #2 (Recommendations for the authors):

(1) Interpunctuation, especially comma placement, should be improved.

We improved interpunctuation, especially comma placement to the best of our knowledge.

(2) Line 63: delete the first "which".

“Which” was deleted.

(3) Lines 81/82: revise sentence.

We revised the sentence to: Accordingly, this raises further questions about the presence of the ED pathway in cyanobacteria and plants.

(4) Line 164: "presumed" instead of "presumes".

The word was changed accordingly.

(5) Line 228: by others? especially in reference 1?

We deleted by others as the reference is given.

(6) Line 240: "or" instead of "no".

“no” was replaced by “or”

(7) Figure 2: The axes are not well visible, and the explanation for the positive control in panel B is missing in the legend.

Axes from figures 2, 3 and 4 were enlarged. For Figure 2B the following information was added in the legend: As a positive control, 0.05 U glucose dehydrogenase from Pseudomonas sp. was added to Δzwf cultures and to purified putative GDH1 (Sll1709). Axes from figures 2, 3 and 4 were enlarged.

(8) The investigation on the general absence/presence of the GDH/GK bypass in cyanobacteria may not be necessary for this study.

We included this data in this manuscript as the mistaken assumption that the GDH/GK bypass exist in Synechocystis lead among other observations to the misinterpretation of an existing ED pathway in Synechocystis. We would therefore prefer to keep these data in the manuscript.

(9) Line 316: delete "on".

“on” was deleted.

(10) Line 352: delete "or".

“or” was deleted.

(11) Lines 352/353: ZWF expression level appears to be reduced accordingly. This should be stated.

We agree that Zwf expression might be lower, however, we are not entirely sure if this is truly valid and would rather test this assumption further as described in the following lines.

(12) Figure 4: The axes are not well visible.

Axes from figures 2, 3 and 4 were enlarged.

(13) Line 363: values are not only normalized to protein content, but also to activity found for the WT.

We now added: The values are normalized to Zwf enzyme activity found in the WT based on protein content.

(14) Line 408: no separate subsection required.

The title for a new subsection was deleted.

(15) Lines 437-439: These are results descriptions, which should not be placed in the legend, but in the main text, as is partially done.

We deleted these result descriptions in the legend.

(16) Lines 441/442: formatting: one or no bracket pair.

The brackets were corrected.

(17) Lines 443/444: refer to Table 2 instead of giving the values in the legend to avoid duplication.

We deleted the values and now refer to Table 2.

(18) Line 503: delete "identified".

We deleted the second “identified” in the sentence and changed the wording to: Apart from four identified cyanobacteria that possess potential EDDs. In addition, we also added the names of the four cyanobacteria that were found including the sequence IDs of the putative EDDs.

(19) Lines 529/530: revise sentence and format.

We added one sentence and revised the following sentence: In contrast to Synechocystis EDA, EDA from Synechococcus prefers OAA over KDPG. The catalytic efficiency of Synechococcus EDA on oxaloacetate is rather low (OAA 0.437 s^-1 mM^-1), however, its activity can be enhanced by NADP⁺(13).

(20) Line 542: revise sentence.

We revised the sentence to: It remains to be investigated whether this reaction could play a role in vivo, with KDPG potentially acting as a regulatory metabolite at low concentrations.

(21) Line 577: The glass tubes used for cultivation should be specified.

We now added the following information: Custom-made glass tubes were placed in a photobioreactor (manufactured by Willi Hilke, Uslar, Germany).

(22) Lines 584-585: unclear, was the resistance cassette not placed in the gene to be deleted?

Yes, the resistance cassette was placed in the gene to be deleted and was fused for homologous recombination to two DNA fragments approximately 200 bp directly upstream and downstream of the gene. This information was now added.

(23) Line 607: cultivation equipment to be specified.

The following information was now added: For the purification of GDH1 from Synechocystis, a 6 L photoautotrophic culture of the P3-His-GDH1 overexpression strain was grown in a 10 L glass flask at 28°C, illuminated with constant light (50 µmol m^-2 s^-1) and gassed with filter sterilized ambient air to an OD₇₅₀ of about 1.

(24) Line 670: GTS should be specified.

Thank you for this hint. This was a typo. GST was meant not GTS. This was now corrected and GST was specified as Glutathione S-Transferase.

(25) Line 679: delete "gluconate kinase (GK) and".

The second gluconate kinase (GK) was deleted and sentence was revised to:

For gluconate kinase (GK) activity measurements in Synechocystis crude cell extracts the GK reaction was enzymatically coupled to 6-phosphogluconate dehydrogenase (GND) reaction, the latter providing NADP⁺ reduction, which was monitored photometrically at 340 nm.

(26) Line 686: again GK activity? Difference unclear. Was the previously described procedure for GND activity determination?

GK activity measurements in Synechocystis crude cell extracts and GK activity measurements with recombinant enzyme that was expressed in E. coli were done in two different labs with different protocols. Therefore, the first description refers to measurements with Synechocystis while the second measurement refers to measurements with E.coli. This is now specified more clearly.

(27) Type/supplier of spectrophotometers and centrifuges used should be given.

Model and brand were now added for the double-beam spectrophotometer (Uvikon 810, Kontron, Augsburg, Germany). As this study was performed in two different labs over a period of 8 years including one lab moving to a new location, it is now impossible to specify all centrifuges that were utilized. Even though we agree that it would be good to provide this information, we now would have difficulties to be specific.

(28) Consider the referencing of published methods to streamline the materials and methods section.

We now streamlined the materials and methods section by deleting repetitions as outlined below. However, as protein expression, protein purification and enzymatic tests were performed in different labs, in some cases several protocols are given.

(29) Line 757: give specifics of anti-rabbit antibody and define PBS-T and PBS-T Cytiva.

Specifics were added to the text.

(30) Lines 761ff: It is not given for all genes used how they were derived. All synthesized?

We now added detailed information for all genes.

(31) Line 762: codon-optimized for? E. coli?

The information was added that genes that were expressed in E. coli were codon-optimized for E. coli.

(32) Lines 782-786: Rationals for experimental strategies do not belong to materials and methods sections, but to results sections.

The part was deleted in the materials and methods section and transferred to the results section.

(33) Lines 818/819: repetitive.

We removed the repetition and refer to the purification method as stated above in the materials and methods section.

(34) Lines 847-851: True for all EDA-type assays? Kinetic parameters are shown in Table 2 rather than Table 1.

Yes, true for all EDA-type assays. We changed the Table number to 2.

(35) Line 863: delete "in".

“in” was deleted

(36) Lines 888-893: sounds repetitive.

The lines were modified accordingly.

(37) Lines 908/909: repetitive.

The repetitive comment on the definition of 1U was deleted.

(38) Lines 928-938: repetition

The repetition was deleted.

References

(1) M. Theune et al., Easy-to-use whole-genome sequencing workflows and standardized practices to uncover hidden genetic variation in <em> Synechocystis </em> PCC 6803 wild-type and knock-out strains. bioRxiv 10.64898/2026.04.08.717167, 2026.2004.2008.717167 (2026).

(2) X. Chen et al., The Entner–Doudoroff pathway is an overlooked glycolytic route in cyanobacteria and plants. Proceedings of the National Academy of Sciences 113, 5441-5446 (2016).

(3) D. Schulze et al., GC/MS-based 13C metabolic flux analysis resolves the parallel and cyclic photomixotrophic metabolism of Synechocystis sp. PCC 6803 and selected deletion mutants including the Entner-Doudoroff and phosphoketolase pathways. Microbial Cell Factories 21, 69 (2022).

(4) A. Makowka et al., Glycolytic Shunts Replenish the Calvin–Benson–Bassham Cycle as Anaplerotic Reactions in Cyanobacteria. Molecular Plant 13, 471-482 (2020).

(5) V. Zaitsev et al., Insights into the Substrate Specificity of Archaeal Entner–Doudoroff Aldolases: The Structures of Picrophilus torridus 2-Keto-3-deoxygluconate Aldolase and Sulfolobus solfataricus 2-Keto-3-deoxy-6-phosphogluconate Aldolase in Complex with 2-Keto-3-deoxy-6-phosphogluconate. Biochemistry 57, 3797-3806 (2018).

(6) J. S. Griffiths et al., Cloning, isolation and characterization of the Thermotoga maritima KDPG aldolase. Bioorg Med Chem 10, 545-550 (2002).

(7) S. E. Evans et al., Plastid ancestors lacked a complete Entner-Doudoroff pathway, limiting plants to glycolysis and the pentose phosphate pathway. Nature Communications 15, 1102 (2024).

(8) J. Moreno-Cabezuelo, G. Gómez-Baena, J. Díez, J. M. García-Fernández, Integrated Proteomic and Metabolomic Analyses Show Differential Effects of Glucose Availability in Marine Synechococcus and Prochlorococcus. Microbiol Spectr 11, e0327522 (2023).

(9) M. Theune et al., Easy-to-use whole-genome sequencing workflows and standardized practices to uncover hidden genetic variation in Synechocystis sp. PCC 6803 wild-type and knock-out strains. bioRxiv 10.64898/2026.04.08.717167, 2026.2004.2008.717167 (2026).

(10) R. Y. Stanier, R. Kunisawa, M. Mandel, G. Cohen-Bazire, Purification and properties of unicellular blue-green algae (order Chroococcales). Bacteriol Rev 35, 171-205 (1971).

(11) D. G. Gibson et al., Enzymatic assembly of DNA molecules up to several hundred kilobases. Nature Methods 6, 343-345 (2009).

(12) M. Boehm et al., Comprehensive study on ferredoxin isoforms in the cyanobacterium Synechocystis sp. PCC 6803. bioRxiv 10.64898/2026.04.08.717189, 2026.2004.2008.717189 (2026).

(13) N. Xie, C. Sharma, K. Rusche, X. Wang, Phosphoketolase and KDPG aldolase metabolisms modulate photosynthetic carbon yield in cyanobacteria. The Plant cell 10.1093/plcell/koae291 (2024).

eLife Assessment

This study applies a relatively novel method, ABR (Automated Behavioural Response system), to an arboreal small mammal, providing some valuable results on the responses of these small primates to both natural predators and human stimuli. However, the study in its current form is incomplete in its framing and methodology and could be improved to appeal more broadly to behavioural biologists.

Reviewer #1 (Public review):

Summary:

The authors test specific but related hypotheses regarding anti-predator responses of wild marmoset groups to predator and human playback sounds triggered to play via a motion sensor on camera-trap devices. The differential responses they observe to human noises and natural predator sounds are interesting, but greater inferences are limited due to a lack of clarity in the methods and analyses.

Strengths:

The authors create an excellent experimental design using a customised ABR system for testing the behavioural responses of wild, social-living, tiny, arboreal primates: pygmy marmosets. Much of the work is described with great transparency, and figures and tables are helpful in facilitating this.

Weaknesses:

The current study requires improvement in three areas, in my opinion, to permit readers to better evaluate the validity and importance of these results.

(1) Improve the framing of the paper:

The current title and justification for this study appear to point to a lack of previous studies testing specific hypotheses (line 51/52: "ABRs have not been applied to hypothesis testing". I find this a rather strange argument to make, given that a quick read through of other ABR papers, cited by the authors (e.g., Kasper et al., 2025, Epperly et al., 2021), are testing predictions set by ecological theory in the cascading effects of predator-prey dynamics. To me, even if these are not explicitly stating "X hypothesis" in their paper, they still appear to be studies guided by implicit hypotheses. To say that previous work with ABR did not test hypotheses is presumptuous, in my opinion. The entire paper would be much better appreciated if the authors could reframe the study for its importance to arboreal mammal/ tropical ecology, anthropogenic effects, and so on. Similarly, the authors should avoid use of phrasing such as "this study is the first direct test of .... " (lines 59/60) and should emphasize the true significance of their work, beyond it being the 'first' of something.

Similarly, on line 87, "demonstrating that the ABR system can be used to generate data for hypothesis testing" should be removed, as firstly sufficient sample size for any study depends on a number of study-specific parameters, and the authors do not actually demonstrate this in my opinion, given that many of their models end up suffering from singular fit. This is due to a lack of sample size, and also because they do not actually do any type of power analysis or something similar to demonstrate that they actually assessed sample size. So again, my suggestion is to reframe the paper to focus on the behavioural ecology and conservation-related impacts rather than this emphasis on methodology.

(2) Methods:

The authors generally do a great job providing sufficient detail on the ABR system and how each experiment was designed. Still, there is room for improvement, as I was confused a number of times. I also would recommend that the authors include a limitations section somewhere which considers the drawbacks of their study, in particular the lack of individual identity for behavioural responses of marmosets (especially given that they used focals, it seems), the groups being in close vicinity of one another/potentially related (?), the specific stimuli used, etc.

Points of confusion for me included what the control was for Experiment 1. Line 93 - 70 videos without playbacks are used (Table 1), but it is not clear how these videos were selected, and it is not a suitable control comparison for assessing the difference in behaviour associated with playbacks (playbacks with control sounds are). At most, these videos will give basal rates of behaviour (like vocalizations, etc.), but then it is not clear why these '70 videos' and how they were chosen to avoid bias. So, for example, in line 105 the authors write that focals were more likely to flee when hearing playback stimuli than in these "control" videos, but this is not convincing. If there was no fleeing after playback of control sounds (cicadas, macaws) - i.e., the true control in this experiment - then this should be the comparison that is emphasized.

Can the authors also clarify how they considered/assessed the sound playback level (normally done in Decibels) and if they did not normalize the sound level across the playback stimuli, why not, and what potential effect this could have on the results (i.e., something else to consider for the limitations section)?

Something else not discussed is the rate of exposure to predator and human noise for these wild monkeys. Are these rates within normal range/expectation for these monkeys? Thinking here of the number of videos captured for each group presumably means exposure to a playback unless 'control' videos were videos where no playback sound was emitted (see question above re: control videos). There were a lot more unsuccessful videos than successful ones that the authors could use, so trying to understand the potential impacts of this (see question re: trial/video # below as well).

(3) Analysis:

A few things are unclear and need more explanation in the way the authors conducted their analyses, although they do well to detail all steps of their statistical methods, which was great.

For assessing model fit, it is not clear what exactly was assessed with the 'performance package' line 467, as the authors do not go on to provide us with any results of the performance/fit. Instead, they tell us that the models did not fit well, with no parameter provided (e.g. lines 481-488). I'm familiar with overdispersion as a parameter that is reported for Poisson models (that does not seem to be provided here). Or by looking at changes in model estimates if one datapoint (and/or one group) is removed after another (with replacement, so keeps sample size static). On line 468/469, it says that model fit was assessed via conditional R2; could the authors provide a citation for this practice, and then provide the R2 parameter for the other models that were used/included in the end?

Also, please standardize how the GLMM results are presented. There should be the estimate, SE, Z or t, then p value. (line 99, 137).

Given the high rates of exposure to playbacks, I think the authors should test trial # (or video #) for a potential habituation effect, with earlier captures more likely to draw stronger responses than later video captures for each group.

Reviewer #2 (Public review):

Summary:

The article describes an interesting methodology to test hypotheses about the impact of anthropogenic noise on a small arboreal primate, the pygmy marmoset. The authors used a motion-triggered combination of camera traps and speakers to play back control sounds, avian predator calls, and anthropogenic noise to test the risk-disturbance hypothesis and the distracted prey hypothesis. In addition, the authors implemented a technique that is usually used for larger mammals and has not been used before for smaller arboreal animals. The authors are careful in their interpretation of the results and do not favor one hypothesis over the other. The authors also elaborate extensively in their discussion on how to improve this kind of data collection in the future.

Strengths:

This study provides a method for rapid data collection while minimizing observer impact. The sample size is comparatively large for a wild animal in a reserve, given the overall observation time. The article also benefits from a solid analysis of the data.

Weaknesses:

Though the authors tested two contrasting hypotheses, the discussion would benefit from more detail on the ecological relevance of the observed behaviors.

Author response:

Reviewer #1 (Public review):

Summary:

The authors test specific but related hypotheses regarding anti-predator responses of wild marmoset groups to predator and human playback sounds triggered to play via a motion sensor on camera-trap devices. The differential responses they observe to human noises and natural predator sounds are interesting, but greater inferences are limited due to a lack of clarity in the methods and analyses.

Strengths:

The authors create an excellent experimental design using a customised ABR system for testing the behavioural responses of wild, social-living, tiny, arboreal primates: pygmy marmosets. Much of the work is described with great transparency, and figures and tables are helpful in facilitating this.

Weaknesses:

The current study requires improvement in three areas, in my opinion, to permit readers to better evaluate the validity and importance of these results.

(1) Improve the framing of the paper:

The current title and justification for this study appear to point to a lack of previous studies testing specific hypotheses (line 51/52: "ABRs have not been applied to hypothesis testing". I find this a rather strange argument to make, given that a quick read through of other ABR papers, cited by the authors (e.g., Kasper et al., 2025, Epperly et al., 2021), are testing predictions set by ecological theory in the cascading effects of predator-prey dynamics. To me, even if these are not explicitly stating "X hypothesis" in their paper, they still appear to be studies guided by implicit hypotheses. To say that previous work with ABR did not test hypotheses is presumptuous, in my opinion. The entire paper would be much better appreciated if the authors could reframe the study for its importance to arboreal mammal/ tropical ecology, anthropogenic effects, and so on. Similarly, the authors should avoid use of phrasing such as "this study is the first direct test of .... " (lines 59/60) and should emphasize the true significance of their work, beyond it being the 'first' of something.

Similarly, on line 87, "demonstrating that the ABR system can be used to generate data for hypothesis testing" should be removed, as firstly sufficient sample size for any study depends on a number of study-specific parameters, and the authors do not actually demonstrate this in my opinion, given that many of their models end up suffering from singular fit. This is due to a lack of sample size, and also because they do not actually do any type of power analysis or something similar to demonstrate that they actually assessed sample size. So again, my suggestion is to reframe the paper to focus on the behavioural ecology and conservation-related impacts rather than this emphasis on methodology.

We agree that the reviewer makes a valid point that while we were focusing on explicit hypothesis testing in our statements that the other papers mentioned are making predictions based off ecological theory. We will remove line 87 and make sure to limit these comments in the manuscript. We will reframe the introduction and discussion to better reflect this and change the verbiage throughout to focus on behavioural ecology, the impacts of anthropogenic noise and the conservation implications of the study as well as the novel arboreal aspect of the work. We will also update the title to better fit this framing of the study.

(2) Methods:

The authors generally do a great job providing sufficient detail on the ABR system and how each experiment was designed. Still, there is room for improvement, as I was confused a number of times. I also would recommend that the authors include a limitations section somewhere which considers the drawbacks of their study, in particular the lack of individual identity for behavioural responses of marmosets (especially given that they used focals, it seems), the groups being in close vicinity of one another/potentially related (?), the specific stimuli used, etc.

We do touch on the limitation of not identifying individuals in the discussion (lines 252-258), but we will draw this into a specific section which will address this and the other limitations mentioned here.

Points of confusion for me included what the control was for Experiment 1. Line 93 - 70 videos without playbacks are used (Table 1), but it is not clear how these videos were selected, and it is not a suitable control comparison for assessing the difference in behaviour associated with playbacks (playbacks with control sounds are). At most, these videos will give basal rates of behaviour (like vocalizations, etc.), but then it is not clear why these '70 videos' and how they were chosen to avoid bias. So, for example, in line 105 the authors write that focals were more likely to flee when hearing playback stimuli than in these "control" videos, but this is not convincing. If there was no fleeing after playback of control sounds (cicadas, macaws) - i.e., the true control in this experiment - then this should be the comparison that is emphasized.

For one group, there were only 13 videos without a playback where a marmoset was present. So, we used these 13 videos for this group, and selected 13 videos at random from the other groups to match sample size. For these groups, we assigned each video without a playback but with a pygmy marmoset a sequential number, and then used a random number generator to select 13 videos for analysis. We refer to these videos as controls as they are negative controls, and the reviewer is correct – they do measure basal levels of behaviour. In contrast, the playback of control sounds is a procedural control (Bui et al., 2022). We will change how we refer to these controls in the manuscript to reflect the types of control they are. We use the comparison between negative controls and videos with playbacks in experiment 1 to assess the impact of the playback procedure itself, though the reviewer is correct that our conclusions would be better supported if we explicitly compared the intervention playbacks with the procedural control. We will add post-hoc tests to make this comparison explicit.

Can the authors also clarify how they considered/assessed the sound playback level (normally done in Decibels) and if they did not normalize the sound level across the playback stimuli, why not, and what potential effect this could have on the results (i.e., something else to consider for the limitations section)?

We edited the audios so that they were at similar volume levels. We will update the methods with this information and touch on this in the updated limitations section discussed above.

Something else not discussed is the rate of exposure to predator and human noise for these wild monkeys. Are these rates within normal range/expectation for these monkeys?

Although we do not have information about exposure rates to predators, we do mention levels of exposure of human noise in our methods section on lines 318-320 and we also touch on this in our discussion lines 206-212.

We will update the methods section to be clearer that all groups are exposed to high levels of anthropogenic noise due to their proximity to the ecotourism lodge and community. We will also expand on this in the discussion as a limitation as having a broader array of groups with varying levels of exposure to humans would allow us to see the broader behavioural reactions to these playback stimuli.

For the predators we mention in the methods section line 382 “All four species have been found in the study area (Barker and Papworth, 2024)” but we will further expand on this to provide information on the density of raptors in the area based on our previous study.

Thinking here of the number of videos captured for each group presumably means exposure to a playback unless 'control' videos were videos where no playback sound was emitted (see question above re: control videos). There were a lot more unsuccessful videos than successful ones that the authors could use, so trying to understand the potential impacts of this (see question re: trial/video # below as well).

The number of videos was the total number of times the camera trap triggered. These included videos triggered by another animal or foliage movement where there was no marmoset present, and includes both videos with and without a playback.

We will make this clearer in our description of the results and will report the number of unsuccessful playbacks.

(3) Analysis:

A few things are unclear and need more explanation in the way the authors conducted their analyses, although they do well to detail all steps of their statistical methods, which was great.

For assessing model fit, it is not clear what exactly was assessed with the 'performance package' line 467, as the authors do not go on to provide us with any results of the performance/fit. Instead, they tell us that the models did not fit well, with no parameter provided (e.g. lines 481-488). I'm familiar with overdispersion as a parameter that is reported for Poisson models (that does not seem to be provided here). Or by looking at changes in model estimates if one datapoint (and/or one group) is removed after another (with replacement, so keeps sample size static). On line 468/469, it says that model fit was assessed via conditional R2; could the authors provide a citation for this practice, and then provide the R2 parameter for the other models that were used/included in the end?

We used various tests from the performance package (e.g. check_overdispersion) to test the fit of different distributional models (e.g. Poisson, negative binomial) to the same data. Although some models were not overdispersed and did not show evidence of zero-inflation, they did have singular fits, and/or a conditional R² of 1.0 suggesting overfitting. We therefore did not choose these models. We will clarify this and provide further details of our approach in the manuscript.

The models for the behaviours not reported did not fit well using any distributional model. These behaviours had very low occurrence (0 seconds in most videos), and so there was very sparse data for generating estimates, and very low variation within / between groups and conditions. Therefore, these models generated the errors ‘Model nearly unidentifiable’ and warnings about singular boundaries. These suggest we would not be able to reliably generate model estimates, so we do not report the results. We will change the manuscript to make this reasoning more explicit.

Also, please standardize how the GLMM results are presented. There should be the estimate, SE, Z or t, then p value. (line 99, 137).

We will update the results reporting to be standardised as the reviewer has suggested.

Given the high rates of exposure to playbacks, I think the authors should test trial # (or video #) for a potential habituation effect, with earlier captures more likely to draw stronger responses than later video captures for each group.

We will include video number in a reanalysis of the data.

Reviewer #2 (Public review):

Summary:

The article describes an interesting methodology to test hypotheses about the impact of anthropogenic noise on a small arboreal primate, the pygmy marmoset. The authors used a motion-triggered combination of camera traps and speakers to play back control sounds, avian predator calls, and anthropogenic noise to test the risk-disturbance hypothesis and the distracted prey hypothesis. In addition, the authors implemented a technique that is usually used for larger mammals and has not been used before for smaller arboreal animals. The authors are careful in their interpretation of the results and do not favor one hypothesis over the other. The authors also elaborate extensively in their discussion on how to improve this kind of data collection in the future.

Strengths:

This study provides a method for rapid data collection while minimizing observer impact. The sample size is comparatively large for a wild animal in a reserve, given the overall observation time. The article also benefits from a solid analysis of the data.

Weaknesses:

Though the authors tested two contrasting hypotheses, the discussion would benefit from more detail on the ecological relevance of the observed behaviors.

We thank the reviewer for their comments and we will update the discussion to add more detail on the ecological relevance of the behaviours that we observed.

References

Bui, S., Madaro, A., Nilsson, J., Fjelldal, P.G., Iversen, M.H., Brinchmann, M.F., Venås, B., Schrøder, M.B. and Stien, L.H. 2022. Warm water treatment increased mortality risk in salmon. Veterinary and Animal Science, 17, 100265. https://doi.org/10.1016/j.vas.2022.100265

eLife Assessment

This valuable study identifies an upstream repressive region in the Saccharomyces cerevisiae DDI2/3 promoter and shows that chemical stress induced by cyanamide or MMS is accompanied by reduced histone abundance and decreased nucleosome occupancy across the DDI2/3 locus. Genetic histone depletion is sufficient to enhance DDI2/3 expression, while MNase-seq and genetic analyses support a model in which the transcription factor Fzf1 both activates transcription through its established promoter-binding function and promotes chromatin changes that relieve nucleosome-mediated repression, helping to explain the unusually strong induction of DDI2/3 relative to other Fzf1 targets. The study is significant, as it provides a conceptually interesting extension of Fzf1-mediated stress regulation. The strength of the evidence is solid, although additional controls and a more cautious treatment of whether Fzf1 directly drives nucleosome eviction versus acting indirectly through transcriptional recruitment or chromatin-remodeling factors would further strengthen the mechanistic conclusions.

Reviewer #1 (Public review):

Summary:

In this manuscript, Du et al. identify a putative URS (nucleotides −709 to −229) that contributes to CY/MMS-induced DDI2/3 transcription in the promoter of S. cerevisiae DDI2 through promoter mapping. They further showed that CY/MMS leads to histone loss, and that genetic depletion of histone triggers DDI2/3 expression in an Fzf1-dependent manner. Using MNase-seq, the authors demonstrate that CY treatment leads to nucleosome loss at the DDI2/3 promoter and coding sequences, and that this chromatin remodeling process requires Fzf1. Based on these results, the authors propose that Fzf1 promotes CY-induced DDI2/3 expression through two distinct mechanisms: as a transcription factor and as a regulator of nucleosome occupancy. This dual mode of regulation may contribute to the exceptionally high induction of DDI2/3 relative to other Fzf1 target genes in response to CY.

Strengths:

This manuscript identifies the URS region at the DDI2 promoter that regulates CY/MMS-induced DDI2 expression. In addition, the authors revealed an important role of nucleosome occupancy in regulating DDI2/3 transcription, proposing the intriguing dual regulation model of Fzf1. Overall, this work furthers our understanding of how Fzf1 mediates the increase of DDI2/3 expression in response to CY/MMS treatment.

Weaknesses:

Overall, the study is of interest, and the data are generally convincing; however, several conclusions would benefit from further experimental validation. Certain controls are necessary for several experiments to improve the strength of evidence. Several major points are listed below:

(1) For Figure 6A and Figure 7A, the authors concluded that there are 'joint effects' of CY treatment and histone depletion. However, it is unclear whether CY treatment acts dependently or independently of histone depletion. As shown in Figure 2, both CY and MMS can cause histone reduction. In addition, the depletion system in the RMY102 strain only depletes about half of the H3 (based on the western blots in Figure 5D, E). It would be necessary to test whether H3 abundance is further depleted in CY-treated RMY102 by Western blot.

(2) Proper controls are missing in Figure 6A and Figure 7A. The authors compare gene expression levels in RMY102 + histone depletion + CY/MMS treatment with RMY102 + non-histone depletion. There are two variables here: histone depletion and CY/MMS treatment. It would be more convincing to include RMY102 YPGal+ CY/MMS treatment (5-40 mM), so that the impact of histone depletion and CY/MMS treatment on Fzf1 target gene expression levels would be clearer.

(3) In lines 242-249 and 269-272, the authors compared RMY102 versus BY4741 to conclude that histone depletion affects dose dependency of CY/MMS treatment. However, RMY102 and BY4741 might have different responses to CY/MMS due to strain background differences. Thus, in line with point 2, showing the expression level curves for non-histone depleting RMY102 treated with different doses of CY/MMS would be necessary.

(4) Figure 4 shows that CY and MMS have differential impacts on cell growth, which is intriguing. However, the rest of the data did not provide further insights in regard to this observation. It might be helpful to speculate possible underlying mechanisms in the Discussion session.

Reviewer #2 (Public review):

Summary:

Previous work established that FZF1 is both necessary and sufficient for activation of FZF1 target genes through the CS2 sequence motif, which is present upstream of FZF1-responsive targets. This study extends that model by demonstrating that, in addition to direct binding of FZF1 to CS2 elements, FZF1 can also promote reduced nucleosome-mediated repression, thereby contributing an additional layer of transcriptional regulation.

The authors investigate why FZF1-dependent transcriptional responses exhibit different magnitudes despite FZF1 binding to CS2 elements with similar affinity. Using promoter constructs derived from the DDI2-3 gene, the authors identify a region upstream of the CS2 element that functions as a repressive regulatory element. Based on this observation and publicly available datasets, the authors propose that this repression may be mediated through nucleosome occupancy.

Strengths:

The authors demonstrate that the DDI2-3 promoter contains positioned nucleosomes and show that chemical stress results in decreased histone protein levels and reduced histone-associated transcripts. They further examine whether histone depletion alone is sufficient to activate the DDI2-3 response and find that reduced histone levels increase expression, although chemical treatment produces an additional increase that remains dependent on FZF1. These findings suggest that FZF1 contributes to reductions in nucleosome occupancy at DDI2-3 and SSU1, revealing a second, potentially independent mechanism by which FZF1 regulates transcriptional responses to chemical stress.

Overall, the authors provide strong evidence that nucleosome occupancy influences the magnitude of FZF1-mediated DDI2-3 responses to chemical stress. This work has important implications for understanding how transcriptional networks evolve to generate highly tuned responses by combining multiple regulatory mechanisms acting on shared molecular components.

Weaknesses:

However, several additional considerations should be addressed. While histone depletion may contribute to differential FZF1-mediated responses, alternative mechanisms may also influence the observed transcriptional differences. For example, YHB1 exhibits basal expression that is independent of FZF1, and SSU1 contains the CS1 regulatory element, which can promote increased expression independently of FZF1 responsiveness. Therefore, differences in promoter architecture and the presence of alternative regulatory sequences may also contribute to differential FZF1 responses and should be discussed.

Additionally, the authors should clarify whether nucleosome depletion is directly mediated by the FZF1 ZF5 domain or occurs indirectly as a consequence of RNA polymerase II (Pol II) recruitment. Although the data presented in Figure 9 are consistent with a direct interaction model, the current evidence does not fully exclude the possibility that Pol II recruitment contributes to subsequent nucleosome/histone depletion. Unless there is direct experimental evidence demonstrating that FZF1 ZF5 independently promotes nucleosome remodeling, this alternative mechanism should be acknowledged and considered in the discussion.

Reviewer #3 (Public review):

Summary:

In the manuscript titled "Dual regulation of chemical stress-induced DDI2/1 3 expression by a transcription factor Fzf1 and nucleosome in Saccharomyces cerevisiae" Du et al have discovered a dual role of Fzf1 in transcriptional control of DDI2/3 during cyanamide (CY) or MMS treatment. While previous literature established that Fzf1 regulates multiple targets (DDI2/3, SSU1, YHB1, and YNR064C) by binding the CS2 consensus sequence, it remained unclear why DDI2/3 uniquely undergoes a massive 1,000-fold induction under cyanamide (CY) stress, whereas the others show only a 20- to 40-fold induction. In this work, the authors showed that Fzf1 functions beyond standard transcriptional activation. Using MNase-seq and a series of promoter truncation mutants, the authors mapped Upstream Repressing Sequences (URS) in the DDI2/3 promoter that are heavily occupied by nucleosomes. The authors showed that Fzf1 is essential for chromatin remodelling and nucleosome eviction (specifically at the -2 nucleosome position) to de-repress the DDI2/DDI3 expression.

Strengths:

The two-tier mechanism of action of Fzf1 in controlling the DDI2/3 expression during CY/MMS stress is compelling and novel.

Weaknesses:

While the authors presented the in vivo MNase-seq data that show Fzf1 is necessary for nucleosome displacement, the current study lacks any in vitro mechanistic proof. As the authors acknowledge, it remains unknown whether Fzf1 directly displaces nucleosomes on its own (perhaps through unmapped post-translational modifications induced by chemical stress) or whether its ZF5 activation domain merely acts as a scaffold to recruit separate chromatin remodelling complexes.

To test nucleosome repression, the authors utilised extreme global interventions, such as deleting the SPT10 gene or halting de novo histone synthesis using a galactose-to-glucose medium shift in the RMY102 strain. While these methods effectively deplete histones and induce DDI2/3 up to 30- to 350-fold, completely depleting cellular histones causes massive, pleiotropic secondary effects across the entire genome, which can obscure specific regulatory relationships.

eLife Assessment

The study addresses a timely problem that is of importance in the cfRNA field, and it provides a potentially valuable resource. The use of a common processing pipeline, publicly available code, and variance-partition analyses are notable strengths. However, several central conclusions, particularly those concerning protocol-level effects, the dominance of technical over biological variation, and the proposed universal QC criteria remain more general than the study design and available data can support. Therefore the evidence is currently incomplete and needs to be improved through the revision process.

Reviewer #1 (Public review):

Summary:

In this manuscript, Tuñí-Domínguez et al. present a large-scale, cross-study meta-analysis of 2,356 cell-free RNA sequencing (cfRNA-Seq) samples. The authors aimed to systematically evaluate the impact of pre-analytical variables and library preparation protocols on biological interpretation. By harmonizing publicly available and internally generated datasets through a uniform bioinformatics pipeline, the study seeks to establish standard quality control (QC) metrics and provide evidence-based guidelines for protocol selection in cfRNA-Seq biomarker discovery.

Strengths:

A major strength of the study is the scale and breadth of the harmonized dataset. The use of a common computational pipeline reduces variation arising from differences in bioinformatic processing and enables more direct comparisons among published datasets. The authors examine multiple complementary dimensions of data quality rather than relying on a single sequencing metric. The inclusion of variance-partition analyses, healthy-control-only analyses, and analyses restricted to samples with low gDNA contamination strengthens the evaluation of technical heterogeneity. The public availability of the analysis code and processing configurations further increases the reproducibility and potential utility of this work. The study convincingly demonstrates that technical and pre-analytical factors are major sources of variation across existing plasma cfRNA-seq datasets.

Weaknesses:

Several central conclusions are broader than the current cross-study design can fully support. Protocol category is strongly associated with dataset, laboratory, sample source, and collection procedure, making intrinsic protocol effects difficult to separate from study-specific effects, particularly for categories represented by only one or a few studies. The conclusion that technical variation generally overwhelms biological variation also requires qualification because diverse diseases and cancer types are combined into broad phenotype categories, and many phenotypes are concentrated within individual datasets. The interpretation and additional value of the proposed NG80 and NP80/NG80 metrics require further support, especially given their variable behavior across datasets. Most importantly, the thresholds used in the proposed universal QC framework are not independently validated and are strongly influenced by library-preparation strategy. The current findings therefore support protocol-dependent comparisons and identification of technical trade-offs more strongly than they support a universal definition of library quality.

Major concerns:

(1) Protocol effects remain difficult to distinguish from study-specific effects.

The manuscript interprets Broad Protocol Category as a major determinant of transcriptomic variation. However, protocol, dataset, laboratory, sample source, and collection procedure are strongly interconnected. Although both Dataset and BPC are included in the variance-partition model, several protocol categories are represented by only a limited number of studies.

This is particularly problematic for WRO, which is represented by a single study. Its apparent characteristics therefore cannot be separated from Reggiardo-specific laboratory, cohort, provider, or sample-processing effects. The authors should clarify the stability of the variance-partition results and, where feasible, provide sensitivity analyses. Alternatively, conclusions based on protocol categories represented by one or very few studies should be explicitly presented as study-specific observations rather than broadly validated protocol properties.

(2) The interpretation and robustness of the proposed library-diversity metrics require further support.

The manuscript attributes the high NG80 values in the Block and Sun datasets to gDNA contamination. However, other datasets with similarly low FSR values, including Wang and Giráldez, do not show comparably high NG80 values. This suggests that gDNA contamination alone is insufficient to explain library diversity and that sequencing depth, fragment length, mapping behavior, or library construction may also contribute.

The NP80/NG80 ratio is conceptually reasonable, but its additional value beyond the RNA-biotype composition shown in Figure 3C is unclear, particularly given the large variability observed in WRR datasets. The authors should further examine the relationships among FSR, NG80, sequencing depth, and protocol characteristics and assess the stability of NP80/NG80. If additional validation is not feasible, the interpretation and generality of these metrics should be moderated.

(3) The conclusion that technical variation generally overwhelms biological variation requires qualification.

The manuscript provides convincing evidence that technical heterogeneity is a major source of variation in the combined cross-study dataset. However, the conclusion that donor phenotype contributes only negligible variation may be broader than the analysis supports.

Phenotypes are reduced to healthy, cancer, and non-cancer disease categories despite substantial biological heterogeneity, and many disease groups are concentrated within particular studies. Consequently, disease-specific biological variation may partly be assigned to the Dataset term. A similar issue affects the cellular-origin analysis, where collection center and phenotype are substantially associated in the Chen dataset, making their individual contributions difficult to distinguish.

Where sample sizes permit, the authors should examine more specific disease categories or perform within-dataset analyses. Otherwise, the conclusions should be narrowed to state that technical variation dominates the present heterogeneous cross-study aggregation, rather than implying that phenotype-associated cfRNA signals are generally negligible.

(4) The proposed universal QC framework is insufficiently justified and protocol-dependent.

Figure 6 defines high-quality libraries using NG80 >1,000 together with FSR >20% or FER >75%. However, the manuscript does not explain how these thresholds were selected or validate them against an independent measure of reproducibility, analytical performance, or biomarker utility.

These criteria are also strongly affected by library-preparation strategy. FER and FSR favor libraries enriched for exonic or spliced RNA, whereas the NG80 cutoff disadvantages WRR libraries containing abundant noncoding transcripts. This is difficult to reconcile with the recommendation of WRR for exploratory transcriptomic and microbial analyses.

The authors should justify the threshold selection and assess its sensitivity and protocol dependence. Ideally, the criteria should be validated against an independent performance endpoint. Otherwise, Figure 6 should be reframed as a descriptive comparison, and protocol- or application-specific guidance should replace a universal binary definition of library quality.

Reviewer #2 (Public review):

Summary:

This manuscript systematically evaluates the impact of experimental workflows on plasma cell-free transcriptome sequencing (cfRNA-seq) data. The authors integrate a large number of cfRNA sequencing datasets from multiple publicly available studies and establish a unified bioinformatics framework to systematically assess the effects of experimental workflows, genomic DNA contamination, library diversity, and preanalytical factors on cfRNA transcriptomic profiles.

Strengths:

This study addresses the technical heterogeneity that may hinder cfRNA biomarker discovery and clinical translation and is of substantial value.

Weaknesses:

The effects of disease phenotype, technical confounding, criteria for library quality, and conclusions regarding DNase treatment require further clarification and validation.

Major Points:

(1) The conclusion that donor phenotype explains only a small fraction of transcriptomic variation requires further support from within-study analyses.

The authors conclude from variance partitioning across all studies that phenotype explains only a small fraction of cfRNA transcriptomic variation. However, the included studies encompass different diseases, while phenotype is simplified into healthy, cancer, and non-cancer disease categories, and the overall transcriptomic variation is strongly influenced by study-specific and experimental workflow batch effects. Therefore, the cross-study pooled analysis may underestimate genuine disease-associated cfRNA differences within individual studies conducted under the same experimental workflow.

Recommendation: The authors are encouraged to perform within-study phenotype analyses in datasets that include both healthy controls and disease samples and have sufficient sample size, and to quantitatively estimate the proportion of transcriptomic variance explained by phenotype. For example, the Zhu dataset includes healthy controls and liver cancer samples, and the authors have already observed relatively clear phenotype-associated clustering between the two groups. The contribution of healthy-versus-liver-cancer phenotype to transcriptomic variation could therefore be quantified within this dataset. If similar results are obtained across multiple independent cohorts, the findings could then be summarized across studies. The authors should also note that a low contribution of phenotype to global transcriptomic variance does not necessarily imply that disease-associated cfRNA signals lack biological or clinical relevance.

(2) Comparison of the relative contributions of technical factors and disease phenotype may be affected by confounding.

Figure 1C shows that phenotype is strongly or even completely confounded with technical variables such as collection center and centrifugation protocol in some cohorts. Nevertheless, the variance partitioning analysis across all samples is used to conclude that technical factors are the primary sources of variation, whereas phenotype contributes little. In the presence of such confounding, technical effects and disease-associated biological effects may not be reliably estimated independently, and this conclusion therefore requires more direct validation.

Recommendation: The authors are encouraged to perform an independent within-study variance analysis in cohorts in which technical variables and phenotype are relatively balanced. For example, in the Moufarrej cohort, phenotype is essentially unconfounded with collection center/centrifugation protocol (Cramer's V = 0). Phenotype and relevant technical variables could be included simultaneously in a within-cohort model to quantify their respective contributions to cfRNA transcriptomic variation. If technical factors still explain a larger fraction of variance in such relatively unconfounded cohorts, this would provide stronger support for the central conclusion of the manuscript.

(3) The use of NG80 as a criterion for defining "high-quality libraries" requires further validation.

The authors use NG80 as a metric of library diversity and further apply NG80 > 1,000 as one criterion for defining high-quality libraries in Figure 6. However, because NG80 is based on gene counts, it may be affected by sequencing depth. In addition, gDNA contamination can artificially increase NG80, whereas genuinely abundant non-coding RNAs in WRR libraries can lower NG80. Therefore, a higher NG80 does not necessarily indicate better overall library quality, and the metric may reflect both technical quality and genuine RNA composition.

Recommendation: The authors are encouraged to re-evaluate NG80 after downsampling samples to a common number of mapped fragments and to examine the relationship between NG80 and sequencing depth. The rationale for the NG80 > 1,000 threshold should also be further justified, and the impact of alternative NG80 thresholds on high-quality library classification and the main conclusions should be assessed. Unless there is evidence that this threshold reliably predicts library reproducibility or biomarker-related information content, library diversity and overall library quality should be clearly distinguished, and NG80 should not be presented as a universal criterion for high-quality libraries.

(4) Conclusions regarding DNase treatment should be interpreted more cautiously.

The Toden study did not explicitly report DNase treatment. The manuscript infers that DNase digestion was performed based on the fact that this study originated from the same laboratory as other studies and used a similar workflow; this inference should not be treated as an established experimental fact. In addition, the authors state that double DNase treatment is the most effective approach among non-EB workflows, but this conclusion is mainly based on cross-study comparisons, in which DNase strategy varies together with laboratory, sample handling, and cohort-specific factors. The current evidence is therefore insufficient to establish that double DNase treatment itself is superior.

Recommendation: The authors are encouraged to label the DNase status of the Toden study as "not reported" or "inferred", unless confirmation can be obtained from the original authors. The conclusion that double DNase treatment is the most effective approach should also be tempered, with explicit acknowledgment that it requires direct parallel validation using the same samples under different DNase treatment strategies.

eLife Assessment

This study presents a valuable patient-derived human cell platform for investigating LMNA-associated cardiomyopathy and identifies cyproheptadine as a compound that improves abnormal calcium handling across multiple LMNA variants in vitro. The experimental approaches are generally rigorous and establish reproducible disease-associated cellular phenotypes; however, the evidence remains incomplete because the mechanistic basis of the observed rescue is not established, and the translational conclusions substantially exceed the presented data.

Reviewer #1 (Public review):

Summary:

In this study, the authors use patient-specific induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) from six patients carrying three distinct pathogenic LMNA variants to investigate disease mechanisms underlying LMNA-associated dilated cardiomyopathy (DCM). The authors report shared abnormalities in nuclear morphology, electrophysiology, calcium handling, and contractility across all patient-derived lines and demonstrate that correction of representative LMNA variants by CRISPR/Cas9 rescues several of these phenotypes. They further develop a high-throughput phenotypic drug screening platform using calcium transient measurements and identify cyproheptadine as the only compound among 1,280 FDA-approved drugs that consistently improves calcium transient abnormalities across all patient-derived lines.

The study addresses an important clinical problem and establishes a technically sophisticated patient-derived screening platform. However, while the experimental work is generally well executed, several of the major biological and translational conclusions are not sufficiently supported by the presented data.

Strengths:

The major strength of this study is the development of a patient-derived functional screening platform using multiple LMNA mutations rather than focusing on a single pathogenic variant. The inclusion of six patient-derived iPSC lines representing three distinct mutations increases the generalizability of the observations and allows identification of disease features that appear reproducible across different genetic backgrounds.

The phenotypic characterization is comprehensive and includes nuclear morphology, transcriptomics, manual and automated electrophysiology, calcium imaging, and impedance-based contractility measurements. Importantly, CRISPR-mediated correction of representative LMNA variants provides convincing evidence that the observed cellular abnormalities are directly attributable to the pathogenic variants.

Finally, the implementation of an unbiased high-throughput drug screen using patient-derived cardiomyocytes represents a valuable technical advance that could facilitate therapeutic discovery in inherited cardiomyopathies.

Weaknesses:

The principal weakness of the manuscript is that the central conclusions substantially exceed what is directly demonstrated by the data.

The manuscript repeatedly concludes that dysregulated calcium handling represents a shared pathogenic mechanism underlying LMNA-associated cardiomyopathy. However, the presented experiments establish only that abnormal calcium handling is a shared cellular phenotype across the studied variants. The data do not distinguish whether calcium dysregulation is a primary disease mechanism or whether it is secondary to the numerous upstream abnormalities already known to result from LMNA dysfunction, including altered nuclear architecture, defective mechanotransduction, chromatin remodeling, and transcriptional dysregulation. This distinction is critical because the manuscript repeatedly interprets correction of calcium handling as correction of the underlying disease process without directly demonstrating this relationship.

A related concern is the interpretation of the drug screening results. The primary screen is entirely based on normalization of calcium transient parameters (CTD75, FWHM, and T75-25). Consequently, the screen identifies compounds capable of correcting calcium cycling rather than compounds that necessarily modify disease biology. Although cyproheptadine subsequently improves impedance-derived contractile parameters, it remains unknown whether treatment rescues other defining features of LMNA cardiomyopathy, including abnormal electrophysiology, nuclear defects, transcriptional alterations, or broader cellular stress responses. Thus, the conclusion that cyproheptadine represents a "novel treatment" for LMNA-associated cardiomyopathy is considerably stronger than the evidence presented.

The mechanistic studies are also insufficient to support the proposed mode of action. The manuscript proposes CHRM2 as the likely mediator of cyproheptadine activity primarily because it is the only appreciably expressed known target in the transcriptomic dataset. However, no functional experiments test this hypothesis. Without genetic or pharmacological interrogation of CHRM2, the proposed mechanism remains speculative. Likewise, alternative mechanisms of cyproheptadine action, including serotonergic signaling, histamine receptor antagonism, direct calcium channel modulation, or antioxidant effects, are not investigated.

Another conceptual weakness is that the manuscript promises to distinguish both shared and variant-specific disease mechanisms but ultimately focuses almost exclusively on shared phenotypes. The transcriptomic analyses remain largely descriptive and are not leveraged to identify mutation-specific biological pathways or explain differences among the three LMNA variants. Given the unique cohort assembled in this study, this represents a missed opportunity to generate broader biological insight into LMNA-associated disease.

The transcriptomic analyses themselves would also benefit from more rigorous interpretation. RNA from multiple independent differentiations was pooled prior to sequencing, limiting assessment of biological variability and reducing confidence in statistical inference. Similarly, many functional analyses emphasize the number of wells, recording sweeps, or individual cells while the number of independent biological differentiations is less prominently presented. Greater emphasis on biological replication would strengthen confidence in the robustness of the findings.

Finally, the translational implications of the work should be interpreted more cautiously. All therapeutic studies are performed in relatively immature two-dimensional iPSC-derived cardiomyocytes. No validation is presented in engineered heart tissues, multicellular cardiac organoids, animal models, or human tissue. The current evidence supports the conclusion that cyproheptadine is a promising in vitro phenotypic modifier rather than an established therapeutic candidate for LMNA-associated cardiomyopathy.

Overall, this study establishes a valuable patient-derived platform for investigating LMNA-associated cardiomyopathy and demonstrates the utility of functional high-throughput screening in identifying compounds that improve disease-associated cellular phenotypes. However, the manuscript currently overstates both the mechanistic significance of calcium dysregulation and the therapeutic implications of cyproheptadine. A more restrained interpretation of the findings together with additional mechanistic validation would substantially strengthen the impact of the work.

Reviewer #2 (Public review):

Summary:

The authors conducted a functional high-throughput drug screening using hiPSC-CMs derived from patients with LMNA-DCM. Cyproheptadine emerged as a therapeutic candidate.

Strengths:

The screen appears well designed.

Weaknesses:

The single candidate that emerged from the screen, cyproheptadine, raises issues with potency. In addition, validation studies that are both expected and necessary for a drug proposed as a novel therapeutic for human cardiomyopathy have not yet been performed. Rigor could be improved once the basic mechanistic and validation studies discussed below have been performed.

eLife Assessment

This study presents a valuable finding on impacts of the common driver mutations APC, KRAS G12D, and TP53 in murine colorectal organoids. In particular, the authors examine how the order of APC and TP53 acquisition influences tumor phenotype. The evidence supporting the claims of the authors is solid. The work will be of interest to scientists working in the field of breast cancer.

Reviewer #2 (Public review):

Summary:

This study addresses an important and timely question in colorectal cancer biology by systematically examining the effects of the common driver mutations APC, KRAS G12D, and TP53 in murine colorectal organoids, with particular emphasis on how the order of APC and TP53 acquisition influences tumor phenotype. These mutations are well known to be frequent, truncal, and often co-occurring in colorectal cancer. While it is increasingly appreciated that mutational order can shape tumor behavior, studies directly comparing the phenotypic consequences of alternative APC-TP53 mutation orders remain rare. This work therefore addresses a relevant and timely question.

Strengths:

A major strength of the study is its focus on previously unexplored biology, combined with the generation of multiple isogenic murine organoid models with controlled mutational sequences. The authors employ careful and robust quality control of the CRISPR-mediated alterations, and the inclusion of both in vitro and in vivo experiments strengthens the relevance of the work.

Weaknesses:

There are, however, several limitations that should be considered when interpreting the findings. First, KRAS G12D activation is used as the initiating alteration, whereas APC loss is generally believed to be the initiating event in most human colorectal cancers. Second, the analysis is restricted to comparing only two mutation orders (KAT versus KTA), which limits the breadth of conclusions that can be drawn about mutation ordering more generally. Finally, key RNA-sequencing and in vivo experiments rely on a limited number of isogenic lines, which constrains interpretability.

The study aimed to systematically investigate how the accumulation and sequence of driver mutations influence colorectal cancer initiation. The data provide intriguing evidence that the relative timing of APC and TP53 loss may impact tumor initiation and survival in a hostile microenvironment. However, given the limited number of biological replicates, these observations should be interpreted with caution and would benefit from further validation.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this study, Li et al. used genetically engineered murine intestinal organoids to investigate how the temporal order of oncogenic mutations influences cell state and tumourigenicity of colorectal epithelial cells. By sequentially introducing Apc and Trp53 loss-of-function mutations in alternate orders within a Kras^G12D background, the authors generated isogenic organoid lines for both in vitro and in vivo characterisation. Bulk RNA-seq reveals expected transcriptional changes with relatively modest differences between the two triple-mutant configurations (KAT vs KTA). The key finding emerges from transplantation assays: while KAT and KTA organoids show equivalent tumourigenic potential in immunodeficient mice, only KAT organoids form tumours in immunocompetent hosts (5/10 vs 0/10), suggesting that mutation order shapes susceptibility to immune-mediated clearance. The experiments are well-executed, and the conclusions are generally supported by the data.

Strengths:

The experimental system is well-designed for the question. By combining a Kras^G12D transgenic background with sequential CRISPR-mediated knockout of Apc and Trp53 in alternate orders, the authors generated truly isogenic organoid lines that differ only in mutational sequence. This is technically non-trivial and provides a clean platform for dissecting order effects, a question otherwise difficult to address experimentally.

The authors performed comprehensive baseline characterisation of these organoids, including morphological and histological assessment, quantification of organoid-forming efficiency and proliferation, and bulk RNA-seq profiling. While these analyses revealed no major differences between KAT and KTA organoids, and the observed enhancement of epithelial stemness upon Apc loss and proliferative advantage conferred by Trp53 loss are largely expected, the systematic nature of this characterisation establishes a useful methodological template for future organoid-based studies.

The authors further investigated the functional impact of mutational order using subcutaneous transplantation assays. By comparing tumour formation in immunodeficient versus immunocompetent hosts, the authors uncover a genuinely unexpected finding: KAT and KTA organoids behave equivalently in the absence of adaptive immunity, but diverge dramatically when immune pressure is applied (KAT: 5/10; KTA: 0/10). This observation is arguably the most compelling aspect of the study and opens an interesting line of inquiry.

We greatly appreciate your comments on this study.

Weaknesses:

The authors acknowledge that initiating with Kras^G12D does not reflect the typical human sporadic CRC trajectory, where APC loss is usually the first event. While this design choice was pragmatic, it means the observed order effects are contextualised within an artificial starting point. It remains unclear whether the Apc/Trp53 order would matter in a Kras-wild-type background, or whether the Kras-driven cellular state is a prerequisite for these phenotypes to emerge.

We agree with the reviewer that initiating tumorigenesis with Kras^G12D does not fully recapitulate the most common trajectory of sporadic human CRC, where APC loss typically occurs first. We had noted this point in the original Discussion and further clarified it more explicitly in the Introduction part of the revised manuscript as shown in Line 97–103.

Our experimental design was intended to establish a controlled and genetically tractable system to interrogate the principle of mutation order effects. In this context, Kras^G12D activation provides a stable oncogenic baseline that facilitates sequential genome engineering and comparison of isogenic lines.

Although APC loss is frequently the initiation event, a recent study has suggested that Kras^G12D priming can reshape the selective landscape for subsequent driver events, including Apc alterations (PMID: 41339549). Consistent with this notion, our data indicate that Kras^G12D activation induces a permissive oncogenic cellular state that may influence the phenotypic consequences of later mutations. We therefore speculate that the Kras^G12D-primed context may contribute to the observed order-dependent effects.

We agree that testing Apc Trp53 order in a Kras-wild-type background would be an important future direction, and we have pointed this out explicitly in the revised Discussion as shown in Line 549–554.

Subcutaneous implantation provides a tractable readout of tumourigenicity, but the cutaneous immune microenvironment differs substantially from that of the intestinal mucosa. Given that the central claim concerns immune-mediated selection, orthotopic transplantation would more directly test whether the observed order effects hold in a physiologically relevant context.

In the present study, we employed subcutaneous transplantation as a widely used platform to assess tumorigenic potential under controlled immune conditions. This approach offers high reproducibility, straightforward tumour monitoring, and has been broadly applied in organoid-based cancer studies in both immunodeficient (PMID: 23273993, 23776211, 32209571, 33055221) and immunocompetent (PMID: 32209571, 33055221, 41672595) settings.

Importantly, our primary goal was to determine whether mutation order influences susceptibility to immune-mediated clearance, rather than to model the full complexity of the intestinal niche. The clear divergence between KAT and KTA specifically in immunocompetent hosts supports the existence of intrinsic mutation order-dependent immune vulnerability.

Nevertheless, we fully agree with the reviewer that orthotopic transplantation would provide a more physiologically relevant immune microenvironment and represents also an important direction for future investigation. We have explicitly discussed this limitation and highlight orthotopic validation as an important future direction in the revised Discussion as shown in Line 563–571.

The ssGSEA comparison involves only 14 ATK tumours, and the key comparisons (Figure 6E) yield borderline significance (p=0.052). More fundamentally, since mutation order cannot be inferred from the clinical samples, the authors are correlating organoid-derived IFN signatures with tumour immunophenotypes without direct evidence that these patients' tumours followed a KAT-like trajectory. The reasoning becomes circular: KAT organoids define the signature used to identify KAT-like clinical tumours.

We thank the reviewer for raising this important point. We would like to clarify that our intention was not to infer the actual mutation order in clinical samples, which indeed cannot be reliably reconstructed from bulk tumour RNA-seq data.

Instead, our goal was to determine whether the transcriptional programs distinguishing KAT and KTA organoids could be observed in human CRC cohorts. In this context, the organoid-derived IFN-related signature was used as a molecular reference to assess potential clinical correlation, rather than to classify tumours by evolutionary trajectory.

We agree that the statistical significance in Figure 6E is modest (p = 0.052), and we have revised the text (Line 478–480) to present this analysis more cautiously as a suggestive trend rather than definitive evidence. We also clarified this limitation explicitly in the revised manuscript (Line 537–542) to avoid overinterpretation.

Furthermore, the most striking finding of the study, that KTA organoids fail to form tumours in immunocompetent hosts while KAT organoids can, lacks a mechanistic follow-up. The transcriptomic differences between KAT and KTA are modest when cultured as monocultures, yet their in vivo fates diverge dramatically. The authors do not address why these subtle intrinsic differences translate into such divergent immune susceptibility, nor do they characterise the immune response adequately (beyond limited CD4/CD8 IHC at tumour peripheries).

We thank the reviewer for this important point. We agree that the mechanistic basis underlying the differential immune susceptibility between KAT and KTA remains incompletely resolved.

A practical limitation of the current study is that KTA grafts failed to establish tumours in immunocompetent hosts, which precluded downstream histological and immune profiling of established lesions. As a result, our in vivo immune characterization of KTA grafts is nearly impossible.

Nevertheless, our transcriptomic analyses indicate that KAT and KTA organoids differ in interferon-response and immune-related programs prior to transplantation, and those differentially expressed genes were consistently preserved in tumour cells derived from immunodeficient hosts. These results suggest the presence of intrinsic tumour-cell-autonomous differences may influence immune recognition or clearance.

We have expanded the Discussion to outline several non-mutually exclusive mechanisms that could account for this phenotype, including altered interferon responsiveness, differential antigen presentation capacity, and changes in tumour cell-intrinsic immune escape programs (Line 527–533). These hypotheses are consistent with the transcriptional differences observed prior to transplantation and provide a framework for future mechanistic investigation. We agree that deeper immune profiling (e.g., immune infiltrate composition, antigen presentation status, and functional immune assays) will be important to fully elucidate the mechanism and represents a key direction for future work.

Reviewer #2 (Public review):

Summary:

This study addresses an important and timely question in colorectal cancer biology by systematically examining the effects of the common driver mutations APC, KRAS G12D, and TP53 in murine colorectal organoids, with particular emphasis on how the order of APC and TP53 acquisition influences tumor phenotype. These mutations are well known to be frequent, truncal, and often co-occurring in colorectal cancer. While it is increasingly appreciated that mutational order can shape tumor behavior, studies directly comparing the phenotypic consequences of alternative APC-TP53 mutation orders remain rare. This work, therefore, addresses a relevant and timely question.

Strengths:

A major strength of the study is its focus on previously unexplored biology, combined with the generation of multiple isogenic murine organoid models with controlled mutational sequences. The authors employ careful and robust quality control of the CRISPR-mediated alterations, and the inclusion of both in vitro and in vivo experiments strengthens the relevance of the work.

We greatly appreciate your comments on this study.

Weaknesses:

There are, however, several limitations that should be considered when interpreting the findings. First, KRAS G12D activation is used as the initiating alteration, whereas APC loss is generally believed to be the initiating event in most human colorectal cancers.

We sincerely thank the reviewer for their insightful comments regarding the initiation of tumorigenesis with a Kras mutation rather than the more canonical Apc loss, which was also raised by the reviewer #1. We fully agree that the Apc-first represents the most prevalent sequence in human colorectal cancer (CRC), We have more clearly explained the rationale for our experimental design in the revised Introduction part as outlined in our response to reviewer #1.

Second, the analysis is restricted to comparing only two mutation orders (KAT versus KTA), which limits the breadth of conclusions that can be drawn about mutation ordering more generally.

We thank the reviewer for this critical concern, which we agree is essential for strengthening the robustness and generality of our findings. However, as a proof-of-concept study of Apc and Trp53 loss, two major oncogenic events in CRC, serves as a biologically meaningful starting point for dissecting order-dependent effects. Although it is of great significance to compare all six possible mutation orders of these three driver genes, generating and thoroughly characterizing all genotypes (with identical replicates) represents a substantial undertaking beyond the scope of this initial study.

Finally, key RNA-sequencing and in vivo experiments rely on a single isogenic line, which substantially constrains interpretability.

The aim of the study was to systematically investigate how mutation accumulation and order influence colorectal cancer initiation. While the data suggest that the relative timing of APC and TP53 loss may be particularly important for tumor initiation, the absence of biological replication makes it difficult to draw robust conclusions. Engraftment efficiency and tumor behavior can be influenced by many factors for a single clone, including additional passenger mutations acquired during culturing, as well as epigenetic differences that are independent of the engineered mutations.

We thank the reviewer for this concern. We apologize that we have not made a clear presentation of our data source. Indeed, for all major in vitro and in vivo assays of double and triple mutants, we analyzed at least two independently derived clones per genotype. These independent clones harbour distinct mutations in target genes and were treated as biological replicates throughout the study.

To improve clarity and transparency, we have revised the relevant figure legends and further provided a Table S5 to explicitly indicate the clonal origin of each data point throughout the study.

eLife Assessment

This important study shows that prenatal alcohol exposure produces lasting changes in amyloid precursor protein processing, including altered APP C-terminal fragments and Notch intracellular domain levels, that persist into adulthood and track with progressive spatial learning and memory deficits, in both wild-type and 3xTg-AD mice. The convincing study was carefully designed and combined biochemical, histological, and behavioral approaches across multiple ages. The work will interest researchers studying fetal alcohol spectrum disorders, Alzheimer's disease risk, and the developmental origins of neurodegeneration. Nonetheless, reviewers identified opportunities to strengthen the conclusions with additional samples and controls.

Reviewer #1 (Public review):

Summary:

The manuscript by Montenegro and colleagues reports a uniquely significant set of compelling results from a carefully designed study. The findings are fundamental and should substantially advance our understanding of whether prenatal exposure to high levels of alcohol produces neural changes that are precursors to the development of Alzheimer-like pathology in an animal model of Fetal Alcohol Spectrum Disorder (FASD). The quest was to test whether prenatal exposure to high-dose alcohol in the mouse would result in selective damage that would result in Alzheimer disease-like cellular disorder and mnemonic impairment. Both outcomes emerged and were exacerbated in relevant transgenic mice. The untoward effect on memory endured and even worsened with age.

Strengths:

The authors noted the importance of using a validated animal model to test their hypotheses related to AD-like outcomes because human postmortem data are unavailable and even in vivo data are limited to younger FASD cohorts. The authors further noted limitations, which also anticipate experiments that could chart the temporal course of the effect and windows of prenatal alcohol exposure that result in damage or resilience. In short, as the authors state on lines 435-7, "These data provide the first experimental evidence that developmental alcohol exposure impacts core proteolytic pathways central to AD/ADRD pathogenesis."

Weaknesses:

Addressing the following would clarify several points in an already well-written paper:

(1) It would be useful to have a timeline of the study, much like the one provided for the water maze protocol. On that timeline, please include the sample sizes examined, ages at exposure, and other pertinent procedures, indicating which animals remained alive for testing, etc.

(2) What were the attrition rates for each study group?

(3) What are the human age equivalents of the maternal mice?

(4) It would be helpful to see a graph of the BECs of each animal relative to the doses given. That would clarify how the alcohol exposure amount and timing are the same and where they are different for all exposed mice, given that the alcohol levels were somewhat different by group, as noted in the Methods. Were these BEC differences at all related to group differences in outcome measures or memory performance?

Reviewer #2 (Public review):

Summary:

In this study, the impact of prenatal alcohol (PAE) on amyloid precursor protein (APP) C-terminal fragments and notch intracellular domain (NICD) levels in adulthood is measured in 3xTg-AD mice.

Prenatal alcohol alters gamma secretase activity with development and aging. This could have implications for Alzheimer's disease risk in populations without inherited Alzheimer's risk genetics.

Strengths:

Strengths include the model, the use of orthogonal approaches, and the rigorous, high-quality data.

Weaknesses:

Some figures lack prenatal alcohol treatment in the 3xTg-AD mice.

Some overstatements should be tempered. For instance, one cannot conclude that the changes in CTFs are driving the changes in learning and memory (as suggested in the last line of the abstract) without a direct intervention testing this. For instance, though PAE caused a more robust learning deficit at 6 mo in WT, the impact on CTFs was less than it was at 3 mo. PAE did not significantly change CTFs or learning/memory in 3xTg-AD mice at 4 months, suggesting the genotype effect takes over at this point. The text should be adjusted to reflect this.

Conclusion:

In summary, this is a rigorous assessment of the long-term impacts of PAE on CTFs and learning/memory in adult WT and 3xTg-AD mice.

Reviewer #3 (Public review):

Summary:

The goal of this study was to test the hypothesis that prenatal alcohol exposure (PAE) can affect Alzheimer's disease (AD) pathogenesis using biochemical proxies, histology, and a behavioral paradigm sensitive to AD-related memory decline. Major strengths include the breadth of techniques used, consideration of different AD-related molecular markers, and the use of different ages as well as appropriate controls. The authors largely achieved their aims to show that PAE does affect amyloid precursor fragments (CTFs) and notch signaling very early on as well as long-term effects in adulthood, which may uncover a previously underappreciated mechanism that may contribute to AD-related neuropathology and behavioral outcomes during lifespan.

Strengths:

(1) Several techniques are used to address molecular, behavioral, and histological PAE-related changes.

(2) There is use of appropriate controls and an AD-relevant mouse model.

(3) Different ages are used to address age-related and long-term effects in AD and control mice.

(4) The novelty of results shows early changes in amyloid-related processes, affected by PAE.

Weaknesses:

(1) It is unclear as to whether there are sex differences, particularly in the adult cohort.

(2) More clarity is needed on sample size per cohort and whether mice that were used for anatomy and biochemical analyses were previously used for behavior. Including a table and noting any overlap would be useful.

(3) In many instances, two-way ANOVAs with treatment (PAE vs vehicle) and genotype as factors will be useful to report (e.g., Figure 1).

(4) In Figure 5 and line 253, it is stated that older mice have more severe deficits, but there are no direct statistical comparisons with younger AD mice.

(5 Lines 270-271 refer to mice as "presymptomatic", but these mice do have behavioral symptoms. Do the authors mean no neuropathology yet? Any data showing lack of robust neuropathology would be useful.

Author response:

Public Reviews:

Reviewer #1 (Public review):

(1) It would be useful to have a timeline of the study, much like the one provided for the water maze protocol. On that timeline, please include the sample sizes examined, ages at exposure, and other pertinent procedures, indicating which animals remained alive for testing, etc.

We agree with the reviewer that a timeline would be helpful for clarifying the PAE exposure paradigm and the subsequent sample collection and processing. Sample sizes vary across the different analyses; therefore, we have indicated the sample size for each experiment in the corresponding figure. To further improve clarity, we will add Author response image 1, which will include a schematic of the overall experimental timeline, including the PAE regimen, collection time points, and sample processing, as shown below.

Author response image 1.

(2) What were the attrition rates for each study group?

We thank the reviewer for raising this important point. There was no attrition of animals within the experimental groups; the number of animals included at the beginning and end of the study remained the same. However, we did observe a reduction in litter size following prenatal alcohol exposure (PAE). In our 3xTg-AD colony, litters typically consisted of approximately 8 pups under control conditions, whereas PAE litters occasionally contained 4–6 pups. Thus, the reduction in animal numbers reflects decreased litter size associated with PAE rather than attrition during the study. We would also like to clarify that the primary scope of this study was not to provide a terminal/end-point analysis of disease progression, but rather to investigate the emergence of Alzheimer’s disease (AD)-related phenotypes during early adulthood following PAE. Accordingly, our longitudinal experimental design focused on identifying the earliest molecular, synaptic, behavioral, and neuropathological alterations that emerge during this period. This approach allowed us to examine whether PAE accelerates or precipitates the onset of AD-related symptomatology in the 3xTg-AD model, rather than following the animals through advanced disease stages. We will clarify this rationale in the revised manuscript.

(3) What are the human age equivalents of the maternal mice?

We appreciate the reviewer’s question regarding the age of the maternal mice. The dams used in our study were young adult females (2 to 3 months of age) at the time of breeding. Because chronological age does not translate linearly between mice and humans, particularly during development and reproductive maturation, we have avoided assigning a precise human-age equivalent. Based on established comparative developmental frameworks and calculations, these animals represent a 20 years old young-adult in the reproductive stage, rather than an advanced maternal-age condition [1]. We will clarify this point in the revised manuscript.

(4) It would be helpful to see a graph of the BECs of each animal relative to the doses given. That would clarify how the alcohol exposure amount and timing are the same and where they are different for all exposed mice, given that the alcohol levels were somewhat different by group, as noted in the Methods. Were these BEC differences at all related to group differences in outcome measures or memory performance?

We appreciate the reviewer’s suggestion to provide a more detailed representation of the BEC data. We agree that displaying the BECs for individual animals would provide additional clarity regarding the consistency of alcohol exposure across groups. We have therefore included the individual BEC values in the new Figure 1. We observed some variability in BECs between the 3xTg-AD and B6129 groups, as noted in the Methods. Importantly, however, the average alcohol consumption was comparable between the two genotypes, indicating that the difference in BECs was not due to differences in the amount of alcohol consumed. All dams in the PAE groups reached BECs above 0.08 g/dL, the commonly used legal blood alcohol concentration limit in the United States, supporting the use of our paradigm as a binge-like alcohol exposure model. We further examined whether the variability in BECs was associated with the differences observed in outcome measures, including memory performance. We did not find evidence that the differences in BECs accounted for the group differences in behavioral or molecular outcomes. Thus, although some intergroup variability in BECs was present, the overall alcohol exposure was comparable, and the observed phenotypic differences were not attributable to differences in alcohol consumption.

Reviewer #2 (Public review):

(1) Some figures lack prenatal alcohol treatment in the 3xTg-AD mice.

We appreciate the reviewer’s observation and agree that the rationale for the different experimental groups across the figures should be clarified. The primary focus of this study is to characterize the effects of prenatal alcohol exposure (PAE) in wild-type B6129 mice, with the 3xTg-AD mice serving primarily as a disease-model reference to determine whether the effects observed following PAE in wild-type animals overlap with or resemble features of AD pathology. Accordingly, the initial figures focus on the effects of PAE in B6129 mice and include the non-exposed 3xTg-AD group as a reference for the AD phenotype. The last two figures specifically address the effects of PAE in the 3xTgAD model, with the 3xTg-AD mice becoming the experimental subject of interest rather than serving solely as a disease reference. For this reason, the PAE-3xTg-AD group is not included in the earlier figures, whereas it is included in the final two figures where the effect of PAE on the AD model is directly evaluated. We will clarify this experimental rationale in the revised manuscript and figure legends.

(2) Some overstatements should be tempered. For instance, one cannot conclude that the changes in CTFs are driving the changes in learning and memory (as suggested in the last line of the abstract) without a direct intervention testing this. For instance, though PAE caused a more robust learning deficit at 6 mo in WT, the impact on CTFs was less than it was at 3 mo. PAE did not significantly change CTFs or learning/memory in 3xTg-AD mice at 4 months, suggesting the genotype effect takes over at this point. The text should be adjusted to reflect this.

We appreciate the reviewer’s careful consideration of this point. We agree that the relationship between APP CTF accumulation and learning and memory deficits should not be interpreted as causal in the absence of a direct intervention experiment. We were careful in choosing the wording throughout the manuscript to describe these findings as associated changes rather than evidence of a causal interaction. Our data demonstrate the presence of APP CTF accumulation and learning and memory deficits following PAE, but they do not establish that CTF accumulation directly drives the behavioral phenotype. We therefore will temper the language in the Abstract and throughout the manuscript to avoid overstatement. We also acknowledge that the relationship between these phenotypes is not necessarily linear across age: although PAE produced a more pronounced learning deficit at 6 months in B6129 mice, the magnitude of APP CTF accumulation was greater at the earlier time point. Importantly, we consider the possibility that the greater APP CTF accumulation observed at earlier ages may represent an early molecular insult whose functional consequences become evident later in life. In this context, the temporal dissociation between the molecular and behavioral phenotypes could be consistent with a “two-hit” model, in which an early-life insult induced by PAE creates or primes a pathological vulnerability that subsequently manifests as cognitive dysfunction with ageing [2]. We recognize, however, that this interpretation remains a hypothesis and would require longitudinal mechanistic studies to establish. This temporal relationship may also contribute to the broader concept of early-life origins of AD/ADRD, suggesting that prenatal environmental exposures may initiate molecular alterations during neurodevelopment that remain detectable or predispose the brain to later-life dysfunction. Similarly, the absence of significant changes in APP CTFs or learning and memory in 4-month-old 3xTg-AD mice suggests that the effects of the AD genotype may become dominant at this stage. Consistent with this interpretation, we state in the Discussion that future studies are needed to identify and experimentally test the direct molecular pathways affected by PAE that ultimately contribute to learning and memory impairment. We will revise the text accordingly to make this distinction clear while highlighting the potential significance of an early molecular insult preceding the later emergence of behavioral phenotypes.

Reviewer #3 (Public review):

(1) It is unclear as to whether there are sex differences, particularly in the adult cohort.

We appreciate the reviewer’s comment regarding potential sex differences. We agree that considering sex as a biological variable is important, particularly for the adult cohorts. To address this point, we will include identifying marks for male and female animals separately in our plots where sample size permits. This will allow the reader to better evaluate potential sex-dependent effects of PAE and to determine whether the observed phenotypes are consistent across sexes. We will also clarify this approach in the revised manuscript.

(2) More clarity is needed on sample size per cohort and whether mice that were used for anatomy and biochemical analyses were previously used for behavior. Including a table and noting any overlap would be useful.

We appreciate the reviewer’s suggestion and agree that greater clarity regarding the sample sizes and use of animals across analyses is important. The sample size for each cohort and experimental group is indicated in the corresponding figures, and we will make this information more explicit in each figure legend to facilitate interpretation. Animals that underwent behavioral testing were subsequently used for biochemical analyses, allowing us to examine molecular changes in the same animals in which behavioral phenotypes were characterized. In contrast, for the neonatal cohort, we performed both anatomical and biochemical analyses, to assess the distribution and extent of APP CTF accumulation across the brain during this early developmental period. This approach was selected to provide a broader assessment of the spatial distribution of APP CTF accumulation at birth. We will clarify these experimental details in the revised Methods and figure legends.

(3) In many instances, two-way ANOVAs with treatment (PAE vs vehicle) and genotype as factors will be useful to report (e.g., Figure 1).

We appreciate the reviewer’s suggestion regarding the use of two-way ANOVA with treatment and genotype as factors. However, we respectfully disagree that this approach is appropriate for all of the analyses presented in this manuscript. Our experimental design and the specific biological questions addressed in each experiment were not uniform across cohorts. In particular, the primary objective of the study was to characterize the effects of PAE in B6129 mice, with the 3xTg-AD mice serving primarily as a disease-model reference, while the effects of PAE in the 3xTg-AD model were specifically examined in the later experiments. Therefore, combining genotype and treatment as factors across all datasets would not always reflect the experimental questions or the structure of the cohorts. In addition, some experiments did not include all four groups, making a two-way ANOVA inappropriate for those analyses. Nevertheless, we agree that a two-way ANOVA may be informative for experiments in which both genotype and treatment are fully represented and the experimental design supports this analysis. We will therefore consider and apply two-way ANOVA, where appropriate, to those datasets, including evaluation of the main effects of genotype and treatment and their interaction. We will clarify the statistical approach and its rationale in the revised Methods and figure legends.

(4) In Figure 5 and line 253, it is stated that older mice have more severe deficits, but there are no direct statistical comparisons with younger AD mice.

We appreciate the reviewer’s observation. We agree that, in the absence of a direct statistical comparison between age groups, the statement that older mice have “more severe deficits” may be too strong. Our intention was to describe the apparent progression of the phenotype across age rather than to imply that we had statistically demonstrated an age-dependent increase in severity. We have therefore revised the text in Figure 5 and at line 253 to use more cautious language, describing the greater magnitude of the observed deficits in older mice without implying a direct statistical comparison between age groups. We agree that a formal conclusion regarding age-dependent progression would require a statistical analysis, which we will include in the revised manuscript.

(5) Lines 270-271 refer to mice as "presymptomatic", but these mice do have behavioral symptoms. Do the authors mean no neuropathology yet? Any data showing lack of robust neuropathology would be useful.

We appreciate the reviewer’s careful observation. We agree that the term “presymptomatic” was not sufficiently precise, particularly because the mice already exhibit measurable behavioral alterations at this age. Our intention was not to suggest that these animals were free of phenotypic abnormalities, but rather that they were at an early stage of disease progression, before the emergence of robust neuropathological features. We have therefore revised the terminology to avoid referring to these mice as “presymptomatic.” Instead, we describe them as being in an early stage of disease development, characterized by emerging behavioral and molecular alterations but without the extensive cardinal neuropathology typically associated with later stages of the 3xTg-AD phenotype. We agree that the distinction between behavioral symptoms and neuropathological progression is important. In this study, our focus was on the emergence of early AD-related phenotypes during young adulthood, rather than on establishing the absence of neuropathology. We have therefore avoided making a definitive claim regarding the lack of neuropathology and have revised the text to more accurately reflect the scope of our data.

Additional References

(1) Dutta, S. & Sengupta, P. Men and mice: Relating their ages. Life Sci. 152, 244–248 (2016).

(2) Gunn, J. S. et al. Exploring the ‘Multiple-Hit Hypothesis’ of Neurodegenerative Disease: Bacterial Infection Comes Up to Bat’. Frontiers in Cellular and Infection Microbiology | www.frontiersin.org 1, 138 (2019).

eLife Assessment

This fundamental work significantly advances our understanding of the circuit-level implementation of predictive processing by elucidating the functional influence between putative prediction error neurons in layer 2/3 and putative internal representation neurons in layer 5. The evidence demonstrating that neither the hierarchical nor the non-hierarchical variant of predictive processing fully accounts for the presented data is convincing. Moving forward, this line of work would benefit from explicitly comparing different theories, thereby clearly articulating the points raised in this paper.

Reviewer #1 (Public review):

Vasilevskaya and Keller test different models of cortical function through the lens of predictive processing, a powerful framework for the brain to learn and predict the statistics of the world via generative internal models. The authors use a clever combination of behavioral perturbations in closed-loop and open-loop visuomotor virtual reality assays, a paradigm the Keller lab pioneered and used effectively in the past decade, in conjunction with two photon imaging of neuronal calcium responses and targeted optogenetic perturbations of activity. They specifically put to test proposed hierarchical vs. non-hierarchical circuit implementations of predictive processing by analyzing the logic of inter-lamina interactions (superficial vs. deep; L2/3 vs. L5/6).

The authors conclude that both versions of predictive processing architectures they analyze are likely invalid and instead formulate an alternative novel model of cortical function based on a recently developed machine learning algorithm for self-supervised learning (joint embeddings of predictive architectures, JEPA) and its further refinements. JEPA borrows elements from predictive processing engaging two encoder networks and training the output of one network to predict the output of the other. In their new model of cortical computations, prediction errors neurons in L2/3 compare the deep layers (L5/6) activity, which is taken as a teaching signal, to a local, L2/3 prediction of this latent representation.

Specifically, the authors build on their previous work and reports from other groups that different sets of L2/3 neurons compute positive prediction errors (fire when sensory stimuli appear unexpectedly with respect to the movements of the animal; e.g., grating onsets in the absence of locomotion) and respectively negative prediction errors (fire when sensory stimuli are absent, while the brain expected them to be present; e.g. mice locomote but visual flow is suddenly halted - visuomotor mismatches). These L2/3 positive and negative prediction error neurons exchange messages with neurons in the deeper cortical layers that, the authors propose, build an internal representation (R) of the sensory stimuli given the animals' movements.

In the hierarchical model, internal representation neurons (R) are supposed to act as a teaching signal for both types of prediction error neurons; the output of the positive prediction error neurons is assumed to suppress activity of R such that the error between the teaching signal and the prediction is minimized; similarly, in the non-hierarchical version, R serves as a prediction for the prediction error neurons, and in turn it receives excitatory drive from the positive prediction error neurons and negative input from the negative prediction error neurons.

The authors find that the functional impact of L5 neurons to L2/3 neurons is not compatible with the non-hierarchical architecture they and other groups proposed, but rather in accordance with the hierarchical model. At the same time, the functional impact of L2/3 neurons (positive vs. negative prediction error neurons) on L5 neurons (internal representation) appears not compatible with the hierarchical model, but rather in accordance with the non-hierarchical implementation.

They further hypothesize that L2/3 prediction error neurons don't use sensory input, but rather the L5 activity as a teaching signal, and test it using perturbations (halts) of optogenetic stimulation of L5 neurons coupled with locomotion (Fig.7).

All in all, the question is topical, and the new model addresses a decades-long quest to develop a unifying model of cortical function. The findings reported here transform our understanding of cortical computations, opening new exciting avenues for future investigation. The experimental design and execution are rigorous; the arguments are clearly laid out (in spite of ample potential for confusion given the numerous loops and sign flips). These include a discussion of why the non-hierarchical model proposed by the same group does not hold, as well as potential caveats in interpreting the results and novel testable proposed experiments emerging from the JEPA-like model.

Comments on revised version

I commend the authors for nicely provided answers and addressing my concerns and clarifying their points on the relationship of their findings to JEPA and current state of understanding.

In particular for Q5 -I meant if there is also a positive correlation between optomotor mismatch response and visuomotor mismatch response when looking only at neurons that the authors identify as PE-?<br /> The authors provided the answer on point.

Overall, I think this is a fundamental study and the strength of evidence for the claims is exceptional as an exemplary use of existing approaches.

Reviewer #2 (Public review):

This manuscript reveals functional connectivity of two different classed of cortical neurons that respond in opposite ways to mismatches between sensory and top-down inputs. These data are very valuable because different theories of information processing in the cortex make different predictions on the patterns of connectivity of these neurons. Therefore, these data strongly constrain possible theories of cortical processing.

Comments on revised version.

I thank the Authors for answering my questions and updating the manuscript.

Congratulations on this important work!

Reviewer #3 (Public review):

Vasilevskaya and Keller set out to experimentally distinguish between two variants of predictive processing: a hierarchical and a non-hierarchical variant. The hierarchical variant assumes a hierarchical organization in which internal representation neurons (believed to be a subset of layer 5 excitatory neurons) serve as a source of a teaching signal for local prediction error neurons as well as for the next higher level of the hierarchy, while simultaneously providing prediction signals to the preceding lower level. In contrast, the non-hierarchical variant posits that these layer 5 internal representation neurons provide local predictions to layer 2/3 prediction error neurons.

The interaction between internal representation neurons and prediction error neurons differs fundamentally between the two variants. In the hierarchical variant, internal representation neurons excite positive prediction error neurons and inhibit negative prediction error neurons, while at the same time being inhibited by positive prediction error neurons and excited by negative prediction error neurons. In the non-hierarchical variant, this pattern of connectivity is reversed.

This work is very exciting, timely, and carefully executed. The authors functionally, and later molecularly, identify layer 2/3 prediction error neurons in V1 and probe their interactions with genetically defined neuron types in cortical layers 5 and 6 using optogenetics. They demonstrate that the functional influence of putative prediction error neurons in layer 2/3 onto layer 5 is incompatible with the hierarchical variant, whereas the influence of layer 5 onto putative prediction error neurons in layer 2/3 is incompatible with the non-hierarchical variant. They then test an alternative hypothesis, in which layer 2/3 responses resemble prediction errors with respect to perturbations of artificial layer 5 activity patterns. To investigate this, they designed an experiment in which optogenetic activation of L5 IT neurons was closed-loop coupled to the mouse's locomotion speed in the absence of visual feedback, allowing them to probe the causal influence of L5 activity on layer 2/3 responses.

Finally, the authors hypothesize that their data are more consistent with a joint embedding predictive architecture (JEPA) and outline experimentally testable predictions arising from this framework.

While the work is overall convincing and provides important insights into the circuit-level implementation of predictive processing, I think the connection to JEPA networks would benefit from a more in-depth discussion of its relationship to recently proposed and implemented models. Below, I address the specific points raised by the authors (flanked by ' ... ' to make the author's statements stand out), in particular in relation to the model proposed by Nejad et al. (2025):

- 'The two proposals indeed share similarities in assuming that bottom-up input for both L2/3 and L5 arrives from thalamus, and that representations formed in L2/3 are used for predicting the activity of L5. However, there are a few important differences between the JEPA implementation proposal formulated here and the Nejad et al. model.

(1) There is no proposed mapping of computations in the Nejad et al. model onto different JEPA networks. We assume that the suggested mapping would be L4 and L5 as encoder networks, and L2/3 as a predictor network? In that case, it is different to our proposal, in which L2/3 is part of the encoder network.'

I think there is some confusion here. In Nejad et al., both L2/3 and L5 function as encoder networks. Each receives sensory input (with L2/3 receiving this input delayed via L4) and computes a latent representation of that input, denoted z_{L2/3} and z_{L5}, respectively. The prediction is obtained by comparing the output of L2/3 with L5 latent representations (via W_{L2/3->L5} * z_{L2/3}). In other words, L5 is the target in the learning objective.

Although, they did not explicitly state the mapping with JEPA, their model has the same fundamental property - predictive learning happens in the latent space. Also, this appears very similar to the roles assigned to L2/3 and L5 in your Figure 9B. The fact that you made this more explicit and the new data included, is in my view, a very interesting contribution. However, from an architectural perspective, it appears that your proposal and the model of Nejad et al. are conceptually very similar, and I do not see a substantial difference between the two. This should be made more clear in the Discussion.

- '2. Our proposal contains explicit prediction error neuron cell types within L2/3, while prediction errors in Nejad et al. are encoded in the gradients, and the layer origin of these signals is hypothesized to be L5 ('the learning-driving error signal originates in L5'). Hence, also the role of L5-L2/3 connection is distinct between the two proposals. In Nejad et al. this connection serves the role of error propagation and update for predictions in L2/3, while in our proposal this connection contains teaching signal (target representations) that are compared to predictions within L2/3. Similarly, the functional role of L2/3-L5 connection is also different, since in Nejad et al, it is supposed to carry predictions of L5 activity, whereas in our proposal we expect it to drive plasticity in L5 encoder.'

Indeed, in Nejad et al., the layer-dependent mismatch responses are modeled as gradients with respect to neuronal activity, and the model does not explicitly include prediction error neurons. However, this appears to be a modeling choice rather than a fundamental aspect of the proposal, and it does not preclude an implementation with explicit prediction error neurons. In fact, the authors explicitly acknowledge this possibility in the Discussion:

"The second approach would be to recast our model within a predictive coding framework... Predictive coding jointly optimizes both model parameters and neuronal activities, which could naturally lead to prediction errors observable in the activity of both L2/3 and L5 neurons. Note that these two views are not mutually exclusive."

While Nejad et al. hypothesize that the learning-driving error signals (that are distinct from their mismatch responses) originate in L5, the abstract loss function itself does not uniquely specify where the underlying comparison between the predicted representation (W_{L2/3 -> L5} z_{L2/3}) and the target representation (z_{L5}) must be implemented. The proposed biological implementation places this computation in L5, but from my understanding, the computational objective itself does not require this specific localization.

That said, I agree that your proposed model introduces a genuine difference. The functional roles assigned to the vertical projections are effectively reversed: in Nejad et al., the L2/3->L5 projection carries the prediction, whereas the L5->L2/3 projection conveys the error/gradient. In your architecture, by contrast, the L5->L2/3 projection carries the teaching signal (target). This is, in my view, a real and testable interpretational divergence that is worth stating clearly.

Therefore, I think the novelty lies less in the computational architecture itself and more in committing to a particular biological implementation-one that adds cell-type-specific detail to an implementation that Nejad et al. hypothesised as being compatible with their framework.

- '3. The difference outlined above also makes it evident that the two proposals should differ in how deep and superficial layers are expected to influence the activity of one another. Indeed, the proposal in Nejad et al. is based on the cortical column idea, and according to eq. 2 and 3 in the Methods, activity in L5 is a function of activity in L2/3, while activity in L2/3 is not a function of activity in L5. Our proposal is based on idea of layers forming parallel networks, where horizontal communication is the dominant mode of cortico-cortical interactions, and activity in deep layers serve as a teaching signal for L2/3. In our case, we expect the opposite - that activity in L2/3 depends on activity of L5, while activity of L5 is not immediately dependent on activity of L2/3 (only via plasticity route). This led us to propose one of direct tests for our framework - silencing L2/3 in a familiar setting should result in no immediate changes to L5 activity and behavior of the animal.'

My reading of Nejad et al. is consistent with your interpretation of the equations. Specifically, Eq. 3 makes L5 activity depend on L2/3 activity (albeit weakly, with a = 0.3), whereas Eq. 2 contains no L5 term, so L2/3 activity does not depend directly on L5 activity. In that model, the L5->L2/3 pathway carries the learning gradient rather than contributing to the activity dynamics. By contrast, in your proposed model, L2/3 activity depends on L5 activity, whereas L5 activity is not immediately dependent on L2/3 activity (except indirectly through learning/plasticity). So, you state that "silencing L2/3 in a familiar setting should result in no immediate changes in L5 activity.<br /> [...].

However, I am unsure how to reconcile this prediction with the results shown in Fig. 6. If I understand the figure correctly, optogenetic activation of Rrad-positive (positive prediction error) L2/3 neurons produces a small increase in L5 activity, whereas activation of Adamts2-positive (negative prediction error) L2/3 neurons produces a decrease in L5 activity. Although these experiments involve activation rather than silencing, they nevertheless suggest that perturbing L2/3 activity can have an immediate effect on L5 activity. Could you clarify how this is consistent with the proposed model? In other words, what aspect of the proposed circuitry makes activation effective while silencing is predicted to have no immediate consequence?

For comparison, Nejad et al. performed a related perturbation analysis in Fig. S15 by scaling the output of L2/3 neurons exhibiting positive mismatch signals (defined through the activity gradients), which increased L5 activity, whereas scaling neurons with negative mismatch signals produced the opposite effect. I am not entirely sure how directly these simulations map onto the experiments shown in your Fig. 6, since the Nejad simulations were performed during mismatch conditions, if I have understood them correctly.

- '4. The proposal in Nejad et al. relies on input reconstruction or variance maximization within the L5 autoencoder network to avoid collapse. Instead, our proposal has no reconstruction objective.'

Nejad et al. only require two encoders (like in JEPA), how these two are learnt can be done in several ways. While Nejad et al. focus on using a reconstruction loss to learn the L5 target, they also show that it works equally well with non-reconstruction objectives. Therefore, I do not think the presence or absence of a reconstruction objective constitutes a fundamental distinction between the two proposals.

As your current work presents a conceptual architecture rather than a fully implemented learning algorithm (in a model), the mechanism that would prevent representational collapse has not yet been defined. From my understanding, every joint-embedding approach must address this issue, whether through reconstruction, variance/covariance regularization, stop-gradient or EMA mechanisms, or other approaches. Thus, the absence of a reconstruction objective (or another anti-collapse mechanism) is not, in itself, a distinguishing feature of the proposed architecture, but rather an as-yet unspecified design choice within the learning objective.

- '5. Lastly, there is time-delay between inputs to L5 and L2/3 that is proposed in Nejad et al., while this is not something inherent to our proposal.'

I agree that the temporal delay introduced by L4 is a key component of the Nejad et al. model and is currently absent from your proposal. However, I would expect temporal delays to emerge naturally in your framework as well, given the multisynaptic and highly parallel organization of cortical circuits. More generally, implementing predictive learning over time (as in JEPA) requires comparing representations at times t and t+1, which seems to require some form of temporal delay. How else would you suggest this is done?

In general, I think the manuscript would benefit from a clearer discussion of its relationship to the model proposed by Nejad et al. (perhaps following the discussion above), including both the shared conceptual claims and the aspects that genuinely differ between the two frameworks. At present, some of the claims are presented as novel, although at least some of these core ideas have already been proposed in Nejad et al.

For example, the authors state: "Thus, we propose that layer 2/3 functions to predict layer 5 activity, not sensory input per se, hence making predictions in the internal representation space, not input space." This appears to be exactly what Nejad et al. proposed as discussed above - in their model L2/3 predicts L5 activity (purely in the latent space), as they state in the abstract.

That said, there are some interesting differences, and I think the community would greatly benefit from making these clear, including the roles assigned to interlaminar connections and the interpretation of the signals carried by these pathways. These differences are interesting and potentially testable, and I think the manuscript would be strengthened by explicitly distinguishing which aspects are in line with the ideas already present in Nejad et al. and which aspects represent new contributions.

Author response:

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

We thank you for the time you took to review our work and for your feedback! We have performed the following additional analyses:

(1) We analyzed the optomotor mismatch response as a function of time spent in the optogenetic closed-loop session.

(2) We show the correlation between optomotor mismatch response and visuomotor mismatch response for functionally identified PE neurons.

The main changes to the manuscript are:

(3) Clarification of our terminology (moved and refined definition of the teaching signals).

(4) A new figure panel summarizing the functional influence patterns we identified.

(5) Clarification of our argumentation for the JEPA-inspired proposal.

All comments are addressed individually in the following.

Public Reviews:

Reviewer #1 (Public review):

Vasilevskaya and Keller test different models of cortical function through the lens of predictive processing, a powerful framework for the brain to learn and predict the statistics of the world via generative internal models. The authors use a clever combination of behavioral perturbations in closedloop and open-loop visuomotor virtual reality assays, a paradigm the Keller lab pioneered and used effectively in the past decade, in conjunction with two-photon imaging of neuronal calcium responses and targeted optogenetic perturbations of activity. They specifically put to test proposed hierarchical vs. non-hierarchical circuit implementations of predictive processing by analyzing the logic of inter-lamina interactions (superficial vs. deep; L2/3 vs. L5/6).

The authors conclude that both versions of predictive processing architectures they analyze are likely invalid, and instead formulate an alternative novel model of cortical function based on a recently developed machine learning algorithm for self-supervised learning (joint embeddings of predictive architectures, JEPA) and its further refinements. JEPA borrows elements from predictive processing, engaging two encoder networks and training the output of one network to predict the output of the other. In their new model of cortical computations, prediction error neurons in L2/3 compare the deep layers (L5/6) activity, which is taken as a teaching signal, to a local, L2/3 prediction of this latent representation.

Specifically, the authors build on their previous work and reports from other groups that different sets of L2/3 neurons compute positive prediction errors (fire when sensory stimuli appear unexpectedly with respect to the movements of the animal; e.g., grating onsets in the absence of locomotion) and respectively negative prediction errors (fire when sensory stimuli are absent, while the brain expected them to be present; e.g. mice locomote but visual flow is suddenly halted - visuomotor mismatches). These L2/3 positive and negative prediction error neurons exchange messages with neurons in the deeper cortical layers that, the authors propose, build an internal representation (R) of the sensory stimuli given the animals' movements.

In the hierarchical model, internal representation neurons (R) are supposed to act as a teaching signal for both types of prediction error neurons; the output of the positive prediction error neurons is assumed to suppress activity of R such that the error between the teaching signal and the prediction is minimized; similarly, in the non-hierarchical version, R serves as a prediction for the prediction error neurons, and in turn it receives excitatory drive from the positive prediction error neurons and negative input from the negative prediction error neurons.

The authors find that the functional impact of L5 neurons on L2/3 neurons is not compatible with the non-hierarchical architecture they and other groups proposed, but rather in accordance with the hierarchical model. At the same time, the functional impact of L2/3 neurons (positive vs. negative prediction error neurons) on L5 neurons (internal representation) appears not compatible with the hierarchical model, but rather in accordance with the non-hierarchical implementation.

They further hypothesize that L2/3 prediction error neurons don't use sensory input, but rather the L5 activity as a teaching signal, and test it using perturbations (halts) of optogenetic stimulation of L5 neurons coupled with locomotion (Figure 7).

All in all, the question is topical, and the new model addresses a decades-long quest to develop a unifying model of cortical function. The findings reported here transform our understanding of cortical computations, opening new, exciting avenues for future investigation. The experimental design and execution are rigorous; the arguments are clearly laid out (in spite of ample potential for confusion given the numerous loops and sign flips). These include a discussion of why the non-hierarchical model proposed by the same group does not hold, as well as potential caveats in interpreting the results and novel testable proposed experiments emerging from the JEPA-like model.

I have several questions about the interpretations of some of the claims and suggestions for potential additional experiments and analyses.

We thank the reviewer for their comments. We address them below.

(1) Some of the pieces of the puzzle remain to be identified and demonstrated: the existence of internal representation neurons in L2/3 and ascertaining that the L5/6 neurons analyzed function indeed as internal representation neurons. The authors find that stimulation of L2/3 positive prediction error neurons enhances activity of L5 neurons...If L5 neurons hold a latent representation that serves as a teaching signal for L2/3 neurons (as the authors posit), wouldn't one expect that the input they receive from the positive prediction neurons be suppressive, such that the error is further minimized?

Not necessarily - this depends on the model one has for how cortex works. In the hierarchical predictive processing model, PE+ neurons are expected to suppress the local internal representation neurons in L5. Our data, however, are not consistent with this model. This is one of the key arguments we build the idea on that JEPA is a better model for cortex than hierarchical predictive processing. In JEPA we would not expect the L2/3 prediction error to update L5 directly, but instead update the prediction of L5 activity (the source of this signal remains to be identified; see our speculations on the origin of prediction signals in the comments to question 3), and drive plasticity in the local L5 encoder network. That something acts like a teaching signal for the L2/3 comparator does not, by itself, determine how the outcome of the comparison influences the source of the teaching signal.

(2) Do the authors envision any specific differences between the representations of the two encoder networks posited to exist in L2/3 and L5 in the JEPA-like implementation? Are they synchronous/offset in their temporal representations, or any other features?

Given theoretical work (Mohammadi et al., 2025), one would expect to find differences in learning rates between the predictor and the encoder networks. Assuming the predictor network is also implemented in L2/3, we would expect to see faster learning rates in L2/3 compared to L5. Implementations inspired by related theoretical works also predict differences in learning rate between the two encoder networks (Grill et al., 2020), again with higher learning rates expected in L2/3 compared to L5. Beyond that, however, we are not aware of any experimentally observable differences one might expect to find. We are hoping the computational community will remedy this soon.

(3) Where is the prediction coming from onto L2/3 neurons? Is it emerging locally in L2/3 from the putative internal representation neurons, or is it long-range - as work from the authors previously proposed? Or a mix of both?

We expect the predictions to come from long-range inputs. In classical (hierarchical) JEPA one would expect these to be the lateral communication within L2/3. In a non-hierarchical implementation that is capable of operating on arbitrary graphs (as would be necessary for it to work in cortex), we suspect that the L2/3 network will use both long-range L2/3 and long-range L5 input. In JEPA terminology: the encoder A network uses information from non-local sources of both networks to predict local activity of encoder B network (assuming that information is statistically useful in predicting that activity).

(4) What is the role of the indiscriminate L4 input that appears to enhance activity of both positive and negative prediction error neurons in L2/3?

The short answer is, we don’t know. We would have indeed expected to find some asymmetry of influence. There are a few options: A) We might be failing to activate a specific interneuron that mediates feedforward inhibition on this pathway by the artificial stimulation of Scnn1a neurons. B) We know that Scnn1a neurons are only a subset of L4 neurons – there are other populations of L4 neurons that might exhibit the opposing influence. C) The Scnn1a population is more than 1 layer of a JEPA away from the comparator and not part of the predictor that forms the actual representation compared against L5 (e.g. L4 could provide an input to L2/3 internal representation neurons, and thus only indirectly influence L2/3 PE neurons). D) The JEPA analogy is wrong.

(5) Does Figure 7D change in a meaningful manner if the authors plot the correlation between optomotor mismatch response and visuomotor mismatch response specifically for the negative prediction error neurons in L2/3 (Adamts-2) rather than for all L2/3 cells sampled?

We might be misunderstanding. If the reviewer means genetically identified negative prediction error neurons (Adamts2), we do not have the data to address this question, as we did not perform any recordings of molecularly defined Adamts2 population in L2/3. If the reviewer means functionally identified negative prediction neurons, this would be the two rightmost data points in Figure 7D (the x-axis in this panel is the visuomotor mismatch response strength we use to functionally identify PE- neurons). These two data points on the right-hand side of the plot correspond exactly to what we classify as negative prediction error neurons throughout the rest of the manuscript (15% of the most responsive neurons to visuomotor mismatch).

Does the reviewer mean, is there also a positive correlation between optomotor mismatch response and visuomotor mismatch response when looking only at neurons that we identify as PE-? If so, the answer is yes (Author response image 1). Interestingly, while there is a positive correlation, there is also nonuniformity in response patterns. We think that this is expected, given that our visuomotor coupling paradigm captures only a very small subspace of stimuli that prediction error neurons are tuned to. Bulk stimulation of L5, in contrast, might work better to separate all PE neurons. In other words, we speculate that L5 stimulation is a better predictor of the functional role of an L2/3 neuron.

Author response image 1.

Optomotor mismatch response as a function of visuomotor mismatch response for neurons that are functionally classified as PE. Red line shows a linear fit estimated with a bootstrap approach.

(6) Do the optomotor mismatch responses in L2/3 neurons depend on how long the closed-loop coupling of optogenetic stimulation of Tlx3 L5 neurons and locomotion speed has been in place for?

No, not that we can measure. We performed an analysis in which we split the optogenetic closed-loop session into two equal parts, early and late. We then quantified the average optomotor mismatch response independently for early and late parts of the session (Author response image 2A). Based on this quantification, we find no evidence of a change as a function of time in the closed-loop session. We also performed a sliding window analysis on a shorter timescale. While it did look like the responses may be smaller in the first few minutes, we do not have sufficient data to address this. None of the differences in response size were significant (Author response image 2B).

Author response image 2.

Optomotor mismatch response as a function of experience with artificial closed-loop coupling. (A) Mean L2/3 population response to optomotor mismatch in the first half (Early MM) and in the second half (Late MM) of the optogenetic closed-loop session. (B) Mean L2/3 population response to optomotor mismatch as a function of time in the optogenetic closed-loop session. Error bars indicate SEM. Differences between the mean values were estimated by hierarchical bootstrap and are not significant.

Reviewer #2 (Public review):

This manuscript reveals the functional connectivity of two different classes of cortical neurons that respond in opposite ways to mismatches between sensory and top-down inputs. These data are very valuable because different theories of information processing in the cortex make different predictions on the patterns of connectivity of these neurons. Therefore, these data strongly constrain possible theories of cortical processing.

We thank the reviewer for their comments. We address them below.

General comments:

(1) The methods of statistical testing are insufficiently described. I did not understand the description in lines 1105-1119. The authors should provide sufficient details so the reader can reproduce their analyses. For example, it may be helpful to provide specific details of the testing procedure for one of the comparisons (e.g. the first comparison in Table S1).

We assume the reviewer is not familiar with hierarchical bootstrapping in general. If so, the explanation below would summarize the procedure. This is the procedure with particular emphasis on its application to neuroscience data is described in the paper we reference in that part of the methods (Saravanan et al., 2020). Given that the analysis has become relatively standard (and is described in the reference provided), we think it might be an overkill to add the full explanation below to the manuscript. In addition to the general procedure, there are only 2 pieces of information relevant to fully reconstructing the analysis:

(1) What are the “levels” (mice, recording sites, neurons, trials)?

(2) What is the number of bootstrap samples used.

Thus, we think all information is already provided in the methods. We now also explicitly added the levels when describing the nested structure of the data in the manuscript to increase clarity.

Hierarchical bootstrap analysis:

Our data are naturally nested: multiple neurons are recorded within a single mouse, and multiple mice are tested within an experimental group. Standard bootstrapping (sampling with replacement from the entire pool of neurons) fails because it assumes all observations are independent. In reality, neurons from the same mouse are more similar to each other than to neurons from a different mouse. The hierarchical bootstrap (or multi-level bootstrap) preserves this nested structure, ensuring your confidence intervals are not artificially narrow due to pseudoreplication.

To illustrate the problem, assume you have 100 neurons from Mouse A and 10 neurons from Mouse B, a simple bootstrap will be heavily biased toward Mouse A. Furthermore, the simple bootstrap ignores the fact that the true variance in your population comes from two sources:

(1) Between-mouse variance (differences in surgery, genetics, or behavior).

(2) Within-mouse variance (differences in tuning or activity between individual cells).

Hierarchical bootstrap addresses this problem, and is implemented as follows: To estimate the mean response while accounting for different sample sizes per mouse, a two-level resampling scheme is used

(1) Resample the higher level (Mice)

First, you account for the variability between animals.

Suppose you have N mice in total.

Randomly draw N mice with replacement from your original pool.

Note: Because this is with replacement, a single mouse’s data might be included multiple times in one bootstrap iteration, while another mouse might be left out entirely.

(2) Resample the lower level (Neurons)

For each mouse selected in Step 1, you must now account for the variability within that specific animal.

Look at the number of neurons actually recorded from that mouse (let’s call it k_i).

Randomly draw k_i neurons with replacement from that mouse’s specific pool of recorded cells.

This step is crucial: you always resample the same number of neurons that were originally recorded for that specific mouse. This maintains the "weight" or "influence" that animal had in the original dataset.

(3) Calculate the resampled statistic

Calculate the mean of all neurons collected in this “bootstrap sample”.

(4) Iterate

Repeat Steps 1–3 many times (typically B = 1,000 or 10,000 iterations).

The distribution of these B bootstrap means represents your sampling distribution and is used to calculate confidence intervals and p-values. 

(2) The authors should clarify how the problem of multiple comparisons was addressed for comparisons performed in multiple moments of time, where significance is indicated by a black bar (e.g. in Figure 2F).

There is no family-wise error correction in cases of comparing response time courses implemented in our analysis. If the reviewer has a good suggestion for how to implement family wise error correction, we would be happy to implement it. We are not aware of anything that is better than what we currently do (the time bin-wise comparison). To briefly explain the problem: In most neuroscience papers, response curves are compared by choosing a time window (e.g. 0.5 to 1 s following a trigger) and calculating mean values of the curves in these windows. This hides the problem of multiple comparisons that arises from the fact that the experimenter is free to choose a response window used for analysis. Note, this also creates a strong incentive to “optimize” choice of an analysis window – a part of the analysis that a reader is typically completely blind to. We could of course also choose an analysis window that sounds reasonable and yields significant differences for our analyses. However, to provide a more unbiased image of the data, we have come to do bin-wise comparisons with a fixed p-value (typically 0.05). This is used in all of our papers at the moment. Given that samples from neighboring timepoints are correlated via a combination of actual responses and a subset of noise sources, the samples are not independent. We now also implemented a correction for spurious positive values by requiring at least 2 neighboring bins to have a p-value below 0.05 to be shown. If the samples were independent, this would mean a false positive rate of 0.0025. Given that they are not, this is a lower bound only. Additionally, any family-wise error correction would be a function of the number of time bins we show in the plot. This would mean that our choice of the time window shown in a plot (-1s to +4s, or +5s, etc.) would change the p-value we consider significant. Thus, there is no explicit family-wise error correction, and we have come to the conclusion that the bin-wise comparison with a fixed p-value is the most unbiased representation of the data we can provide. The alternative would be to additionally plot z-scores or the p-values as a function of time, but in our experience these types of plots are even harder to read for readers not used to it.

(3) It would be helpful to add a figure in the Discussion summarising the functional connectivity suggested by all experiments.

We now added a panel that summarizes the functional connectivity observed in our experiments to Figure S10. 

(4) Throughout the manuscript, the authors use the term "teaching signals", but I am unclear what they mean by it: after reading the definition in lines 45-46, I thought that they corresponded to values (as they are compared to sensory signals). Later (428-430), the text suggests that they correspond to error neurons. But then lines 605-607 say it is not an error signal. The authors should define teaching signals very precisely or remove this term.

The formal definition of the teaching signal was in footnote 1 of the manuscript. We assume the reviewer may have missed this. We now moved this definition into the main text to increase clarity.

We use the term teaching signal to mean exactly this definition throughout the manuscript. We suspect, a second source of confusion may come from ambiguity in regards to anatomical vs. functional definitions. We have attempted to emphasize that the definition of teaching signal is a functional one, not an anatomical one (as is the case for ‘prediction’ – predictions are functionally defined, not anatomically – hence it makes sense to ask questions of the form “what are the potential sources of predictions” etc.). A teaching signal is a signal that is compared against a prediction (the ‘ground truth’ the prediction is compared and trained against). In different circuit implementations of predictive processing, different inputs function as predictions and teaching signals. In the hierarchical implementation, activity of the internal representation neuron at the lower level serves as a teaching signal for a prediction signal that is formed by the internal representation neuron from the higher level. In the non-hierarchical implementation, external inputs from the thalamus or other cortical areas serve as teaching signals for the respective prediction signals that are formed by internal representation neurons. This terminology is most intuitive when thinking from the perspective of a prediction error neuron, since a prediction error neuron computes the difference between two inputs signals – one of which functions as a teaching input, and the other one as a respective prediction. Hence, lines 428-430 specify that layer 5 input onto prediction error neurons of layer 2/3 serves as a teaching input.

Reviewer #2 (Public review):

Vasilevskaya and Keller set out to experimentally distinguish between two variants of predictive processing: a hierarchical and a non-hierarchical variant. The hierarchical variant assumes a hierarchical organization in which internal representation neurons (believed to be a subset of layer 5 excitatory neurons) serve as a source of a teaching signal for local prediction error neurons as well as for the next higher level of the hierarchy, while simultaneously providing prediction signals to the preceding lower level. In contrast, the non-hierarchical variant posits that these layer 5 internal representation neurons provide local predictions to layer 2/3 prediction error neurons.

The interaction between internal representation neurons and prediction error neurons differs fundamentally between the two variants. In the hierarchical variant, internal representation neurons excite positive prediction error neurons and inhibit negative prediction error neurons, while at the same time being inhibited by positive prediction error neurons and excited by negative prediction error neurons. In the non-hierarchical variant, this pattern of connectivity is reversed.

This work is very exciting, timely, and carefully executed. The authors functionally, and later molecularly, identify layer 2/3 prediction error neurons in V1 and probe their interactions with genetically defined neuron types in cortical layers 5 and 6 using optogenetics. They demonstrate that the functional influence of putative prediction error neurons in layer 2/3 onto layer 5 is incompatible with the hierarchical variant, whereas the influence of layer 5 onto putative prediction error neurons in layer 2/3 is incompatible with the non-hierarchical variant. They then test an alternative hypothesis, in which layer 2/3 responses resemble prediction errors with respect to perturbations of artificial layer 5 activity patterns. To investigate this, they designed an experiment in which optogenetic activation of L5 IT neurons was closed-loop coupled to the mouse's locomotion speed in the absence of visual feedback, allowing them to probe the causal influence of L5 activity on layer 2/3 responses.

Finally, the authors hypothesize that their data are more consistent with a joint embedding predictive architecture (JEPA) and outline experimentally testable predictions arising from this framework.

We thank the reviewer for their comments. We address them below.

While the work is overall convincing and significantly advances our understanding of the circuit-level implementation of predictive processing, there are a few weaknesses that should be addressed or discussed:

(1) The authors define putative positive prediction error neurons as the 15% of neurons most responsive to grating onset and putative negative prediction error neurons as the 15% most responsive to visuomotor mismatch. While this selection would be expected to overlap with negative and positive prediction error neurons, the criterion is not sufficiently stringent (independent of the exact percentage chosen). In particular, classification of a neuron as a prediction error neuron should ideally be accompanied by evidence that it does not exhibit a significant increase in activity when the prediction matches the sensory input or teaching signal.

We understand the reviewer’s intuition. We can indeed use other stimuli to identify prediction error neurons, like the relative suppression of responses in closed-loop running onset vs open-loop running onsets. This was the reason behind including Figure S1, to show that our selection results in expected pattern of running onset responses. We don’t typically use the running onset responses as they have an additional confound we have not fully understood. This is that running onset always tends to result in an increase of calcium activity in all neurons. This could have a variety of reasons: A) Contamination of hemodynamic occlusion signals (blood vessels tend to constrict at running onset, making it appear like an increase in calcium activity) – see Yogesh et al., 2025. B) The virtual coupling in our VR is not good enough to provide a true “closed loop” experience. Humans typically notice lags of larger than 30ms – in our VR it is approximately 100 ms. C). Running onset in head-fixed animals is not accompanied by a vestibular input. D) Predictive processing is wrong. We tend to think it is a combination of the three.

We can also use combinations of the two criteria to select neurons - if the reviewer has a specific selection criteria in mind (top XXX% MM responsive AND top XXX% closed-loop suppressed, etc.) we are happy to repeat the analysis for that specific set of criteria, but the fundamental problem that we are using a functional response to select these neurons does not go away. We know that our functional selection criteria mean we select a population of neurons that is enriched for prediction error neurons. If the enrichment is too weak, we would expect to find no effects in terms of functional influence. It is hard to explain, however, how a weak enrichment could result in a strong effect on functional influence. Our arguments in more lengthy form, for why the visuomotor mismatch is a good stimulus to identify negative prediction error neurons can be found here: Attinger et al., 2017; Jordan and Keller, 2020; Leinweber et al., 2017; O’Toole et al., 2023; Vasilevskaya et al., 2022; Zmarz and Keller, 2016.

(2) The authors "speculate that the prediction error responses in layer 2/3 may not be computed with respect to sensory input, but with respect to layer 5 activity as a teaching signal." However, it is unclear how this perspective differs from earlier statements in the manuscript. In the Introduction, the authors note that "these signals, typically referred to as sensory signals, we will refer to as teaching signals," and later describe the hierarchical variant as one "in which internal representation neurons act as a source of the teaching signal." Given this framing, it is difficult to identify what is conceptually novel in the updated view. Is the key distinction that layer 2/3 neurons are now proposed to generate predictions in an internal representation space rather than in sensory input space, as briefly suggested in the Discussion? Or are the authors introducing a distinction between an external (sensory) and an internal (cortical) teaching signal? If so, this distinction should be made explicit. Clarifying this point would considerably strengthen the manuscript.

There might be a misunderstanding regarding our usage of the term teaching signal. In hierarchical predictive processing the teaching signal is typically referred to as a sensory signal, as e.g. in: “prediction error neurons compare predictions to sensory input”. In non-hierarchical predictive processing, or far away from the sensory input (think prefrontal cortex), or for cross-modal interactions “sensory” input is misleading. Also, in non-predictive-processing type models (like JEPA), sensory input has a different functional role. Thus, we operationally define teaching signal as the signal that is compared against the prediction by prediction error neurons.

The two primary options we are comparing are:

(1) Is the teaching signal to the L2/3 comparator a bottom-up input to V1 (as one would expect in predictive processing). 

(2) Is the teaching signal to the L2/3 comparator L5 input (as one would expect in JEPA).

Our data argue in favor of option 2. We have rephrased parts of the manuscript to try to make this clearer. 

(3) The authors propose that "L2/3 neurons predict L5 activity, hence making predictions in the internal representation space rather than the input space," and further suggest that, since both deep and superficial cortical layers receive thalamic input, the cortex may function like a JEPA. This idea appears closely related to the model introduced by Nejad et al. (2025), which effectively implements a JEPA-like architecture: L5 activity serves as a target against which L2/3 predictions are compared in a selfsupervised manner, with both L5 and L2/3 (via L4) receiving thalamic input. It would be helpful for the authors to clarify how their framework differs from that model, and to specify the key conceptual or mechanistic distinctions between the present proposal and the approach described by Nejad et al.

The two proposals indeed share similarities in assuming that bottom-up input for both L2/3 and L5 arrives from thalamus, and that representations formed in L2/3 are used for predicting the activity of L5. However, there are a few important differences between the JEPA implementation proposal formulated here and the Nejad et al. model.

(1) There is no proposed mapping of computations in the Nejad et al. model onto different JEPA networks. We assume that the suggested mapping would be L4 and L5 as encoder networks, and L2/3 as a predictor network? In that case, it is different to our proposal, in which L2/3 is part of the encoder network.

(2) Our proposal contains explicit prediction error neuron cell types within L2/3, while prediction errors in Nejad et al. are encoded in the gradients, and the layer origin of these signals is hypothesized to be L5 (‘the learning-driving error signal originates in L5’). Hence, also the role of L5-L2/3 connection is distinct between the two proposals. In Nejad et al. this connection serves the role of error propagation and update for predictions in L2/3, while in our proposal this connection contains teaching signal (target representations) that are compared to predictions within L2/3. Similarly, the functional role of L2/3-L5 connection is also different, since in Nejad et al, it is supposed to carry predictions of L5 activity, whereas in our proposal we expect it to drive plasticity in L5 encoder.

(3) The difference outlined above also makes it evident that the two proposals should differ in how deep and superficial layers are expected to influence the activity of one another. Indeed, the proposal in Nejad et al. is based on the cortical column idea, and according to eq. 2 and 3 in the Methods, activity in L5 is a function of activity in L2/3, while activity in L2/3 is not a function of activity in L5. Our proposal is based on idea of layers forming parallel networks, where horizontal communication is the dominant mode of cortico-cortical interactions, and activity in deep layers serve as a teaching signal for L2/3. In our case, we expect the opposite - that activity in L2/3 depends on activity of L5, while activity of L5 is not immediately dependent on activity of L2/3 (only via plasticity route). This led us to propose one of direct tests for our framework – silencing L2/3 in a familiar setting should result in no immediate changes to L5 activity and behavior of the animal.

(4) The proposal in Nejad et al. relies on input reconstruction or variance maximization within the L5 autoencoder network to avoid collapse. Instead, our proposal has no reconstruction objective.

(5) Lastly, there is a time delay between inputs to L5 and L2/3 that is proposed in Nejad et al., while this is not something inherent to our proposal.

We expect that the most useful future models should move beyond JEPA, with the emphasis on nonhierarchical models capable of operating on arbitrary graphs. We think cortex functions according to principles of a JEPA (predictions in latent space), and that the role of cell types and the exact computational organization remain to be constrained.

REFERENCES

Attinger, A., Wang, B., Keller, G.B., 2017. Visuomotor Coupling Shapes the Functional Development of Mouse Visual Cortex. Cell 169, 1291-1302.e14. https://doi.org/10.1016/j.cell.2017.05.023

Grill, J.-B., Strub, F., Altché, F., Tallec, C., Richemond, P.H., Buchatskaya, E., Doersch, C., Pires, B.A., Guo, Z.D., Azar, M.G., Piot, B., Kavukcuoglu, K., Munos, R., Valko, M., 2020. Bootstrap your own latent: A new approach to self-supervised Learning. https://doi.org/10.48550/arXiv.2006.07733

Jordan, R., Keller, G.B., 2020. Opposing Influence of Top-down and Bottom-up Input on Excitatory Layer 2/3 Neurons in Mouse Primary Visual Cortex. Neuron 108, 1194-1206.e5. https://doi.org/10.1016/j.neuron.2020.09.024

Leinweber, M., Ward, D.R., Sobczak, J.M., Attinger, A., Keller, G.B., 2017. A Sensorimotor Circuit in Mouse Cortex for Visual Flow Predictions. Neuron 95, 1420-1432.e5. https://doi.org/10.1016/j.neuron.2017.08.036

Mohammadi, A.G., Halvagal, M.S., Zenke, F., 2025. Understanding cortical computation through the lens of joint-embedding predictive architectures. https://doi.org/10.1101/2025.11.25.690220

O’Toole, S.M., Oyibo, H.K., Keller, G.B., 2023. Molecularly targetable cell types in mouse visual cortex have distinguishable prediction error responses. Neuron 111, 2918-2928.e8. https://doi.org/10.1016/j.neuron.2023.08.015

Saravanan, V., Berman, G.J., Sober, S.J., 2020. Application of the hierarchical bootstrap to multi-level data in neuroscience. Neurons Behav. Data Anal. Theory 3, https://nbdt.scholasticahq.com/article/13927-application-of-the-hierarchical-bootstrap-tomulti-level-data-in-neuroscience.

Vasilevskaya, A., Widmer, F.C., Keller, G.B., Jordan, R., 2022. Locomotion-induced gain of visual responses cannot explain visuomotor mismatch responses in layer 2/3 of primary visual cortex. https://doi.org/10.1101/2022.02.11.479795

Yogesh, B., Heindorf, M., Jordan, R., Keller, G.B., 2025. Quantification of the effect of hemodynamic occlusion in two-photon imaging of mouse cortex. eLife 14, RP104914. https://doi.org/10.7554/eLife.104914

Zmarz, P., Keller, G.B., 2016. Mismatch Receptive Fields in Mouse Visual Cortex. Neuron 92, 766–772. https://doi.org/10.1016/j.neuron.2016.09.057

eLife Assessment

This important study provides new insights into the patterns of organelle inheritance in the protozoan parasite Toxoplasma gondii. The authors introduce an innovative dual-labeling approach to distinguish maternally inherited from de novo synthesized organelles, representing convincing evidence that different organelles follow distinct inheritance fates during parasite replication. Future studies will be needed to determine whether the residual body functions as a central recycling hub, as the current data are also consistent with alternative models.

Reviewer #1 (Public review):

Summary:

This work asks the question of how different organelles and structures in the apicomplexan parasite Toxoplasma gondii are recycled and/or segregated to the daughter cells during cell replication. In particular, they consider an unusual cell structure called the residual body that links replicating cells during the intracellular infection stage of this parasite. The residual body has historically been considered a 'dumping ground' for unnecessary relics of the mother cell during division, but this notion is increasingly being revised. Indeed, cell replication in Toxoplasma is often misinterpreted as cell division (cytokinesis), but in fact, the cell replicates its organelles and structures to multiple 10s of copies in seemingly distinctly formed daughter cells, but cytokinesis is delayed for many such cycles and typically only occurs simultaneously with parasite egress from its host cell. The residual body is, in fact, the connection between these pre-cytokinetic replicated daughters, and effectively, this is still a single cell at this stage. The authors have previously shown that an actin network extends through the residual body between these daughter cells, and ER and mitochondria common to all cells are also linked through this structure. This study examining the fates of organelles during cell replication is timely for continuing our understanding of how this fascinating component of the cell participates in these processes. The authors use Halo-tags as their principal tool to track discrete populations of proteins, labelling their organelle locations, and this provides beautiful insight into these processes.

Strengths:

Using dyes conjugated to Halo tags this work elegantly tracks the fates of proteins synthesised by an original 'mother' cell over several replication cycles of pre-cytokinetic 'daughters'. Using this tool, they show that some organelles are made intact just once and that some of these can be subsequently sorted to the daughters (micronemes and rhoptries) while others are dismantled (IMC) and the daughters must make their own. A third set of organelles (largely synthesis, sorting and metabolic compartments) are divided and inherited, and new daughter-synthesised proteins are added to the preexisting maternal proteins in these structures. A role for actin and myosin is clearly demonstrated for micronemes and rhoptries, and this correlates with their relatively late inheritance into the developing daughters. Overall, this work gives clarity to the behaviours of several cell structures during replication and paves the way to better understanding the mechanisms that drive the differences between structures and the universality of these processes in other apicomplexan parasites. In particular, this study shows that the residue body is a region of the cell syncytium that organelles can be actively transported from. Therefore, it is a space that can actively contribute to the segregation of the late segregating micronemes and rhoptries.

Weaknesses:

In addressing the question of residual body participation in sorting of organelles, a clear definition of this structure is required including when and where it is delineated from the posterior of a mother cell during the formation of daughter structures. The authors' definition is as follows: 'The RB originates from the collapse of the maternal parasite during daughter cell budding and occupies the space previously occupied by the mother cell.' As such, a clear marker of the mother cell 'collapse' is required, but such a marker is not identified or used in the study to separate what might be considered an active part of the mother cell during early daughter formation, and the residual body. This might seem like moot a point, but it would help to give clarity to notions of recycling and 'reservoirs'. Mother cells retain their active invasion apparatus until very late in daughter formation and the need for micronemes and rhoptries to be released from this service late in the process might explain why they are only then trafficked to the cell posterior and then into the daughters. So, is this a distinct 'residual body' body function/reservoir or just a spatial constraint of this sequence of daughter formation? The authors elegantly show that MyoF is necessary for segregation of micronemes and rhoptries into daughters, and that MyoF depletion leads to accumulation of these organelles within the residual body. Moreover, restored expression of MyoF can then recover these organelles. This clearly demonstrates the activity of the residual body as part of the syncytium space that participates in the maintenance of the vacuole. But does it imply that this space necessarily handles all inherited micronemes and rhoptries as a 'trafficking hub'? My concern with the lack of a clear definition could provide some misinterpretation or overinterpretation of the contribution residual body.

A further, remarkable conclusion is that maternal micronemes are evenly segregated into daughters through an active process for 'balanced microneme inheritance'. The proportion of maternal micronemes is quantified up to the 8-cell stage and shown to be not significantly different between cells. But would this result be expected with random assortment at this stage? The authors model the probability of a 32-cell stage vacuole occurring with each daughter having within 0-3 maternal micronemes and this is considered unlikely. However, the authors neither present the modelling for the 8-cell stage or show quantification of 32-cell vacuoles. They do show some images of large vacuoles, but it is not possible to determine the distribution of maternal micronemes in these images. A regulated process of segregation would require a complex mechanism where some form of microneme counting would be required to create the proposed balance. It is, therefore, important to have strong data supporting such a hypothesis, but this is not currently presented.

Reviewer #2 (Public review):

Summary:

Toxoplasma gondii is an obligate intracellular parasite and the causative agent of toxoplasmosis. Parasite invasion of host cells, intracellular replication, and subsequent egress, which results in destruction of the infected cell, are central to pathogenicity. This manuscript focuses on understanding how maternal resources, specifically cellular organelles, are shared between daughter parasites during cell division. Many organelles are present as a single copy, making their division and inheritance essential for successful replication. In T. gondii, our understanding of how organelles are divided during cell division remains limited, and this study helps address this important knowledge gap.

Strengths:

The major strength of this study is the use of a Halo-based pulse-chase assay to characterize patterns of organelle inheritance and to monitor protein synthesis, turnover, and movement. This approach will be of considerable interest to the field. Using this method, the authors identify three major modes of organelle inheritance:

(1) Organelles present in multiple copies (such as micronemes and rhoptries) are partitioned between daughter parasites, with additional contributions from newly formed vesicles. Newly synthesized and pre-existing material remain as distinct populations within the cell.

(2) Single-copy organelles, such as the Golgi and apicoplast, are expanded through the incorporation of newly synthesized material before division.

(3) Cytoskeletal structures are synthesized de novo during each round of cell division.

These findings provide a more refined understanding of organelle inheritance and demonstrate that secretory organelles are not generated entirely de novo during each round of division, as was previously thought.

The paper places particular emphasis on the fate of maternal micronemes and rhoptries during division. The data show that (1) during division in wild-type cells, maternal micronemes and rhoptries are detectable in the residual body (RB); however, the majority of these organelles are localized within the parasite body, either at the apical or basal ends of the daughter parasites (Fig. 6). (2) In the absence of the myosin motor MyoF, micronemes and rhoptries accumulate in the residual body and are not properly trafficked to the daughter cells. Upon restoration of MyoF protein levels, these organelles redistribute to the daughter cells, although in an uneven manner.

Weaknesses:

The second half of the paper focuses on a more detailed characterization of microneme and rhoptry recycling. The authors strongly argue that the RB is a central hub for recycling micronemes and rhoptries; however, this conclusion is not fully supported by the data. For example, the authors state:

Line 227:<br /> "Notably, after endodyogeny was completed, M-MIC2 was redistributed from the RB to the apical tip of the daughter cells (Figure 6A, 11:30, 16:00 h), confirming that the RB serves as a temporary reservoir during microneme recycling (Periz et al., 2019)."

Line 231:<br /> "In approximately 90% of parasites undergoing replication, M-RON2 was integrated into daughter rhoptries prior to mother cell collapse and formation of the RB (Figure 6B, 4:15-4:30 h and 10:30-10:45 h). Like M-MIC2, M-RON2 was occasionally detected in the RB, though less prominently, suggesting more rapid, tightly regulated, or more efficient recycling due to their lower number."

Line 326:<br /> "However, we show that the RB temporarily stores maternal secretory organelles, such as micronemes and rhoptries, which are later redistributed to daughter cells in a MyoF-dependent manner (Figure 9B)."

Thus, the model that all microneme and rhoptry trafficking is RB-dependent is based primarily on the MyoF depletion phenotype (which results in RB accumulation) together with the observation that a relatively small amount of maternal microneme and rhoptry material is detectable in the RB of wild-type parasites. Although the authors' interpretation-that recycling is RB-dependent-is one possible explanation, alternative models are not discussed.<br /> For example, an alternative possibility is that the majority of micronemes and rhoptries are trafficked directly from the apical end of the mother parasite to the daughter cells without passing through the RB. In this scenario, only a subset of the organelles would enter the residual body, perhaps reflecting imperfect trafficking efficiency rather than an obligatory recycling step. Loss of MyoF would impair this trafficking pathway, resulting in the accumulation of secretory organelles within the RB. In other words, RB accumulation could be a consequence of MyoF depletion rather than evidence that all trafficking in wild-type parasites normally proceeds through the RB.

This alternative interpretation seems particularly relevant for the rhoptries, given that the authors themselves state that "M-RON2 was integrated into daughter rhoptries prior to mother cell collapse and formation of the RB."

Other comments:

Figure S10C<br /> To determine whether microneme degradation occurs in the RB, the authors quantified the fluorescence intensity of individual micronemes in control parasites and following auxin washout, showing that after redistribution the fluorescence intensity of individual vesicles is unchanged. However, this is not the appropriate analysis to address the question being asked. To conclude that micronemes are not degraded, the authors would need to quantify the total fluorescence intensity within the entire vacuole. For example, if half of the micronemes were degraded, the remaining micronemes would be expected to retain the same fluorescence intensity as those in the control parasites. Thus, unchanged fluorescence intensity of individual vesicles does not exclude the possibility that degradation has occurred.

Reviewer #3 (Public review):

Summary:

Knoerzer-Suckow et al. explore the mechanisms of organelle inheritance during endodyogeny in Toxoplasma gondii using an innovative dual-labeling approach to track the distribution of maternal organelles into daughter parasites. They can clearly distinguish between maternal and daughter-derived organelles using their dual-labeling Halo Tag approach. They reveal that different organelles are trafficked to daughter parasites in three broad patterns they have binned into groups. Their findings reveal a role for MyoF in the inheritance of micronemes and rhoptries, and notably, they observe that the inner membrane complex (IMC) is not recycled. Instead, the IMC undergoes a pronounced relocalization to the posterior of the maternal cell, where it is likely targeted for degradation.

Strength:

The data surrounding their MyoF knockdown experiments, IMC degradation, and trafficking of MIC2 after auxin washout are convincing. These data add to the knowledge of how organelle inheritance occurs in T. gondii, increasing the field's understanding of endodyogeny.

Weakness:

The inability to achieve higher temporal resolution due to phototoxicity precluded tracking of single micronemes, thus it remains possible that some micronemes follow a path similar to rhoptries and enter daughter cells before development of the residual body while others are recycled via the residual body.

Author response:

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

We sincerely thank the reviewers and the Reviewing Editor for their careful evaluation of our manuscript and for their constructive and insightful comments. Their suggestions have helped us to improve the clarity, rigor, and presentation of our work. In response to these comments, we have substantially revised the manuscript and performed several additional analyses and experiments, as summarized below.

Major additions and modifications made during revision

In response to the reviewers' comments, we have substantially revised the manuscript and performed several additional analyses and experiments:

New analyses

- Quantification of MyoF recovery following auxin washout using MyoF-mAID-HA immunofluorescence (Figure 8B, Figure S10D).

- Quantification of maternal MIC2 fluorescence intensity following 24 h MyoF depletion and subsequent redistribution after auxin washout (150 micronemes per condition; Figure S10A,B).

- Pearson correlation analysis of ANKER1-Halo and HDEL-GFP localization (Pearson's R = 0.92 ± 0.04; n = 30 parasites).

- Additional probability-based analysis supporting regulated microneme inheritance.

- Expanded analysis of microneme redistribution across larger replication stages (Figure S5D).

New figures

- Figure S6: Dual-labelling analysis of additional Group 2 organelles (ER, apicoplast, glideosome).

- Figure S10A, B: MIC2 fluorescence intensity analysis following RB retention and redistribution.

- Figure S10D: Correlation between MyoF recovery and phenotype rescue.

- Figure S11: Schematic overview of quantification and analysis workflow.

Additional experimental efforts

- Generation of a MIC2-Halo / IMC1-mKATE / Cb-Emerald parasite line to improve visualization of RB-associated trafficking.

- Multiple attempts to perform higher-temporal-resolution live-cell imaging. However, prolonged acquisition resulted in severe phototoxicity, replication arrest, and parasite death, preventing reliable long-term recordings.

Textual and methodological revisions

- Expanded Materials and Methods section with detailed descriptions of fluorescence quantification, colocalization analyses, and statistical procedures.

- Re-evaluation of statistical analyses using two-tailed tests throughout.

- Revision of manuscript text to clarify the evidence supporting RB-associated trafficking and to better acknowledge current limitations.

Public Reviews:

Reviewer #1 (Public review):

Summary:

This work asks the question of how different organelles and structures in the apicomplexan parasite Toxoplasma gondii are recycled and/or segregated to the daughter cells during cell replication. In particular, they consider an unusual cell structure called the residual body that links replicating cells during the intracellular infection stage of this parasite. The residual body has historically been considered a 'dumping ground' for unnecessary relics of the mother cell during division, but this notion is increasingly being revised. Indeed, cell replication in Toxoplasma is often misinterpreted as cell division (cytokinesis), but in fact, the cell replicates its organelles and structures to multiple 10s of copies in seemingly distinctly formed daughter cells, but cytokinesis is delayed for many such cycles and typically only occurs simultaneously with parasite egress from its host cell. The residual body is, in fact, the connection between these pre-cytokinetic replicated daughters, and effectively, this is still a single cell at this stage. The authors have previously shown that an actin network extends through the residual body between these daughter cells, and ER and mitochondria common to all cells are also linked through this structure. This study examining the fates of organelles during cell replication is timely for continuing our understanding of how this fascinating component of the cell participates in these processes. The authors use Halo-tags as their principal tool to track discrete populations of proteins, labelling their organelle locations, and this provides beautiful insight into these processes.

Strengths:

Using dyes conjugated to Halo tags, this work elegantly tracks the fates of proteins synthesised by an original 'mother' cell over several replication cycles of pre-cytokinetic 'daughters'. Using this tool, they show that some organelles are made intact just once and that some of these can be subsequently sorted to the daughters (micronemes and rhoptries) while others are dismantled (IMC) and the daughters must make their own. A third set of organelles (largely synthesis, sorting, and metabolic compartments) is divided and inherited, and new daughter-synthesised proteins are added to the preexisting maternal proteins in these structures. A role for actin and myosin is clearly demonstrated for micronemes and rhoptries, and this correlates with their relatively late inheritance into the developing daughters. Overall, this work gives clarity to the behaviours of several cell structures during replication and paves the way to a better understanding of the mechanisms that drive the differences between structures and the universality of these processes in other apicomplexan parasites.

Weaknesses:

In addressing the question of residual body participation in sorting of organelles, it would be useful to clearly define this structure and when and where it is delineated from the posterior of a mother cell during the formation of daughter structures. This might seem like a moot point, but it would give clarity to notions of recycling and 'reservoirs'. Mother cells retain their active invasion apparatus until very late in daughter formation, and the need for micronemes and rhoptries to be released from this service late in the process might explain why they are only then trafficked to the cell posterior and then into the daughters. So, is this a distinct 'residual body' body function/reservoir or just a spatial constraint of this sequence of daughter formation? In subsequent cell replications (4, 8, 16... stages), is there a separation between the residual body that links them all and the posterior of each new 'mother cell', and if so, when is this distinction lost? This is important because without a definition, we might be confusing different processes.

We thank the reviewer for this excellent and thoughtful question. The residual body (RB) emerges at the end of the first replication cycle, where it is delineated by the basal complex and persists as an IVN-associated compartment connecting all daughter parasites through both plasma membrane and cytoplasm. Previous EM and live-cell studies, including ours, have shown that the RB is not a passive remnant but a dynamic structure dependent on F-actin and unconventional myosins, supporting recycling, inter-parasite connectivity, and synchronous growth (Delbac et al., 2001; Muñiz-Hernández et al., 2011; Frénal et al., 2017; Periz et al., 2017).

In the present study, the MyoF reversibility experiment provides strong support for a model in which RB functions as an active recycling hub. Upon MyoF depletion, maternal microneme and rhoptry proteins accumulate within the RB. Following restoration of MyoF expression, this material is redistributed to daughter organelles. We interpret this reversible phenotype as evidence that the RB represents a distinct and regulated trafficking intermediate rather than simply a by-product of late daughter cell formation.

We agree with the reviewer that mother cells retain a functional invasion apparatus until very late during daughter formation, and that the delayed release of micronemes and rhoptries likely contributes to their late trafficking toward the cell posterior. However, our data indicate that once released, these organelles transit through a defined RB compartment that actively participates in their recycling rather than merely reflecting positional constraints. This has been previously well illustrated for micronemes, which are trafficked along F-actin filaments within the residual body (Periz et al., 2019).

At later rounds of replication (4, 8, 16 parasites), previous studies have demonstrated the presence of multiple residual body centres within the same vacuole. However, the precise temporal and structural distinction between the RB linking parasites within the vacuole and the posterior of newly formed mother cells remains insufficiently resolved and is beyond the scope of the present study. Importantly, available ultrastructural and live-cell imaging supports the persistence of shared RB compartments connecting parasites within a vacuole, arguing against a simple conflation of posterior membranes and residual body material.

While the primary aim of the current work was to investigate the RB's role in organelle recycling, we fully agree that a more precise definition of when and how the RB is formed, remodelled, and ultimately resolved during successive replication cycles will be essential to distinguish recycling from spatial constraints. We have revised the Discussion to better acknowledge this limitation and to avoid overinterpreting the role of the RB in organelle inheritance.

Are rhoptries/micronemes that originate in one 'mother' able to be sorted to the 'daughters' from a distinct mother in this syncytium? If so, this would make it a sorting centre, but otherwise we could be just capturing the activities at the posterior of any given cell during replication. The authors' further thoughts on this would be very interesting.

We agree with the reviewer that our current data do not definitively demonstrate whether rhoptries or micronemes originating from one “mother” parasite can be redistributed to daughters derived from another mother within the same syncytial vacuole. Nevertheless, our MyoF chase experiments are consistent with a model in which the RB/IVN functions as an active recycling and sorting hub rather than simply representing posterior trafficking events associated with individual parasites.

Upon MyoF depletion, maternal micronemes accumulated within the RB. Following restoration of MyoF expression, these accumulated micronemes were subsequently redistributed to daughter parasites. This reversible redistribution is more consistent with an active recycling process than with passive accumulation alone.

To further support this interpretation, we expanded the analysis presented in Figure S5 by including additional vacuoles and larger replication stages (new panel D). These analyses show that maternal micronemes are redistributed broadly and relatively evenly among daughter parasites. We additionally performed a probability-based analysis demonstrating that the recurrent and homogeneous redistribution patterns observed are highly unlikely to arise from stochastic capture events occurring independently at the posterior end of each parasite during replication. Together, these analyses support the interpretation that microneme redistribution is a regulated process.

Direct demonstration of recycling between all parasites within a vacuole would require a system allowing simultaneous differential labeling of (i) daughter parasites derived from a specific mother cell and (ii) the maternal organelles originating from that same mother during a subsequent replication cycle. To our knowledge, such an approach is not currently technically feasible. Nevertheless, our live-cell imaging experiments provide additional support for communal redistribution, as microneme material accumulated within the RB was subsequently observed redistributing, albeit unevenly, across multiple tachyzoites within the same vacuole.

The Group 2 structures are described as those that are divided between daughters and receive newly synthesised proteins that add to the maternal protein of these compartments. While this is a logical conclusion for several that are mentioned, where the maternal protein signal is seen to be depleted with replication (including for the apicoplast, ER, glideosome, and Golgi). Data for the addition of new proteins to these existing structures is actually only presented in direct support of this for the Golgi.

We thank the reviewer for this important clarification. We initially selected the Golgi as a representative example because its morphology and restricted localization provide the clearest visualization of the dual-labeling dynamics. However, the same experimental approach was applied to all Group 2 organelles analyzed in this study. To address the reviewer's concern more directly, we have now included a new supplementary figure (Figure S6) showing that the same pattern is also observed for the apicoplast, ER, and glideosome.

We would also like to clarify that the maternal protein signal is not lost during replication but instead becomes progressively diluted as these organelles expand, are partitioned into daughter parasites, and incorporate newly synthesized proteins. The Golgi was originally highlighted because these dynamics are most readily visualized in this compartment; however, the same principle applies to all Group 2 organelles analyzed in this study, as now illustrated in Figure S6.

Reviewer #2 (Public review):

Summary:

Toxoplasma gondii is an obligate intracellular parasite and the causative agent of Toxoplasmosis. Parasite invasion into host cells, intracellular replication, and then egress, which results in the destruction of the infected cell, is central to pathogenicity. This manuscript focuses on understanding how maternal resources (in this case, cellular organelles) are shared between daughter parasites during cell division. Many organelles are single copy, meaning that division and inheritance by the daughters is crucial for successful replication. The major strength of this study was the use of a Halobased pulse chase assay to characterize patterns of organelle inheritance. The results show that both microneme and rhoptries (secretory vesicles) previously thought to be synthesized de novo are inherited by daughter parasites. Thus, this paper adds new insight to our understanding of cell division in this important parasite.

Strengths:

This study demonstrated that pulse labeling of proteins can be used to monitor protein synthesis, turnover, and movement. This approach will be of great interest to the field. Using this method, the authors demonstrate three main modes of organelle inheritance.

(1) Organelles, where there are multiple copies (such as secretory vesicles, micronemes, and rhoptries), are divided between the daughter parasites, with additional contribution of newly formed vesicles. New and old material remain as separate entities in the cell.

(2) Single-copy organelles, which are expanded to include newly synthesized material prior to division, such as the Golgi and apicoplast.

(3) Cytoskeletal structures that are synthesized anew during each round of division. These studies provide more refined insight into patterns or organelle inheritance and demonstrate that secretory organelles are not made de novo during each round of division as previously thought. The paper has a logical flow, and overall, the data is presented in a clear and organized fashion.

Weaknesses:

(1) Descriptions of methodology and statistical analysis were incomplete.

We agree with the reviewer that the description of the methodology and statistical analyses required further clarification. To address this, we have added a new supplementary figure (Figure S11) illustrating the experimental workflow, quantification strategy, and analysis pipeline. We have also expanded the Materials and Methods section to provide detailed descriptions of the experimental design, fluorescence quantification procedures, statistical analyses, and the number of biological replicates. These revisions provide a clearer and more comprehensive description of the methodology and data analysis.

(2) There are inconsistencies between the data in Figures 1 and 5. In Figure 1, a small amount of maternal IMC is visible in stage 2 parasites. Although this is a ~90% reduction, these parasites should be quantified as parasites with material IMC. However, the graph in Figure 5C indicates that no material parasites have GAPM1a, given that graph 5C is a binary measure (present vs. absent), one would expect a non-zero percent of parasites to have maternal material.

We agree with Reviewer 2 that, based on the raw fluorescence signal, one might expect a non-zero percentage of parasites to retain maternal IMC material after the first replication. The apparent discrepancy between Figures 1 and 5 reflects our thresholding strategy rather than inconsistent data.

Figure 5C presents a binary analysis (presence versus absence) using a threshold calibrated from stage 1 parasites and applied uniformly across all markers. Under this criterion, the residual GAPM1a signal after the first replication falls below the detection threshold, resulting in 0% positive vacuoles. Although normalization to stage 2 parasites would detect this weak residual signal, such a protein-specific threshold would compromise direct comparison across the dataset.

To clarify this point, we have updated the Figure 5C legend to explain the analytical approach and the asterisk associated with GAPM1a. The residual maternal IMC signal visible in Figure 1 represents a rare example selected to illustrate the remaining ~10% signal and is consistent with the absence of detectable maternal IMC1 after replication in Figures 2C and 5E.

(3) The conclusion from Figure 6 was not justified based on the data. I agree with the author's conclusion that the accumulation of micronemes and rhoptries in the residual body was timedependent. In Figure 6A, the signal observed in the residual body at times 6:30, 13, and 14 hours is not observed in subsequent time points. However, the fate of these micronemes and rhoptries is unclear. It cannot be concluded that these vesicles are recycled back to the mother. They could also have been degraded. In fact, the graphs of microneme inheritance in Figure 2B show a decrease in maternal signal from 100% to 80% between stages 1 and 2, indicating that some microneme degradation is taking place.

We agree with the reviewer that Figure 6 alone does not definitively establish the fate of micronemes and rhoptries accumulating within the residual body (RB), and that both recycling and degradation remain possible interpretations. Our conclusion that maternal micronemes are predominantly recycled is therefore based on the integration of Figure 6 with our MyoF depletion and recovery experiments, additional quantitative analyses, and previous work demonstrating F-actin-dependent microneme trafficking through the RB (Periz et al., 2019).

Consistent with this model, MyoF depletion results in the accumulation of maternal micronemes within the RB, whereas restoration of MyoF expression following auxin washout leads to their redistribution across multiple tachyzoites within the same vacuole (Figure 8). Furthermore, maternal microneme signal remains detectable even after prolonged MyoF depletion (up to 48 h) and multiple rounds of replication (Figures 7 and 8), arguing against extensive degradation.

To further address this possibility, we quantified the fluorescence intensity of individual maternal MIC2-positive micronemes retained within the RB after 24 h of MyoF depletion and following redistribution after auxin washout (150 micronemes per condition). No significant difference in fluorescence intensity was observed compared with control maternal micronemes (Figure S10A,B), indicating that maternal microneme signal is preserved during RB retention and redistribution.

We therefore interpret the decrease in maternal microneme signal observed between stages 1 and 2 in Figure 2B primarily as a consequence of redistribution and dilution rather than degradation, consistent with the stable fluorescence intensity of individual micronemes (Figure 3). Regarding rhoptries, we note that the majority (~90%) are incorporated into daughter parasites before budding is complete, limiting their accumulation within the RB and suggesting that RB-associated trafficking primarily reflects redistribution rather than bulk degradation.

(4) To convincingly demonstrate that the redistribution of micronemes and rhoptries was due to recovery of MyoF protein levels after auxin washout, a Western blot should be performed to show MyoF protein levels over time. In addition, the decrease in mMIC2 protein levels in the residual body in Figure 8F should be measured and normalized for photobleaching. Both apical and basal signals appear to be reduced over the time course of imaging.

We agree with the reviewer that demonstrating MyoF recovery following auxin washout is important. Rather than performing a Western blot, we monitored MyoF recovery by immunofluorescence using the HA tag in the MyoF-mAID-HA strain, allowing direct correlation between MyoF reappearance and microneme redistribution at the single-vacuole level. These data are now included in Figure 8B, with the corresponding MyoF presence–phenotype association analysis presented in Figure S10D.

Regarding photobleaching, we agree that fluorescence loss during time-lapse imaging is an important consideration. However, in this experiment, changes in fluorescence intensity reflect not only photobleaching but also biological redistribution of micronemes and movement of parasites in and out of the imaging plane. In the absence of a stable internal reference fluorophore, applying a standard photobleaching correction could therefore introduce additional inaccuracies. For this reason, we did not quantify fluorescence intensity during the redistribution phase.

Instead, to assess whether maternal microneme signal is lost during RB retention and redistribution, we quantified the fluorescence intensity of individual maternal MIC2-positive micronemes following 24 h of MyoF depletion and subsequent auxin washout (Figure S10A,B). No significant difference was observed compared with control maternal micronemes, supporting the conclusion that redistribution occurs without substantial loss of the maternal microneme pool.

Reviewer #3 (Public review):

Summary:

Knoerzer-Suckow et al. explore the mechanisms of organelle inheritance during endodyogeny in Toxoplasma gondii using an innovative dual-labeling approach to track the distribution of maternal organelles into daughter parasites. They can clearly distinguish between maternal and daughterderived organelles using their dual-labeling Halo Tag approach. They reveal that different organelles are trafficked to daughter parasites in three broad patterns, which they have binned into groups. Their findings reveal a role for MyoF in the inheritance of micronemes and rhoptries, and notably, they observe that the inner membrane complex (IMC) is not recycled. Instead, the IMC undergoes a pronounced relocalization to the posterior of the maternal cell, where it is likely targeted for degradation.

Strengths:

The data surrounding their MyoF knockdown experiments, IMC degradation, and trafficking of MIC2 after auxin washout are compelling. These data add to the knowledge of how organelle inheritance occurs in T. gondii, increasing the field's understanding of endodyogeny.

Weaknesses:

(1) The evidence provided to support the claim that microneme and rhoptry inheritance specifically traffics through the residual body does not sufficiently substantiate the claim. The temporal resolution of the imaging is inadequate to precisely trace the path of microneme and rhoptry inheritance. From the data shown in the manuscript, it can be concluded that at least some of the micronemes and rhoptries might be recycled through the residual body, but it is unclear whether many or most of these organelles do so.

We thank the reviewer for this important comment and refer also to our response to Reviewer 1 above.

Previous work has demonstrated F-actin-dependent trafficking of micronemes within the residual body (RB) (Periz et al., 2019). Consistent with these findings, our data support a model in which RB-mediated trafficking contributes to maternal microneme recycling. We acknowledge, however, that the temporal resolution of our imaging does not allow continuous tracking of every individual organelle throughout the entire replication process.

In contrast, our observations indicate that the majority of maternal rhoptry material is incorporated into daughter cells before replication is complete and therefore does not necessarily transit through the RB under normal conditions (Figure 6). Nevertheless, rhoptry inheritance remains dependent on the actin–MyoF trafficking machinery, as MyoF depletion results in the accumulation of rhoptry material within the RB (Figures 6 and 7).

Taken together, our data support a model in which the RB serves as an important recycling hub for maternal micronemes and can, under conditions of impaired trafficking, also transiently accommodate rhoptry material. However, our imaging resolution does not allow us to conclude that all microneme or rhoptry inheritance obligatorily transits through the RB, and we have revised the manuscript to reflect this limitation more explicitly.

(2) The absence of specific markers for the residual body brings into question whether microneme inheritance occurs through a discrete residual body or simply via the basal end of the maternal parasite. The authors need a robust way to visualize and define the residual body to claim that micronemes and rhoptries are specifically transported through this structure.

We agree with the reviewer that the absence of a dedicated residual body (RB) marker remains a limitation and that such a tool would improve the precision of our analyses. To date, no specific RB marker has been identified (see also our response to Reviewer 1). The most reliable proxy currently available is the F-actin chromobody, which labels the dense F-actin network associated with the RB. Using this approach, previous work demonstrated F-actin-dependent trafficking of micronemes within the RB (Periz et al., 2019).

Building on these findings, our data support a model in which RB-associated trafficking contributes to maternal microneme recycling, whereas rhoptries are more frequently incorporated directly into daughter cells without obvious RB transit. In addition, functional perturbation of the actin–MyoF transport machinery, through MyoF depletion and subsequent recovery, supports the interpretation that the RB represents a discrete actin-associated compartment involved in organelle redistribution. Nevertheless, we acknowledge that our current imaging resolution does not allow us to determine the extent to which all microneme or rhoptry inheritance occurs through the RB, and we have revised the manuscript accordingly.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Comments for revision where either the clarity or accuracy could be improved:

(1) The methods could do with some further detail with respect to the fluorescence intensity measurement. For example, where the Z-series was taken, and for measurements, were maximum projects taken or single Z-planes? Were all measurements made on unprocessed images, or was any deconvolution, etc, undertaken?

We updated the Materials and Methods for better clarity and now read as: “Parasites were labeled as described above and allowed to replicate for 24 h on HFF-coated Ibidi live-cell dishes. Approximately 15 fields of view were imaged using Z-stacks spanning 3 μm centred on the vacuoles. For each replication stage (1, 2, 4, and 8 parasites per vacuole), individual tachyzoites were sampled across multiple vacuoles.

Maximum-intensity projections were generated from non-deconvolved images. Fluorescence intensity (FI) was quantified as the maximum grey value measured within regions of interest (ROIs) drawn on individual tachyzoites from each vacuole stage. ROIs excluded overlapping parasites, neighbouring vacuoles, and regions with atypical signal intensity. For each biological replicate, up to 25 tachyzoites per replication stage were analyzed, and the mean FI value was calculated for each stage. The highest mean FI observed among the stages within a replicate was defined as 100%, and FI values for the other stages were expressed relative to this maximum. Relative FI values were then averaged across three independent biological replicates. Data are presented as mean ± SD.”

(2) Line 93: Is more JF549 added at each of stages 2, 4, and 8? I assume so to get the progressive increase, but it would help to clarify this here.

As illustrated in Figure 1a, the second dye is added only once, after 24 h of replication, and is then washed off before imaging. It is not maintained throughout the replication steps. The observed increase in signal is related to the presence of newly synthesized (de novo) proteins generated during replication. The Halo tags of these new proteins are initially free of ligand, as no ligand is present during replication. During the second labelling step, all free Halo tags can bind the dye. The signal intensity at each replication step therefore reflects the amount of de novo material present, which is higher at step 4 than at step 2, because the proteins generated de novo during stage 2 are also present in stage 4. The signal will be determined by both the number of newly produced molecules and their concentration at the same localization.

(3) Line 103: Check that the Carruthers and Sibley, 1997, ref for Tic20 in the apicoplast is correct. I don't think this could be correct given the date.

The reference it will be corrected to G van Dooren et al. 2008.

(4) Figure legends: It would be useful to state what form of microscopy was used in each figure.

Following the reviewer's advice the legends has been updated.

(5) Figure 2, S2: How are single organelles tracked, such as rhoptries? I'd assume that a cell will gain new de novo organelles as well, and that this would reduce the signal per cell. Stage 2 has a rhoptry signal in both daughters, so I'd expect the signal to be roughly half for the whole cell, unless the authors can resolve individual rhoptries (which would surprise me with this microscopy). If individual rhoptries were resolved, how was this done, and what was the confidence in this (were there controls?)

We thank the reviewer for this important point. We do not resolve individual rhoptries with the imaging conditions used in this study. Instead, fluorescence intensity was measured as the maximum grey value within a representative region of interest (ROI) encompassing the apical rhoptry signal while excluding overlapping parasites and regions with atypical fluorescence intensity. The same ROI selection strategy was applied consistently across all replication stages, allowing direct comparison with stage 1 parasites.

The analyses presented in Figures 3 and S2 show an increasing proportion of tachyzoites lacking detectable maternal rhoptry signal as replication progresses, while the fluorescence intensity of the remaining maternal signal remains relatively stable. Together, these observations are consistent with the redistribution of intact maternal rhoptries rather than a progressive loss of rhoptry fluorescence. 

(6) Line 150: it is unclear what is meant by 'regulated partitioning'. The more equal inheritance of micronemes versus rhoptries might not indicate a 'regulated partitioning' but just a more uniform distribution, given the larger number of micronemes versus rhoptries.

We agree with the reviewer that this statement required clarification, and we have revised the text accordingly. Our intention was to emphasize that microneme inheritance appears to be a regulated process, rather than to directly compare it with rhoptry inheritance. The observed differences between these organelles are likely influenced, at least in part, by their different abundances.

Across successive rounds of replication, daughter parasites consistently inherit comparable amounts of maternal micronemes, even at later replication stages (Figure S4). Given that a single mother parasite contains approximately 30–40 micronemes, whereas successive rounds of endodyogeny can generate up to 32 daughter parasites, a purely stochastic segregation would be unlikely to produce the relatively uniform distribution observed (~1–2 maternal micronemes per tachyzoite). To support this interpretation, we performed an additional probability-based analysis, which indicates that the observed redistribution patterns are unlikely to arise by chance alone. We therefore interpret these findings as supporting the existence of mechanisms that promote balanced microneme inheritance during parasite replication.

(7) Line 184: How is the maternal signal measured without detecting the internal daughter signal? Is this an average signal for the full parasite, or just for a cross-section of the IMC? And if the latter, how are the different profiles of mother and daughter accounted for? Also, the abbreviation in the brackets doesn't make sense here.

We thank the reviewer for this important point. We have revised the Materials and Methods section to provide a clearer description of the fluorescence intensity (FI) measurements and added a new supplementary figure (Figure S11) illustrating the analysis workflow.

Briefly, FI measurements were performed on maximum-intensity projections generated from Z-stack images without deconvolution. A representative region of interest (ROI) was selected, and the maximum grey value was used for quantification. This approach minimizes variability arising from differences in ROI size and provides a robust metric for comparison across replication stages.

Daughter cell fluorescence was measured using the same approach while excluding overlapping signals from neighboring daughter cells and the maternal IMC. Maternal and daughter signals were distinguished based on their spatial localization and fluorescence labeling. Finally, the abbreviation in brackets has been corrected for clarity.

(8) Line 191: Subheading a bit unclear. Distinct from other organelles, or are miconeme and rhoptry pathways distinct from each other?

We agree with the reviewer and have updated the subheading to “Whole-organelle inheritance of micronemes and rhoptries occurs via distinct recycling pathways”

(9) Line 193: The site of disassembly of the IMC (suggested RB here) might not be the same as the site of degradation. I suggest using 'disassembly' instead here.

We agree that “disassembly” is an appropriate term to describe the breakdown of the IMC at the residual body (RB). However, we also believe that the RB represents the primary site of IMC degradation, for two reasons. First, if IMC material were not degraded at this site, we would expect to detect Halo-positive signal elsewhere following IMC collapse, which we do not observe. Second, transport of IMC material to an alternative degradation site would be required, but no IMC-positive vesicles are observed, arguing against significant redistribution. Together, these observations support the conclusion that the RB is both the site of disassembly and degradation of maternal IMC.

(10) Line 215: The conclusion for a difference in timing of microneme and rhoptry segregation is not clearly supported by the data presented. Also, if there are more micronemes than rhoptries, then the frequency of observing a microneme being trafficked through the RB would need to be higher than for rhoptries if the mechanisms were the same. So, a difference in frequency here cannot be used to argue for a different mechanism.

We agree with the reviewer and the text have been edited to soften our conclusion. 

(11) Line 234: 'segregation' might be a better term than 'recycling' here because it is actually the sorting into daughter cells that is the important process.

The text have been edited

(12) Line 235: I don't think this can be what the authors intend to say. If the maternally-inherited rhoptries are not trafficked through the RB (every time), then how do they get into the daughters? Perhaps this is a case where a clear definition of the RB is required.

Our observations indicate that maternally inherited rhoptries are frequently incorporated into daughter cells before collapse of the mother cell and establishment of the residual body (RB). Although F-actin is enriched within the RB, an actin network is also present throughout the parasite cytoplasm, where MyoF is likewise localized. We therefore propose that, unlike micronemes, rhoptries do not necessarily transit through the RB during every replication cycle but can be incorporated directly into developing daughter cells while still relying on the same actin–MyoF-dependent trafficking machinery.

(13) The MyoF Rescue, the experimental plan is not fully described in order to be clear. If the endomembrane architecture was disrupted by MyoF depletion, and this secondary effect caused the segregation phenotype, restoration of MyoF might also simply restore the endomembrane system. So a direct role for MyoF doesn't seem to have been tested in this case.

We appreciate the reviewer's concern that the segregation phenotype could, in principle, arise indirectly from disruption of endomembrane architecture following MyoF depletion. However, although Golgi morphology is altered in MyoF-depleted parasites, its core functions appear largely preserved. This is supported by the normal biogenesis of de novo micronemes, their correct targeting to the apical pole, their efficient secretion, and the previously reported preservation of parasite invasion. In addition, Golgi markers are not detected in the residual body, where maternally inherited micronemes accumulate, arguing against Golgi-mediated trafficking as the primary cause of the segregation phenotype.

Taken together, these observations support the interpretation that the segregation defects are more likely to reflect a direct role of MyoF in organelle trafficking and inheritance than a secondary consequence of generalized disruption of endomembrane organization.

(14) Line 279: Why call it a checkpoint? What is the evidence for its presence here being sensed before a further process is activated, which is what a checkpoint does?

We agree the reviewer that checkpoint is a misleading term and have been updated to trafficking hub. 

(15) Line 287 confuses replication of the daughters from cytokinesis, which only happens when each cell loses cytoplasmic connectivity with the other.

We will clarify this point. In Toxoplasma gondii, cytokinesis represents the final step of daughter cell formation, during which the two fully assembled daughter parasites separate from the mother cell following collapse of the maternal cytoplasm. Historically, the residual body was proposed to arise simply as leftover material from this process. However, multiple studies have now shown that residual body formation is an active and regulated process, dependent on specific cytoskeletal and trafficking factors. Importantly, although cytokinesis marks the physical separation of daughter cells from the mother, parasites within a vacuole remain connected via the residual body and continue to share cytoplasmic and plasma membrane components until egress. 

(16) Line 298: I don't think there is direct evidence of degradation in the RB. There might be disassembly, but degradation implies proteolysis, which hasn't been tested for.

We agree with the reviewer that our data do not provide direct biochemical evidence of proteolysis within the residual body (RB) and primarily demonstrate disassembly of the maternal IMC at this site. However, several observations are consistent with local degradation. Following IMC collapse, we do not detect Halo-positive signal elsewhere in the parasite, nor do we observe IMC-positive vesicles or other structures that would suggest transport to a distinct degradation compartment.

In addition, previous work identified the E3 ubiquitin ligase CSAR1 as a mediator of protein turnover within the RB, supporting the idea that this compartment is associated with degradation-related processes (O'Shaughnessy et al., 2023). While we cannot formally demonstrate proteolysis, these observations support a model in which IMC disassembly is closely coupled to local degradation within the RB.

(17) The paragraph structure gets a bit confusing at times. See single sentence paragraph, Line 224. Does this sentence justify its own paragraph?

The text has been edited.

(18) Make sure Toxoplasma gondii is in italics throughout.

The text has been edited

(19) Line 279 cites Figure 10. But there is none.

The text has been edited

(20) I advocate introducing a few new acronyms, like DCs. I find that this ultimately reduces the ease with which readers read the work if they don't learn them all quickly.

We agree that excessive use of acronyms can negatively impact readability. In the present manuscript, all abbreviations used in the text are introduced at their first occurrence in the Introduction, including DCs (line 32), IMC (line 41), PV (line 35), ER (lines 38–39), RB (line 50), and IVN (line 49). We have carefully limited the use of abbreviations to commonly used terms in the field and to those that recur frequently throughout the manuscript, with the aim of balancing clarity and readability. Nevertheless, we are happy to reduce or remove specific abbreviations if the reviewer feels this would further improve clarity.

Reviewer #2 (Recommendations for the authors):

(1) Descriptions of methodology and statistical analysis were incomplete as follows:

(1a) It was unclear how the fluorescence intensity measurements (used to evaluate inheritance vs. new synthesis) were carried out. The y-axis on the graph is labeled average fluorescence intensity (% of max intensity). However, it does not state what was averaged (average fluorescence per vacuole?) and what was max intensity (max pixel intensity in each image or time point with the highest average intensity, relative to the other time points?)

We agree with the reviewer that the original description of the fluorescence intensity (FI) measurements lacked clarity. We have therefore revised the Materials and Methods section and added a new supplementary figure (Figure S11) illustrating the analysis workflow.

Briefly, vacuoles were imaged as Z-stacks, and maximum-intensity projections were used for analysis. FI was quantified as the maximum grey value measured within representative regions of interest (ROIs) drawn on individual tachyzoites, rather than as an integrated fluorescence signal across the vacuole. This approach minimizes variability arising from differences in ROI size and allows direct comparison between replication stages.

For each biological replicate, up to 25 tachyzoites per replication stage were analyzed. The mean FI for each stage was normalized to the highest mean value within that replicate, and data from three independent biological replicates were subsequently averaged.

(1b) Given the uncertainties with how these measurements were performed, it is difficult to interpret the data. For example, one would expect that the fluorescence intensity of newly synthesized IMC1 in 8-parasite vacuoles would be 4 times higher than that of a 2-parasite vacuole; however, based on the graph in Figure 1B, the measured increase was only 30%.

We agree that the original description of the fluorescence intensity (FI) measurements required further clarification and have revised the Materials and Methods accordingly. As the reviewer correctly notes, a fourfold increase in FI between 2- and 8-parasite vacuoles would be expected if total IMC fluorescence across the entire vacuole had been measured. However, this was not the parameter quantified.

Instead, FI was measured as the maximum grey value within representative regions of the daughter IMC, providing a per-cell rather than a whole-vacuole measurement. Using this approach, FI increases between the 2- and 4-parasite stages and then reaches a plateau.

This behavior is consistent with the biology of IMC biogenesis. Although the total amount of IMC per vacuole increases with parasite number, the amount of IMC protein incorporated into each daughter parasite remains relatively constant. Consequently, once daughter IMCs are fully assembled from de novo-synthesised material, additional rounds of replication increase the total IMC content per vacuole but not the fluorescence intensity measured for individual parasites.

(1c) T. gondii replicates in an asynchronous manner, so that at the 24-hour time point, a single dish can contain vacuoles containing 2, 4, and 8 parasites. This should be stated explicitly so readers unfamiliar with T. gondii's growth patterns can understand how the experiment was performed.

We agree with the reviewer and have updated the text line 93. “As Toxoplasma gondii replicates in an asynchronous manner, after 24 of replication, vacuoles containing 1, 2, 4, and 8 parasites can be observed in a single dish.”

(1d) Colocalization package in Fiji used for ANKER1-Halo with HDEL-GFP and MIC2/RON2 with CbEmeraldFP should be specified.

We thank the reviewer for this suggestion. Following this recommendation, we performed Pearson correlation analysis for the ANKER1–HDEL-GFP experiment using the Coloc 2 plugins of FiJi. ANKER1Halo and HDEL-GFP showed a strong spatial correlation (Pearson's R = 0.92 ± 0.04, n=30 parasites from three independent biological replicates), supporting localization of ANKER1 to the ER.

We note, however, that this analysis should be interpreted as evidence for co-distribution within the same organelle rather than direct molecular colocalization, as ANKER1 is a transmembrane protein whereas HDEL-GFP labels the ER lumen.

For all the rest of our analysis, no automated colocalization package or plugin in Fiji was used for the analyses involving ANKER1-Halo with HDEL-GFP or MIC2/RON2 with Cb-EmeraldFP. Colocalization was assessed manually across all experiments by inspecting both full Z-stacks and maximum-intensity projections to ensure robust spatial overlap.

For MIC2 and RON2, the presence of signal within the Cb-Emerald–positive filament was scored as either cytoplasmic, on the residual body or absence of colocalisation.

In total, more than 300 and 500 vacuoles were analyzed for MIC2 and RON2–Cb-Emerald colocalization respectively (stable expression of both markers), and more than 150 vacuoles were analyzed for ANKER1-Halo and HDEL-GFP colocalization (transient expression of HDEL-GFP).

This information has now been added to the Methods section.

(1e) Statistical methods should be described on an experiment-by-experiment basis. The authors should justify why a one-tailed t-test was conducted. A two-tailed t-test seems more appropriate.

We agree that statistical methods should be clearly justified on an experiment-by-experiment basis. All the statistical analysis have been performed using two tails and updated in the figures.

(1f) In Figures 5E and 5F, using boxes to indicate the exact areas of the cell that were used in the fluorescence intensity measurements, rather than arrows, would make this data easier to interpret.

The figure has been updated.

Reviewer #3 (Recommendations for the authors):

The current time-lapse images and videos do not clearly demonstrate microneme movement from the maternal parasite apical end to the residual body and back to the apical end of daughter parasites. As such, the route by which micronemes enter daughter parasites remains inconclusive. To strengthen their claims, the authors should employ higher temporal resolution imaging to definitively capture the movement of micronemes from the maternal apical region into the daughters. From the current data, it also seems plausible that the micronemes may be trafficked into the daughters through the conoid as well, as there is no evidence provided showing a movement of micronemes away from the apical end of the maternal parasite before being present in the daughter parasites.

We agree with the reviewer that higher temporal resolution imaging would provide a more definitive view of microneme trafficking. However, long-term live imaging of replicating Toxoplasma gondii requires a compromise between temporal resolution and parasite viability. In our experiments, images were acquired every 15–30 min over periods of up to 16 h, as more frequent acquisition consistently induced phototoxicity and prevented completion of parasite replication.

Despite this limitation, our imaging reliably tracked maternal micronemes over successive rounds of endodyogeny and consistently showed microneme signal associated with the residual body. These observations are in agreement with previous high-temporal-resolution studies, which demonstrated F-actin-dependent microneme trafficking within the residual body over shorter imaging periods (Periz et al., 2019).

We cannot formally exclude the possibility that some micronemes are transferred directly to daughter parasites through the apical end. However, together with previous studies showing enrichment of F-actin at the basal region of developing daughter cells rather than at the apical tip (Periz et al., 2017), our observations support a model in which RB-mediated trafficking contributes to maternal microneme inheritance.

The lack of a clear residual body marker needs to be addressed, as the distinction between the basal end of the maternal cell and a bona fide residual body must be explicitly defined to substantiate the major claim of the study. As it stands, it remains unclear whether micronemes and rhoptries as a whole travel through the residual body to be transported into the daughter parasites.

We agree with the reviewer that a marker specific to the residual body would strengthen this study. Unfortunately, no such marker has been identified to date. The F-actin chromobody is currently the best available proxy, as previous studies have shown that the F-actin network is enriched within the residual body (Periz et al., 2017; Kellermeier et al., 2024). Moreover, high-resolution live-cell imaging has previously demonstrated F-actin-dependent microneme trafficking within this compartment (Periz et al., 2019). We have revised the manuscript to more clearly acknowledge this limitation.

The conclusions drawn from the actin colocalization data in Figure 6C are based entirely on fixed samples, despite all experimental tools being compatible with live-cell imaging. Supplementing the fixed imaging with live cell data would increase its biological relevance. Published studies have shown that fixation of the actin chromobody results in the loss of resolution of an appreciable amount of the cytosolic F-actin network, and while the localizations analyzed here are primarily along the periphery, since the quantification and text make claims about the colocalization within the cytosol, this potential loss of cytosolic F-actin becomes an issue as there may be more actin available for analysis that is lost due to fixation within the cytosol of the parasites.

We thank the reviewer for raising this important point and agree that conventional fixation can compromise preservation of the F-actin network. However, we used the same fixation protocol described by Periz et al. (2019), which allows reliable visualization of the RB-associated F-actin network. We have also corrected the description of the fixation protocol in the Materials and Methods.

Fixation was necessary to image entire vacuoles with sufficient spatial resolution and signal-to-noise ratio for the volumetric analyses presented in Figure 6C. Although some loss of cytosolic F-actin cannot be excluded, this would be expected to reduce, rather than artificially increase, the detection of organelle–actin associations.

Importantly, previous live-cell imaging studies demonstrated F-actin-dependent microneme trafficking (Periz et al., 2019), and our observations are consistent with these findings. Moreover, the defects observed following MyoF depletion provide independent functional evidence that the trafficking events described here rely on the actin–MyoF transport machinery.

The statement of colocalization should be backed up by quantitative coefficients like Pearson's coefficient.

We thank the reviewer for this helpful suggestion. Following this recommendation, we performed a Pearson correlation analysis of ANKER1-Halo and HDEL-GFP using the Coloc 2 plugin in Fiji. ANKER1-Halo showed a strong spatial correlation with HDEL-GFP (Pearson's R = 0.92 ± 0.04, n = 30 parasites from three independent biological replicates), supporting localization of ANKER1 to the ER. As ANKER1 is a transmembrane protein and HDEL-GFP labels the ER lumen, this analysis should be interpreted as evidence of co-distribution within the same organelle rather than direct molecular colocalization.

In contrast, we do not consider Pearson's coefficient appropriate for evaluating the association of micronemes or rhoptries with F-actin. These organelles are predominantly concentrated at the apical pole and, when associated with F-actin, are typically positioned along rather than directly overlapping the filaments. Consequently, Pearson's coefficient would underestimate these biologically relevant associations. We therefore relied on morphological and spatial criteria, which we consider more appropriate for assessing organelle–cytoskeleton interactions.

In addition, from the methods and presented figure images, specifically in Figure 6C for RON2, how the cytosolic and residual body actin is separated is difficult to discern, as there is a clear residual body actin signal overlapping a parasite. The methods for how this was separated and analyzed should be clearer to remove doubts about how this area was measured, as the current description raises concerns about the counting of the residual body actin within the cytosol.

We agree with the reviewer that the distinction between cytosolic and residual body (RB)-associated F-actin required further clarification. All analyses were performed manually, as described in our response to Reviewer 2 (comment 1d). The RB was identified by the presence of thick, bundled F-actin filaments at the basal pole that formed a continuous structure connecting parasites within the vacuole, whereas cytosolic F-actin was defined as the thinner filamentous network within the parasite body.

No automated or threshold-based segmentation was used because the marked differences in filament morphology and fluorescence intensity make reliable thresholding difficult and prone to misclassification. Manual annotation based on spatial localization and filament morphology was therefore considered the most appropriate approach. We have clarified these criteria in the Materials and Methods section.

Line comments:

(1) 113: round to rounds - "did not obtain maternal organelles after successive rounds of replication...".

The text has been updated

(2) 146: grammatical, "As consequence a progressive" -> "As a consequence", or "Consequently".

The text has been updated

(3) 164-166: "Autonomous duplication" implies the separation and duplication of the Golgi occurs on its own, i.e., without any outside intervention, when we know from Carmeille et al. 2021 and Figure 7C here that the Golgi becomes fragmented over rounds of division in the absence of MyoF. I think this is primarily a word choice error with "autonomous".

We agree with the reviewer and the word autonomous has been removed

(4) 184: The wording suggests that DC's refers to daughter IMC's, when DC has already been given as an abbreviation for daughter cells previously.

The text has been updated to correct this error

(5) 189: de novo is not italicized.

The text has been updated

(6) 192: The data shown so far do not show that the RB plays a selective role in organelle recycling.

The text has been edited to fit better our results “Our data suggest that the organelles trafficking through the residual body (RB) have different fate”

(7) 208: State that it depends on F-actin, but never show that it is dependent on F-actin through actin disruption, such as cytochalasin D treatment or a specific conditional disruption of F-actin.

We agree with the reviewer that we did not repeat F-actin disruption experiments (e.g., cytochalasin D or jasplakinolide treatments) in this study. These experiments were performed in our previous work, where pharmacological disruption of F-actin was shown to impair microneme trafficking (Periz et al., 2019). We therefore chose not to repeat these assays.

Instead, the present study provides complementary evidence by demonstrating that depletion of Myosin F (MyoF), a motor that uses F-actin as a transport track (Kellermeier et al., 2024), disrupts the trafficking of both maternal micronemes and rhoptries. Together, our previous F-actin perturbation experiments and the MyoF depletion data presented here support the interpretation that these trafficking events depend on the actin–MyoF transport machinery. 

(8) 209: This suggests that the chromobody was transiently expressed in the RON2-Halo line, but the methods suggest MIC2-Halo and RON2-Halo were integrated into a parasite line stably expressing Cb-Emerald.

The text has been edited.

(9) 229: The section is confusing with the mention of (now maternal). If I understand correctly, the point being made is that the de novo synthesized MIC2 at stage 2 is now the maternal MIC2 for stage 4, but coloring-wise within the figure, the now maternal MIC2 at stage 4 from stage 2 would still be green. The methods suggest these images were all taken simultaneously, and not at specific timepoints of the same vacuole, so the now maternal line remains confusing.

The text has been revised for clarity and now reads: “Because Toxoplasma gondii replicates asynchronously, vacuoles at different replication stages coexist within the same culture after 24 h. The second labeling step marks all proteins synthesized since the beginning of the experiment, allowing discrimination between proteins present in the original mother parasite and those synthesized during subsequent replication cycles. As daughter parasites form, they inherit material from their mother, such that proteins synthesized during one replication cycle become maternal proteins in the next. Under MyoF depletion, these newly synthesized protein pools accumulate within the residual body instead of being redistributed to daughter parasites during subsequent rounds of replication (Figure 7, stage 4).”

(10) 238: The sentences here indicate that Golgi inheritance occurs without issue: "golgi inheritance remained unaffected by MyoF depletion". But, it is evident from the images shown that the Golgi is extremely fragmented, with many more Golgi fragments by stage 8 than there are parasites. This could be solved by rewording and including a line along the lines of "in accordance with the results found in Carmeille et al. 2021".

The reference to this article is already stated later in the text now line 299-303 “This active role is further supported by the dependence of RB-mediated recycling on F-actin and the class XXII myosin MyoF, which we show to be essential for retrieval of maternal MIC2 and RON2 but dispensable for Golgi inheritance although we noticed a fragmentation of the Golgi, which has been described to depend on MyoF (Carmeille et al., 2021).”

(11) 319: missing a comma, "Many of these, particularly...".

The text has been edited.

(12) 437: Images -> imaged.

The text has been edited.

(13) 439: a fresh media -> and fresh media.

The text has been edited.

(14) 439: Replication -> replicate.

The text has been edited.

(15) 440: images -> imaged.

The text has been edited.

Other grammatical issues within the methods:

(1) Figure 5C: Within the y-axis label of Figure 5C, there is an asterisk with no asterisk explanation within the legend.

We thank the reviewer for pointing out this oversight. The figure legend has been updated to include an explanation of the asterisk, providing a clearer understanding of the results.

(2) Figure 5E: The addition of an IMC1 label to match the magenta color would be helpful to readers.

The figure has been updated

(3) Figure 6D: No statistics showing significance of colocalization.

We thank the reviewer for this comment. In Figure 6D, we report the frequency of observed colocalization between organelles (MIC2 and RON2) and F-actin filaments, based on manual analysis across >300 vacuoles for MIC2 and >500 vacuoles for RON2. Specifically, we observed cytoplasmic F-actin colocalization in 94% of vacuoles for MIC2 and 79% for RON2, and residual body F-actin colocalization in 85% of vacuoles for MIC2 and 11% for RON2. This analysis is descriptive and is not intended to compare MIC2 versus RON2 quantitatively; rather, it illustrates the general association of these organelles with F-actin filaments. Standard statistical measures such as Pearson correlation are not meaningful in this context because the organelles are punctate and primarily localized along filaments rather than overlapping continuously. Importantly, the percentages reported represent the fraction of vacuoles in which colocalization can be observed, not the percentage of colocalization between Cb-Emerald and MIC2/RON2 within individual vacuoles.

(4) Figure 7: The legend title of Figure 7 is at the end of the legend of Figure 6.

The text has been edited.

eLife Assessment

This study presents fundamental and significant findings that bovine mammary epithelial cells can be infected with both avian and human influenza A viruses, providing a potential site for viral reassortment. The evidence to support these claims is compelling. The work will be of interest to virologists and evolutionary biologists working on cross-species transmission of viruses and pandemic preparedness.

Reviewer #2 (Public review):

The authors use a library of influenza A viruses from different strains, classified in lab-adapted, human, avian, and swine according to the animal from which they were isolated. They propose that the cow mammary gland serves as a mixing vessel for influenza A viruses. As a first approach, the authors assess susceptibility to infection across different cell types, including continuous and primary cell lines, bovine mammary cells, and mammary explants. All these cells support polymerase activity. Then, they analyzed changes in the bovine virus's viral fitness relative to an avian precursor. The authors use single-gene replacement to study whether and which RNP segments improve viral transcription. As part of this section, they also test IFN-specific antagonism by NS1 to assess the input of segment 8. Quantitative glycomic analysis was performed on the continuous bovine mammary cell line to demonstrate the presence of both a2,3 and a2,6, which is consistent with their observation that these cells can be co-infected with human and avian IAVs simultaneously. The main question, however, is: what is the glycome in the explants, or directly from tissues?

Overall, the manuscript is clearly written and provides new insights into the behaviour of the cattle isolate, now compared with a representative group of model or precursor HAs of different origins.

It would be great if a consistent nomenclature for the IAV strains could be used in the study. There is a mix of origin (Texas), animal from which the virus was isolated (mallard), or abbreviations that do not follow guidelines (IAV07). Are the USSR and Udorn not lab-adapted?

The experimental setup includes bovine mammary primary and continuous cells, as well as mammary explants. Some of the most significant differences, for example, in viral fitness studies and co-infection experiments, are observed in these explants. Perhaps there could be some additional focus on this observation. The implications in comparison to the results obtained in cultured cells could be described. How will the human and other HA subtype viruses fare in the explants?

Comments on revised version.

The authors have satisfactorily addressed the reviewers' comments.

Reviewer #3 (Public review):

Summary:

This excellent manuscript by Pinto, Sharp, and colleagues examines bovine tissue tropism for influenza viruses. They find that bovine flu, as well as other strains, have strong replication in mammary tissue. They also map the genetic changes to influenza that improve replication in bovine cells. Overall, the study is well designed and executed and the results are very timely.

Strengths:

(1) The experiments are well-controlled.

(2) The figures are well-constructed and easy to follow.

(3) The Methods and legends are detailed, with sufficient information.

Comment on revised version.

The authors have strengthened the manuscript by addressing comments from the three reviewers and I have no additional concerns/suggestions.

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

Summary:

Here, Pinto and colleagues set out to investigate whether the cow udder is a potential mixing site for the influenza virus. The authors have demonstrated that bovine mammary epithelial cells can be infected with both avian and human influenza A viruses, supporting the idea that the cow udder may be a potential site for reassortment. Furthermore, they demonstrate that the bovine-adapted IAV replicates to similar titers in avian epithelial cells when compared to an AIV precursor virus. Thus, suggesting there is no fitness trade-off, and confirms the potential for spill-back of the cattle B3.13 into poultry, which has already been observed. Overall, I believe the authors achieved their aims. However, there are instances in which the results do not entirely support the conclusions (noted in weaknesses). Given the ongoing questions surrounding highly pathogenic avian influenza A virus in dairy cows, this work provides valuable evidence for the potential of the cow udder as a site of reassortment. These findings highlight the need for surveillance of influenza A virus incursions into livestock species, particularly cows. Some specific strengths and questions regarding weaknesses have been outlined below.

Strengths:

(1) The authors use a diverse range of cell types and influenza A virus strains, as well as a wide range of techniques to address the questions at hand.

(2) The use of cells from multiple bovine breeds for the MAC-T, bMEC and explants suggests the phenomenon is not unique to a single breed.

(3) The results suggesting there is no fitness trade-off for Cattle Texas in an avian host are interesting, and confirm the potential for spill-back of the cattle B3.13 into poultry, which has been observed.

Weaknesses:

I have listed my complete questions/concerns below. However, there are two main weaknesses of the article in its current state. Firstly, there is no apples-to-apples comparison in terms of determining a preference for IAV to infect the cow udder over other organs (Q4). The mammary gland and respiratory tract are represented by epithelial cells, but for other organs, fibroblasts were chosen. I think the fairer comparison would be to compare epithelial cells from different organs to demonstrate a preference for the mammary gland. Secondly, the main premise of the article relies on bMEC and MAC-T (primary and immortalised mammary epithelial cells), facilitating higher viral growth than the cells from other organs. Yet throughout the article, a 10x higher dose of IAV is used in the bMEC cells compared to everything else (Q6). This raises the question of how much of the results are due to a preference for the mammary epithelial cells, and how much is simply due to the increased dose.

(Q4) When we set out to test if cow mammary gland cells were particularly susceptible to IAV infection compared to other bovine cell types, we used what was available in the Roslin Institute – a mix of primary and continuous cells from various anatomical sites: three epithelial cell types (two mammary, one respiratory tract) two immune cell types and four sets of fibroblasts from various organs. Given the representation of different anatomical sites, cell types and differentiation statuses, we considered this a suitably diverse panel with which to characterise infection dynamics of a broad range of IAVs, before more focussed investigations using the bMEC and explant tissues. Both mammary epithelial cell types grew our library of influenza challenge strains significantly better than the BAT-II respiratory epithelial cells, as well as the two immune cell types and all four fibroblast populations. Of the fibroblast cells, those derived from the brain grew IAV significantly better than the skin and turbinate fibroblasts, while blood-derived macrophages grew virus significantly better than the lymphocytes and non-brain fibroblasts. So there are “apple-to-apple” comparisons as well as apple-to-pear comparisons that give significant differences. We therefore think that our conclusions (in the abstract) that mammary cells are particularly replication competent for IAV, (at the end of the introduction) that “a wide range of cow-derived cells are susceptible” and that (in the results section) that “mammary cells showed the highest susceptibility” are justifiable. However, we agree that testing a wider variety of epithelial cells would be useful and have added text to the Discussion (lines 224-228) to acknowledge this.

(Q6) We used a higher MOI for bMECs because test experiments with WT PR8 and the Cattle Texas 6:2 reassortant virus showed that MOI 0.01 infections gave more variable results than those run at MOI 0.1, perhaps because of the intrinsic variability of mixed primary cell populations. However, the end-point titres between the two conditions were not significantly different, so we therefore chose to go with the higher MOI. Accordingly, we do not think this choice is a confounding issue. This explanation (line numbers 340-345) and a new Supplementary Figure 11 showing the results of the two MOI tests have been added to the manuscript.

Reviewer #2 (Public review):

The authors use a library of influenza A viruses from different strains, classified in lab-adapted, human, avian, and swine according to the animal from which they were isolated. They propose that the cow mammary gland serves as a mixing vessel for influenza A viruses. As a first approach, the authors assess susceptibility to infection across different cell types, including continuous and primary cell lines, bovine mammary cells, and mammary explants. All these cells support polymerase activity. Then, they analyzed changes in the bovine virus's viral fitness relative to an avian precursor. The authors use single-gene replacement to study whether and which RNP segments improve viral transcription. As part of this section, they also test IFN-specific antagonism by NS1 to assess the input of segment 8. Quantitative glycomic analysis was performed on the continuous bovine mammary cell line to demonstrate the presence of both a2,3 and a2,6, which is consistent with their observation that these cells can be co-infected with human and avian IAVs simultaneously. The main question, however, is: what is the glycome in the explants, or directly from tissues?

We report quantitative glycomics for the primary bovine mammary epithelial cells as well as the continuous line the referee highlights. However, we agree with R2 that a detailed glycomic analysis of primary bovine mammary tissue would allow a better understanding of the actual glycosylation status in vivo. This has been undertaken by the authors and is available as a bioRxiv preprint. This is now cited (ref 25) in the relevant part of the results (line 184-185)

Overall, the manuscript is clearly written and provides new insights into the behaviour of the cattle isolate, now compared with a representative group of model or precursor HAs of different origins.

It would be great if a consistent nomenclature for the IAV strains could be used in the study. There is a mix of origin (Texas), animal from which the virus was isolated (mallard), or abbreviations that do not follow guidelines (IAV07). Are the USSR and Udorn not lab-adapted?

We chose the abbreviated names for a variety of reasons. Partly from common usage (e.g. PR8, Udorn), partly for consistency with other already published papers from the FluTrailMap consortia (e.g. Cattle Texas; Dholakia et al 2026), partly to make diversity obvious in certain figures (e.g. H3N1, H5N2 etc) and partly to avoid confusion between viruses that originate from the same geographic area (e.g. AIV07, AIV09, H5N8-20 etc which are all A/Ck/England/isolate numbers). Overall, we found it more confusing to use the expanded nomenclature. Re AIV07 which the referee criticises for not following naming guidelines – if this is a reference to the EURL nomenclature, AIV07 is the abbreviation for the specific virus A/Chicken/England/053052/2021, our representative virus for EURL genotype EA-2020-C, as we say in the text. This nomenclature has now been added to Table 1, to provide a fuller cross-reference for all the names.

As to whether USSR and Udorn are lab-adapted – that depends on definitions. There is a continuum of adaptive changes and/or sequence drift starting from the very first growth cycle of an isolate in the laboratory. The viruses we define here as lab adapted are ones that have been deliberately adapted to other host species or which have very long passage histories in multiple laboratory systems resulting in known functionally significant changes; for example, one lineage of PR8 was passaged 77 times in mice, 717 times in cell culture, 30 times in chick embryos, 5 times in ferrets and a further 50 times in chick embryos (https://www.medscape.com/viewarticle/812621_3?form=fpf), rendering it unarguably lab-adapted. We admit that A/USSR/77 and A/Udorn/307/1972 are probably further along this adaptive pathway than more recent isolates such as A/Norway/3433/2018, but are unaware of any specific reason that would put them into our lab-adapted category.

The experimental setup includes bovine mammary primary and continuous cells, as well as mammary explants. Some of the most significant differences, for example, in viral fitness studies and co-infection experiments, are observed in these explants. Perhaps there could be some additional focus on this observation. The implications in comparison to the results obtained in cultured cells could be described. How will the human and other HA subtype viruses fare in the explants?

We agree that this is an important and interesting question, and had already tested the strains we used for co-infections: human seasonal pdm09 H1N1 “Norway” and low pathogenic avian influenza “H3N1”, in the mammary explants. Both replicate the avian virus to 20-fold higher titres. We have added this information to the revised manuscript as new Figure panels S6E-H, called out on line 200-201 of the results.

Reviewer #3 (Public review):

Summary:

This excellent manuscript by Pinto, Sharp, and colleagues examines bovine tissue tropism for influenza viruses. They find that bovine flu, as well as other strains, has strong replication in mammary tissue. They also map the genetic changes to influenza that improve replication in bovine cells. Overall, the study is well designed and executed, and the results are very timely.

Strengths:

(1) The experiments are well-controlled.

(2) The figures are well-constructed and easy to follow.

(3) The Methods and legends are detailed, with sufficient information.

Weaknesses:

(1) A comparison to human cells would strengthen the overall impact of the results. Are human mammary cells also uniquely susceptible to influenza? Are bovine mammary cells special in some way?

This is an interesting question, but we have not tested mammary gland cells from humans (or any other species of mammal). We have however reported elsewhere (Dholakia et al., Nat Commun. 2026 Jan 16;17(1):1603. doi: 10.1038/s41467-026-68306-6.) that Cattle Texas grows well in a variety of human respiratory cells. Here, we are considering the bovine mammary organ as a potential reassortment site for IAVs because of the ongoing viral mastitis epidemic in US dairy cattle; human mammary organs seem unlikely to create a similar opportunity.

(2) For the virus infection studies with segment 8 swaps, it should at least be noted that some of the phenotypes could be driven by NEP.

We agree; as Table S1 indicates, NEP has two changes (one shared with NS1) between AIV07 and our B3.13 isolate, so we should not have conflated segment and NS1. We have changed the text to acknowledge this throughout the results (lines 127, 137 and 149) and in the discussion (line 243-244).

(3) The data demonstrating that bMEC can support co-infection are compelling and important, but would be strengthened with a comparison from a different cell type or species. Do mammary cells uniquely support higher co-infection?

We have data showing that co-infection also occurs in the continuous MAC-T udder cell line and have now included these data in a revised Figure 4D (described/called out on lines 198-206). We have not tested bovine cells from other organs for co-infection potential as they do not seem to be significant sites of infection in vivo.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

(1) How nasal turbinate and cardiac fibroblasts are acquired/cultures is missing from the methods.

Apologies for the omissions and thank you, because rectifying this brought to light an error in cell naming. The cells originally called bovine cardiac fibroblasts were in fact a second independent preparation of skin fibroblasts. We have corrected the labelling in Figs 1, 2 and S2. The nasal turbinate cells were bought in from ATCC (code CRL-1390). This information has now been added to the methods (line 310).

(2) Please specify what cell types make up the 2D enteroids, the mammary explants and the nasal turbinates.

The composition of the 2D enteroids is described in detail in reference [60]. In precis, they are comprised predominantly of epithelial cells, including Paneth, goblet and enteroendocrine cells, as well as stem cells. This information has been added to the methods (lines 374-376) The mammary explants include duct epithelium, connective and muscle tissue, defined by H&E staining of cut sections (see new Figure S12 and text added to the methods on lines 384-386). We have also added the person who did the histology (Rebecca Ross) as an author and to the credit taxonomy (line 877). The nasal turbinate cells appear to be predominantly fibroblast morphology (Methods line 310-311).

(3) Epithelial cells are the main target for IAV, so why were fibroblasts chosen as a comparison to mammary epithelial cells? This needs justification in text. Without justification, how much can be attributed to the results being mammary-specific, rather than epithelial-specific? The brain (choroid plexus epithelium), heart (epicardium) and skin all contain epithelial cells.

We think this query is what the referee calls “Q4’ in the public part of their review. Please see our answer above.

(4) Figure 1, S1, S2 and S3 seem to suggest that mammary cells are more susceptible to IAV infection than cells from other organs. But Figure 2 demonstrates that when it comes to Cattle Texas and AIV07, most of the cell types show high viral titers. If the question is about whether the cow udder is the primary mixing site, would it not be more relevant to investigate which cell type facilitates the best growth of the potential precursor viruses, similar to Figure 4A?

Figs 1, S1, 2 and 3 all use “full” viruses whereas Fig 2 uses 6:2 reassortants between PR8 and the HPAIVs for biosafety reasons. WT PR8 replicates well in most of the bovine cells tested (Figs S1-3) so we do not see any contradiction. Figure 2 examines the contributions the internal genes make to replication in bovine cells. Re the question over the udder being a potential mixing site – this is where the virus is replicating in the real world; at least in part because of the transmission mechanism, but also because the mammary gland epithelium is highly susceptible to infection, as we show.

(5) It is misleading to compare the bMEC infection of MOI 0.1, to all other infections of MOI 0.01. This is a consistent problem throughout the article - Figures 1, S1, 2, 4. Why was the bMEC infection at a 10x greater dose? The main premise of the article relies on bMEC and MAC-T (primary and immortalised mammary epithelial cells), facilitating higher viral growth than the cells from other organs. If we compare the MOI 0.01 experiments alone, then the evidence relies on the immortalised MAC-T cells, compared to primary cell types. In this case, how much can be said about it being mammary specific, rather than immortalised vs primary? I do wonder, for example, how the epithelial nasal turbinates or type II pneumocytes would compare to the primary mammary epithelial cells if they were at the same MOI.

Please see our answer to this query earlier in the rebuttal.

(6) Why are the 2D enteroids excluded from Figure 1?

We had limited supplies of a difficult-to-grow cell model, so we only used them to test the 6:2 viruses (Fig 2).

(7) The colour scheme for Figure 3 is confusing. In Figure 3A, blue indicates European ancestry, and yellow represents North American ancestry. However, in Figure 3B, these colours now mean something different. To a reader, when there is a colour-coded schematic, it is instinctual to think that this then corresponds to the following panel(s). Since consistently throughout the article, yellow has been used for Cattle Texas, and blue has been used for AIV07, I would suggest choosing different colours to represent European and North American ancestry in Figure 3A

We’ve changed the figure as the referee suggests and modified the Fig 3 legend accordingly (line 895)

(8) I am unsure about the conclusions drawn from the results of Figure 3B. In the results, it is framed as trying to determine which segments contributed to the improved activity of Cattle Texas compared to AIV07. In lines 110-111, "... PB2 or PA from AIV07 significantly decreased Cattle Texas minireplicon activity". If PB2 is indeed significant, there is a missing yellow asterisk in Figure 3B.

Apologies, there was indeed a missing asterisk on the figure; now added.

Given the significance of PA, why was it not investigated in terms of growth kinetics similar to Figure 3C? Was it overlooked because it doesn't have a North American ancestry? The results of Figure 3B suggest that the 4 amino acid mutation in PA has significantly contributed to changes in polymerase activity.

The PA changes do indeed matter for minireplicon activity – the key change is K497R, as detailed in our related publication in Nat Comms (citation 17). However, it is less important than changes in PB2, and the PA segment swap by itself has little effect on overall virus replication.

(9) Similarly, in Figure 3C and lines 116-117, the error bars on the graph are overlapping at 48 hours, suggesting no difference in overall replication. The kinetics are slowed for AIV07 seg1-3, but not for AIV07 seg 1, indicating PB1 does not have an effect. This would then suggest that something in segment 2 or 3 contributes to the slowed kinetics in Figure 3C, which, from the Figure 3B results, is unlikely to be due to PB2. While it was not reassorted, the PA segment is potentially the driver, with its 4 amino acid mutations. I think it is worth performing growth kinetics with and without these 4 amino acid changes in PA.

We agree that visually on a log₁₀ scale, the titres of the “WT” 6:2 Cattle Texas and 5:2:1 segment 1 reassortment appear close, but the average titres are 5 and 7-fold different at 24 and 48h respectively, while a 2-way ANOVA with Dunnet’s multiple comparison post-test gives statistical significance at 48h. We have added this information to the figure and its legend (lines 905-906).

(10) In Figure 3C and E, why did you choose to perform the growth kinetics in the immortalised cell line, when you have access to primary cells? The primary cells would be a more accurate representation of what happens in situ.

The primary cells were difficult to work with and only available intermittently, so we used what was available at the time.

(11) In lines 145-147, "thus overall, the reassortment event that replaced segments 1, 2 and 8 alongside drift adaptations in segment 3 may have contributed to the ability of the B3.13 genotype virus to infect cattle". This is not clearly supported by the evidence presented. In terms of segment 1/PB2, the growth kinetics of Figure 3C have overlapping error bars at 48 hours. Where is any evidence presented for the role of segment 2/PB1? There is no change in Figure 3B.

The referee is correct, calling out seg2 here was an error; we have revised the text (line 144).

Segment 3 is overlooked in Figure 3 (as highlighted in Q9 and 10), and shows no difference in Figure S4.

Please see response to Q8; we think segment 3 contributes via PA adaptation, not via PA-X.

(12) In line 146 "... drift adaptation in segment 3". Make it clear here that you are talking about genetic drift. However, is this likely to be genetic drift? The 4 amino acid mutations are shown to have a significant impact on polymerase activity in Figure 3B, and in Figure 3C, PA potentially contributes to the reduced kinetics. When there are amino acid mutations that correspond to a beneficial phenotypic change, attributing this to drift alone rather than host adaptation is strange.

Yes, wording clarified (line 145). “Drift” was used to distinguish it from reassortment but we agree this was an incorrect term in the context.

(13) Figure 4A MAT-C cells: this is ostensibly the same experiment as Figure 3C in terms of the Cattle Texas and AIV07 viruses. If this is the case, how can you explain the difference in kinetics and overall titer? In Figure 3C, Cattle Texas reaches 10^6, and in Figure 4A it reaches almost 10^9. That's almost 3 log difference. Similarly, in Figure 3C Cattle Texas reaches 10^3, but in Figure 4A it reaches 10^6, a 3-log difference. At 24 hrs, they have roughly a 3-log difference between them in Figure 3C, but in Figure 4A this difference is much smaller. As far as I can tell, these are the same viruses, same dose and same cell model. The t0 titer is also vastly different between the two experiments.

The experiments were done at different times (several months apart, so different cell passage numbers and/or serum batches) and by different people. We have no explanation other than biological variability. However, both groups of experiments include genuine biological replicates done over the course of 2-3 weeks, so in our view represent coherent tests within themselves.

(14) In Figure 4, why weren't the a-2,6 a-2,3 proportions analysed for the explants and/or used for the co-infection experiment? It showed the greatest difference between the Cattle Texas and precursor viruses in Figure 4A.

Our data on the proportion of 2,6 and 2,3 SA in bovine udder tissue are now available in a separate preprint (now cited as [25] in our MS). We did not use the explants for co-infection experiments because it would have been technically difficult to read out the outcome by flow cytometry.

(15) Figure 4D requires a supplementary figure demonstrating the gating strategy, including one of the samples as an example.

We have compiled a figure of this and added it as new Figure S13 (called out line 556).

(16) In Figure 4D, why was an MOI of 5 chosen instead of the MOI of 0.01 used throughout the article for MAC-T cell infection? An MOI of 5 (so in a co-infection, a total of 10 virus particles per cell) is completely overloading the cells. At this dose, 10% of the cells were able to be co-infected, but how representative is this of a real co-infection scenario? While it demonstrates it is possible, it potentially remains highly unlikely, similar to the discussion around the swine respiratory tract in lines 235-240.

We had also performed the co-infections at lower MOI (1) with very similar results – this is now included in Figure 4D. Furthermore, we redid the experiments at a lower MOI of 0.05 and still see co-infection; this now replaces the MOI 5 data in Figure 4D. The text has been revised accordingly (lines 202-205)

(17) In Figure 4D, why were immortalised cells used when primary bMEC and mammary explants are available? Primary cells would provide more convincing evidence for the potential of the cow udder to be a mixing vessel. Considering that throughout the paper, a 10x higher viral dose is used in the bMEC culture, I wonder if you would need a significantly higher MOI than 5 to produce similar results in a co-infection experiment. The bMEC also has a more even a-2,3 to a-2,6 ratio compared to MAC-T in Figure 4C.

The bMECs in the original figure are primary cells. In response to other queries, we now include data from the immortalised MAC-T cell line as well.

Reviewer #2 (Recommendations for the authors):

Figure 1A, the coloured underline to discriminate continuous and primary cells is lost upon printing... perhaps another way is better?

We have changed the primary cells to italic text to make the distinction clearer.

We have made some other minor changes to wording to correct grammatical errors or improve clarity as we went through.

eLife Assessment

This article provides a computational study to understand how evolutionary selection could introduce structural priors into neural networks to improve learning. This useful investigation provides incomplete evidence that such joint learning is possible and has some counterintuitive features. However, the extent to which the models studied are distinct from those studied in the literature previously remains unclear.

Reviewer #1 (Public review):

In this article, the authors set out to understand how evolutionary selection could introduce structural priors into neural networks that act as an inductive bias to accelerate learning. To do so, the authors propose an evolutionary conditioning (ED) algorithm and analyse its properties. I find the conceptual framing of the paper very interesting, and it addresses an important question. Since the paper adopts a mostly theoretical approach with no comparison to empirical biological data, I do have a couple of concerns regarding the setup of the computational framework/method, which I think is incomplete and limits how much we can conclude from the current results.

Major Concerns:

(1) The evolutionary conditioning (EC) algorithm proposed works as fine-tuning training, plus propagation of the best parent network's weights to the next generation with added Gaussian noise. Conceptually, I find this to be a fairly implausible mechanism for evolution, since it requires carrying the entire set of network weights at some precision. The authors themselves point out that direct weight transfer could be problematic in the introduction.

(2) More importantly, I would like the authors to conduct a baseline / null model comparison, in which the "evolution" process consists simply of training a neural network for a small number of iterations, adding Gaussian noise, and repeating. The resulting network at each step serves as the "generations", which is then trained further. The same learning speed and dynamics analyses should be applied to this null model. What I am getting at is that I am not sure to what extent the EC algorithm can be thought of as "running a few iterations of SGD" and chaining them together; how much work is the selection process in the GA actually doing?

(3) I am also unclear on why the EC algorithm does not improve throughout learning. Is this behavior the result of applying only a small number of fine-tuning steps? Presumably, with longer fine-tuning, the individual networks in the middle generations would also improve in performance?

(4) The EC algorithm applied to a single problem seems somewhat artificial in its setup. I would conceptualize evolution as learning a prior that conditions the network for a range of survival-related tasks. A more realistic setup would apply EC to an ensemble of tasks and then examine its impact on learning a specific task afterwards.

Reviewer #2 (Public review):

Summary:

This paper studies the interplay of evolutionary and in-lifetime learning. The authors develop a neural network model in which initial weight configurations evolve under selective pressure, while fitness is determined by the network's performance after a learning period. They show that such a network displays very distinct learning dynamics from those trained by either gradient descent or genetic algorithms alone: in particular, they do not learn the task, but they show evidence of learning-to-learn and unusual representational structure.

Strengths:

(1) The writing, figures, and presentation of ideas were clear.

(2) The question of how evolution on initial weights combines with learning from within-lifetime experience to structure a learning trajectory seems interesting.

(3) The analysis of existing experiments was well-done, highlighting that though these networks did not really learn, they show latent learning structure that makes the network perform better from less data.

(4) The interplay between Baldwin & learning dynamics seemed novel and interesting, presenting many attractive puzzles.

Weaknesses:

First, the authors point to an important distinction between performing evolutionary selection on the weights pre- or post- lifetime training, the latter of which is Lamarckian. They argue, correctly, that their model is interesting because it selects on the weight initialisation, unlike, for example, Shuvaev et al. However, my understanding is that a long line of papers beginning perhaps with Hinton & Nowlan also do non-Lamarckian evolution: Hinton & Nowlan have unspecified weights (denoted '?' in the paper) that can be inherited and then learnt. Is this not exactly inheritance of initial conditions (in this case, whether learnable or not)? This novelty is a primary motivation of the paper, whereas to me it seems it was already apparent in Hinton & Nowlan, and developed further in what seems to be a long line of uncited literature (see next paragraph). As such, this paper's conclusions seem poorly positioned within the existing state of knowledge/literature.

Second, the algorithm is framed as novel, but I think it is a rediscovery. This framework is very close to MAML, in which an initial weight configuration is optimised by gradient descent to be good after a few steps of fine-tuning (Finn et al., 2017). The authors' approach differs in using a genetic algorithm to perform the training of the initial weights, avoiding some of the computational complexities of MAML, especially after long fine-tuning. In this, the authors have, I think, rediscovered ES-MAML, MAML where the inner optimisation loop is gradient descent, while the outer is genetic (Song et al., 2020). Other similar work is "Meta-Learning by the Baldwin Effect" (Fernando et al., 2018).

Further, within Fernando et al. there is a rich literature review, almost none of which are cited by the authors. I point especially to Keesing & Stork, 1990, which appears to show a strong dependence of Baldwin-like improvements on the amount of data, something this paper also shows but explores less thoroughly.

To summarise my critique thus far: I think the literature already answers the main concern of the motivation (i.e. non-Lamarckian neural network evolution and learning), I think it has already discovered this particular algorithm, and I think past work has more thoroughly analysed behaviours similar to those presented in this paper. Without positioning correctly within this literature, the more general contribution of the paper is hard to establish. The true novelty of the authors' analysis seems to be the emphasis on Saxe et al.-like learning dynamics and its interplay with the Baldwin effect, but I am not certain of this without knowing the literature better.

Regarding experiments, there was an interesting effect where the EC networks didn't learn but did show latent learning (Figure 2, Figure 3), which sped up later learning (Figure 4). There were a few details I was surprised by on which I would appreciate clarity:

(1) The main result has basically no headline learning under EC. This will clearly be very dependent on parameters (e.g. if you add or remove enough training steps, the algorithm becomes SGD/GA, which both show learning). It seems like a natural analysis would examine this (e.g. a plot of final performance of EC after 4000 generations with different per-generation learning budgets).

(2) It is then shown that after 4000 generations EC can learn very quickly to perform the semantic task perfectly, at least within 200 generations (Figure 4C, and perhaps much sooner, Figure 4D, Figure 4F last panel; it was hard to say. This and the previous point seem somewhat inconsistent; was it just that Figure 3 used only 100 fine-tuning steps while the perfect-task-performing networks in Figure 4 required somewhere between 100 and 200? This seems to point to extreme parameter dependence. More broadly, how should I square this inconsistency/near-inconsistency?

(3) Figure 3k, and especially Figure 4f bottom right panel, seem to show networks that correctly separate all stimuli but cannot classify them. Should I understand this as networks learning to just push apart all pairs of datapoints without structure?

(4) If I understood the genetic algorithm correctly, only three individuals from each population seeded the next generation. This seems another important parameter to tune, since I think it is far lower than standard evolutionary work, but I am not sure.

Finally, the paper most interested me as a neural network learning puzzle: how can the network perform so badly, yet lead to such different post-fine-tuning results? The paper pointed to these as 'distinct' learning phenomena without explaining what was causing those differences. The only way I could square these results in my head was as above: that the 4000 generations pushed the initial representation to represent all datapoints differntly, effectively changing the learning problem gradient descent faces from one with a lot of structure (the semantic task) that leads to stepwise learning, to one in which it was basically linear regression on a set of well separated stimuli without the structure necessary for stepwise learning. Since the paper focuses so much on learning dynamics, and studies a task where such things can be precisely probed, it would have been nice to pin down exactly what was happening slightly more.

eLife Assessment

This valuable study examines whether neurally-derived non-decision time estimates can guide decision models to more reliable parameter estimates. Convincing evidence for their value is obtained from state-of-the-art model-fitting methods applied to two empirical datasets. However, the premise relies on very strong assumptions regarding the degree to which the neural metrics directly index non-decision time and comparisons across models that appear to differ in more ways than just the non-decision time constraint.

Reviewer #1 (Public review):

Summary:

This paper proposes a non-decision time (NDT)-informed approach to estimating time-varying decision thresholds in diffusion models of decision making. The manuscript motivates the method well, outlines the identifiability issues it is intended to address, and evaluates it using simulations and two empirical datasets. The aim is clear, the scope is deliberately focused, and the manuscript is well written. The core idea is interesting, technically grounded, and a meaningful contribution to ongoing work on collapsing thresholds.

Strengths:

The manuscript is logically structured and easy to follow. The emphasis on parameter recovery is appropriate and appreciated. The finding that the exponential NDT-informed function produces substantially better recovery than the hyperbolic form is useful, given the importance placed on identifiability earlier in the paper. The threshold visualisations are also helpful for interpreting what the models are doing. Overall, the work offers a well-defined, methodologically oriented contribution that will interest researchers working on time-varying thresholds.

Weaknesses / Areas for Clarification:

A few points would benefit from additional clarification following the previous revision:

Returning to one point from my original review. The applications to empirical data describe 6 models, including the FT-DDM with across-trial variability described as a benchmark. Yet the modelling results for both studies (Tables 2 and 3) show only 5 models, omitting the benchmark model. I acknowledge the comment about this in the response letter (no other FT-DDM models in the main text have across-trial variability). Nevertheless, for this to serve as a benchmark, the modelling results really should be reported in Tables 2 and 3 with the other 5 models, and the model goodness of fit figures currently shown in Appendix 7 incorporated into Figures 10 and 13 of the main text. Unfortunately, as it currently reads, it looks as if something is being obscured, which I don't believe is the intention. This won't change the primary conclusions, but it will increase transparency and ease of interpretation.

Many thanks for introducing Appendix 1 to the revised manuscript. Additional motivation for the central issue is always helpful. However, I'm not (yet) convinced the simulation study in Appendix 1 achieves the intended aim. The simulation study shows that NDT misspecification leads to poorer recovery of other CT parameters (i.e., if a generating value is perturbed and then held fixed at that perturbed value during estimation, recovery of the remaining parameters deteriorates). This demonstrates the consequences of fixing NDT incorrectly, but it does not seem to address the central claim of the manuscript: that NDT and the CT threshold parameters trade off when they are estimated simultaneously. I had expected the simulation study to estimate NDT alongside the remaining model parameters, mirroring the estimation procedure used in the empirical analyses. Such a simulation would directly test whether the two sets of parameters compensate for one another during estimation, and whether this results in poor recovery of the NDT parameters.

Reviewer #2 (Public review):

Summary:

The authors use simulations and empirical data fitting in order to demonstrate that informing a decision model using noisy single-trial estimates of an underlying fixed non-decision time can guide the model to more reliable parameter estimates, especially when the model has collapsing bounds.

Strengths:

The paper is well written and motivated, with clear depth of knowledge in the areas of neurophysiology of decision-making, sequential sampling models, and in particular, the phenomenon of collapsing decision bounds.

Two large-scale simulations are run to test parameter recovery, and two empirical datasets are fit and assessed; the fitting procedures themselves are state-of-the-art, and the study makes use of a very new and well-designed ERP decomposition algorithm that provides single-trial estimates of the duration of diffusion; the results provide inferences about the operation of decision bound collapse - all of this is impressive.

Weaknesses:

This is an interesting and promising idea, but a very important issue is not clear: it is an intuitive principle that information from an external empirical source can enhance the reliability of parameter estimates for a given model, but how can the overall BIC improve, unless it is in fact a different model?

Comment on revised version.

Thanks to the authors for their responses and inclusion of additional analyses and simulations. Thanks, in particular for clarifying a crucial detail, that the ndt-informed model actually assumes, like the uninformed model, that there is no variability in the non-decision time, and the idea is that the variable single-trial measurements of non-decision time are noisy estimates of an underlying, constant ndt. The revised paper itself has not made this clear - for example, throughout the Intro, there is no statement that the behavioural model assumes a trial-invariant ndt, and line 231 still calls tau the 'mean' non decision time, implying there is a distribution rather than an invariant single value in the behavioural model.

One implication of the above is that if the lognormal sigma is purely measurement noise that does not relate to actual variation in the underlying decision process generating behaviour, then the HMP latencies should not relate to behaviour, e.g. shorter latencies predicting shorter RT. I assume that even if the authors did find such a relationship, the principle still stands that a model with fixed ndt is more accurately fit when there are single-trial ndt estimates whose mean provides a constraint on that ndt value, than without such measurements. Still, given ndt variability is a core feature of many decision models, the authors could comment on whether the strategy would work in theory for a model with ndt variability (in the behavioural part), where the single trial estimates would then presumably reflect a mix of measurement noise and genuine ndt variability.

Another more important implication is that since it is in fact the same model being compared with and without the HMP data guiding the fixed ndt estimate, the reason the fit quality improves with HMP-information is not because it is a better model per se (it is the same model) but because without the HMP guidance, the search algorithm somehow gets lost and fails to find the 'optimal' parameter vector. That is, the parameter vector (just the parameters that relate to the behavioural model itself, not the HMP lognormally-distributed noise associated with VEP measurements) identified as optimal in the HMP-informed version of the model exists in the parameter space of the model without HMP information, but it is just not found? I raised this implication before, and it is still not clear whether it applies. I'm sorry to press on it, but it is critical for readers to understand why it is that neural information can improve overall model fit. Again, the enhancement of parameter recovery (like in Nunez 2025) makes sense, but the enhancement of the "model's fit to behavioural data" does not, without pointing to a deficiency in the search algorithm / fitting procedure.

The authors state in their replies that the onset of bound collapse is set at accumulation onset and imply that setting it instead at stimulus onset could "mathematically resolve the issue" but they don't do it because it is implausible. It is in fact not only plausible but clearly evidenced in empirical data - collapsing bounds are implemented neurally through urgency signals, and these can begin to dynamically build toward threshold well before, let alone at, stimulus onset. There is nothing bizarre about this - we can prepare movements without sensory input, and indeed even if choosing actions based on a sensory discrimination, motor preparation can launch well before the sensory evidence (e.g. Stanford, Salinas et al 2010) and this in effect collapses the bound on cumulative evidence for triggering action before any evidence actually arrives. So, Urgency/bound-collapse does not need to be triggered by a stimulus; it can start in anticipation of the stimulus. It seems critical, therefore, for the authors to clarify this point - does re-defining the onset of the collapse at stimulus onset remove the trade-off and render unnecessary the neurally-informed ndt estimation?

Related to this, it is still not clear how bias in the estimation of nondecision time would not be a problem. What if, for example, it is the end of the N2 rather than the peak of the N2 that marks accumulation onset, and/or there is an additional fixed motor time that adds to the N2-based marker to make the full nondecision time that applies in the underlying decision process. By definition (and I think this is essentially what the authors' new simulations verify), because of the trade-offs, this bias would simply be absorbed in shifted estimates of theta and lambda describing the bound collapse function. But wasn't the whole point of the exercise to more accurately estimate those parameters? The obvious implication is that the parameter-estimation accuracy of the ERP-informed model is determined by the accuracy with which the proposed ERP marker directly pinpoints the full nondecision time without bias, but this is not at all obvious in the paper as written. Importantly, in the example scenario I describe above where accumulation onsets when N2 ends, there may still be a perfect correlation of N2 peak latency with underlying ndt across trials - they could still be very strongly "linked" statistically, but we can't know what size offset might be involved.

Reviewer #3 (Public review):

The current paper addresses an important issue in evidence accumulation models: many modelers implement flat decision boundaries because the collapsing alternatives are hard to reliably estimate. Here, using simulations the authors demonstrate that parameter recovery can be drastically improved by providing the model with additional data (specifically, an EEG-informed estimate of non-decision time). Moreover, in two empirical datasets it is shown that those EEG-informed models provide a better fit to the data. The method seems sound and promising and might inform future work on the debate regarding flat vs collapsing choice boundaries. As an evidence-accumulation enthusiast, I am quite excited about this work, although for the more broader audience the immediate applicability of this approach seems limited because it does require EEG data (i.e. limiting widespread use of the method or e.g. answering questions about individual differences that require very large N).

Comments on revised version.

I thought the authors carefully addressed my concerns.

Author response:

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

Reviewer #1 (Public review):

Summary:

This paper proposes a non-decision time (NDT)-informed approach to estimating timevarying decision thresholds in diffusion models of decision making. The manuscript motivates the method well, outlines the identifiability issues it is intended to address, and evaluates it using simulations and two empirical datasets. The aim is clear, the scope is deliberately focused, and the manuscript is well written. The core idea is interesting, technically grounded, and a meaningful contribution to ongoing work on collapsing thresholds.

Strengths:

The manuscript is logically structured and easy to follow. The emphasis on parameter recovery is appropriate and appreciated. The finding that the exponential NDT-informed function produces substantially better recovery than the hyperbolic form is useful, given the importance placed on identifiability earlier in the paper. The threshold visualisations are also helpful for interpreting what the models are doing. Overall, the work offers a well-defined, methodologically oriented contribution that will interest researchers working on time-varying thresholds.

We appreciate the positive and constructive feedback. We have addressed your comments in the revised manuscript, as detailed in our responses to the specific comments below.

Weaknesses / Areas for Clarification:

A few points would benefit from clarification, additional analysis, or revised presentation:

(1) It would help readers to see a concrete demonstration of the trade-off between NDT and collapsing thresholds, to give a sense of the scale of the identifiability problem motivating the work.

Thank you for this constructive suggestion. We conducted a new simulation study in which we considered the non-decision time as a fixed parameter and estimated the remaining parameters of the collapsing threshold model. In this simulation study, we contaminated the non-decision time with noise at six levels (i.e., 0%, 2%, 4%, 6%, 8%, and 10%). The simulation results showed that increasing the noise level in non-decision time worsens the estimation of the starting threshold and decay rate parameters, indicating a trade-off between non-decision time and collapsing threshold parameters. The results of this simulation are presented in Appendix 1 in the new version of the manuscript.

(2) Before moving to the empirical datasets, the manuscript really needs a simulation-based model recovery comparison, since all major conclusions of the empirical applications rely on model comparison. One approach might be to simulate from (a) an FT model with across-trial drift variability and (b) one of the CT models, then fit both models to each of the simulated data sets. This would address a longstanding issue: sometimes CT models are preferred even when the estimated collapse in the thresholds is close to zero. A recovery study would confirm that model selection behaves sensibly in the new framework.

We are grateful for this constructive comment. The revised manuscript includes a model recovery study (see Simulation study 2, in particular Table 1 and the accompanying text). In this simulation, as the reviewer suggested, we generated data from FT-DDM with across-trial variability in drift rate and CT-DDM with hyperbolic and exponential collapsing thresholds. Then, for each generated dataset, we fitted two models and classified the datasets based on goodness-of-fit and estimated decay rate. The results show that incorporating the decay rate, which can be reliably estimated by the NDT-informed modeling framework, significantly improves the precision of model recovery.

(3) An additional subtle point is that BIC is defined in terms of the maximised log-likelihood of the model for the data being modelled. In the joint model, the parameter estimates maximise the combined likelihood of behavioural and non-decision-time data. This means the behavioural log-likelihood evaluated at the joint MLEs is not the behavioural MLE. If BIC is being computed for the behavioural data only, this breaks the assumptions underlying BIC. The only valid BIC here would be one defined for the joint model using the joint likelihood.

We thank the reviewer for raising this important methodological point. We agree that, strictly speaking, the behavioral log-likelihood evaluated at the joint maximum likelihood estimates is not guaranteed to be equal to the behavioral maximum likelihood estimate. Therefore, if one were to interpret BIC_Behavior for joint (NDT-informed) models as a conventional BIC derived from a purely behavioral maximum likelihood fit, this would indeed violate the standard assumptions underlying BIC. Our intention, however, was not to claim that BIC_Behavior represents a formally valid BIC in the strict information-theoretic sense. Rather, we used it as a diagnostic measure to assess how well the jointly estimated parameters account for the behavioral data relative to the uninformed models. Importantly, in our datasets, the behavioral log-likelihood evaluated at the joint estimates is higher than that obtained from the uninformed models. This suggests that the additional constraint introduced by the non-decision time information helps guide the optimization procedure toward parameter regions that provide a better account of the behavioral data. In other words, uninformed models appear to converge to suboptimal parameter estimates, but the joint modeling framework helps regularize the estimation process and yields better estimates of the optimal parameters. We fully acknowledge that this use of BIC_Behavior for the joint models departs from standard practice in the cognitive modeling literature. For this reason, we rely primarily on the joint likelihood–based model comparison as the formally valid criterion. The behavioral BIC is reported only to provide additional intuition regarding the goodness of fit on behavioral data and is used exclusively to compare NDT-informed models with their uninformed counterparts under identical evaluation criteria.

Also, to warn readers about this limitation, we included the following statement in the section where we defined BIC measures:

“It is worth noting that comparing joint and behavioral models using BIC_Behavior is uncommon and constitutes a limitation of the model comparison study.”

(4) Table 1 sets up the Study 1 comparisons, but there’s no row for the FT model. Similarly, Figures 10 and 13 would be more informative if they included FT predictions. This matters because, in Study 1, the FT model appears to fit aggregate accuracy better than the BIC-preferred collapsing model, currently shown only in Appendix 5. Some discussion of why would strengthen the argument.

In the revised manuscript, we have included the results for NDT-informed FT-DDM in the main text. However, we kept the FT-DDM with drift variability in the appendix, since none of the models in the main text include drift rate variability.

(5) In Figure 7, the degree of decay underestimation is obscured by using a density plot rather than a scatterplot, consistent with the other panels of the same figure. Presenting it the same way would make the mis-recovery more transparent. The accompanying text may also need clarification: when data are generated from an FT model with across-trial drift variability, the NDT-informed model seems to infer FT boundaries essentially. If that’s correct, the model must be misfitting the simulated data. This is actually a useful result as it suggests across-trial drift variability in FT models is discriminable from collapsing-threshold models. It would be good to make this explicit.

Regarding the visualization of decay rate estimation in Figure 7, we would like to clarify that, in the data-generating process for the FT model with across-trial drift variability, the decay rate is fixed at zero. That is, unlike the other parameters, there is only a single true value on the x-axis. If we were to present the decay rate recovery using a standard scatter plot (true vs. estimated values), all points would lie vertically above the single true value (zero), resulting in a vertical strip of points. While such a plot would technically be consistent with the other panels, it would not clearly convey the distributional properties of the estimated decay rates, specifically, which values are more likely under model misidentification. For this reason, we chose a density plot to more transparently illustrate the distribution of inferred decay rates when the true generating process is an FT model. We believe this representation more effectively communicates the extent and structure of mis-recovery. To avoid confusion, we have added a clarifying footnote in the revised manuscript explaining why a density plot was used in this specific panel.

Regarding distinguishing fixed-threshold models from collapsing-threshold models, as suggested in your comment 2, we conducted an additional simulation, and the results showed that when using the NDT-informed diffusion model, we can distinguish between both models with high accuracy. Specifically, we showed that incorporating the estimated decay rate value provided by the NDT-informed modeling approach can significantly improve the model recovery accuracy.

(6) Given the large recovery advantage of the exponential NDT-informed function over the hyperbolic one, the authors may want to consider whether the results favour adopting the former more generally. Given these findings, I would consider recommending the exponential NDT-informed model for future use.

Consistent with the reviewer’s argument, we included the following text in the general discussion:

“Importantly, the exponential collapsing threshold exhibited substantially better parameter recovery and superior model recovery performance, suggesting that this specification may be preferable in future cognitive modeling applications.”

(7) In Study 2 (Figure 13), all models qualitatively miss an interesting empirical pattern: under speed emphasis, errors are faster than corrects, while under accuracy emphasis, errors become slower. The error RT distribution in the speed condition is especially poorly captured. It would be helpful for the authors to comment, as it suggests that something theoretically relevant is missing from all models tested.

Thank you for mentioning this point. Because the models considered in the main text do not include across-trial variability in the starting point, the model cannot predict the fast error pattern observed in the speed condition of the second study. In the revised manuscript, we included a note on this point:

“The NDT-informed models’ predictions are depicted in Figure 13. This figure shows that the FT-DDM overestimates the last RT quantiles for both correct and incorrect responses in both speed and accuracy conditions. However, the qualitative predictions of CT-DDMs align more closely with the empirical data. It is also worth noting that all the considered computational models misfit the incorrect responses in the speed condition. This misfit is to be linked to the presence of fast errors, specifically in the speed condition. Including the starting-point variability parameter in the model enables the model to predict fast errors. However, as the aim here was not merely to fit the data with the best possible model, but to test the NDT-informed modeling framework, we did not include starting-point variability in the model.”

(8) The threshold visualisations extend to 3 seconds, yet both datasets show decisions mostly finishing by 1.5 seconds. Shortening the x-axis would better reflect the empirical RT distributions and avoid unintentionally overstating the timescale of the empirical decision processes.

In the new version of the manuscript, we shortened the x-axis in these plots. See Figures 9 and 12 in the new version of the manuscript.

Reviewer #1 (Recommendations for the authors):

(1) The manuscript should explicitly state how critical the log-normal assumption for NDT is, and whether there are caveats if it doesn’t hold.

We agree with this suggestion. Therefore, we conducted an additional simulation study in which we assumed a normal distribution for non-decision time measurements and replicated the main results reported in simulation study 1 (see Appendix 8 in the new version of the manuscript). These simulation results reveal that independent of distributional assumptions on non-decision time measurements, constraining non-decision time can improve the estimation of the collapsing threshold.

(2) On page 14, one paragraph refers to five models and another to six. It is unclear which is correct.

We are sorry for the confusion. In the new version of the manuscript, we addressed this issue.

(3) In the Discussion: Hawkins & Heathcote (2021) found that NDT estimates in the TRDM recover well, but the timer-offset parameter does not (and hence is set to a fixed value). It would be interesting to test whether the NDT-informed approach could extend to that parameter. The TRDM can also generate error RT distributions that are faster than correct RT distributions, which is the qualitative pattern that none of the present models capture in the speed condition of Study 2.

We appreciate the reviewer’s suggestion, as it would be another important demonstration of the framework we built in the paper. Nevertheless, given the number of models already considered in the manuscript and the focus on collapsing threshold diffusion models, we believe that this addition would blur the focus of the present manuscript. Therefore, we have decided not to include TRDM in the manuscript. Nevertheless, we have addressed the reviewer’s concern regarding the non-decision time estimation in the TRDM in the revised manuscript:

“Notably, this model can also predict faster error responses than the correct response, the pattern that is observed in the speed condition of Study 2. However, the authors reported poor parameter recovery for the onset of the timing process (Hawkins and Heathcote, 2021). Thus, the reliability issue here specifically concerns the estimation of the shift parameter of the timing accumulator. Informing the model with external estimates of non-decision time might, therefore, improve parameter recovery in this model as well.”

Reviewer #2 (Public review):

Summary:

The authors use simulations and empirical data fitting in order to demonstrate that informing a decision model on estimates of single-trial non-decision time can guide the model to more reliable parameter estimates, especially when the model has collapsing bounds.

Strengths:

The paper is well written and motivated, with clear depth of knowledge in the areas of neurophysiology of decision-making, sequential sampling models, and, in particular, the phenomenon of collapsing decision bounds.

Two large-scale simulations are run to test parameter recovery, and two empirical datasets are fit and assessed; the fitting procedures themselves are state-of-the-art, and the study makes use of a very new and well-designed ERP decomposition algorithm that provides single-trial estimates of the duration of diffusion; the results provide inferences about the operation of decision bound collapse - all of this is impressive.

We appreciate your feedback and comments. Below, we provided a response for each comment.

Weaknesses:

(1) This is an interesting and promising idea, but a very important issue is not clear: it is an intuitive principle that information from an external empirical source can enhance the reliability of parameter estimates for a given model, but how can the overall BIC improve, unless it is in fact a different model? Unfortunately, it is not clear whether and how the model structure itself differs between the NDTinformed and non-NDT-informed cases. Ideally, they are the same actual model, but with one getting extra guidance on where to place the tau and/or sigma parameters from external measurements. The absence of sigma (non-decision time variance) estimates for the non-NDT-informed model, however, suggests it is different in structure, not just in its lack of constraints. If they were the same model, whether they do or do not possess non-decision time variability (which is not currently clear), the only possible reason that the NDT-informed model could achieve better BIC is because the non-NDT-informed model gets lost in the fitting procedure and fails to find the global optimum. If they are in fact different models - for example, if the NDT-informed model is endowed with NDT variability, while the non-NDT-informed model is not - then the fit superiority doesn’t necessarily say anything about an NDT-informed reliability boost, but rather just that a model with NDT variability fits better than one without.

To respond to this comment, we would like to note that the structural difference between NDT-informed and uninformed models is the assumption about non-decision time. In principle, the behavioral parts (i.e., parameters related to the diffusion part) of both NDT-informed and uninformed models are identical. However, during the estimation, the non-decision time in the NDT-informed model is subject to an additional constraint imposed by the neural data. In other words, in the uninformed model, non-decision time is estimated using the behavioral data by maximizing the likelihood of a CT-DDM. However, the NDT-informed model incorporates an additional data type, resulting in a different likelihood function (see Equation (3)). Specifically, in the NDT-informed model, we make an additional assumption regarding the non-decision time: it is set to the mean of the trial-level non-decision time measurements distribution derived from neural data (Equation (3) specifies the joint model structure). This additional assumption constrains the search space of the non-decision time parameter in the NDT-informed model. As in the main text, we assumed that non-decision time measurements obtained from neural data are log-normally distributed, where sigma is the shape parameter of the log-normal distribution. Therefore, sigma can represent the variability of the non-decision time measurements, and it is not the trial-to-trial non-decision time variability parameter. Indeed, the non-decision time parameter is fixed across trials in both models. Also, it should be clear from Equation (3) that sigma only appears in the second term of the joint likelihood, which corresponds to non-decision time measurements and not the CT-DDM term. Therefore, the behavioral parts of the NDT-informed models and Uninformed models are identical, and the NDT-informed models include one additional parameter (i.e., sigma) corresponding to the variability in non-decision time measurements.

Also, to explain how constraining non-decision time improves the BIC, we would like to clarify that constraining non-decision time using an additional data source constrains the search space for non-decision time, thereby leading to better parameter identification and, consequently, a better fit to behavioral data. We included the following text in the discussion section to make this explicit.

“This improvement likely reflects more accurate parameter estimation enabled by the additional information. In other words, constraining the non-decision time using neural measurements led the optimizer to estimate the CT-DDM parameter more accurately and, as a result, improve the fit to empirical data.”

(2) One reason this is unclear is that Footnote 4 says that this study did not allow trial-to-trial variability in nondecision time, but the entire premise of using variable external single-trial estimates of nondecision times (illustrated in Figure 2) assumes there is nondecision time variability and that we have access to its distribution.

We are sorry for the confusion. To respond to this comment, we would like to highlight that this modelling approach does not include any mechanism for across-trial variability in the behavioral part, as the likelihood of the choice behavior (see Equation (3)) does not include any variability parameter. As mentioned before, sigma represents the variability in non-decision measurements. However, the model does not propagate the across-trial variability of the neural data on the behavior side (unlike the models in Ghaderi-Kangavari et al. 2023). In other words, in this approach, none of the diffusion model’s parameters include across-trial variability, and we have only considered and estimated neural data variability in the NDT-informed model. Also, to improve the manuscript’s coherence, we removed this footnote.

(3) It is good that there is an Intro section to explain how the tradeoff between NDT and collapsing bound parameters renders them difficult to simultaneously identify, but I think it needs more work to make it clear. First of all, it is not impossible to identify both, in the same way as, say, pre- and postdecisional nondecision time components cannot be resolved from behaviour alone - the intro had already talked about how collapsing bounds impact RT distribution shapes in specific ways, and obviously mean (or invariant) NDT can’t do that - it can only translate the whole distribution earlier/later on the time axis. This is at odds with the phrasing “one CANNOT estimate these three parameters simultaneously.” So it should be first clarified that this tradeoff is not absolute. Second, many readers will wonder if it is simply a matter of characterising the bound collapse time course as beginning at accumulation onset, instead of stimulus offset - does that not sidestep the issue? Third, assuming the above can be explained, and there is a reason to keep the collapse function aligned to stimulus onset, could the tradeoff be illustrated by picking two distinct sets of parameter values for non-decision time, starting threshold, and decay rate, which produce almost identical bound dynamics as a function of RT? It is not going to work for most readers to simply give the formula on line 211 and say ”There is a tradeoff.” Most readers will need more hand-holding.

We are grateful for this comment. In response to this comment, which was also partly mentioned by the first reviewer, we first highlight that, in the presence of a nonlinear collapsing threshold, the effect of non-decision time is no longer linear, as it forces the threshold to take a specific value at the final stopping point. To illustrate the tradeoff between imprecise non-decision time estimation and collapsing threshold estimation, we conducted a simulation study. The results for the simulation study are presented in Appendix 1. In this study, we contaminated the non-decision time with noise at six levels (i.e., 0%,2%,4%,6%,8%, and 10%). The simulation results showed that increasing the noise level in non-decision time worsens the estimation of the starting threshold and decay rate parameters, indicating a trade-off between non-decision time and collapsing threshold parameters.

Regarding the second point, we would like to clarify that in this paper, consistent with other works on collapsing threshold, we assumed that the collapsing starts with evidence accumulation and not with stimulus onset, as the decision makers need a short amount of time for perceiving and encoding the stimulus (i.e., perceptual encoding time). Fixing the collapsing onset to the stimulus onset introduces an ad hoc assumption into the model (that the encoding time is zero), which is cognitively implausible. Therefore, although this assumption can mathematically resolve the issue, it is not cognitively plausible. However, one important point we did not consider in the previous version is the two-stage accumulation process models in which the collapsing onset occurs later than the evidence-accumulation onset. We have discussed the estimation of such models as a limitation in the general discussion:

“The collapsing threshold dynamics considered in this work (e.g., exponential and hyperbolic) impose a monotonically decreasing threshold over time. Although these dynamics are theoretically well motivated (Fudenberg et al., 2018; Frazier and Yu, 2007) and have been employed in several previous studies (e.g., Olschewski et al., 2025; Milosavljevic et al., 2010; Voskuilen et al., 2016), some research has proposed delayed-collapsing threshold models (e.g., Diederich and Oswald, 2016), in which the onset of threshold collapse does not coincide with the onset of evidence accumulation. Estimating such delayed-collapsing models may require more than simply constraining non-decision time, as the onset of threshold collapse must also be identified. A promising approach for addressing this challenge is the HMP method, which may provide additional temporal information about distinct cognitive processing stages. In particular, HMP may allow the onset of threshold collapse to be estimated as a separate cognitive stage. Future research should therefore investigate the estimation of multi-stage evidence accumulation models (e.g., Diederich and Oswald, 2016; Diederich and Colonius, 2021) within the HMP framework.”

(4) A lognormal distribution is used as line 231 says it “must” produce a right-skew. Why? It is unusual for non-decision time distribution to be asymmetric in diffusion modeling, so this “must” statement must be fully explained and justified. Would I be right in saying that if either fixed or symmetrically distributed nondecision times were assumed, as in the majority of diffusion models, then the non-identifiability problem goes away? If the issue is one faced only by a special class of DDMs with lognormal NDT, this should be stated upfront.

We would like to clarify that this assumption is about the non-decision time measurements and not about the across-trial variability parameter of non-decision time. Although for computational convenience, non-decision time is often assumed to follow a normal distribution in diffusion modelling literature, some empirical studies have shown that the perceptual encoding time and total approximated non-decision time follow a right-skewed distribution. Moreover, as HMP estimations of non-decision time are usually right-skewed, we formalized the model with a log-normal distribution. However, we would like to note that the identifiability issue in collapsing-threshold diffusion models is not related to the distributional assumption over the non-decision time measurements, as poor parameter recovery of the collapsing threshold was also reported in Evans et al. (2020). To show that the distributional assumption about the non-decision time measurements does not affect the results, we conducted an additional simulation study in which we assumed a normal distribution for non-decision time measurements and showed that constraining non-decision time improves parameter recovery for collapsing threshold parameters. The results for this simulation are presented in Appendix 8 in the new version of the manuscript. We also revised line 231 as follows (see line 236 in the new version):

“Empirical studies on non-decision time measurement usually have reported a right-skewed distribution for their measurements (e.g., Weindel, 2021; Weindel et al., 2025). For instance, the measured perceptual encoding time and motor execution time reported by Weindel et al. (2025) are right-skewed. Therefore, we assume that the observed non-decision time measurements Z_n follow an approximate log-normal distribution, which is right-skewed (we will discuss how this distributional assumption can affect the results later). Thus, we model the non-decision time measurements Z_n, using a log-normal distribution with parameters µ and σ_z. The available measurements (i.e., observed data) at trial n can be represented as

follows:”

(5) In the simulation study methods, is the only difference between NDT-informed and non-informed models that the non-NDT-informed must also estimate tau and sigma, whereas the NDT-informed model “knows” these two parameters and so only has the other three to estimate? And is it the exact same data that the two models are fit to, in each of the simulation runs? Why is sigma missing from the uninformed part of Figure 4? If it is nondecision time variability, shouldn’t the model at least be aware of the existence of sigma and try to estimate it, in order for this to be a meaningful comparison?

As mentioned in the response to your first comment, the difference between NDT-informed and uninformed models is the access to an additional source of data related to non-decision time, and sigma represents the shape parameter of the non-decision time measurements distribution. Therefore, sigma belongs only to the NDT-informed model, and, as in the uninformed model, there is no additional data source, so the model does not include sigma.

(6) I am curious to know whether a linear bound collapse suffers from the same identifiability issues with NDT, or was it not considered here because it is so suboptimal next to the hyperbolic/exponential?

Thank you very much for this comment. The main reason we did not include linear collapsing threshold models was the assessment by Evans et al. (2020), which indicated that, with sufficient trials, these models can be estimated reasonably well. However, to investigate whether constraining non-decision time can also improve the estimation of linear collapsing threshold models, we conducted an additional simulation study, which is reported in Appendix 3 in the new version of the manuscript. The simulation results confirm those reported by Evans et al. (2020) and show that the parameters of the uninformed linear models can be identified using more than 500 trials. Constraining the non-decision time using the NDT-informed diffusion modeling framework still improves parameter estimation in the linear collapsing threshold model and reduces the required number of trials for reliable estimation to 250. Especially, the estimation of the starting threshold improves significantly.

(7) The approach using HMP rests on the assumption that accumulation onset is marked by the peak of a certain neural event, but even if it is highly predictive of accumulation onset, depending on what it reflects, it could come systematically earlier or later than the actual accumulation onset. Could the authors comment on what implications this might have for the approach?

Thank you for mentioning this point. We first would like to point out that Weindel et al. (2024) showed that HMP can predict the underlying generative distribution of cognitive states with high precision and without systematic bias. Second, it is worth highlighting the results reported in Appendix 5 (i.e., “Bias in non-decision time measurements”). In this appendix, we discussed how bias in non-decision time measurements (i.e., systematic underestimation or overestimation) affects the estimation of the collapsing threshold. Particularly, see Figure 3 in Appendix 5. To make these results clearer in the main text, we included the following paragraph at the end of the results section in simulation study 1:

“Additionally, we examined the effect of systematic bias in non-decision time measurement on parameter recovery. Appendix 5 presents the parameter recovery simulation results in the presence of biased non-decision time measurement (i.e., systematically underestimated or overestimated). The results revealed that, even in the presence of biased non-decision time measurement, the actual generating parameters show a high correlation with the estimated parameters. Underestimation in non-decision time leads to overestimation in the starting threshold and decay rate. Conversely, overestimation in non-decision time leads to underestimation in both the starting threshold and non-decision time.”

(8) Figure 7: for this simulation, it would be helpful to know the degree to which you can get away with not equipping the model to capture drift rate variability, when the degree of that d.r. variability actually produces appreciable slow error rates. The approach here is to sample uniformly from ranges of the parameters, but how many of these produce data that can be reasonably recognised as similar to human behaviour on typical perceptual decision tasks? The authors point out that only 5% of fits estimate an appreciable bound collapse but if there are only 10% of the parameter vectors that produce data in a typical RT range with typical error rates etc, and half of these produce an appreciable downturn in accuracy for slower RT, and all of the latter represent that 5%, then that’s quite a different story. An easy fix would be to plot estimated decay as a scatter plot against the rate of decline of accuracy from the median RT to the slowest RT, to visualise the degree to which slow errors can be absorbed by the no-dr-var model without falsely estimating steep bound collapse. In general, I’m not so sure of the value of this section, since, in principle, there is no getting around the fact that if what is in truth a drift-variability source of slow errors is fit with a model that can only capture it with a collapsing bound, it will estimate a collapsing bound, or just fail to capture those slow errors.

Thank you for this comment. We would like to first note that the aim of this section is to illustrate that NDT-informed modeling enables us to distinguish between CT-DDM and FT-DDM with drift variability (i.e., the two competing models that are relatively hard to distinguish). Therefore, we changed the name of this section to “Simulation study 2: Model recovery”. In the new version of the manuscript, we included a cross-model fitting simulation and a model recovery simulation in this section. The estimated decay rates in the cross-model fitting study indicate that, when the underlying generating model is FT-DDM with drift variability, parameter estimation using the NDT-informed approach yields precise inference. This is due to the estimated decay rate, which is very close to zero. The model recovery results also confirmed that incorporating the decay rate value into the model inference improves precision.

Moreover, to address the concern regarding the slow-error pattern in the simulated data, we examined the relationship between the slow–fast accuracy difference and the estimated decay rate. Specifically, we computed the difference between the accuracy of responses with response times below the median (ACC1) and those above the median (ACC2), and plotted this difference against the estimated decay rate. Author response image 1 presents the resulting scatter plot, with color indicating the drift rate. As shown in Author response image 1, incorrect inferences about the decay rate primarily occur at high drift rates. This pattern emerges because high drift rates produce very fast responses, leaving little time for the threshold to meaningfully collapse. Consequently, the behavioral signatures of collapsing-threshold and fixed-threshold models become increasingly similar under high drift conditions.

Author response image 1.

Illustration of the difference between the accuracies of responses with response time below the median and those with response time above the median against the estimated decay rate. Colour shows the drift rate value.

Reviewer #2 (Recommendations for the authors):

(1) Abstract: improves fit to behaviour in what way? Reliability or absolute quant fit to behavior? I.e., is it just helping constrain it so it finds the global opt?

(2) Line 72 - Are these “neuroimaging” studies? Perhaps use the broader term “neuroscience.”

(3) Line 95 - It’s important because it may confuse readers how something dynamic like a collapsing bound could be resolved with, say, fMRI.

(4) Line 85 – “greater variability”... in what?

(5) Line 87 -Revise to avoid misconstruing a time on task effect - e.g., “higher error rate for trials with longer RT” is more explicit.

(6) Line 89-94 - It’s not clear what findings are being referenced here.

(7) Line 103 - External biases such as priors and relative value?

(8) Line 175 - Clarify this applies to any ddm, not just ctddm.

(9) Line 268 - Explain what portion N200 accounts for.

(10) Line 327 - Is this equation supposed to be for Delta-X, as opposed to X(t+delta-t)? If you want X(t+delta-t) on the LHS, then X(t) must be added to the RHS.

We are very thankful for such a precise evaluation and constructive comments. We addressed all the comments raised by the reviewer in the revised manuscript.

Reviewer #3 (Public review):

Summary:

The current paper addresses an important issue in evidence accumulation models: many modelers implement flat decision boundaries because the collapsing alternatives are hard to reliably estimate. Here, using simulations, the authors demonstrate that parameter recovery can be drastically improved by providing the model with additional data (specifically, an EEG-informed estimate of nondecision time). Moreover, in two empirical datasets, it is shown that those EEG-informed models provide a better fit to the data. The method seems sound and promising and might inform future work on the debate regarding flat vs collapsing choice boundaries. As an evidence-accumulation enthusiast, I am quite excited about this work, although for a broader audience, the immediate applicability of this approach seems limited because it does require EEG data (i.e., limiting widespread use of the method or e.g., answering questions about individual differences that require a very large N).

We are very grateful for your positive evaluation and your comments.

Reviewer #3 (Recommendations for the authors):

This is a very decent study, very well written and properly executed. Most of my comments below are suggestions for the authors as to how to make the story more compelling.

(1) I think the authors can do more to explain why the NDT-informed models fit better to empirical data. If the NDT-estimates are equal to the ground truth, then isn’t it possible that such models win simply because they have one parameter less? However, this is not what the authors are claiming, though, on l.567 it says that the fit is better because parameters are better estimated. I think it should be possible to dissociate these two accounts.

Thank you for this important comment. For clarification, both NDT-informed and uninformed models estimate the non-decision time parameter, and neither treats it as equal to the ground truth. However, the difference between these two models is that, in the NDT-informed model, the non-decision time is subject to an additional constraint; therefore, the number of behavioral parameters (i.e., parameters of the diffusion part) is identical for both NDT-informed and uninformed collapsing threshold models. Although the number of behavioral parameters is identical in the NDT-informed model, one parameter is constrained, thereby limiting the model’s complexity and flexibility compared to uninformed models. Consequently, the improvement in the fit can only be attributed to better parameter estimation in the NDT-informed models, resulting from constraining the non-decision time using neural measurements. To clarify that in the manuscript, we included the following text in the revised version:

“This improvement likely reflects more accurate parameter estimation enabled by the additional information. In other words, constraining the non-decision time using neural measurements led the optimizer to estimate the CT-DDM parameter more accurately and, as a result, improve the fit to empirical data.”

(2) Given that so much of the writing focuses on the conclusion that collapsing boundary models ¿ flat models, it is very odd that there is no comparison to a flat boundary model reported in the text. Why is the model in Supplement 5 not just included in the main text (and in Table 1)? This would make it so much easier for the reader.

To address this comment and also the similar point raised by Reviewer 1, we included the NDT-informed FT-DDM in the results section and compared the FT-DDM with CT-DDMs with respect to BIC_Joint. However, we retained the other FT-DDM in the appendix because this model includes across-trial variability in drift rate, whereas the models in the main text do not.

(3) I would have appreciated a bit more background about the importance of the number of trials per participant. Given that the proposed method requires collecting EEG data (which is time and labour-intensive) I wonder to what extent you get similar improvements in parameter recovery by collecting more data per participant (which is usually cheap). Put differently, I would appreciate an additional matrix in Figure 5 for vanilla models.

The new version of Figure 5 in the revised manuscript now contains the goodness of parameter recovery for the uninformed CT-DDMs for different numbers of trials. As the simulation results suggest, even with 1000 trials, the parameters of the uninformed CT-DDMs are still not reliably identifiable. Therefore, these results suggest that although increasing the number of trials can slightly improve the parameter recovery of the CT-DDMs, it cannot fully resolve the reliability issue in their parameter recovery.

In addition, it is worth clarifying that, although the paper focuses on extracting non-decision time from the EEG signal using the HMP method, as discussed in the general discussion, the NDT-informed approach can also be employed with purely behavioral methods.

(4) L116-117: minor detail: I don’t think it’s fair to write that it’s an open question whether or not thresholds collapse. I think it’s fair to say that this is hard to show, and that the conditions under which it appears are unclear; but saying that it’s still unclear whether this occurs at all seems unfair with regard to previous work.

Thank you for mentioning this point. We revised the mentioned sentence as follows in the new version of the manuscript:

“This issue is particularly critical given that the conditions under which individuals adjust their decision thresholds during a single trial remain an open question.”

(5) Figure 5: Minor detail: It would be useful to mention in the figure or caption how many trials underlie these simulations, which is now somewhat buried in the text.

As Figure 5 shows sensitivity to the number of trials, and the number of trials is explicitly mentioned in the figure, we assume that the reviewer intended Figure 6. In the revised manuscript, we explicitly specify the number of trials for the results reported in Figure 6. We simulated 500 trials for each noise level in this graph.

(6) Figure 10 and similar figures: I tried to figure out how corrects and incorrects differ, but couldn’t see the difference. Can the authors use something more colorblind friendly (hope I didn’t give up on being anonymous here)?

We are so sorry for the inconvenience. In the revised manuscript, we used different symbols to distinguish correct and incorrect data.

eLife Assessment

This important study investigates the peptide-binding principles by promiscuous chicken MHC molecules. The data from crystallography, mass-spec, and modeling are compelling. This paper will be of broad interest to immunologists and those interested in vaccine development.

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:

Combining in vitro refolding, SEC-based assembly assays, peptide-library screening, MALDI-TOF, LC-MS/MS, structural analysis and immunopeptidomics, this manuscript investigates the peptide-binding principles of the promiscuous chicken MHC-I molecule BF2*21:01.

Strengths:

Although the peptide motif of BF2*21:01 is highly complex, this manuscript identified several principles, including a preference for 10-mer peptides, co-variation between P2 and Pc-2, effects of P3 and Pc-3, and a strong cellular preference for Leu at Pc. The results are important for avian MHC biology and poultry vaccine epitope prediction.

Reviewer #2 (Public review):

Summary:

The study presents an in-depth analysis of the peptide repertoire bound by a promiscuous chicken MHC molecule using mass spectrometry, x-ray crystallography and modelling. While the MHC can bind a very diverse set of peptides, the authors have found some new rules that govern peptide binding to this MHC that could help to build a predictive model to study the repertoire of pathogen-derived peptides.

Strengths:

The study uses a range of well performed experiment across multiple techniques and provides an in-depth analysis of the peptide repertoire, including peptide sequences, length, preferred residues, stability and MHC presentation.

Author response:

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

eLife Assessment

This important study investigates the peptide-binding principles of promiscuous chicken MHC molecules. The data from crystallography, mass spectrometry, and modeling are convincing. However, the presentation would benefit from streamlining and clear links between data and conclusions. This paper will be of broad interest to immunologists and those interested in vaccine development.

Overall, we are delighted and grateful to the eLIFE editors and the two reviewers for the careful and thoughtful assessments and reviews of our paper. We are glad that the strengths of the paper were apparent and appreciated. And of course, every paper has weaknesses, especially for a story as complex as this one.

We made only minor changes to accommodate the reviewer comments, along with additions for which we only became aware upon this submission of a revised manuscript. In particular, we shortened the title and abstract to fit what is usual for an eLIFE paper, added Key Resources table with accompanying references, changed the numbering of the figures throughout the manuscript to ensure that each page represented a figure (rather than panels of a figure), moved the figure legends from the embedded figures to a list near the end of the manuscript, and split the supplemental spreadsheet into two renamed Data Source files.

Before answering the comments and questions directly, perhaps a few points would help clarify why the paper is as it is.

First, the experiments cover over three decades of work, with the first gas phase sequencing results done in 1992. Unlike some of the chicken class I alleles which immediately gave completely clear stringent motifs (B4, B12 and B15 in Wallny et al 2006 PNAS, B19 in Han et al 2023 J Immunol), we harvested nothing but confusion from the B21 class I results (Fig. 1). Initially, we thought that the lack of a clear motif for B21 was due to multiple well-expressed class I molecules but only one dominantly-expressed class I molecule was found (Wallny et al 2006 PNAS, Shaw et al 2007 J Immunol) and, to our surprise, bacterially-expressed BF2*21:01 heavy chain and b₂-microglobulin refolded with two synthetic peptides without sequence in common, and the crystal structures showed that this molecule remodeled the binding site to accommodate two such disparate peptides (Koch et al 2008 Immunity). This was the beginning of our understanding of the spectrum of class I alleles from promiscuous generalists to fastidious specialists, which we have explored in a series of further papers (in particular, Chappell et al 2015 eLIFE, Tresgaskes et al 2016 PNAS, Kaufman 2018 Trends Immunol, Tregaskes and Kaufman 2022 Mol Immunol).

Second, over these many years, we continued to explore the binding properties of BF2*21:01 in ever more detail, resulting in the current manuscript. We learned only slowly how to probe this unexpected promiscuity, unprecedented in the MHC literature, so that the experiments proceeded with our best understanding at the time, including taking advantage of new approaches as they become available. Each experiment built on the previous set of experiments and each brought us closer to an understanding.

Third, having amassed a collection of data, we chose eLIFE exactly because it allows us to present the entire story from beginning to end without compromise, not just the highlights with the major points illustrated by a few main figures and with the supporting data in many supplementary figures. We include all the data, because it is all part of the story, and so interested researchers to look at the data from their own perspective. Although mostly we provide bar graphs, we include spreadsheets for the raw data (or close to them) for the final experiments (illustrated by Figs. 10 and 14-22) in the two source data files, so these can be assessed easily by others in the field, perhaps using approaches that we may not feel competent to perform.

Public Reviews:

Reviewer #1 (Public review):

Summary:

Combining in vitro refolding, SEC-based assembly assays, peptide-library screening, MALDI-TOF, LC-MS/MS, structural analysis and immunopeptidomics, this manuscript investigates the peptide-binding principles of the promiscuous chicken MHC-I molecule BF2*21:01.

Strengths:

Although the peptide motif of BF2*21:01 is highly complex, this manuscript identified several principles, including a preference for 10-mer peptides, co-variation between P2 and Pc-2, effects of P3 and Pc-3, and a strong cellular preference for Leu at Pc. The results are important for avian MHC biology and poultry vaccine epitope prediction.

Weaknesses:

The manuscript is sometimes difficult to follow because the authors present a large amount of peptide-library, structural and immunopeptidomics data. without always clearly explaining how these datasets support the proposed simplifying principles.

We are delighted and grateful to the reviewer 1 for the careful and thoughtful comments and questions concerning our manuscript. We are glad that the strengths of the paper were apparent and appreciated, and acknowledge the weaknesses that come with such a complex story with experiments performed over decades.

Major Issues - Points Requiring Clarification or Additional Support:

(1) Line 282-301, 537-545）

The immunopeptidomics conclusions are mainly based on one B21 cell line with one biological replicate and at least two technical replicates. Given the complexity of the BF2*21:01 peptide repertoire, this is a major limitation. The authors should either provide additional biological replicates or clearly state this limitation in the Abstract, Results and Discussion.

This limitation is clearly stated in lines 537-545, as part of a paragraph covering the various ways in which the data presented in this manuscript could be improved. In fact, we have performed immunopeptidomics of several different B21 cell types, with many replicates and found similar data as presented, giving us confidence in our interpretations. However, these other experiments belong in different stories, so it is not appropriate that the data be reported in this manuscript.

(2) (Lines 290-313）

The B21 cell preparations contain both BF2 and the lowly expressed BF1 molecule. Some peptides, especially 8-mers or peptides with atypical motifs, may derive from BF1*21:01. The authors should clarify how BF2*21:01-bound peptides were distinguished from possible BF1-derived peptides, or interpret the immunopeptidomics motif more cautiously. The authors should also provide or cite evidence confirming the B21 haplotype identity of the cell line and chicken materials used for immunopeptidomics.

The concern about the contribution of BF1*21:01 to the immunopeptidomics is clearly stated in the manuscript, both lines 290-313 and as part of the paragraph describing the limitations of the experiments (lines 542-543). In fact, the expression of BF1 molecules has long been known to be less than 10% of BF2 molecules at the RNA level, and much less at the protein level (Wallny et al 2006 PNAS, Shaw et al 2007 J Immunol). The proportion of 8mers identified by immunopeptidomics is also low (Fig. 14), and it is not impossible that most 8mers are due to BF1*21:01. We have used assembly assays with peptide libraries, immunopeptidomics and a crystal structure to determine the peptide motif for typical BF1 molecules, of which BF1*21:01 is one and found it may contribute to 8mer peptides but very seldom to longer peptides. This work is unpublished but gives us confidence that the characteristics of BF2*21:01 are not misrepresented by the data in this manuscript.

The sources of the chicken samples and the cell lines are described in detail under Materials and Methods (lines 577-590), citing relevant publications.

(3) (Lines 217-221, 243-253)

The authors acknowledge that MALDI-TOF cannot reliably distinguish peptide combinations with identical or similar masses, nor determine residue positions in some cases. Therefore, MALDI-TOF results should not be overinterpreted as precise evidence for residue preference. The authors should clearly indicate which conclusions are supported by LC-MS/MS.

As described, the experiments follow each other in temporal sequence, so that we started with single peptides, then peptide libraries that varied in one position, then peptide libraries that varied in two positions first analysed by MALDI-TOF and later by LC-MS/MS. The final experiment (Fig. 10, with the original data in the supplementary spreadsheet) directly compares MALDI-TOF and LC-MS/MS results for six peptide libraries, so that the strength of the evidence for residue preference is clear. Throughout the manuscript, we do our best to not to overstate conclusions based on the data of any particular experiment.

(4) (Lines 297-301, 316-330)

The authors suggest that longer peptides may bulge in the middle or extend out of the groove at the C-terminal end. The rationale for the C-terminal extension is not clearly explained. Why is the C-terminal extension considered rather than the N-terminal extension? If the binding register is uncertain, long peptides should be analyzed separately from canonical-length peptides.

When the first sequence of a chicken class I cDNA was determined, an immediate mystery was why one of the so-called invariant residues that coordinate the N- and C-termini of the bound peptide is not conserved (Kaufman et al 1992 J Immunol). In fact, this residue Tyr at position 86 in HLA-A2 and the equivalent position in all mammalian classical class I molecules is an Arg in the classical class I molecules of all non-mammalian vertebrates and is common with class II molecules (Kaufman et al 1995 Semin Immunol). Similar to class II molecules, this Arg in chicken class I molecules allows the peptide to extend out of the C-terminus, as shown by a crystal structure (Xiao et al 2018 J Immunol). The concern that we might be misidentifying the C-terminal amino acid was the basis for the analysis in Figs. 23 and 24, but in the absence of crystal structures, we are not able to provide a final answer this question. Perhaps relevant is the fact that a chicken class II molecule can bind exactly the same peptide in two conformations, one with a canonical 9mer core and the other with an unexpected 10mer core (Goryanin et al 2026 J Virol).

By contrast, N-terminal extensions are only found for some class I alleles and thus far depend on the substitution of small amino acid sidechains for W166 (Li et al 2011 J Virol for bovine, Ma et al 2020 J Immunol for Xenopus, Wei et al 2022 J Immunol for ovine). Thus far, no chicken BF2 sequences have this substitution, consonant with the many crystal structures, including those for BF2*21:01 (Koch et al 2008 Immunity, Chappell et al 2015 eLIFE, this manuscript). However, in unpublished data, we find that most BF1 sequences have sequence differences that could allow N-terminal extensions, although we have no crystal structures to support this possibility.

(5) (Lines 406-439)

In vitro assembly assays show that several hydrophobic residues can be tolerated at Pc, whereas immunopeptidomics shows a strong Leu preference at this position. The authors should clarify whether this Leu preference reflects intrinsic BF2*21:01 binding specificity, TAP-mediated peptide transport, antigen processing, peptide loading, or a cell-line-specific effect. Additional experimental support, such as TAP transport analysis, would strengthen this conclusion.

The preference for Leu at the final position of the peptide by immunopeptidomics of the B21 cell line is strong but not absolute and is certainly affected at the least by the length of the peptide (Figs. 23 and 24). Unpublished immunopeptidomics results (mentioned above) show that this is not a cell line-specific result. The evidence from assembly assays of various peptides is that several hydrophobic amino acids are tolerated with sufficient stability of BF2*21:01 that they are detected in the assay (Figs. 3, 5, 9 and 10). Thermostability assays (Fig. 6) show that peptides with these same hydrophobic amino acids are stable to at least body temperature of chickens. These experiments show that such stability is peptide-dependent (that is, whether a particular amino acid is tolerated depends on the stability conferred by the rest of the peptide). Finally, peptide translocation assays using B21 cells have been done (Tregaskes et al 2016 PNAS) and show that peptides with several hydrophobic amino acids can be pumped into the lumen of the endoplasmic reticulum. However, the assays are with single synthetic peptides, so the data are not extensive enough to separate the effects of the final amino acid from the rest of the peptide. Certainly, peptides with amino acids other than Leu at the C-terminus can be translocated. So, it is not yet clear at which point the preference for Leu at the C-terminus of the peptide arises.

(6) (Lines 172-178, 243-279, 442-457)

The structural analysis explains some residue combinations, such as Arg at P2 with Glu at Pc-2 or Trp at Pc. However, the structural interpretation is not fully integrated with the large-scale peptide library and immunopeptidomics results. Representative high- and low-frequency combinations should be discussed structurally.

Six crystal structures show that BF2*21:02 remodels the binding to accommodate a variety of anchor residues (Koch et al 2008 Immunity, Chappel et al 2015 eLIFE). These crystal structures are representative of sequences found by the immunopeptidomics from very frequent (H-E at roughly 15% 8-12mers) to moderately frequent (E-L at roughly 6% 8-12mers) to infrequent (N-F, A-D and E-D at roughly 1.5%, 1.6% and 0.7% 8-12mers) based on Fig. 18. All but one of the structures has Leu at the C-terminus, with the last one having Val which is found but not frequently by immunopeptidomics.

Similar numbers are found by LC-MS/MS of double-substitution libraries of the two original peptide sequences in Fig. 10 with H-E found frequently (8.1% in P390, 3.8% in P498) and the others infrequently (0.1, 0.9, 1.0, 0.3% in P390, 0, 1.4, 1.0, 0.3% in P498), as calculated from the numbers in the Supplementary data spreadsheet. As discussed in the manuscript, for single-substitution peptide libraries of the two original peptides, Ile/Leu at the C-terminus was very frequent but at the same or slightly less level as Phe, with Met less frequent and Val even less so (Fig. 7).

In addition, there are two more structures along with models explicitly testing some substitutions (Fig. 5). Attempting more current modelling approaches, we found AlphaFold 3 was unable to correctly predict most of the conformations that are found in the crystal structures of BF2*21:01, so we don’t feel confident in using them to predict unknown structures of this kind.

(7) The inference of co-variation between P2 and Pc-2, as well as the modulatory effects of P3 and Pc-3, should be better explained. At present, some conclusions appear to be based mainly on residue-frequency patterns, and the logical connection between these observations and the proposed binding principles is not always clear. Statistical analyses, such as mutual information, chi-square tests or permutation tests, and representative structural explanations would strengthen this conclusion.

We endeavored to do our best to explain the data, our interpretations and our reasoning, so we apologise if we have not managed to be as clear as might be desired. We have included as close to raw data as possible for the LC-MS/MS and MALDI-TOF (Fig. 10) and for the immunopeptidomics (Fig. 14 and 18) in the Supplementary Data spreadsheet, exactly so that competent practitioners can carry out further analyses (including the sophisticated statistical tests mentioned).

Reviewer #2 (Public review):

Summary:

The study presents an in-depth analysis of the peptide repertoire bound by a promiscuous chicken MHC molecule using mass spectrometry, x-ray crystallography and modelling. While the MHC can bind a very diverse set of peptides, the authors have found some new rules that govern peptide binding to this MHC that could help to build a predictive model to study the repertoire of pathogen-derived peptides.

Strengths:

The study uses a range of well performed experiment across multiple techniques and provides an in-depth analysis of the peptide repertoire, including peptide sequences, length, preferred residues, stability and MHC presentation.

Weaknesses:

The data overall support the analysis and conclusion well. The only caveat is linked to Figure 4, which does not describe the stability of the peptide-MHC complex, but instead shows refold yield, and the two are not always linked.

We are grateful for the clear understanding of the strengths of the work. With regards to Fig. 4, we agree with the reviewer that there are differences in refold yield but that measure may not be correlated with stability of the peptide-MHC complex. However, we were basing our interpretation of stability on the position and quality of the monomer peak, as illustrated by the trace in Fig. 2, in which a sharp peak at the monomer position represents a stable complex (as seen for the 10 and 11mer peptides) and later peaks represent unstable complexes falling apart during the chromatography (as seen for the 7, 8 and 9mer peptides).

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Minor Issues: Editorial and Data Presentation Modifications

(1) Lines 53-62, 155-170, 303-314: The terms Pc, Pc-2 and Pc-3 should be clearly defined early in the manuscript and figure legends.

The Abstract introduces the abbreviations as “peptide positions P₂ and P_c-2” followed by P_C, which are standard usage and seem clear. The first usage in the text is “anchor residues at three positions, with co-variation between the anchor residues at P₂ and P_c-2” which again seems clear, particularly in the context of the text and the figure. However, a parenthetical description has been added to read “…anchor residues at three positions, with co-variation between the anchor residues at P₂ and P_c-2 (position 2 and the position two before the C-terminus) …”. Given these usages, it seems unlikely that the reader will fail to understand P_C and P_C-3.

(2) Lines 255-279: The term "peptide backbone" should be defined as the fixed sequence context outside the randomized positions, if this is what the authors mean. Suggest clarifying the meaning of "peptide backbone".

The phrase reads “double-substitution libraries based on four peptide backbones”, which in context of the figures seems clear. However, a parenthetical description has been added: “The examination of double-substitution libraries analysed by MALDI-TOF was expanded to other backbones (that is, other sequences in which two positions were randomised): …”.

(3) Several figures are complex. The authors should add brief take-home messages to figure legends.

Every figure legend starts with a (sometimes quite long) take-home message. It is not clear what more should be added.

(4) A concise summary table of the proposed binding rules, including preferred peptide length, P2, Pc-2 and Pc preferences, and the effects of P3/Pc-3, would be useful for readers.

A concise summary table of binding rules would be very helpful, but the rules are complex, both qualitative and quantitative. For example, immunopeptidomics shows that 10mers are preferred, but that fails to capture the quantitation. The co-variation of P₂ and P_c-2, which is the strongest and best characterized correlation, nevertheless is complex since the occupancy of P₂ by different amino acids (presumably independently of P_c-2) varies considerably. At our current level of understanding, it is hard to imagine a table that is both concise and precise. With time and more data, perhaps code for quantitative prediction might be constructed (dare one suggest machine learning…), which is the hope of presenting all the available data in one place.

(5) The Results section contains a large amount of peptide-library, structural and immunopeptidomics data. The authors should improve the logical flow and add clearer transition sentences to explain how each dataset supports the proposed simplifying principles.

The reviewer is of course correct that any written text can be improved (although each critic may have a different idea about which part should be fixed), but we have done our best with the material and time available. We could respond productively to a more detailed critique.

(6) Some statements in the Abstract and Discussion should be softened, especially those related to in vivo peptide preferences, BF2*21:01 promiscuity and peptide prediction, given the limited biological replication and uncertainty in peptide assignment.

The statements in the both the Abstract and Discussion are very general, reflecting what we believe to be careful interpretations based on the data. We could respond productively to concerns about specific claims.

Reviewer #2 (Recommendations for the authors):

Overall, the data presented in this study are interesting; however, it is complex, and some results could be merged and simplified, as well as the figures. The data provide an in-depth analysis, using mass spectrometry, of the interplay between the different positions of the peptide and the residues favourable to bind within the antigen-binding cleft.

(1) From the abstract, the concept of "promiscuous generalists and fastidious specialists" is not explored after or defined within the results.

The Abstract introduces the concept of promiscuous generalists and fastidious specialists to provide the basis for exploring the peptide-binding specificity of the most promiscuous class I known, BF2*21:01. This overall concept is described in enormous detail in several publications cited in the current manuscript, but it is not particularly germane to the analyses.

(2) From the abstract "These simplifying principles may eventually allow predictions of pathogen peptides", I'm not sure how "simplifying" the principles are with the data, if anything, it does show a rather complex interplay between the different residues of the peptide that enable the MHC to bind a large and diverse number of peptides.

Compared to any combination of anchor residues being permitted at equal frequency, there are clear preferences which the experiments identify. Instead of an enormous range of possible peptide lengths, roughly 50% of peptides are 10mers. Of course, the structural reasons behind these results are certainly complex and likely must be understood in detail in order to attempt peptide predictions. 

(3) Line 48. "less well-expressed". Less than what? Do you mean the level of expression was lower? And if yes, are there values of fold change for comparison?

The sentence in the Abstract reads “Chicken BF2 alleles … are less well-expressed on the cell surface … while certain human HLA-B alleles … are well-expressed …”. Read as a complete sentence, the meaning is clear. This difference for chicken BF2 alleles has been quantified as reported in several publications cited in the current manuscript, ranging from 3-5 fold for peripheral blood lymphocytes to ten-fold for erythrocytes (Kaufman et al 1995 Immunol Rev, Chappel et al 2015 eLIFE), with similar numbers for a few HLA-B alleles on human peripheral blood lymphocytes and monocytes (Chappell et al 2015 eLIFE).

(4) Line 149. "with individual peptides" which peptides are we referring to here?

This introductory sentence to a paragraph outlines the general method used for the experiments in this section of the Results, “refolding in vitro … with individual peptides.” Which “individual peptides” are described in the following paragraphs, with the next section of the Results using “refolding in vitro … with peptide libraries”.

(5) Line 170. If 9 mers and below are not stable, which is not really quantified or shown with the data on Figure 4, why is refolding material observed for peptides with different lengths of 9 aa and below on Figure 4? A lower yield of refolded material can have a different origin, and there is no association between stability and refold yield. The notion of stability here probably needs to be changed, as it does not apply to the data.

As described in our response to a similar concern above, we are not basing our interpretation of stability on the quantity of refolded material, but on the position and quality of the monomer peak, as described clearly in the legend to Fig. 4: “The original 10mer (REVDEQLLSV) and 11mer (GHAEEYGAETL) peptides refold with BF2*21:01 to give stable monomers as do 11mer and 10mer derivative peptides, but 9mer, 8mer or 7mer peptides give heavy chain only peaks” and “The peptides 11mer GHAEEYAETL (top panel), 10mer REVDEQLLSV (middle panel), 11mer GHAEAAAAETL and 10mer GHAEAAAETL (bottom panel) gave sharp monomer peaks, while the 9mer GHAEAAETL, 8mer GHAEAETL and 7mer GHAEETL gave a delayed broad peak indicative of unstable binding or heavy chain.” This concept is illustrated by the trace in Fig. 2 (discussed in the text at the beginning of this section of the Results), in which a sharp peak at the monomer position represents a stable complex, and later peaks represent unstable complexes falling apart during the chromatography or free heavy chains.

(6) Line 175. As the 3BEV structure had a P2-His, is a comparable structure expected?

This sentence reads “Structures with amino acid substitutions in the 11mer peptide GHAEEYGAETL bind with Asp at P_c-2 and either His or Arg at P₂ (5AD0 and 5ADZ), comparable to the original structure (3BEV) (Fig. 5A).” Minor changes in positions and orientations of individual amino acid sidechains are expected and are clear from the crystal structures presented in Fig. 5A, but are overall comparable to the original peptide in 3BEV, which has a His at P₂ and a Glu at P_c-2.

(7) Line 175 "modelling the substituted". How was the modelling done?

The sentence reads “Modelling the substituted peptide with Arg at P₂ and Glu at P_c-2 shows a steric clash that can explain why this peptide did not refold with BF2*21:01 (Fig. 5A).” The legend to Fig. 5A states that “modelling done as detailed in Materials and Methods”, but apparently that section was omitted. A section has been added now to the Materials and Methods which states “Beginning with known crystal structures, modelling was carried out using PyMol with the protein mutagenesis Wizard, the rotamer toggle, show bumps and show surfaces.”

(8) Line 176 "shows a steric clash" with what? Figure 5 is too small to see, and there is no label on the residue to follow where the steric clash is coming from.

As stated in the legend to Fig. 5, “Structures were determined for GHAEEYGAETL (3BEV), GHAEEYGADTL (5AD0) and GRAEEYGADTL (5ACZ), which all refolded successful to give stable monomers, while GRAEEYGAETL did not (see Fig. 3), all models of which showed steric clashes (one depicted, red arrow).” In the model shown, the clash is between R9 of the BF2*21:01 with the Glu at peptide position 9. Parenthetically, this depiction of the key MHC residues for the co-variation has been used repeatedly, starting with Chappell et al 2015 eLIFE. The picture can be zoomed to make it large enough to see.

(9) Line 247 "many combinations". It is not clear here if combinations are referring to a set of double-substitutions or different peptides?

Each peptide in the library has a different double-substitution, so the meaning is the same either way. The point of Fig. 10 is to compare the identification of peptides from LC-MS/MS (in which each peptide is identified exactly) with the identification of sets of peptides from MALDI-TOF (in which the order of the double substitution is not clear, as well as the exact amino acid in the cases of I/L and Q/K). As the figure shows, the numbers are generally very similar, but this sentence describes the percentage of cases for which only one method or the other identified a peptide.

(10) Lines 252-253. The conclusion of the MS data that the LC-MS/NS is more accurate and sensitive than MALDI-TOF is not very surprising. What was the rationale for using both?

Our examination of the binding properties of BF2*21:01 for peptides and peptide libraries developed over a long time-span, so in the beginning we looked at single peptides with size exclusion chromatography peaks as the measure, then single- and later double-substitution libraries, first by MALDI-TOF and later by LC-MS/MS. Each set of experiments built on the previous work, so that together they tell the story. For example, after we optimized the use of double-substitution libraries, we were worried about the effects of temperature, so we tested that by MALDI-TOF. While repeating the temperature experiment once we optimized the LC-MS/MS approach might have yielded some additional data, there were other questions to answer.

(11) Line 272-273 "support the idea that P3 is an important position within the peptide (Figures 8-10) despite not contacting the MHC molecule" The structure of 3BEV clearly shows interaction between the P3-Ala and the Tyr156 of the MHC. Residues that are fully or partially buried in the MHC cleft almost all contact the MHC molecule. Maybe I've missed something, but I think this statement is inaccurate.

We agree with the reviewer that nearly every peptide residue contacts the MHC molecule (but of course some much more than others). The statement is now changed in the text to read “support the idea that P₃ is an important position within the peptide (Figures 8-10) despite not being an anchor residue."

(12) Line 294. Figure 14 clearly shows that the number of 8 and 9-mer peptides eluted is at the same level as the 11mer and above, so how does the data fit with the statement that 9mer and shorter peptides are not stable with the MHC?

Fig. 4 shows that 7, 8 and 9mer derivatives of the original 11mer failed to refold to give a single peak of stable monomers, while Fig. 6 shows that the 10mer derivative of the original 11mer was more thermostable than both the 9mer derivative and the original 11mer. The reason why the 9mer yielded so little monomer in Fig. 4 while giving enough to test by thermostability in Fig. 6 is no longer remembered. A key point is that these results are peptide sequence-specific, so it is not impossible to imagine stable binding of an appropriate 8mer (or perhaps even a 7mer). Another uncertainty, mentioned in the Results and Discussion, is that BF1*21:01 molecules bind primarily 8mers, and the contribution of peptides from BF1*21:01 is not known with certainty.

(13) Line 316. What was the rationale for choosing peptides > 12aa to see if there is some overhang or bulge? As even 11-12mer can exhibit such features.

We were just looking for any obvious patterns, but we didn’t find any.

(14) The section starting at line 366 would have benefited from some structure prediction or modelling to illustrate the findings.

We would have been delighted to model peptides, but we have used AlphaFold3 to compare the models to our crystal structures for seven chicken class I alleles (including BF2*21:01) with one or more peptides. The models sometimes fit the experimental data but they often didn’t, often by a wide margin, and with BF2*21:01 the worst (presumably because the system is not trained on MHC molecules which utilise charge transfer). Therefore, we do not feel confident in using any such modelling approaches except in the most defined situations (such as illustrated in Fig. 5).

(15) Typo - Alleles should be italic, and in vitro as well.

Alleles of genes are in italics, but alleles of proteins are not. To write in vitro in italics is customary, which we have corrected.

(16) Figures

(a) Some of the figures could be merged together.

Of course, any presentation can be improved, but which figures we should merge (some already being three pages in length) is not clear. We could productively respond to more detailed suggestions.

(b) Figure 1. I can't see the different colours mentioned in the figure legend

Our apologies if the colours are not clear enough, but they are present only as an aid (as elsewhere in the manuscript), with red D and E, blue H, K and R, green N, Q, S and T, and all other amino acids black.

(c) Figure 12. Is this figure only with 11mer peptides?

Figure 12 shows the percentage of peptides with particular amino acids at P₂ and P_c-2 for three 11-mer double-substitution peptide libraries, the sequences of which are written on the graphs and described in the figure legends.

(d) Table 1. The name of the protein should be added to the table.

OK.

eLife Assessment

This study makes a valuable contribution by broadening the range of eukaryotic model systems and establishing Blastocystis, the most prevalent microeukaryote in the human gut, tractable for reverse-genetics investigations. The presented imaging data are convincing and informative. The work should interest readers studying host-microbe interactions in the human gut, as well as those developing new systems for eukaryotic research.

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 reviewers' suggestions.]

Summary:

This paper presents a toolkit for the transformation of Blastocystis. The authors have screened a number of selectable agents, promoters and reporter genes and present their findings. This resource will be of immense use to those in Blastocystsis field, as well as those seeking to establish transformation tools in other species where such tools do not yet exist. Establishing new transformation tools is extremely challenging, and the authors have done an excellent job.

Strengths:

The authors have carried out a systematic screen of promoters, reporter genes and selectable agents. They have screened numerous for each, and all the data is presented. It is good to see when things did not work as well as when things did - so this data set is extremely useful indeed.

Comments on previous version.

The authors have revised their manuscript to clarify that molecular analyses have not yet occurred and have resolved the technical/publication issues with the figures. I look forward to seeing these tools used in future publications to answer important questions in Blastocystsis research.

Reviewer #3 (Public review):

Summary:

The primary objective of this study was to establish a practical and functional framework for propagation of stable transgenic cell lines of Blastocystis, a common animal gut microeukaryote. Although the work focused on Blastocystis ST7-B, a subtype with relatively low prevalence in humans, this choice is justified by its association with more frequent negative health effects. Beyond their relevance to the medical field, the methodological advances described here have the potential to also expand cell biology studies of this anaerobic organism, including its unusual mitochondria and redox metabolism.

Strengths:

Prior to this work, genetic tools for Blastocystis were very limited, relying on a single strong promoter-terminator combination. The authors successfully expanded the available promoter set across a range of expression strengths by testing two dozen variants in luciferase-based assays. Critically, they developed an integrated workflow from a modular transgenic construct design to an expanded inventory of molecular components (promoters, reporters), optimized DNA delivery, stepwise antibiotic resistance-mediated clonal selection and propagation, and to reporter validation. The evaluation of several anaerobiosis-compatible labeling strategies for live (and fixed) cell optical imaging will be particularly useful, with the SNAP-tag system appearing especially promising for Blastocystis.

Author response:  

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1 (Public review):

Comments on revised version.

The authors have revised their manuscript to clarify that molecular analyses have not yet occurred and have resolved the technical/publication issues with the figures. I look forward to seeing these tools used in future publications to answer important questions in Blastocystis research.

We are grateful to Reviewer 1 for their careful and constructive assessment of our revised manuscript. We are pleased that the revisions have satisfactorily addressed their previous concerns, and we sincerely appreciate their recognition of the value of our work.

Reviewer #3 (Public review):

Comments on revised version.

The revised version provides sufficient clarity and appropriate visual presentation. Some confusion evidently arose due to my misunderstanding, so I thank the authors for their comprehensive clarifications and patience.

We are grateful to Reviewer 3 for their careful reading of the revised manuscript and for the additional constructive suggestions. We appreciate that these recommendations are aimed at improving clarity, accessibility, and presentation, and we have addressed them as detailed below.

Recommendations for the authors:

Reviewer #3 (Recommendations for the authors):

(1) Table 1

(1a) The description of the column "Promoter Lengths Tested" is a bit confusing because it actually shows acronyms of the used promoters. Suggestion for clarity: either keep the name and mention just the lengths, or rename to "Promoter abbreviation" or "Promoter acronym".

We thank the reviewer for pointing this out. To avoid confusion, we have renamed this column “Promoters Tested”, which better captures the information presented. The column contains both promoter identifiers and the corresponding promoter lengths, which are further clarified in the table notes.

(1b) It is not entirely clear based on the formatting of the table, which columns are relevant for "Blastocystis ST4-WR1" and which for "Blastocystis ST7-B".

We thank the reviewer for this helpful suggestion. We have revised the formatting of Table 1 to more clearly distinguish the data corresponding to Blastocystis ST7-B and Blastocystis ST4-WR1. Specifically, we have added a vertical divider between the relevant column groups to improve readability and make the species-specific information easier to follow.

(2) Figure 2:

If the authors wish to preserve the decorative colouring in the parts B and C, it would be better to at least keep the datapoint and boxplot hues consistent for individual conditions (or plot the datapoints in the same, neutral hue throughout, e.g., black or grey). Currently, this is only the case in the part C, but not in the part B.

We thank the reviewer for this useful suggestion. We have revised Figure 2B to improve visual consistency between the data points and boxplots while preserving the overall colour scheme of the figure. This adjustment improves readability without changing the data or its interpretation.

(3) Figure 3: With the modifications everything is clear. However, two issues are now apparent.

(3a) After the part A was modified, the arrow colours changed (had been yellow and red, now are cyan and magenta), but the description in the legend remained (yellow and red), so the text should be corrected. [By the way, the original colours worked well.]

We apologise for overlooking this inconsistency. The Figure 3 legend has now been corrected to match the revised arrow colours in panel A.

(3b) The care taken by the authors to make the images more accessible is really greatly appreciated, but being a person on the colourblind spectrum, I can attest that the issue is not always just about red and green discrimination: the greyscale and blueish green used in the part A photo in particular are discernible only at huge magnification (to some people). If the signal in the part A were of the same hue as in the part B (yellowish green instead of blueish green), it would readily pop out. For that matter, what is the reason for the use of a different colour profile of the UnaG fluorescence in these two images?

We sincerely thank the reviewer for this important accessibility-related comment. The difference in colour profiles between panels A and B was intentional because the images were acquired using different imaging modalities, as indicated in the figure legend. We wanted to avoid implying that the two panels were generated under identical imaging conditions. However, we appreciate the reviewer’s point that this distinction should not come at the expense of readability. We have therefore adjusted the display of the UnaG signal in panel A to improve contrast and visibility. 

(4) Conclusion:

The expression "bringing endogenous regulatory part discovery, namely the identification of native promoter and terminator elements" feels a bit clunky. What the part "endogenous regulatory part discovery" alludes to is now clearer, but consider reformulating it to "discovery of endogenous regulatory elements". This would make it clear without the need to add the explanation "namely the identification of native promoter and terminator elements". [It is now clear that the accumulation of noun adjectives and the significance of the word "part" was what originally blurred the overall meaning.

We thank the reviewer for this valuable suggestion. We agree that the original phrasing was unnecessarily clunky and have revised the sentence for clarity and flow. Lines 682–684 now read:

“By integrating endogenous promoter and terminator discovery, DNA delivery, selection, clonal recovery, and reporter validation into a single pipeline, we provide a flexible foundation for routine transgene-based studies.”

eLife Assessment

This is a fundamental study on the sensory roles of cerebrospinal-fluid-contacting neurons (CSF-cNs) in mammals, revealing how the apical extension is used as an amplifier of chemical changes in content of the CSF. Specifically, the authors show compelling evidence that PKD2L1 is predominantly a pH-sensing channel in CSF-cNs and link its apical localization to dual phasic and sustained responses underlying CSF chemosensation.

Reviewer #1 (Public review):

This study by Vitar et al. probes the molecular identity and functional specialization of pH-sensing channels in cerebrospinal fluid-contacting neurons (CSFcNs). Combining patch-clamp electrophysiology, laser-based local acidification, immunohistochemistry, and confocal imaging, the authors propose that PKD2L1 channels localized to the apical protrusion (ApPr) function as the predominant dual-mode pH sensor in these cells.

The work establishes a compelling spatial-physiological link between channel localization and chemosensory behavior. The integration of optical and electrical approaches is technically strong, and the separation of phasic and sustained response modes offers a useful conceptual advance for understanding how CSF composition is monitored.

Comments on revised version:

I thank the authors for their extensive revisions and detailed responses to the reviewers' comments. The manuscript has been substantially improved, and most of the major concerns raised in the initial review have been adequately addressed. In particular, the additional analyses of PKD2L1 channel activity, the incorporation of physiologically relevant pH conditions, the clarification of ASIC involvement, and the expanded Discussion have significantly strengthened the study.

Major scientific concerns largely addressed:

Quantification of PKD2L1 channel activity<br /> The authors appropriately addressed my previous concerns regarding the use of Po as the sole measure of channel activity. The inclusion of additional parameters such as apparent Po, open time, nmax, holding current, and membrane charge provides a more robust assessment of PKD2L1 activity and substantially strengthens the conclusions.

Physiological relevance of pH modulation<br /> The inclusion of experiments at pH 6.5 and the additional analyses of holding current and resting membrane potential are valuable additions. These experiments considerably improve the physiological relevance of the study.

ASIC contribution<br /> The additional pharmacological experiments using ASIC blockers are helpful and support the conclusion that the photolysis-evoked response in the apical process is predominantly mediated by PKD2L1 channels.

Functional implications<br /> The expanded Discussion regarding Ca2+-dependent signaling, neurosecretion, and the potential physiological roles of CSFcNs considerably improves the manuscript.

Remaining concerns:<br /> Continued overstatement regarding "exclusive" localization and function:

Although the authors softened some statements in the revised manuscript, the term "exclusive" remains in several key locations, including the title.

For example:<br /> "PKD2L1 channels segregated to the apical compartment are the exclusive dual-mode pH sensor..."

The data clearly demonstrate strong enrichment of functional PKD2L1 channels in the apical process. However, the available evidence does not fully justify the term "exclusive," particularly because:

- PKD2L1 immunoreactivity is still detectable outside the apical process.<br /> - ASIC-mediated responses are present in CSFcNs.<br /> - The authors themselves use more appropriate terminology such as "predominantly located" in the Discussion.

Therefore, I recommend replacing "exclusive" with more conservative terminology such as:

- predominant<br /> - predominantly localized<br /> - enriched<br /> - functionally segregated

throughout the manuscript, including the title, Abstract, Introduction, Results, and Discussion.

Use of the term "tonic current"

The manuscript continues to use the term "PKD2L1 tonic current."

While the dibucaine-sensitive holding current is clearly present, the precise mechanism generating this current remains uncertain. Indeed, the authors themselves acknowledge in the Discussion that:

- an alternative conducting state may exist, or<br /> - unresolved brief channel openings may account for the current.

Therefore, the data support the existence of a sustained PKD2L1-associated current, but do not yet definitively establish a distinct tonic gating mode of the channel.

I therefore recommend replacing:

"tonic current"

with a more neutral expression such as:

- sustained current<br /> - PKD2L1-associated holding current<br /> - sustained PKD2L1-mediated current throughout the manuscript.

Continued use of "off-current" and "off-response":<br /> The revised manuscript has improved considerably in this regard. However, the terms "off-current" and "off-response" still remain in portions of the text and figure legends.

Because the manuscript itself demonstrates that the response reflects recovery from transient acidification rather than a separate OFF signaling mechanism, these terms remain potentially misleading.

I recommend replacing them with terminology such as:<br /> - photolysis-evoked PKD2L1 current<br /> - recovery current<br /> - proton-removal-induced current

throughout the manuscript, including figure legends.

Minor editorial corrections<br /> Figure 1Bd Please change: "po" to "Po" for consistency with standard channel physiology nomenclature.<br /> Figure 1Ca Please add units (mV) to the voltage labels shown on the left side of the traces.<br /> Figure 3E Please change: "Norm po" to "Norm Po".<br /> Figure 4Fb Please replace: "sec" with "s" to conform with SI unit conventions.

The authors have addressed the majority of my previous concerns and the manuscript has been substantially improved. The remaining issues are primarily related to terminology and overinterpretation rather than experimental deficiencies.

Reviewer #2 (Public review):

Summary:

Cerebrospinal fluid contacting neurons (CSF-cNs) are GABAergic cells surrounding the spinal cord central canal (CC). In mammals, their soma lies sub-ependymally, with a dendritic-like apical extension (AP) terminating as a bulb inside the CC.

How this anatomy-soma and AP in distinct extracellular environments-relates to their multimodal CSF-sensing function remains unclear.

The authors confirm in the GATA3:GFP mice where these cells are labeled that CSFcNs exhibit prominent spontaneous electrical activity mediated by PKD2L1 (TRPP2) channels, non-selective cation channels with ~200 pS conductance modulated by protons and mechanical forces.

They investigated PKD2L1 pH sensitivity and its effects on CSFcN excitability. They uncovered that PKD2L1 generates both phasic and tonic currents, bidirectionally modulated by pH with high sensitivity near physiological values.

Combining electrophysiology (intact and isolated AP recordings) with elegant laser-photolysis, they show functional PKD2L1 channels localize specifically to the apical extension (AP).

This spatial segregation, coupled with PKD2L1's biophysical properties (high conductance, pH sensitivity) and the AP's unique features (very high input resistance), renders CSFcN excitability highly sensitive to PKD2L1 modulation. Their findings reveal how the AP's properties are optimised for its sensory role.

Strengths:

This is a very convincing demonstration using elegant and challenging approaches (uncaging, outside out patch of the AP) together to form a complete understanding on how these sensory cells can detect so finely the changes of pH in the CSF.

Weaknesses:

Not weaknesses, there are only minor requests to complete the beautiful study.

(1) The apical extension's response to removal of acidification is nicely illustrated in Figure 4C,G. There's something puzzling there: while the response to Glutamate is immediate, the channel responses to H+ is extremely delayed by 100ms - 2s, and even sometimes came in bursts separated by few hundreds of ms. H+ diffuse even faster than glutamate. Why is that?

I don't quite understand how the response is so delayed & how to explain the recurring bursts of channel opening in the figure panel ?

- The authors should show in Fig 4C,G the traces for 1-2 s before uncaging occurs so we can appreciate whether such events occur as well in baseline and discuss this further in revisions.

- Could the authors use a fluorescent pH sensor to monitor pH in the extracellular space and in the cell ?

- Could the authors investigate whether in the apical extension, PKD2L1 channels are mainly at the outer membrane in the apical extension OR whether many channels are located in inner membranes ?

(2) Suppl Fig 4 is very cool and should be moved to main figure. The coupling of Soma and AP is very tight, yet there is a clear difference in targeting of channels that respond to cues in the CSF. In the context of an intact spinal cord, we can wonder how and when the contribution from ASIC in the some would be relevant to physiology. Can the authors think of experiments with an intact central canal to test the sensitivity and condition of recruitment of pH sensing in the soma (ASIC) versus the apical extension (PKD2L1)?

(3) The Reissner fiber is missing after slicing the spinal cord. From our observations in fish, the fiber being under tension triggers lots of activity in CSF-cNs (Bellegarda et al Elife 2023) that also relies on PKD2L1 (Bohm et al NC 2016; Sternberg et al NC 2019). Could the authors discuss the contribution of the Reissner fiber to the PKD2L1 mediated modulation of CSFcN excitability ? Could the authors conceive a way to slice along the anteroposterior axis (sagitally) the spinal cord to keep the Reissner fiber in the central canal when recording CSF-cN apical extension ?

Author response:

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

Public Reviews:

Reviewer #1 (Public review):

This study by Vitar et al. probes the molecular identity and functional specialization of pH-sensing channels in cerebrospinal fluid-contacting neurons (CSFcNs). Combining patch-clamp electrophysiology, laser-based local acidification, immunohistochemistry, and confocal imaging, the authors propose that PKD2L1 channels localized to the apical protrusion (ApPr) function as the predominant dual-mode pH sensor in these cells.

The work establishes a compelling spatial-physiological link between channel localization and chemosensory behavior. The integration of optical and electrical approaches is technically strong, and the separation of phasic and sustained response modes offers a useful conceptual advance for understanding how CSF composition is monitored.

Several aspects of data interpretation, however, require clarification or reanalysis-most notably the single-channel analyses (event counts, Po metrics, and mixed parameters), the statistical treatment, and the interpretation of purported "OFF currents." Additional issues include PKD2L1-TRPP3 nomenclature consistency, kinetic comparison with ASICs, and the physiological relevance of the extreme acidification paradigm. Addressing these points will substantially improve reproducibility and mechanistic depth.

Overall, this is a scientifically important and technically sophisticated study that advances our understanding of CSF sensing, provided that the analytical and interpretative weaknesses are satisfactorily corrected.

(1) The authors should re-analyze electrophysiological data, focusing on macroscopic currents rather than statistically unreliable Po calculations. Remove or revise the Po analysis, which currently conflates current amplitude and open probability.

We agree with the reviewer that the Po analysis has strong limitations, particularly in experiments where the recording times are short, like when extracellular pH is changed either by photolysis (Figure 4D) or puff applications (Figure 3Aa). In order to circumvent that problem and not to rely only on Po estimations, we used alternative methods as well, including the analysis of the current membrane charge that we have used extensively during the manuscript (Figures 3A and 4D, for example) or the analysis of the event latencies (Figure 4G). Nevertheless, single-channel recordings clearly contain information that is not included in the macroscopic current analysis. We intend to stress in the revised version that the elementary current amplitude is conserved by manipulations such as pH changes, leaving the total number of channels (N) and the channel open probability (Po) as possible culprits for the current changes. Since these changes are rapid and reversible, it is likely that N stays constant and that Po changes. In order to address the reviewer’s concern, we propose the following changes/reanalysis: i) to report in each condition the minimum N (maximum observed openings; for example, in Figure 3Aa the minimum N goes from 4 in control conditions to 1 during the puff of the pH 6.4 solution). This method (to estimate N by counting the maximum number of open channels), while imperfect, provides a tentative estimate of Po; ii) following the previous point, we propose to reword the text (and images) and use the expression “apparent Po” instead of “Po”; iii) to report the fraction of time that the channels remain open. Also, we acknowledge that some traces are confusing (Figure 3Aa, top) as they seem to show macroscopic currents. We will modify those figures by plotting the amplitude histograms (as in Figure 1Bb) in order to show unambiguously that current recordings from CSFcNs only show single-channel activities.

(2) PKD2L1-TRPP3 nomenclature should be clarified and all figure labels, legends, and text should use consistent terminology throughout.

We agree with the reviewer that the nomenclature concerning polycystin members is confusing. In this manuscript we have followed the nomenclature that has been proposed in a recent, comprehensive review on polycystin channels by Palomero, Larmore and DeCaen (Palomero et al. 2023), where the authors refer to the channels by their gene name. In this review, the authors indicate that PKD2L1 channels correspond to TRPP2 (formerly TRPP3, their table 1). In another, recent review on TRP channels, however, the authors refer to the PKD2L1 channel as TRPP3 (Zhang et al. 2023). In order to avoid any confusion we will remove from the text any reference to the TRPP nomenclature and stick to the PKD2L1 name.

(3) The authors should reinterpret the so-called OFF currents as pH-dependent recovery or relaxation phenomena, not as distinct current species. Remove the term "OFF response" from the manuscript.

We concur with the reviewer that the term “OFF response” is not very helpful from the biophysical perspective and conveys the idea that it is another current. We will remove the term “OFF response” or “OFF current” in the revised manuscript and replace it by the term “photolysis-evoked PKD2L1 current”. Also, we will condense two sections (“The proton-induced current is an off-current” and “The off-current is mediated by the activation of PKD2L1 channels”) into a single new section (“The photolysis-induced current is mediated by PKD2L1 channels”), as separating the description of the photolysis-evoked PKD2L1 current compromises its description. Finally, we will rewrite the discussion to better describe this current.

(4) Evidence for physiological relevance should be provided, including data from milder acidification (pH 6.5-6.8) and, where appropriate, comparisons with ASIC-mediated currents to place PKD2L1 activity in context.

This is partly addressed in Figure 3. The data there indicates that PKD2L1 channels are very sensitive to pH variations around physiological pH. In order to make this conclusion stronger, we will add to the figure the EC50 values drawn from the fittings. In terms of the ASIC-mediated currents, one of our main conclusions is that ASIC channels are not present in the ApPr, as the effects of proton photolysis in the ApPr and are not blocked by ASIC channels blockers. Our results indicate that PKD2L1 channels are the exclusive pH sensitive channels in the ApPr, while ASIC channels are probably the acid-sensitive channels in the soma, although we have not studied the latter in detail. Following this and the editor’s comments, the subsection “the involvement of ASICs” in the Discussion has been modified.

(5) Terminology and data presentation should be unified, adopting consistent use of "predominant" (instead of "exclusive") and "sustained" (instead of "tonic"), and all statistical formats and units should be standardized.

Following the suggestions of the reviewer, an exhaustive work will be performed to unify terminology, data presentation and correct the text following the reviewer’s suggestions.

(6) The Discussion should be expanded to address potential Ca²⁺ -dependent signaling mechanisms downstream of PKD2L1 activation and their possible roles in CSF flow regulation and central chemoreception.

This is indeed a very interesting and currently unresolved point in the physiology of CSFcNs. Published data indicate that calcium flowing into the cell through PKD2L1 channels is a key regulator of apical process physiology: on the one hand, PKD2L1 channels are calcium permeable and at the same time, they are inhibited by intracellular calcium (DeCaen et al. 2016). Also, ultrastructural data indicate that the ApPr is rich in mitochondria and tubulo-vesicular structures resembling the Golgi apparatus (Bjugn et al. 1988; Bruni et Reddy 1987), which are intracellular organelles that contribute to calcium homeostasis. Altogether, this evidence suggests that intraApPr calcium concentration needs to be finely regulated, both in space and in time, in order for the ApPr to fulfil its physiological roles. Based on what has been published in the literature, we can speculate that these calcium signals can be decoded by several systems: i) calcium can act as the second messenger linking the activation of the multimodal PKD2L1 channels to changes in CSFcNs excitability, which in turn regulate spinal neuronal networks controlling locomotor activity; ii) calcium could initiate neurosecretion of different molecules from the ApPr to the central canal (as has been proposed by the Wyart group in the zebrafish in the context of bacterial infections(Prendergast et al. 2023)); iii) calcium could activate the Hedgehog signaling pathways (as has been shown by (Delling et al. 2013)); iv) calcium could modulate CSF flow directly (by modulating ciliary activity) or indirectly (by modulating ependymal cells ciliary activity through paracrine interactions). Resolving these downstream pathways is essential to fully define the role of CSFcNs as integrators of CSF homeostasis. We will expand this issue in the Discussion of the revised ms.

Reviewer #2 (Public review):

Summary:

Cerebrospinal fluid contacting neurons (CSF-cNs) are GABAergic cells surrounding the spinal cord central canal (CC). In mammals, their soma lies sub-ependymally, with a dendritic-like apical extension (AP) terminating as a bulb inside the CC.

How this anatomy-soma and AP in distinct extracellular environments relate to their multimodal CSF-sensing function remains unclear.

The authors confirm that in GATA3:GFP mice, where these cells are labeled, that CSFcNs exhibit prominent spontaneous electrical activity mediated by PKD2L1 (TRPP2) channels, non-selective cation channels with ~200 pS conductance modulated by protons and mechanical forces.

They investigated PKD2L1 pH sensitivity and its effects on CSFcN excitability. They uncovered that PKD2L1 generates both phasic and tonic currents, bidirectionally modulated by pH with high sensitivity near physiological values.

Combining electrophysiology (intact and isolated AP recordings) with elegant laser-photolysis, they show that functional PKD2L1 channels localize specifically to the apical extension (AP).

This spatial segregation, coupled with PKD2L1's biophysical properties (high conductance, pH sensitivity) and the AP's unique features (very high input resistance), renders CSFcN excitability highly sensitive to PKD2L1 modulation. Their findings reveal how the AP's properties are optimised for its sensory role.

Strengths:

This is a very convincing demonstration using elegant and challenging approaches (uncaging, outside out patch of the AP) together to form a complete understanding of how these sensory cells can detect the changes of pH in the CSF so finely.

Weaknesses:

The following do not constitute weaknesses; rather, they are minor requests that this reviewer considers would complete this beautiful study.

(1) It would be nice to quantify further the relation in spontaneous as well as in acidic or basic pH between the effects observed on channel opening and holding current: do they always vary together and in a linear way?

Following the reviewer’s suggestion, we have performed a Spearman’s rank correlation test, which shows a significant correlation between the changes in the apparent open probability and holding current (paired experiments; ctrl vs pH 6.4 pressure applications; p < 0.05, Spearman r = 0.72 and critical value = 0.67). The Pearson correlation coefficient calculated on the same data set = 0.63 and the critical value is 0.632, which indicates that the correlation is not linear. We will add this analysis to the manuscript.

(2) Since CSF-cNs also respond to changes in osmolarity (Orts Dell Immagine 2013) & mechanosensory stimulations in a PKD2L1 dependent manner (Sternberg NC 2018), it would be nice to test the same results whether the same results hold true on the role of PKD2L1 in AP for pressure application of changes in osmolarity.

This is a very important point. As the reviewer mentions, previously published experimental evidence indicates that CSFcNs are also sensitive to osmolarity changes and mechanical stimulation in a PKD2L1-dependent manner. It is therefore reasonable to assume that, as for the pH sensitivity, osmotic and mechanical sensitivity depends on channels segregated to the ApPr. For the mechanosensitivity, the spatial segregation could be tested by “touching” either the ApPr or the soma with a piezo-controlled blunted pipette (see, for example, Hao et al. 2013). However, the sensitivity to osmotic changes is much more difficult to assess, as pressure application does not have enough spatial resolution to discriminate among compartments in such a compact cell such as the CSFcNs. In theory, a highly spatially localized osmotic jump could be reached with photolysis, but a caged compound releasing many osmotic particles simultaneously should be used. In typical photolysis experiments, a localized osmotic jump is produced, but it is very low (in the order of 1 to 2 mOsm).

In mice, like in fish (Sternberg et al, NC 2018), we can observe throughout the figures that a large fraction of the channel activity occurs with partial and very fast openings of the PKD2L1 channel. I recommend the authors analyse the points below:

(a) To what extent do these partial openings of the channel contribute to the changes in holding current and resting potential?

As the reviewer indicates, these partial and very fast openings are a characteristic of PKD2L1 single-channel activity that seems to be present in different species. However, estimating what is the exact contribution of these events to the sustained current would require a detailed model of the channel that it is still lacking. Indeed, the exact mechanism by which CSFcNs show this prominent sustained current is unknown and should definitely being addressed in future works.

(b) In the trace from the outside out AP, it looks like the partial transient openings are gone. Can the authors verify whether these partial openings are only present in somatic recordings?

The outside-out recordings from the ApPr also show some partial openings (please look at the upper trace in Figure 4Db). We will specifically mention this important point in the revised version of the ms.

(3) Previous studies have observed expression of metabotropic Glutamate receptors in CSF-cNs (transcriptome from Prendergast et al CB 2023). The authors only used blockers for ionotropic glutamate receptors in their recordings: could it be that these metabotropic receptors influence the response to uncaging of MNI-Glu when glutamate is co-released with a proton?

We thank the reviewer for pointing out the presence of metabotropic glutamate receptors in CSFcNs. However, our evidence indicates that there is no contribution of metabotropic receptors when uncaging MNIglutamate because: i) the response obtained when uncaging MNI-gLGG (where there is no glutamate release; Figure 5Ab) and ii) the response obtained when uncaging protons from DPNIGABA (a GABA cage that has similar photochemistry than MNI cages which also release a proton when photolysed; data not shown), are the same. Indeed, in both experiments (MNI-gLGG or DPNI-GABA uncaging) a clear photolysis-evoked PKD2L1 current can be observed.

(4) In the outside out patch of the AP, PKD2L1 unitary currents appear rare. Could it be that the disruption in the cilium or underlying actin/myosin cytoskeleton drastically alter the open probability of the channel?

Although we have not quantified it, the reviewer is right that the opening frequency of PKD2L1 channels in the outside-out patches is lower than in the whole-ApPr recordings. We interpreted this difference as a difference in channel number. However, another plausible interpretation is that, as the reviewer suggests, the biophysics of the channels are affected because the protein is taken out from its normal ionic environment and/or loses important interactions with regulatory proteins.

(5) Could the authors use drugs against ASIC to specify which ASIC channels contribute to the pH response in the soma?

As described in the manuscript, we did perform experiments with ASIC channel blockers, although we did not attempt to characterize the specific ASIC channel involved in the somatic response. Based on what has been published in the literature, we used both psalmotoxin-1 (which blocks ASIC1 channels) and APETx2 (which blocks ASIC3 channels). The presence of ASIC1 channels in mice CSFcNs has been shown by (Orts-Del’Immagine et al. 2012; Orts-Del’Immagine et al. 2016), while the presence of ASIC3 in the lamprey CSFcNs has been shown by (Jalalvand et al. 2016). When we puff an acidic solution aiming at the soma, we can record an inward current that is blocked by psalmotoxin-1, although there is always a small component remaining (as originally shown by Orts-Del’Immagine in the aforementioned articles); however, we have not attempted to block this small component that remains after psalmotoxin-1 bath application.

(6) This is out of the scope of this study, but we did observe in fish a very rarely-opening channel in the PKD2L1KO mutant. I wonder if the authors have similar observations in the conditions where PKD2L1 is mainly in the closed state.

We have never seen such kind of openings in our recordings (when the channel is closed or in the presence of dibucaine).

Bjugn, R, H K Haugland, et P R Flood. 1988. “Ultrastructure of the mouse spinal cord ependyma.” Journal of Anatomy 160 (octobre): 117‑25.

Bruni, J. E., et K. Reddy. 1987. “Ependyma of the Central Canal of the Rat Spinal Cord: A Light and Transmission Electron Microscopic Study”. Journal of Anatomy 152 (juin): 55‑70.

DeCaen, Paul G., Xiaowen Liu, Sunday Abiria, et David E. Clapham. 2016. “Atypical Calcium Regulation of the PKD2-L1 Polycystin Ion Channel”. eLife 5 (juin): e13413. https://doi.org/10.7554/eLife.13413.

Delling, Markus, Paul G. DeCaen, Julia F. Doerner, Sebastien Febvay, et David E. Clapham. 2013. “Primary cilia are specialized calcium signalling organelles”. Nature 504 (7479): 311‑14. https://doi.org/10.1038/nature12833.

Hao, Jizhe, Jérôme Ruel, Bertrand Coste, Yann Roudaut, Marcel Crest, et Patrick Delmas. 2013. “Piezo-Electrically Driven Mechanical Stimulation of Sensory Neurons”. In Ion Channels, édité par Nikita Gamper, vol. 998. Methods in Molecular Biology. Humana Press. https://doi.org/10.1007/978-1-62703-351-0_12.

Jalalvand, Elham, Brita Robertson, Peter Wallén, et Sten Grillner. 2016. “Ciliated Neurons Lining the Central Canal Sense Both Fluid Movement and pH through ASIC3”. Nature Communications 7 (janvier): 10002. https://doi.org/10.1038/ncomms10002.

Orts-Del’Immagine, Adeline, Riad Seddik, Fabien Tell, et al. 2016. “A Single Polycystic Kidney Disease 2-like 1 Channel Opening Acts as a Spike Generator in Cerebrospinal Fluid Contacting Neurons of Adult Mouse Brainstem”. Neuropharmacology 101 (février): 549‑65. https://doi.org/10.1016/j.neuropharm.2015.07.030.

Orts-Del’immagine, Adeline, Nicolas Wanaverbecq, Catherine Tardivel, Vanessa Tillement, Michel Dallaporta, et Jérôme Trouslard. 2012. “Properties of Subependymal Cerebrospinal Fluid Contacting Neurones in the Dorsal Vagal Complex of the Mouse Brainstem”. The Journal of Physiology 590 (16): 3719‑41. https://doi.org/10.1113/jphysiol.2012.227959.

Palomero, Orhi Esarte, Megan Larmore, et Paul G. DeCaen. 2023. “Polycystin Channel Complexes”. Annual Review of Physiology 85 (Volume 85, 2023): 425‑48. https://doi.org/10.1146/annurev-physiol-031522-084334.

Prendergast, Andrew E., Kin Ki Jim, Hugo Marnas, et al. 2023. “CSF-Contacting Neurons Respond to Streptococcus Pneumoniae and Promote Host Survival during Central Nervous System Infection”. Current Biology 33 (5): 940-956.e10. https://doi.org/10.1016/j.cub.2023.01.039.

Zhang, Miao, Yueming Ma, Xianglu Ye, Ning Zhang, Lei Pan, et Bing Wang. 2023. “TRP (Transient Receptor Potential) Ion Channel Family: Structures, Biological Functions and Therapeutic Interventions for Diseases”. Signal Transduction and Targeted Therapy 8 (1): 261. https://doi.org/10.1038/s41392-023-01464-x.

Recommendations for the authors:

Reviewing Editor Comments:

Both reviewers were very impressed with your work and definitely feel it is a scientifically important and technically sophisticated study that advances our understanding of CSF sensing. They, however, request some re-analysis of the data and discussion with minimum new experiments, if any. I think, if feasible, this will improve the quality of the study and would look forward to receiving a revised version.

Reviewer #1 (Recommendations for the authors):

(1) Figure 1 - Molecular identity and localization of PKD2L1

Major

Nomenclature clarity.

Please clarify the distinction between PKD2L1 and TRPP3. Several parts of the text and figure labels appear to conflate these names. PKD2L1 corresponds to the TRPP3 subfamily member and should not be interchanged with PKD2/TRPP2. Please confirm and update the nomenclature consistently throughout the manuscript (text, figure labels, and captions).

We have addressed this issue in response to reviewer 1. There seems to be some confusion in the literature concerning the nomenclature of PKD2L1 channels as in some recent publications the PKD2L1 channels are still named as TRPP3. However, the nomenclature of PKD2L1 channels or TRPP2, was updated in 2016 (Wu, Sweet and Clapham, Pharmacological Reviews, 2010). As indicated in response to reviewer 1, we have removed from the text any reference to the TRPP nomenclature and stuck to the PKD2L1 name.

Physiological meaning of apical restriction.

Expand the discussion of why apical-restricted localization matters. Specifically, address how segregation to the ApPr could support directional sensing of CSF flow and/or detection of localized pH gradients.

Following the reviewers and editor’s comments, we have revised the discussion in order to take this and other comments into account.

Open-state annotation (O3).

Please include the O3 state in panels Ba and Bb; the figure clearly shows an additional open level consistent with O3.

The editor is right in that there is another state that presumably corresponds to O3. Following the editor’s recommendation, we now indicate this 3rd level and add a short sentence explaining this in figure 1 legend.

Minor

Indicate the ROI definition and background-subtraction method used for fluorescence quantification (ApPr vs soma).

Not applicable for this figure.

Ensure the same intensity scale (lookup table and range) is used across panels to enable direct comparison.

Done.

(2) Figure 2 - Electrophysiological characterization of ApPr and somatic recordings Major

Definition of "PKD2L1-dependent current."

Define this term precisely at its first appearance. Specify whether it denotes currents inhibited by dibucaine, abolished in PKD2L1-knockout preparations, or both.

Done

Statistical power of single-channel analysis.

The number of observed openings (< 1000 events) is too low to estimate open probability (Po) reliably. Please re-analyze the data using macroscopic current traces rather than Po-based kinetics.

Confounded Po analysis.

The current Po analysis mixes current amplitude and Po in the same calculation, conflating independent variables. Re-evaluate or remove this analysis.

Unknown channel count.

Because the number of channels in each patch is unknown, Po and "closed probability" values cannot be interpreted meaningfully. Focus instead on the averaged macroscopic current density.

General analytical validity.

The single-channel analyses in Figure 2 are not interpretable under these experimental conditions. Closed-time distributions and Po-based metrics (e.g., "Po1," "P2") depend critically on channel number and event sampling. Moreover, the manuscript applies essentially the same Po methodology across conditions (Po1 vs P2), which adds no mechanistic resolution and risks circular interpretation.

Actionable recommendation:

Remove Po- and closed-time-based analyses from Figure 2 and from the manuscript as a whole. Reanalyze the data using metrics that remain valid when the channel number is unknown:

Macroscopic current analysis (leak-subtracted current density, I-V relationships, activation time constants).

Single-channel conductance only (amplitude histograms and unitary slope conductance), without attempting Po or dwell-time inference.

Report filtering bandwidth and sampling rate, and restrict statistical treatment to these robust parameters.

Following the reviewers (see above) and editors’ recommendations, we have reanalyzed the data in order to avoid the analysis based on Po and Pc. We have instead calculated from the recordings other 2 parameters, n_max (the maximum number of channels that open simultaneously during a 500 ms time window) and the total open time of a single channel during the same 500 ms time window. The main text, figures and corresponding figure legends, and the Materials and Methods section have been changed accordingly. Notably, the Po and Pc analysis were removed from Figures 3, 4 and Supplementary Figure 3, and replaced by the above-mentioned parameters. Also, the fact that the recordings are not long enough to calculate Po is now specifically mentioned in the Materials and Methods section, lines 785 to 790. In addition, the analysis in Figure 3Ce has been redone so that the activity of the channel as a function of pH is now plotted as the normalized apparent Po (relative to the apparent Po value at pH 7.4).

Minor

State whether input-resistance values (1.8-4.4 GΩ) were leak-subtracted and series-resistance-compensated.

As already mentioned in the methodology section (line 693), series resistance was not compensated for during the experiments. We have now added a sentence in the methodology section indicating that in the voltage range that was chosen for the analysis of the input resistance, no voltage-dependent conductance was activated (lines 809 to 812).

Ensure unit consistency: use Po or normalized Po rather than frequency (Hz) throughout.

(3) Figure 3 - pH-evoked currents and kinetics

Major

Invalid Po analysis (Fig. 3Ca-Ce).

The Po- and Pc-based single-channel analyses in panels 3Ca-3Ce should be deleted. As noted earlier, the event count is insufficient, and the number of active channels in each patch is unknown. Under these conditions, Po and Pc values have no quantitative meaning and could mislead readers. These panels do not contribute additional mechanistic insight beyond the macroscopic current data and therefore, should be removed. If retained for illustrative purposes, they must be explicitly labeled as representative traces without any statistical quantification.

As we mentioned above, we removed the Po and Pc analysis from the manuscript.

Minor

Present regression equations and r² values for the linear fits shown in Fig. 3D.

Done (lines 947951).

Confirm that all axes include units and identical scaling between conditions for direct comparison.

Done.

(4) Figure 4 - Laser photolysis and local stimulation experiments

Major

Laser timing annotation.

Clearly mark laser-pulse timing (e.g., arrow or shaded region) on all current traces to facilitate interpretation.

We thank the editor for pointing out the inconsistencies in terms of the laser pulse timing. To indicate the laser pulses, we have now added an arrowhead in cases where a single sweep is shown (for example, Figure 3D), and an arrowhead and a dotted magenta vertical line in cases where multiple sweeps are shown (for example, Figure 3C).

pH calibration within the laser spot.

Provide quantitative calibration of pH changes induced by laser photolysis, including information on spot size, local diffusion, and estimated pH recovery kinetics.

This is an important point and we thank the editor for mentioning it. We have now completed the subsection untitled “photolysis” where we provide information on the lateral and axial dimensions of the photolysis laser spot used in this work (lines 734 to 737). We have also rewritten part of Figure 5A legend to highlight the fact that the experiments presented there (photolysis on top of the ApPr and next to it) are compatible with a high spatial resolution of proton release (lines 1023 to 1026).

On the other hand, we have attempted to perform pH calibrations in the setup using the pH-sensitive dye pyranine (or HPTS: 8-Hydroxypyrene-1,3,6-trisulfonic acid). HPTS is a very useful tool for pH calibrations in the physiological range: its pKa value is close to 7.2, and it can be used as a ratiometric dye (its fluorescence is pH-independent at 405–410 nm and pH-dependent at 450 nm). Unfortunately, when trying to perform a calibration under the conditions of a real experiment,

where photolysis occurs in a tiny volume (approximately 1 µm³ in a total bath volume of more than 1 ml), we encountered the following problem, which made it impossible to obtain any useful data: the 405 nm uncaging pulse bleaches the dye, and any useful information is lost. Also, our imaging system is not fast enough to follow the pH change. As it is discussed in the Materials and Methods section, subsection “Estimation of the pH drop induced by photolysis” (line 814), the fast protonation of bicarbonate indicates that the pH change induced by the photolysis recovers in the submillisecond time range.

Repeated stimulation effects.

Discuss whether repeated photolysis induces adaptation or desensitization of PKD2L1 currents, and indicate whether current amplitude decreases across successive trials.

This issue is now specifically mentioned in the Materials and Methods section, lines 739 to 741.

Invalid interpretation of the "OFF response."

The interpretation of the so-called "OFF response" in Figure 4C is not supported by the presented data. There is no evidence for a bona fide OFF current, and the literature cited does not demonstrate such a phenomenon for PKD2L1 alone. Rather, previous studies implicate PKD1L3-dependent mechanisms in similar biphasic responses. Please reconsider the cited references and remove claims of an OFF current attributed to PKD2L1.

Done.

Actionable recommendations:

Do not use the term "OFF response" throughout the manuscript. Recast these transients as pH dependent recovery or relaxation of current following cessation of acidification.

Done. We have performed extensive rewriting and reorganization of the Results and Discussion in order to take into account both the reviewer’s and editor’s comments. Please also take a look at comment #3 of Reviewer 1 and point 11 below.

Include continuous-illumination controls (sustained local acidification) to test whether a steady state current is maintained. This will clarify whether the post-stimulus transient reflects recovery kinetics rather than a distinct current species.

We thank the editor for suggesting this experiment. However, continuous laser illumination is a difficult manipulation and does not necessarily lead to an acidification of the illuminated volume. Indeed, with continuous illumination the cage is lost from the illumination spot and needs to be replaced by diffusion from the non-illuminated volume, leading to non-homogeneous concentrations. Also, the chances of inducing photo damage are higher. We thus designed a similar experiment where instead of performing continuous illumination we photolysed with short and high frequency trains in order to produce a long-lasting acidification. The results of these experiments have been added to the manuscript as part of the results section and in Figure 5H. Similarly to what is seen with single illuminations, the photolysis trains induce a current that appears almost exclusively at the end of the train, implying that the current is indeed a PKD2L1-dependent recovery current.

Align the current time course with measured or estimated local pH (or calibrated proxy) to demonstrate causal coupling and avoid implying a separate conductance.

We have added the calculated pH change to the inset of Figure 4C as an example.

Revise the schematic/model figure and textual description accordingly, restricting the framework to phasic vs sustained activation modes without invoking a separate OFF current for PKD2L1.

Done.

Minor

Include scale bars, sample numbers (n), and laser parameters (duration, power) in all panels.

In order not to make the figure and the panels very heavy in the original version, we tried to limit the number of scale bars. We have now performed some modifications, added the missing scale bars, and changed the figure legend in order to take into account the editor’s comments. We have also corrected a few values that were wrongly reported.

Standardize p-value formatting (e.g., p = 6 × 10 ⁶) throughout the figure and legend.

Done.

(5) Figure 5 - Single-channel recordings

Major

Mixed parameters (current amplitude and Po).

The current analysis improperly mixes single-channel current amplitude and Po within the same figure, conflating distinct parameters. These quantities must be analyzed and presented separately, or the Po data should be removed entirely if not independently supported.

Insufficient event count.

Given the very limited number of observed openings, Po-based statistics are not meaningful. Please report only representative single-channel traces and corresponding amplitude histograms without attempting quantitative Po estimation.

Minor

Convert frequency (Hz) values to Po for consistency with earlier analyses, or remove frequency metrics altogether if Po analysis is omitted.

Figure 5 does not include Po or event frequency analysis, so we think there must be a misquotation of the figure. However, the Po issue has already been addressed before and alternative analysis have been proposed.

(6) Introduction

The introductory paragraph mentions the "five senses" as a framing concept. However, this statement lacks scientific grounding in the context of CSF-contacting neurons and chemosensory physiology. The traditional "five senses" classification is not an evidence-based neurophysiological framework and may be misleading to readers. I recommend removing or rephrasing this part, focusing instead on molecular and cellular mechanisms of sensory transduction (e.g., chemical, mechanical, and pH sensing) rather than on classical sensory categories.

Following the editor’s recommendation, we have removed this part.

The manuscript refers to PKD2L1 using the term TRPP2 in some parts of the introduction. This is incorrect, as PKD2L1 corresponds to TRPP3, not TRPP2. Please correct this nomenclature and ensure consistent use of "PKD2L1 (TRPP3)" throughout the entire manuscript to avoid confusion with the distinct PKD2/TRPP2 protein, which belongs to a different subfamily with separate physiological roles.

We thank the editor for pointing this out. As we mentioned in the responses to the “public reviews”, the literature is confusing, so we decided to remove from the manuscript any mention to TRPP channels.

(7) Discussion

The current Discussion reads largely as a descriptive summary of results and lacks conceptual depth. It does not effectively integrate the biophysical properties of PKD2L1 with its physiological role as a neuronal pH sensor, nor does it develop a broader interpretation relevant to CSF homeostasis or chemoreception.

Following the reviewers and editor’s recommendations, we have now added a new section in the Discussion untitled “PKD2L1 downstream signaling mechanisms”.

(8) Insufficient biophysical analysis

The discussion of channel gating and pH dependence is superficial and does not explore the energetic or structural mechanisms underlying proton sensitivity. The authors should analyze their data in the context of known PKD/TRPP family biophysics-for example, protonation sites, subunit composition, or gating kinetics-and explain how these confer bidirectional (acidic vs alkaline) sensitivity within physiological ranges.

In this work, we studied the pH sensitivity of PKD2L1 channels in the context of CSFcN sensory physiology. From a pure biophysical perspective, the pH sensitivity of PKD2L1 channels has been studied by multiple groups; however, it is still unknown how the gating of the channel responds to pH changes, although it can be proposed that some polar residues in the protein regulate the state of the pore. Likewise, the mechanism of the “off-response” is also unknown. To the best of our knowledge, there is only one article in which the authors have attempted to relate pH, PKD2L1 channel structure, and function. In this work (Su et al., Nature Communications 2018), the authors compare PKD2L1 channels with another pH-sensitive member of the TRP family, TRPML3, whose structures at pH 7.4 and 4.8 are known (Zhou et al., Nature Structural and Molecular Biology, 2017). We have rewritten some sentences of the Discussion in order to be more specific about the pH dependence of PKD2L1 channels and its proposed mechanisms.

(9) Weak physiological context

The manuscript does not adequately address how PKD2L1 functions as a true physiological pH sensor. The discussion should connect channel activity to realistic CSF pH fluctuations (6.8-7.6) and to relevant physiological or pathophysiological conditions (e.g., respiratory acidosis, neurogenic regulation of CSF composition). Without this, the relevance of large, artificial acidification (pH 3-3.5) remains unclear.

We have added a new section in the Discussion where we speculate on how PKD2L1 channels may be activated in physiological and pathophysiological conditions. However, we would like to insist here that the main goal of the photolysis experiments (which induce short and large acidifications) was to assess the spatial segregation of PKD2L1 channels. We now mention this point specifically and also speculate on the conditions that could eventually give rise to the “recovery” current.

(10) Over-interpretation of unsupported points

The paragraph describing voltage propagation from the ApPr to the soma/axon is speculative and unsupported by any data in the manuscript. Please delete this section entirely, including the citation to Orts-Del'immagine et al., unless new electrophysiological evidence is added.

We think this point (the propagation of signals originating from the ApPr to the soma) is important in the context of our work, so we have decided to make new experiments in order measure directly the degree of coupling between the 2 compartments. To do that we made simultaneous, current-clamp and voltage-clamp recordings from the ApPr and the soma. In these conditions we were able to measure experimentally and for the first time both the coupling coefficient and coupling conductance, which confirm that the propagation of voltage signals from the ApPr to the soma is extremely efficient. These new results are now described in a new subsection and in a new Figure 6.

(11) Clarify the role of "OFF currents."

The Discussion repeatedly refers to an "OFF response," but this phenomenon is not experimentally demonstrated for PKD2L1 alone. It likely represents pH-dependent recovery rather than an independent current. All discussion of "OFF currents" should be removed or reformulated accordingly.

Following the editor and reviewer’s comments, we have deleted the term “off response” and “off currents” from the ms and have replaced them with the term “recovery current”. We have also changed the discussion accordingly.

(12) Integration with ASICs and compartmental sensing

While the manuscript briefly mentions ASIC involvement, it does not articulate how PKD2L1- and ASIC-mediated signals might complement each other in different compartments (ApPr vs soma). The authors should discuss the potential division of labor between these sensors and how such compartmentalization enhances pH detection in CSFcNs.

Following the editor’s comments, we have rewritten the part of the subsection ‘the involvement of ASICs’ in the Discussion.

(12) Broadened physiological perspective

The Discussion should close by considering Ca²⁺ -dependent downstream pathways activated by ⁺ PKD2L1 and their implications for CSF flow regulation, neurosecretion, and central chemoreception. These translational aspects would substantially improve the impact and readability of the manuscript.

Done

Overall, the Discussion must evolve from a descriptive narrative to a mechanistically and physiologically integrative synthesis, highlighting why PKD2L1 is not merely present in the ApPr but is a key molecular transducer linking ionic microenvironment to neuronal excitability.

As it has been detailed above, we have performed several changes in the Discussion that follow the reviewer’s and editor’s recommendations.

eLife Assessment

This valuable work introduces HSSM, a flexible and modular Python toolbox for hierarchical Bayesian estimation of sequential sampling models. By combining model simulation, neural likelihood approximation, established inference backends, formula-based regression, and support for trial-level covariates, HSSM could become a widely used resource in cognitive and computational neuroscience. Evidence for the framework's scope and core functionality is convincing, but support for its usability, computational performance, and reliability is incomplete. The contribution would be substantially strengthened by an executable end-to-end analysis, clearer exposition for non-specialists, parameter and model recovery demonstrations, guidance on diagnostics and model comparison, quantitative benchmarks, and a clearer delineation of HSSM's scope relative to broader claims about neurocognitive modelling.

Reviewer #1 (Public review):

Summary:

This article describes a new software package, HSSM, for simulating and fitting sequential sampling models. The package consists of three modules - one for simulating the models, one to train neural networks on mappings from behavior to parameter values, and one for combining these tools to fit particular models to users' data.

Strengths:

This is a very detailed description of a new package that is building on an already highly successful package. It promises to become a go-to software package for cognitive modelers, experimentalists, and practitioners.

Weaknesses:

I have only a few critiques of the article, and some are a matter of taste:

(1) I think it would be helpful for the authors to take the reader through one complete example at the end of the article, including loading in a dataset, fitting it with a regression model, checking model convergence, reading out parameter values, and doing posterior predictive checks (etc). It would be helpful to see it all in one place to get a sense of how much code is required to go through the whole process. One or two actual examples would help readers who are not already familiar with HDDM.

(2) The article assumes a certain level of computing and modeling expertise, which somewhat limits its reach. There are many abbreviations and references to other software tools that a reader might not be familiar with. Such readers might feel like HSSM is beyond their reach. Below is a non-exhaustive list of such undefined or unexplained terms:<br /> DDM, API, LBA, fMRI, EEG, JAX, PyTorch, ONNX, MCMC, VI, MAP, PyMC, PyTensor, NUTS, ArviZ, WAIC, LOO, KDE, CLI, YAML, GUI, LBA, RDM, QP.

(3) I found some of the figures/listings to be unpolished, unhelpful, and/or unnecessary. For instance, Figure 3 and Listing 3 seem to just be zoomed-in pieces of Figure 2. In Figure 4, what is the meaning of the little globe traveling between the within-trial and across-trial rows? Why is Figure 5 a figure and not a listing? Does HSSM not produce these plots directly? I have the same question for Figure 7. Figures 8 and 9 seem unnecessary to me, but perhaps they serve a function that I'm missing. Perhaps they could be turned into supplements for Figure 2.

Reviewer #2 (Public review):

Summary:

This manuscript introduces HSSM, a Python-based toolbox for fitting cognitive models, with a specific focus on sequential sampling models. The toolbox brings together several components: a model construction interface, surrogate likelihoods, sampling tools, an inference backend, formula-based regressions, and tools for validation and visualization.

One of the key advantages is that HSSM relies on well-established open-source packages. This ensures both a robust foundation and also opens a potential for future (community-driven) development. While HSSM is not the first publicly available toolbox for fitting sequential sampling models, it introduces several novel features that will be very valuable to researchers in the field.

Strengths:

The biggest strength of HSSM is its flexibility and ease of use for hierarchical modeling. Many existing toolboxes work as closed systems that are hard to modify. In contrast, HSSM's modular design allows it to be used either as a stand-alone tool or to pick out specific components to integrate into existing pipelines. In addition, the toolbox combines simulation-based inference, surrogate likelihoods, and formula-based regression. This opens up a lot of new modeling possibilities and makes it easy to incorporate trial-by-trial neural or physiological covariates alongside standard RT and choice data.

Weaknesses:

The paper provides a high-level overview of the toolbox, rather than a didactic walk-through that shows how to use it in practice. Additionally, despite being framed as a broad toolbox for "neurocognitive modeling", HSSM currently focuses on sequential sampling models. While these models are widely used, they represent only a small slice of neurocognitive modeling as a whole. Additionally, the toolbox is currently in beta phase, and lots of planned extensions are not implemented yet. Finally, there is currently no information on general performance benchmarks.

eLife Assessment

This study provides a useful single-cell atlas of the Clytia hemisphaerica planula, complemented by an updated medusa dataset, ultrastructural analyses and in situ validation of expression patterns. The cross-stage comparison and cluster-similarity framework offer a promising basis for investigating cellular diversification across the life cycle. The evidence is currently incomplete due to documentation deficits and insufficient cross-referencing with prior work. It should be relatively easy to address these issues, which should make this a valued resource for cnidarian researchers and for colleagues studying cell-type evolution.

Reviewer #1 (Public review):

Summary:

The manuscript further explores the single-cell atlas of Clytia hemisphaerica by incorporating the planula larva. It compares the cell clusters with the previously established atlas of the medusa. It identifies similarities and differences between the two life stages.

Strengths:

The manuscript provides an important set of single-cell data that have not been assessed previously: the Clytia planula. The data is further supplemented with high-quality EM-based histology and an extensive in situ hybridisation of selected genes.

Weaknesses:

The detailed analysis does not go deep into the comparison between stages, nor does it provide an analysis of genes within the clusters; it could be described as remaining overall rather superficial.

Reviewer #2 (Public review):

Summary:

The generation of alternate stages in the life cycle of a single species requires vast remodeling of the cellular complement of the individual during metamorphosis from one stage to another. In this paper, the authors provide a detailed description of single-cell RNA-seq data derived from the planula stage of the hydrozoan model Clytia hemispherica and compare this to an expanded dataset from the medusa stage to assess changes in transcriptomic identity of cell types between these two phases of the life cycle. The paper further includes valuable TEM data illustrating fine anatomy of cell types present at both stages investigated, and documentation of the retention of epithelial polarity from the planula through to the polyp stage, using a reporter line.

Strengths:

The study provides a solid and convincing transcriptomic characterization of planula cell types (including in situ validations and a planula-to-polyp mapping of epithelial polarity), and introduces a potentially valuable method for evaluating cluster similarity.

Weaknesses:

The work suffers from insufficient documentation of methodological approaches and missing code, lack of clarity regarding clustering resolution and nomenclature (thereby hindering cross-referencing with prior papers), and unclear plans for public, fully annotated data release.

Full Review:

The single-cell transcriptomic data analyzed include both previously published and newly generated data: two additional medusa libraries and two additional planula libraries were generated and integrated with the data from https://doi.org/10.1126/sciadv.abh1683, and https://doi.org/10.1126/sciadv.adv1159. The original release of the planula dataset in their 2025 Science Advances paper did not include analyses of all cell types. Here the authors provide this analysis for the planula stage. However, as both the number of clusters and the nomenclature of the clusters changed, this leads to some confusion and inability to cross-reference the two papers. There is no explanation given for the re-processing of the planula dataset in the current paper, and the fact that only some of the data is new is buried in the supplement, which is not referenced in the main document, while the text within the main article suggests that the entire dataset is new. The fact that the dataset in the current analyses contains fewer cells than presented in their Science Advances paper further adds to this confusion. The current paper would benefit from greater transparency in the origin of the data analyzed.

The authors do try to apply the same nomenclature for the updated medusa dataset that is present in their 2021 Science paper. For example, the previously identified 'bioluminescent cells' are identified as 'gas-m8'. A look-up table that has all of the cluster id's cross-referenced would be useful (i.e. new: 8 = gas-m8 = previous: 28 = BC = "Tentacle GFP cells"). The inability to easily cross-compare with the published data is a major weakness of the current work and would benefit greatly from consistency between the three papers. Indeed, the clustering resolution is quite different across all three papers, and the current work does not adequately address how the clustering resolution was selected here. As an updated atlas, one would expect the entire transcriptomic diversity to be included here, so that the previous work can be transferred to the updated genomic mapping resource used in the current work. Nonetheless, presenting a unified nomenclature for moving forward would benefit the community as a whole and would increase the impact of the current work substantially.

The paper also includes new TEM data of the planula cell types. The authors attempt to correlate transcriptomic profiles with these anatomical data through in situ hybridizations that provide spatial distribution of the profiles. While the TEM data are valuable to catalog the presence of cells with different morphologies within the planula, the association with the transcriptomic profiles is somewhat speculative. These valuable anatomical data should be provided at a high enough resolution to zoom in and see the details, and further description could be provided. For example, the paper states that vacuolated cells are characteristic of the basal gastrodermal cells adjacent to the mesoglea; please identify the vacuoles in Figure 4f/h for the reader.

A novel method for reconstructing cluster similarity relationships is applied to grouping clusters into cell categories within the same life cycle stage, and also for matching cell types between stages. This is a valuable contribution to the field that is worthy of further evaluation. This is, however, difficult, as the methods for which DESeq2 was applied ("see code for details") are not present in the provided code, nor is it adequately described how the "binary matrix of marker gene presence/absence" was constructed. Similarly, there are additional details of other parts of the data analysis that are missing from the provided code, and the provided supplementary material is not referenced in the main document. More rigorous documentation of the methods is warranted.

The description of the transcriptomic profiles present in the planula is solid, and the attempt to associate these profiles with anatomic locations and putative morphology provides a foundation onto which further studies can be developed. Mapping of the planula ectoderm through to the polyp stage is also an important step forward in characterizing the life cycle, and the evidence for the retention of the oral/aboral ectodermal axis is convincing. The paper falls short in describing the updated medusa dataset and could benefit from a minor restructuring of the paper. Introducing the new medusa data only after the planula dataset is fully described would mediate the shallower treatment of the updated medusa dataset, where only 22 of the original 36 transcriptomic states are recovered. In this way, the focus will shift onto the cross-life cycle stage comparisons, and it could be argued that the lower resolution of the medusa dataset is justified in order to simplify the comparisons.

It will be essential that the datasets that are presented in this work be made available for public exploration in a fully annotated format. It is currently unclear how the authors intend to do this; however, there are many repositories available for this. The UCSC Cell Browser hosted at cells.ucsc.edu is one very good option if the authors do not wish to develop an interactive tool themselves. It is imperative that the gene annotations which correspond to the dataset, and the cluster annotations that are presented in this paper, are available and easily connected to the released dataset.

eLife Assessment

This important study presents an ultrastructural atlas of extracellular vesicles and non-vesicular particles in Drosophila olfactory sensilla, using cryofixation-based serial block-face electron microscopy to catalogue roughly 7,800 particles across four antennal regions. The evidence is compelling: the cryofixation minimizes fixation artifacts, and the rigorous, well-validated methodology and morphometric analyses provide strong support for the structural classifications and spatial distributions, though some conclusions require moderation or additional evidence. The work provides a potentially foundational resource for researchers studying extracellular-particle signaling in sensory organs.

Reviewer #1 (Public review):

Summary:

Using cryofixation and serial block-face electron microscopy (SBEM), P. Vijayakumar and K. Cauwenberghs characterize extracellular vesicles (EVs) and non-vesicular extracellular particles (NVEPs) within native Drosophila olfactory sensilla. The study provides a unique and valuable dataset comprising approximately 7,800 extracellular particles, systematically describing their morphology, size, density, and distribution. The ultrastructural analysis across different sensillum classes offers insights into the potential biogenesis and functions of these extracellular particles.

Strengths:

Cryofixation preserves EVs and NVEPs within native tissue conditions. The detailed quantification of a very large dataset provides a unique source of information on extracellular particle number, categories, and distribution. The expertise of the group in the method and the tissue explored, as well as their detailed quantification, provide confidence in the dataset and observations.

Weaknesses:

Major comments

(1) As the authors state, the rare observation of EV budding or MVB release events suggests that these are transient processes, whereas EVs and NVEPs are retained for relatively long periods within the sensillum lumen. The current analyses may overinterpret steady-state vesicle abundance as differences in vesicle production.

(2) Given the above conclusion, differences between sensillum classes may be somewhat overstated.<br /> a) The absolute number of EVs and NVEPs per sensillum is highly variable, even within the same sensillum class (Figure 3D). For example, a substantial proportion of coeloconic sensilla have an empty lumen (Figure 3C). Consequently, expressing the data as ratios or proportions (Figures 3B, 4C, and 4E) may exaggerate differences between sensillum classes and should therefore be interpreted with caution.<br /> b) The rate of EV/NVEP production is unknown. For a similar rate of production across sensillum classes, Figure 3E suggests that the differences in lumen morphology and size may largely explain variation in EV density and distribution.

That said, I agree that ab1 sensilla display a striking enrichment of large cargo-filled EVs compared with the other sensillum classes, while coeloconic sensilla display enrichment in small dense filled EVs (Figure 4C). Together, large and cargo-filled EV observation provides strong support for differences in EV biogenesis between ab1 sensilla and the other sensillum classes. In that context, I also think the EV size distribution shown in Figure 4 - Figure Supplement 1 should be moved into the main figure, as it demonstrates that the majority of EVs in the ab1 lumen are relatively large and are therefore likely to represent microvesicles. Could you clarify why ab1 sensilla are only included in Figure 4 and not analysed in Figure 3?

(3) Approximately 10% of ORNs appear to be degenerating in 6-8-day-old flies, which seems unexpectedly high. This contrasts with the relatively infrequent occurrence of auxiliary cell apoptosis or complete sensillum degeneration. In these "degenerating ORNs", the authors state that the hallmarks of ORN apoptosis are restricted to the dendrites. As hallmarks, they state dendrite truncation, fragmentation and blebbing. Rather than apoptosis, I wonder whether these observations might instead represent ciliary truncation and ectosome shedding, followed by degradation of the shed ciliary membrane into EVs. Ciliary truncation and ectosome shedding, followed by ciliary regrowth, are dynamic processes that have been described across multiple species. This interpretation could explain large EVs that remain in the lumen long after the cilium has regenerated. It would reconcile this article with the general agreement that cilia are a prime site for the budding of EVs across species. Additional evidence supporting apoptosis of the ORNs would help distinguish between these possibilities. Otherwise, I believe the author should reconsider their interpretation.

Reviewer #2 (Public review):

Summary:

This paper presents a large structural survey of extracellular vesicles (EVs) and non-vesicular extracellular particles (NVEPs) in the olfactory sensilla of Drosophila melanogaster. Using high-pressure freezing and serial block-face SEM, the authors avoid many of the artifacts associated with conventional fixation and analyze more than 7,800 particles across 352 sensilla. The manuscript maps the distribution of these particles, describes their morphological heterogeneity, and examines their likely origins across different sensillum classes in both normal and degenerating tissue.

Strengths:

The strongest aspect of the paper is the imaging. Preservation is painstakingly controlled. The cryofixation appears to preserve the sensillum lymph in a more convincing native state than standard preparation methods, giving this work gravitas. Further, the authors characterized thousands of particles, further making this data strong.

The figures are strong. They are clear, easy to read, and generally well designed; I think people will use this paper as a model for how to present complex data in a concise and straightforward manner. The manuscript is careful in how it presents the dataset and does not overinterpret the descriptive observations. As an ultrastructural resource, this paper will be useful to the field. The identification of auxiliary support cells as major secretory sites, together with the striking accumulation of EVs in degenerating tissue, will provide a useful starting point for future work.

Weaknesses:

The main point that could use more clarification is the vesicle categorization. In particular, the distinction between "dense," "cargo-filled," and "double EVs" is not always easy to follow from a biological perspective. Some additional discussion of how the authors think these categories relate to one another, and whether they are intended as purely morphological groupings or as distinct biological classes, would strengthen the manuscript.

Reviewer #3 (Public review):

Using cryofixation-based serial block face electron microscopy of several subregions of the Drosophila antenna, the authors segment and assemble a high-resolution atlas of the anatomical structure, density, and spatial distribution of extracellular vesicles (EVs) and non-vesicular extracellular particles (NVEPs) in different Drosophila olfactory sensilla types. This systematic and thorough description is an important prerequisite to understanding the function of extracellular particles in intercellular signaling in the nervous system. Additionally, they describe examples of putative biogenesis events (budding/fusion), as well as neuronal and axonal cell degeneration events and measure the changes in particle accumulation in these altered microenvironments.

Overall, this is a significant and comprehensive analysis and represents an invaluable resource to this burgeoning field. The authors assemble an important dataset and their claims match the level of evidence provided.

Strengths:

(1) The authors use segmentations from four different patches of the antenna to provide a systematic ultrastructural survey of extracellular particles in native insect sensilla. The dataset captures the diversity of sensilla types and reconstructs ~7800 particles.

(2) We commend the authors for making the EM volumes available in the public Cell Image Library with accession numbers. It would be helpful to the community to also make the segmentations for this great resource easily accessible.

(3) The sample preparation technique appears to minimize typical artifacts associated with chemical fixation, as evidenced by the high reported sphericity of EVs.

Specific points:

(1) In Figure 3C, the authors should include a continuous measure of particle distribution in the sensilla. Currently, the authors define three categories of particle localization. In the five examples shown in Figure 3A, the spatial distribution of these particles appears quite distinct across classes. For example, EVs/NVEPs in large and small basoconic sensilla are largely restricted to the area proximal to the base, with a limited number located more distally. In contrast, intermediate sensilla show a marked concentration of particles more distally.

(2) The conclusion of different EV ratios across sensillum classes stems from a Kruskal-Wallis of p = 0.0476, with none surviving pairwise comparisons. This is not a strongly supported conclusion and is probably better characterized as a trend.

(3) The statement "selective enrichment of large, cargo-filled vesicles within the ab1 lumen suggests specialized EV-mediated communication adapted to the coordination demands of this neuronal population" seems speculative for a Results section without supporting functional evidence. It would seem better suited for the Discussion.

(4) Figure 4: Criteria for defining the classes of EVs.<br /> a) The authors should explain the rationale for classifying EVs using relative density rather than absolute density? We would expect EVs with similar contents to have similar electron density (similar darkness in the images). Would classifying them relative to the background, which itself might vary across sensilla or regions, create a possible confound, especially when comparing across sensillum classes?<br /> b) The two example images (in Figure 4A) of the cargo-filled EVs appear to have different densities themselves. Do the cargo-filled ones also display systematic differences in density and, if so, why is this another class instead of being a subcategory within the dense and lucent classes (i.e. dense with/without cargo, lucent with/without cargo)? The dense and lucent classes are defined by their density, whereas this is a more structural property.<br /> c) Regarding "Double" and "Ball-and-Socket" EVs, does the density vary between the two particles involved (e.g., does the inner structure consistently differ in density from the outer)?

(5) What was the rationale for the 200μm and 1000μm size cutoffs? A continuous distribution of maximum particle sizes would provide a clearer understanding of the data.

eLife Assessment

The authors investigate the activation mechanism of a gasdermin pore-forming effector of Callorhinchus milii, which is a shark that is one of the most basal members of the cartilaginous fishes. The valuable findings show that GSDMA/B can be cleaved by a lipopolysaccharide-activated caspase here termed CmiCASP1, analogous to the cytosolic LPS detection pathway already established in mammals. The strength of evidence supporting these conclusions is solid; however, the data indicating the functional consequences of these proteins remain incomplete.

Reviewer #1 (Public review):

Summary

The authors present a valuable study of the gasdermins and caspases encoded by Callorhinchus milii, a shark that is one of the most basal members of the cartilaginous fishes. C. milii encodes GSDME and PJVK as well as another gene here called GSDMA/B (which has also been termed GSDMEc in other work). This latter gene is the ancestral gene for bird/reptile/amphibians GSDMA that, in turn, is the ancestral gene to mammal GSDMA, B, C, and D. Prior work had shown that more ancient animals have only GSDME and PJVK, and in these animals caspase-3 and caspase-1 can both independently cleave GSDME. Prior work had also shown that in birds/reptiles/amphibians, GSDMA is cleaved by caspase-1, and GSDME is only cleaved by caspase-3. Here, the authors demonstrate that the more ancient C. milii gene is similar to the bird/reptile/amphibian GSDMA in that it is cleaved by caspase-1. They further demonstrate that C. milii does not encode inflammasomes that would activate CmiCASP1, and instead this caspase is an LPS sensor through its CARD domain analogous to mammal caspase-4/5/11. The data supporting these conclusions are convincing, and could be strengthened by primary cell studies from C. milii in future studies. They further provide evidence that this gasdermin can kill bacteria directly, but the data supporting this conclusion are incomplete.

Strengths:

The data demonstrating that CmiCASP1 is an LPS sensor via its CARD domain is thorough and convincing.

The data demonstrating that CmiCASP1 cleaves and activates GSDMA/B and that this causes pyroptosis is also thorough and convincing.

Weaknesses:

I think that the gene/protein referred to in this paper as GSDMA/B was in prior publications called GSDMEc (doi 10.3389/fcell.2022.952015). Is this correct? If not, the relationship or lack thereof to GSDMEc needs to be described. If the authors wish to rename the gene, this needs to be justified and discussed clearly. Also, a gene name with a slash is not typical and was initially confusing to me as it made me think the authors were referring to two different genes.

The authors do not have data from primary cells from C. milii to demonstrate that the LPS sensing by CmiCASP1 and the pyroptosis induction by GSDMA/B is relevant in the native cell types. This is a common limitation in publications that seek to study diverse animals where tools may not be available. This issue can be studied in future publications.

The ability of gasdermins to kill bacteria is controversial.

This bactericidal effect was first shown by the cited article Liu et al. 2016 from Judy Lieberman's lab. I reviewed that manuscript at Nature, and I implored the authors to remove that data from the paper because the experimental design had a high risk of not being physiologically relevant. Indeed, my lab had previously published that bacteria survive the process of pyroptosis and they must be killed by secondary efferocytic phagocytes attracted to the pyroptotic corpse (doi 10.1084/jem.20151613). I have continued to consider whether gasdermins could kill bacteria over the decade since that 2016 paper, and wrote a detailed argument describing how this is unlikely to be physiologically relevant in a recent review article (see Box 3 in doi 10.1038/s41564-026-02272-z).

In the author's current manuscript, the experiments performed show a very mild effect in the linear range of a reduction of perhaps 20% of the control bacteria in Figure 5A. This is a minimal effect compared to antimicrobial peptides, which will reduce colony-forming units by 99.9%. Take a look at the magnitude of effect in Figure 1 of an example paper looking at polymyxin or colistin killing of Acinetobacter (doi: 10.1128/AAC.00756-12), where CFUs are reduced by about 3 logs (1000-fold) in 30 minutes. The magnitude of effect in Figure 5A is not even 2-fold. Further, it would be very challenging to determine whether the concentration of gasdermin protein used in the assay is equivalent to the concentration that exists in cells. The methods section needs to be clearer to explain how many effective cell lysates of 293T cells were exposed to how many bacteria, because these concentrated lysates are of unspecified concentration.

Regarding cardiolipin binding, this is a lipid that has a small head group attached to 4 lipid chains, resulting in a cone-like shape with the polar groups at the cone tip and the lipids forming the wide cone base. As such, it creates a larger lipid area than polar area, thus naturally creating a curved membrane shape such that cardiolipin is in the leaflet of the interior of a curvature. Thus, in the mitochondria, it exists in the inner membrane in the mitochondria and allows for the bends that form the cristae, where it faces the surface that is concave (nicely diagrammed in Figure 1 of doi 10.3390/biom16010071). Similarly, cardiolipin enriches in the inner leaflets of membranes at the poles of rod-shaped bacteria to allow for the curvature of the membrane at the poles. Therefore, cardiolipin is not exposed in the outer leaflet of the outer membrane of Gram-negative bacteria; instead, the primary lipid in the outer leaflet is LPS.

A competing mechanism that could explain the results is that the opening of gasdermin pores in eukaryotic cell plasma membranes occurs concomitant with the generation of ROS from mitochondria. This could occur by gasdermins inserting into mitochondria, as supported by DOI: 10.1038/s41419-025-07760-4. The resulting ROS production due to mitochondrial dysfunction could cause the bactericidal toxicity seen in the cell extracts.

Reviewer #2 (Public review):

The authors investigate the mechanism by which a gasdermin pore-forming effector of the cartilaginous fish Callorhinchus milii, GSDMA/B (CmiGSDMA/B), is activated. This potentially provides information on the ancestral function of gasdermins, a class of proteins broadly important in human health and disease. By reconstituting components of this system in vitro using transfection models, they show that GSDMA/B is activated by cleavage by the caspase-1 homolog CmiCASP1, which directly senses lipopolysaccharide. This mechanism is broadly similar to the non-canonical pathway in mammals, wherein caspase-4/5/11 cleaves GSDMD upon cytosolic LPS sensing. The conclusions of these interactions are mostly well supported by data, but some aspects need clarification, and based on the experimental approaches, some of the broader interpretations have limitations that should be considered and further discussed.

A more detailed analysis and discussion on the differences between caspases with regard to their LPS-binding capacity would be valuable for comparison. The analysis of Figure 4A and 4B effectively shows that there are similarities between CmiCASP1 and some of the studied mammalian caspases. However, part of this analysis is to make the point that some caspases do not bind LPS, and it would benefit from the inclusion of additional relevant LPS-insensitive caspases to show the connection between the chondrichthyan caspase residues highlighted and LPS-binding dependence. Modeling the LPS binding site (such as in Figure 1F) would further help clarify whether these are appropriately positioned for coordination, or for non-conserved residues, if there are alternate binding modes thought to have biological relevance.

The authors note that two different cleavage products are formed, with variable function, which is of interest. The results of Figure 2b suggest that the 241A mutation (blocking the 30 kDa product) increases processing to the larger 35 kDa product, while the 288A mutation decreases processing of the 30 kDa product (also blocking the 35 kDa form). Paired with the lysis data (Figures 2C-2E), its not clear that the 35 kDa product is anything but inactive, but this is quite different from the observations in the experiments with each form (Figure 3N-3Q). A more detailed kinetic and stoichiometric analysis between full-length, N241, and N288 would be important for clarifying the potentially interesting observation of N288 inhibition of N241.

The mechanism of bacteriocidal activity proposed in the final model and by the experiments of Figure 5 would benefit from further development to support the claim. The experiments do not adequately address whether, during pyroptosis, there is release of N241-like fragments that can kill bacteria. Figure 3G would indicate that it stays in the cell, either in the membrane or mitochondria, and it's not clear there would be circumstances where it could be extracted from it to then target bacteria. Figure 5B might require additional explanation and analysis, but the appearance of similar colonies between conditions would appear to support that there is not measurable antibacterial activity. More rigorous support would come from differences in bacterial killing by knockout Callorhinchus cells, but a minimal step to demonstrating the relevance would be MIC assays, and connecting the effective concentration with one that could naturally occur in Callorhinchus.

Broadly, the methods of reconstitution of components of this system demonstrate the sufficiency of LPS for activating Casp1, and Casp1 for activating GSMDA/B. However, in more established models, it is clear that there are inhibitors, feedback mechanisms, alternative pathways, and regulation that could render these interactions irrelevant in Callorhinchus. For example, it's not clear where Casp1 and GSMDA/B are ever expressed in the same cell, at quantities sufficient for this mechanism, or that Casp1 doesn't induce more rapid death by acting on something other than GSDMA/B, or that Casp1 is irrelevant because GSDMA/B can be activated more readily by another mechanism. Therefore, while the insights into the evolution of the individual factors of GSDMA/B and Casp1 are interesting and of potential value to the field, reconstituting choice components by transfection of human HeLa and HEK293 cells introduces limitations to how far these experiments can be interpreted as a system. The abstract, for example, states this is a "pyroptosis pathway in cartilaginous fish". However, for all the interest of these data in the evolution of these proteins, the evidence falls short of this. It establishes a biological potential, but it's not clear this is an active pathway in fish.

Reviewer #3 (Public review):

In this manuscript, the authors focused on Callorhinchus milii GSDMA/B (CmiGSDMA/B) and its upstream inflammatory caspase, CmiCASP1, and revealed that LPS directly engages the CARD domain of CmiCASP1, triggering its activation, which subsequently promotes the proteolytic cleavage of CmiGSDMA/B, yielding two N-terminal fragments with opposite functions. Moreover, consistent with GSDMD, the functional N241 of CmiGSDMA/B can mediate pyroptosis and exhibit bactericidal activity against Gram-negative bacteria in vitro. Based on these observations, the authors clarified that they uncovered an ancestral LPS-sensing CASP1-GSDMA/B axis in cartilaginous fish; however, several issues should be addressed.

(1) The evidence for direct and functional LPS sensing by CmiCASP1 remains insufficient. Although the authors propose that LPS directly binds the CARD domain of CmiCASP1 to trigger a non-canonical inflammasome-like pathway, the current support mainly comes from pull-down, competition, and in vitro cleavage/activity assays. These results are suggestive but do not yet establish a direct, specific, and physiologically relevant interaction. Additional quantitative binding and specificity analyses are needed to exclude indirect association, aggregation, or other assay artifacts. Therefore, the claim that CmiCASP1 functions as a bona fide direct LPS sensor appears overstated at this stage.

(2) The proposed antagonistic role of N288 is not yet convincingly supported. While the "dual-fragment antagonistic regulation" model is interesting, it currently relies mainly on overexpression/co-expression, co-IP, and localization analyses, which do not clearly distinguish a physiological inhibitory mechanism from a non-specific dosage or sequestration effect. Stronger support would require evidence for the relative generation and timing of N241 and N288, as well as quantitative data showing that N288 interferes with N241 membrane targeting, oligomerization, or pore formation. Testing the effect of selectively blocking D288 cleavage in the full-length protein would also strengthen this conclusion. At present, the antagonistic model remains premature.

(3) The physiological and evolutionary claims are stronger than the available evidence. Although the study shows that CmiCASP1 can cleave CmiGSDMA/B and that this module can be reconstituted in heterologous mammalian systems, these data do not demonstrate that such a pathway operates in elephant shark cells or tissues under physiological conditions. A similar concern applies to the antibacterial assays, which use HEK293T lysates rather than purified N241, making it difficult to exclude contributions from host-derived factors. The authors should either provide more direct evidence in a relevant chondrichthyan context or substantially tone down the evolutionary and physiological interpretations.

(4) The inhibitor data do not convincingly demonstrate suppression of CmiCASP1 activation. Although the authors state that Z-VAD-FMK blocks CmiGSDMA/B cleavage and pyroptotic phenotypes, Figure 1L and Figure 2H do not clearly show that CmiCASP1 activation or processing itself is inhibited. If CmiCASP1 remains processed in the presence of the inhibitor, it becomes unclear whether Z-VAD-FMK blocks CmiCASP1 activation, catalytic activity, or only downstream substrate cleavage. This point should be clarified with more direct biochemical evidence.

(5) The dosage control for GSDM-derived proteins in the antibacterial assays is unclear. In Figure 5, antibacterial activity is tested using HEK293T lysates or concentrated supernatants containing full-length CmiGSDMA/B, N241, or N288, but it is not clear how protein amounts were normalized across conditions. Differences in expression, stability, or recovery could substantially affect the apparent antibacterial activity. The authors should clarify how input was controlled and ideally provide quantitative normalization or matched-concentration assays to support the comparison.

eLife Assessment

This important study presents a new approach to individualised functional brain mapping using multi-task batteries, and introduces a toolbox to support its implementation. Convincing evidence from simulations and empirical analyses across multiple datasets demonstrates clear advantages over traditional single-task approaches and provides practical guidance on task selection. The extent to which these improvements translate into more accurate recovery of individual-specific functional boundaries beyond well-characterised systems remains to be established.

Reviewer #1 (Public review):

Summary:

In this well-written and well-presented manuscript, Arafat and colleagues describe the proper use and advantages of multi-task batteries to understand the organization of the human brain. The authors present both simulation and empirical results suggesting a substantial advantage in using many short tasks vs a single localizer in identifying specific task-engaged regions. The natural question arises as to which tasks should be used within the battery, and how they should be organized. The authors address this question by demonstrating a data-driven strategy for task selection that outperforms random selection, and they further demonstrate the advantages of highly interspersed tasks over a more typical one-task-per-run strategy.

Strengths:

In general, I find this work highly compelling. The topic itself should be of high interest to the majority of researchers conducting human functional neuroimaging studies. The manuscript itself is comprehensive and sound. The analyses and data are truly excellent, with only a few, relatively minor issues that can be improved. The authors do an exceptional job of laying out the motivation and logic for almost every analysis and conclusion in the manuscript.

I want to point specifically to the potential impact of this work. While the claims made here are very appropriately constrained to the conclusions that can be drawn from the actual analyses, their impact is potentially far-reaching. By the end of this manuscript, we are left with a set of ideas that in effect overturns 2-3 decades of received knowledge about how functional neuroimaging tasks should be designed to optimally understand the organization of the human brain.

Thanks to this paper, I personally will be rethinking how I design all of my fMRI studies in the future after reading this work. The authors are to be commended for this excellent contribution to the literature.

Weaknesses:

I struggled to understand the motivation and logic of the "connectivity modeling" section of the analyses.

Reviewer #2 (Public review):

Summary:

This paper presents theoretical and empirical insights into the use of multi-task batteries for precision functional brain mapping and offers practical guidelines for optimal task design. Specifically, the authors evaluate differences between single-contrast and multi-task localizers, explore data-driven strategies for battery selection, such as minimizing collinearity, and compare grouped and interspersed stimulus-presentation designs. Through a combination of simulations and analyses of empirical fMRI data, the study provides a systematic set of recommendations for improving the reliability and specificity of individualized functional mapping.

Strengths:

Traditional functional mapping has long relied on single-contrast localizers or resting-state fMRI. However, there is growing recognition that diverse batteries of general tasks can yield more detailed functional maps with higher signal-to-noise ratios (SNRs). This manuscript systematically evaluates these advantages using both simulations and empirical data. The contribution is timely and provides the community with not only a theoretical justification for multi-task designs but also practical tools, in the form of the MultiTaskBattery toolbox, for implementing them.

Weaknesses:

Although the results are robust, they are largely consistent with existing expectations in the field, and the conceptual novelty or "surprise" factor is therefore somewhat limited. Nevertheless, synthesizing these findings into a coherent set of design recommendations provides significant value to researchers.

Additionally, there appears to be a slight mismatch between the content of the manuscript and its designated article type. Although the manuscript was submitted as a "Tools and Resources" article, its extensive empirical analyses and theoretical evaluation make it read more like a "Research Article." I defer this categorization to the Editor's judgment.

Finally, the authors use inter-subject overlap as a primary metric for validating the accuracy of functional mapping (Figure 3). However, given that genuine inter-individual variability in brain organization is a central premise of precision mapping, greater overlap across subjects may not necessarily indicate more accurate individual-level localization. A more detailed analysis or discussion of how to distinguish measurement noise from genuine individual differences would make the paper more comprehensive and strengthen its overall contribution.

Reviewer #3 (Public review):

Summary:

This study introduces a principled framework for optimizing multi-task batteries for individualized functional brain mapping. Through simulations and empirical validation, the authors show that selecting tasks to maximize differences in regional response profiles can substantially improve the identification of functional brain regions. The work represents a valuable methodological advance for precision functional mapping, although some assumptions underlying the broader applicability of the framework would benefit from further discussion.

Strengths:

The manuscript addresses an important methodological challenge in precision functional mapping using a rigorous combination of theoretical analyses, simulations, and empirical validation. The framework is practical and well supported by open-source software and a publicly available task library, making it readily accessible for adoption and further development by the research community. The manuscript is well written, logically structured, and clearly presents both the methodological framework and its practical implementation.

Weaknesses:

(1) The abstract and introduction emphasize the application of the framework to individualized brain parcellation. While the presented analyses convincingly demonstrate improved prediction of held-out task responses using atlas-guided parcel assignments, they do not directly validate whether the optimized task batteries improve the estimation of an individual's true functional boundaries. The empirical validation relies on atlas-defined parcel identities as the reference standard, yet substantial inter-individual variability in the location and extent of functional regions - particularly within association cortex - has been well documented. Consequently, improved recovery of atlas-defined parcel labels does not necessarily imply more accurate recovery of an individual's functional organization. It would therefore be valuable to clarify this distinction in the abstract and discussion and to discuss how inter-individual variability may influence the interpretation and generalizability of the parcellation analyses.

(2) The framework assumes that informative task batteries can be designed to distinguish neighboring functional regions. While this is compelling for well-characterized systems with distinct functional response profiles, it is less clear how the approach generalizes to finer-scale subdivisions within association cortex (e.g., subnetworks), where neighboring regions may exhibit highly similar task-response profiles and their functional roles remain incompletely understood. In these settings, the relevant functional dimensions may not yet be known, making it difficult to design optimized task batteries a priori. It would therefore be valuable for the authors to discuss how the framework could be extended to such cases.

(3) More generally, the framework assumes that the sampled task space adequately captures the functional dimensions that differentiate cortical regions. However, particularly within the association cortex, neighboring regions may exhibit similar task-response profiles while differing in the information they represent, their interactions with other regions, the computations they perform, or their cortical layer-specific response patterns. In such cases, the dimensions that best distinguish cortical organization may not be fully reflected in task-response profiles alone, but instead become apparent through complementary approaches such as representational analyses, task-evoked or resting-state connectivity, computational modelling, or laminar response profiles. It would therefore be valuable to discuss how the proposed framework relates to these complementary perspectives.

(4) Many task batteries inherently contain tasks that vary substantially in cognitive demand. Given that task difficulty is itself a major organizational axis in association cortex, it would be helpful for the authors to discuss how the optimization framework accounts for this. Specifically, could differences in task difficulty drive regional differentiation, even when tasks probe similar underlying cognitive processes? If so, how does the framework distinguish between organizational differences arising from a common demand axis and those reflecting more specific functional specializations?

(5) The Discussion places the proposed framework in the broader context of precision functional mapping and refers readers to a companion paper demonstrating advantages over resting-state ("inside-out") approaches. Given that these comparisons motivate several of the broader recommendations made in the Discussion, it would be helpful to provide a brief summary of the main findings of the companion paper here. This would allow readers to better understand the basis for these conclusions without relying on a separate manuscript.

eLife Assessment

This useful study builds on the previously identified terminal Schwann cell (tSC) marker Col20a1 to generate a novel Col20a1-CreERT2 knock-in mouse line that enables the selective genetic labelling and ablation of tSCs at the neuromuscular junction. This model represents a new tool for distinguishing tSCs from myelinating Schwann cells. The authors conclude that loss of tSCs does not cause major alterations in the overall structure of the neuromuscular junction, but that tSC ablation intrinsically compromises the presynaptic reserve pool. However, the study remains incomplete because the evidence provided does not adequately support these conclusions.

Reviewer #1 (Public review):

Summary:

This manuscript's major strength is the identification of Col20a1 as a novel marker for terminal Schwann cells (tSCs) and the generation of the Col20a1-CreERT2 knock-in mouse line, which represents a valuable new genetic tool for studying tSC biology at the neuromuscular junction. However, the central conclusion that terminal Schwann cells regulate presynaptic vesicle homeostasis is not sufficiently supported because the electrophysiological and ultrastructural analyses are based on limited sample sizes. Increasing the number of animals and NMJs analyzed would substantially strengthen the conclusions. In addition, further validation of Col20a1 expression using RNAscope or immunostaining, together with comparisons to established tSC markers such as Kir4.1 and NG2, would help establish its specificity. Finally, the developmental appearance of Col20a1-positive axonal Schwann cells is intriguing and warrants further investigation to determine whether these cells migrate and differentiate into terminal Schwann cells during postnatal development.

Strengths:

The authors identified Col20a1 as a specific marker of terminal Schwann cells (tSCs) at the neuromuscular junction (NMJ) in mice and generated a Col20a1-CreERT2 knock-in mouse line for in vivo labeling of tSCs. This represents a novel and significant technical advance for the NMJ field, providing a valuable genetic tool for studying the development, maintenance, and function of terminal Schwann cells in vivo.

Weaknesses:

The major weakness of this manuscript is that the sample sizes are too small to support the authors' conclusion that "Terminal Schwann Cells Regulate Presynaptic Vesicle Homeostasis but Not Neuromuscular Junction Integrity in Mice."

Although the authors state that "Quantification of the tdTomato-positive NMJ ratio showed an ablation efficiency of approximately 80%, with only ~20% of NMJs retaining escaper tSCs" (page 9), they do not provide the sample size or sufficient quantitative information for the Col20a1-tdTomato/DTA ablation experiment shown in Figure 4. This information is essential for evaluating the robustness and reproducibility of the ablation strategy.

The electrophysiological analyses are also based on very limited sample sizes. According to Figure 6, mEPP recordings were obtained from 10 NMJs from 3 control mice and 14 NMJs from 5 tSC-ablated mice. The EPP recordings were based on similarly small numbers (control, n = 10 NMJs from 3 mice; tSC-ablated, n = 14 NMJs from 5 mice). Thus, only approximately 2-3 NMJs were analyzed per tSC-ablated mouse on average. Given the inherent variability among individual NMJs and animals, these sample sizes are insufficient to support broad conclusions regarding the effects of terminal Schwann cell ablation on synaptic transmission.

In addition, the authors describe the phenotype as "leaky" presynaptic spontaneous release, but this terminology is not defined and lacks mechanistic explanation. It is therefore unclear what specific physiological alteration the authors intend to describe.

The sample sizes for the electron microscopy analyses (Figure 7) also appear to be limited, making it difficult to determine whether the reported changes in synaptic vesicle distribution are representative or statistically robust. Because the central conclusion relies heavily on these electrophysiological and ultrastructural data, the evidence presented is not sufficient to support the claim that terminal Schwann cells regulate presynaptic vesicle homeostasis. At present, this conclusion is overly broad and not adequately supported by the available data.

Minor comments:

(1) While several figures contain high-quality NMJ images (e.g., Figures 1B and 4), the image quality in other figures should be improved. For example, the S100B immunostaining appears overexposed in some panels, making it difficult to distinguish individual Schwann cells or visualize the boundaries between adjacent cells.

(2) Some neuromuscular junctions shown in Figure 2C appear to be partially denervated. The authors should clarify whether these represent normal variability, effects of the experimental manipulation, or imaging artifacts.

(3) The authors should specify the muscle preparation used in Figure 3, as this information is necessary for interpreting the results and comparing them with previous studies.

Reviewer #2 (Public review):

Summary:

Two types of Schwann cells (SCs) ensheath motor axons - myelinating SCs along the axonal length and terminal SCs (tSCs) that cover nerve terminals at the neuromuscular junction (NMJ). Therefore, the NMJ is, like other synapses, tripartite, with specialized presynaptic, postsynaptic, and glial cells. Many studies have shown that tSCs play roles in the development and function of the NMJ, but for some of these, interpretation is difficult because it is hard to manipulate tSCs without also manipulating myelinating SCs. To circumvent this problem, Kong et al. make use of a gene selectively expressed in tSCs, Col20a1 (Figure 1), to generate a knock-in mouse line, Col20a1-CreER, that gives them genetic access to tSCs. They cross this to Cre-dependent lines that mark tSCs with a red fluorescent protein (Figures 2 and 3) or ablate them by expression of diphtheria toxin along with the fluorescent protein (Figure 4). They show that ablation at postnatal day (P) 10 does not affect the overall structure or function of the NMJ (Figures 4 and 5). It does, however, affect some aspects of neuromuscular transmission over the following few weeks (Figures 6 and 7). Long-term effects cannot be studied by this method, however, because terminal SCs are replaced, presumably from the preterminal population (Figure 8).

Strengths:

The work is done to a high technical standard, including detailed characterization of the knock-in model. Results are presented clearly and illustrated beautifully. The finding that some early reports of synaptic alterations may result from concurrent loss of axonal SCs is important in rethinking the role of tSCs.

Weaknesses:

(1) The authors claim that tSCs are dispensable for some aspects of NMJ maturation, including synapse elimination (called pruning here), formation of "pretzel-like" postsynaptic topology, and generation of junctional folds in the postsynaptic membrane (lines 223 and 363). However, this conclusion is based on injection of tamoxifen to initiate tSC ablation at P10, which is necessary because Col20a1 is expressed in some preterminal SCs at earlier times. It presumably takes a few days for CreER to translocate to the nucleus and activate the toxin transgene, and some more time for the toxin to be generated and act. This is problematic because synapse elimination and other aspects of maturation mentioned occur during the first two postnatal weeks and are largely complete by P14. Therefore, one cannot conclude that these aspects "proceeded normally despite the loss of tSCs....".

(2) Effects on synaptic transmission are modest at best, being significant at a level of p<0.05 but not p<0.01 (Figure 6E, G, H and most of L). Effects on vesicle density are more robust (Figure 7).

(3) The authors use red fluorescent protein from the Col20a1 to label tSCs, and antibodies to S100b to label all SCs. This is appropriate in normal muscle and soon after tSC ablation. At later times, however, the NMJ is repopulated by S100+ Col20a1- SCs (Figure 8B). It is therefore important to show when this repopulation begins, because a modest recovery of SC coverage could have a big effect. For example, Figure 4C quantifies loss of NMJs with residual RFP+ cells but not S110+ cells; both should be quantified at this and slightly later stages.

Reviewer #3 (Public review):

Summary:

This manuscript reports a novel genetic model, Col20a1-CreERT2 knock-in mouse, to target terminal Schwann cells (tSCs) at mouse neuromuscular junctions in a cell-type-specific and temporal manner. The authors analyzed multiple publicly available single-cell transcriptome databases to identify Col20a1 as the tSC marker. The authors crossed Col20a1-CreERT2 and Rosa26-LSL-tdTomato to label tSCs successfully. In addition, the authors generated Col20a1-CreERT2; Rosa-tdT/ diphtheria toxin subunit A (DTA) to specifically ablate tSCs and analyze the role of tSCs in motor behavior, neuromuscular junction electrophysiological function, and the histology and ultrastructure of neuromuscular junctions.

Strengths:

The Col20a1-CreERT2 x Rosa26-LSL-tdTomato mice successfully labeled the tSCs at NMJs and reported the developmental distribution of the Col20a1-positive cell population. The Col20a1-CreERT2; Rosa-tdT/DTA mice successfully ablated tSCs, which did not cause changes in gross neuromuscular junction architecture, neuromuscular synapse physiology, or motor behavior.

Weaknesses:

The conclusion of this manuscript will be strengthened by additional analysis showing time-course data of tSC ablation and replacement by non-recombined Schwann Cells. Currently, it is not clear when and how long the tSCs are ablated, which makes it difficult to interpret the data and phenotype. Detailed review comments are provided to the authors in the "recommendations for the authors" section.

eLife assessment

The findings documented here are interesting and valuable with respect to revealing genetic players that may underlie AD. However, there are concerns related to the analysis pipeline and the basic assumptions underlying the modelling. There is an overarching lack of controls in terms of comparator diseases and/or positive and negative control findings; hence, the main claims are only partially supported, and the study is incomplete.

Reviewer #1 (Public review):

Summary:

The authors build a reusable, disease-agnostic pipeline that retrieves GWAS risk genes from the GWAS Catalog by ontology terms, filters them, and maps them onto Human Protein Atlas (HPA v24) co-expression modules at three biological scales (tissue/organ, brain region, cell type). Enrichment is assessed by a consensus of Fisher's exact test with Benjamini-Hochberg correction and 10⁶-iteration Monte Carlo simulation. Applied to AD, DLB/PD, and FTD/ALS, the analysis reports convergent neuronal-module enrichment across all three diseases, AD-specific enrichment in liver- and immune-associated modules, and DLB-specific enrichment in ciliary modules, followed by a DrugBank-based survey of compounds targeting module gene products.

Strengths:

(1) The core premise is sound and well-motivated: risk-gene lists are hard to interpret because most variants are low-penetrance and broadly expressed, and projecting them onto a multiscale expression atlas is a reasonable route from statistical association toward tissue/cell context.

(2) The pipeline is delivered as reusable, open code (GitHub) built entirely on public inputs (HPA, GWAS Catalog, DrugBank), which is a significant contribution to the field and provides opportunities for replication and expansion.

(3) The dual-enrichment design (fold enrichment with FDR correction plus a 10⁶-iteration Monte Carlo empirical null) is more defensible than any single test, and the "consensus" logic is sound.

(4) The multiscale framing (organ → brain region → cell type), with UMAP module projections, is genuinely helpful and makes the mapping legible to broad scientific backgrounds.

(5) The authors are commendably restrained on one key point: they explicitly report that most risk genes are broadly expressed and not brain-selective, rather than overstating neuronal specificity.

(6) The AD liver/immune convergence is nicely discussed and integrated, and the NAFLD-AD discussion (Kupffer-cell/hepatocyte Aβ clearance, locus coeruleus noradrenergic parallels, "type 3 diabetes") is thorough and well-referenced, even where it remains speculative.

Weaknesses:

(1) Gene-to-variant mapping via the author-reported gene field is unclear. The Methods assign genes using the GWAS Catalog author-reported gene field, which predominantly reflects the nearest gene to the lead SNP and may not be the effector gene; it is also inconsistent across studies and different time periods of publication (i.e., changing methodologies in genomics and GWAS procedures). Because every downstream result depends on the gene set, this choice likely introduces noise and bias into all enrichment, specificity, and drug claims. More modern practices link variants to genes via fine-mapping plus eQTL/pQTL colocalization, or integrative scores. At minimum, the sensitivity of the main signatures to nearest-gene versus colocalization-based assignment should be demonstrated.

(2) The suggestive threshold (p < 1×10⁻⁵) trades specificity for coverage in the analysis that requires specificity. Relaxing from 5×10⁻⁸ substantially raises the false-positive fraction of the gene set. This is defensible for exploratory coverage in under-powered DLB/FTD, but the headline claims concern disease specificity (limited gene-set overlap; distinct signatures). Non-overlap among partially false-positive lists can lead to biological specificity conclusions that may not actually exist. A genome-wide-threshold sensitivity analysis is needed to show the signatures persist. Also, see concerns below about the DLB designation.

(3) "Disease specificity" is confounded by GWAS power. AD GWAS (e.g., Bellenguez; Kunkle; Sherva ~205,500 cases) vastly outpower DLB and FTD discovery. The gene counts (453/278/219) and the minimal three-way overlap (only two genes) track sample size and locus density as much as biology. The claim of "disease-specific genetic architectures" should be tempered and ideally power-matched (e.g., subsampling AD, or restricting to comparable effective N) before specificity is asserted. Also, see concerns below about the DLB designation.

(4) Linkage disequilibrium structure at gene-dense loci is not addressed and may inflate the lipid/liver signal. The overlap genes named as driving the liver/lipid theme include APOC1, APOC2, and APOE, all of which reside within the same chromosome-19 linkage disequilibrium (along with TOMM40). Counting co-regulated, physically clustered genes from one association signal as independent risk genes risks inflation of enrichment for lipid/lipoprotein modules. The five "liver-enhanced" genes flagged in the text again lead with APOE. Evaluation of one gene per independent signal is essential before the liver/lipid signature can be interpreted as multi-gene convergence rather than a single strong gene driving the effect.

(5) The enrichment background is not clear. Risk genes are filtered to HPA brain-detected transcripts, but the Methods do not state whether the Fisher/Monte Carlo background (N_Total) is likewise restricted to brain-expressed/HPA-detected genes or reflects all protein-coding genes, which may introduce bias. Specific information on the Fisher/Monte Carlo should be provided to ensure that the enrichment background is the same. If they are not, additional analyses should be performed to ensure robustness of the findings when restricted to brain-expressed only or the broader background.

(6) Cross-scale "consensus" is not independent confirmation. The same genes reappear across tissue, brain, and cell modules, so agreement across scales is partly built-in rather than corroborating. The neuronal-signature counts (82 AD / 62 DLB / 53 FTD; 186 "unique" genes) should be accompanied by a clear statement of how much cross-scale evidence is non-redundant. Also, see concerns below about DLB designation.

(7) Temporal claims are overstated. Using control HPA tissue avoids end-stage confounds but, by construction, cannot capture disease-state programs central to AD. More importantly, framing these modules as "baseline vulnerability hotspots that precede clinical neurodegeneration" is not tested, as nothing here is longitudinal. This assumption is presented as a finding and should be reworded as a hypothesis.

(8) DLB and PD should not be merged, and the cilia-associated risk genes cannot be termed causal based on the study design. The supporting literature is almost entirely PD (Schmidt iPSC-NPCs from sporadic PD; LRRK2 PD striatum), yet DLB and PD are merged, and the signature is branded "DLB." These should not be merged, particularly as it relates to sporadic PD. Although there are similar genetic risk factors (i.e., GBA, SCNA), of which GBA is unfortunately not mentioned in the manuscript, there are major genetic differences in sporadic PD and even PD with dementia (PDD) and DLB. It is not clear why PDD was not incorporated.

Cross-sectional expression overlap with GWAS genes cannot establish that cilia-associated risk genes are "causative vulnerabilities rather than secondary effects." Please soften to association and rename to reflect the synucleinopathy grouping. Further, additional discussion of important differential genes in this category (i.e., APOE and GBA) is needed, and the pathological description of DLB is incomplete in the introduction (i.e., focuses only on synuclein).

(9) The drug/repurposing analysis rests on a weak targeting rationale. Most of the 1,777 drugs target co-expression-module neighbors of risk genes, not risk genes themselves, so "substantial repurposing reservoir" likely overstates the impact. The anesthetics example (sevoflurane/halothane/desflurane hitting GABA_A subunits and ATP2B2) is a near circular argument, as anesthetics necessarily engage neuronal ion channels. This finding does not independently confirm the neuronal signature.

Statin-dementia and hydroxychloroquine/amodiaquine links are pre-existing, and the Tirzepatide→cilia→DLB inference is highly speculative. This section should be labeled hypothesis-generating, with direct-risk-gene targets separated from module-neighbor targets.

(10) Interpretation leans heavily on nominal (p < 0.05) modules. Several of the most novel claims (parts of the liver and immune signatures) rest on nominally significant modules that do not survive FDR (itself set leniently at p_adj < 0.1). The text should make consistently explicit which claims are FDR/Monte-Carlo-supported versus nominal-only, and de-emphasize conclusions resting solely on the latter.

Reviewer #2 (Public review):

Summary:

Genes associated with risk for a specific disease commonly have widespread expression and functions across the body. Surveying patterns in these effects may reveal novel mechanisms, organs, and systems implicated in a disease, amongst other associations that are truly independent. In this work, Husen and coauthors use the Human Protein Atlas to explore such associations in Alzheimer's disease (AD), Lewy body dementia (DLB), and Frontotemporal dementia (FTD). Focusing on human non-disease tissue expression may avoid the effects of disease progression obscuring initial vulnerabilities. However, associations in non-diseased tissues do not necessarily reflect mechanisms causally related to the diseases themselves.

The work describes patterns of enrichment of genes across tissue types, brain regions, and cell types. A relatively small set of classes of each show enrichment for disease. While neural signatures are unsurprisingly prevalent, these classes are largely distinct across the three diseases. Alzheimer's disease is linked to liver and central and peripheral immune cells, while DLB shows interesting enrichments associated with cilia, which are linked to an existing literature. Results from the drug repurposing approach are then presented, with 1777 drugs linked to protein products of any of the modules enriched by the risk genes using DrugBank, categorised according to key signatures.

The authors developed R code (the HPA GeneSet Explorer) to automate the production of multi-system summaries of the organs, brain regions, cells and gene modules associated with traits and diseases and associated gene sets, within the HPA. Risk gene sets for the three dementia types were derived from the GWAS Catalog.

Strengths:

While many studies of how risk genes contribute to disease take a narrow approach focusing on organs, cell types, and processes already associated with a disease, it is a sensible approach to start with a system-agnostic approach that assesses tissues that are not ostensibly affected by disease. Here, this approach reveals a range of associations for 3 neurodegenerative diseases, identifying disease-associated modules and drug candidates that might be prioritized for subsequent confirmatory inference across biological scales. Results highlight key organs and cell types, most of which have established associations with the disease. Perhaps the most intriguing results are the links of DLB to cilia-related processes, which can be linked to some prior reports of DLB/PD but are not a core element of current theories of pathogenesis.

Weaknesses:

A difficulty with broad, multi-dataset surveys of disease associations is the need to distinguish novel and robust patterns - even if they lack causal evidence - from those that are unsurprising or do not stand out statistically. The work is exploratory in nature, but it is often hard to know how strong the evidence is for particular observations reported.

The work combines nominal, FDR<0.1, and Monte Carlo-based inference (with no apparent multiple assessment control across all tested modules) p-values throughout the paper, with patterns of effects of nominal significance. In some places, modules appear to be retained if they meet any of these criteria, muddying inference. This makes it difficult to weigh the different reported associations. Results report numbers of risk genes showing nominal p<0.05 enrichment across gene modules and biological scales - it is difficult for the reader to determine null expectations for false positives here. Similarly, it is unsurprising that thousands of drugs can be linked to the risk genes and their signatures using nominal significance.

The results have limited mechanistic specificity. The modules identified often reflect biological processes implicated in the diseases. This provides some validation of the approach, but the modules are often broadly defined, providing little mechanistic insight. For example, many aspects of ciliary biology may overlap with DLB, but can the HPA provide more specific insight? More generally, it is difficult to determine how much relevance that enrichment in non-disease tissue has for disease processes. Similarly, it is hard to determine whether overlap of drug targets from DrugBank with these modules realistically increases their prioritization.

Methodologically, there could be more detail. The paper - in particular the methods - is partially presented as a tool/pipeline paper, but thorough descriptions of the HPA models that are employed and modules reported for the analyses should still be presented in detail. The drug repurposing approach is described in a couple of sentences without a precise reference to the tool or statistical methods.

Reviewer #3 (Public review):

Summary:

The manuscript presents the development of a software tool and a computational workflow for the comparison and biological interpretation of GWAS results among three neurodegenerative diseases - AD, PD and FTD - using the data from the Human Protein Atlas. The multi-scale content analysis produces a representation of GWAS results highlighting enrichment at the level of brain regions, organ systems/tissues and cell types as well as in terms of molecular pathways. The manuscript argues that DLB, AD and FTD have differential 'modules' revealed by the procedure.

Strengths:

The system is leveraging vast knowledge sources including both the GWAS catalog and the HPA, and it is integrating these data. This synthesis of information is useful. It is making use of large investment in data generation and data warehouses in order to yield interpretation of epidemiologic results from GWAS in biological terms that may yield insight into disease processes and differences between diseases. The multiscale analysis recognizes the various biological lenses at which the implications of GWAS can be evaluated.

Weaknesses:

There is a lack of controls and/or disease comparators in the study. It is hard to assess that the workflow is performing 'as expected' without a set of positive or negative controls - or at least comparators - to gauge the performance of the tool. The statistical methods are simplistic and rely on Fisher's exact tests, UMAP analyses and clustering with little justification. There is a lack of power analysis and specification of the number of genes required for the procedure to 'work'. The mapping of GWAS hits to genes is simplistic and may in many cases be erroneous, and the implications of errors in these mappings are not considered. Many of the hits are not brain-specific - highlighting the complexity of gene function and the pleiotropic nature of gene activity. Moreover, the mapping between 'tissue enrichment' and 'tissue that is the functional driver of the GWAS signal' may be a logical flaw in the reasoning of the authors. Just because a tissue - such as liver enrichment in AD - is enriched in the GWAS gene-mapping analysis does not mean that that tissue found to be enriched is in fact the functional tissue that gave rise to the GWAS signal. Genes have different isoforms, functions, and regulatory mechanisms in different parts of the organism in different parts of development. In a phrase, the enrichment observed could be correlative and not causative, and in fact the enrichment could be driven by some hidden variable not considered. The etiologic tissue for neurodegenerative disease is the brain. The findings are not necessarily surprising or novel in the distinction between PD, AD, and FTD. Finally, the figures are perhaps not the best way to present results. There are many small pie charts that lack interesting results; the figures are in general hard to read and could use refinement in terms of fonts.

eLife Assessment

This important study shows how hunger alters avoidance of harmful heat in C. elegans by reconfiguring the activity of key sensory neurons. The evidence is convincing, with well-designed behavioural, genetic, and imaging experiments that support the main conclusions. The work will be of interest to neuroscientists studying how internal states shape sensory processing and behaviour across species.

Reviewer #1 (Public review):

This study by Thapliyal and Glauser investigates the neural mechanisms that contribute to the progressive suppression of thermonociceptive behavior that is induced under conditions of starvation. Several previous studies have demonstrated that when starved, C. elegans alters its preferences for a variety of sensory cues, including CO2, temperature, and odors, in order to prioritize food seeking over other behavioral drives. The varied mechanisms that underlie the ability of internal states to alter behavioral responses are not fully understood, however there is growing evidence for a role by neuropeptidergic signaling as well as capacity for functionally distinct microcircuits, formed by distinct internal states, to trigger similar behavior outcomes.

Within the physiological range of C. elegans (~15-25C), starvation triggers a profound reduction in temperature-driven thermotaxis behaviors. This reduction involves the recruitment of the amphid sensory neuron pair AWC. The AWC neurons primarily act to sense appetitive chemosensory cues, however under starvation conditions begin to display temperature responses that previous studies have linked to the reduction in thermotaxis navigation. Here, Thapliyal and Glauser investigate the impact of starvation on thermonociceptive responses, innate escape behaviors that are triggered by exposure to noxious temperatures above 26C or rapid thermal stimuli below 26C. They compare the strength of thermonociceptive behaviors, specifically heat-triggered reversals, in worms experiencing either early food deprivation (1 hour off food) or prolonged starvation (6 hours off food). Their experiments demonstrate a progressive loss of heat-triggered reversals that is mediated by AWC and ASI neurons, as well as both glutamateric and neuropeptidergic signaling.

At the level of neural activity, this study reports that the transition from early food deprivation to prolonged starvation reconfigures the temperature-driven activity of AWC neurons from mostly excitatory to a heterogenous mix combining excitatory and inhibitory responses. This finding is interesting in light of previous work that reported the opposite transition in temperature-driven AWC responses when comparing well-fed worms to those kept from food for 3 hours. Specifically, these differences highlight the differences between temperature responses within the C. elegans physiological temperature range (previous studies) and their noxious temperature response (this study). This study also identifies neural and genetic mechanisms that contribute to differences in thermonociceptive responses at +1 versus +6 hours starvation; interestingly, these mechanisms are also partially distinct from those that contribute to differences in negative thermotaxis behaviors in well-fed and +3 hours starvation worms. A limitation of this manuscript is that these differences are not particularly acknowledged or addressed, other than the hypothesis that independent mechanisms underlie negative thermotaxis versus thermonociceptive stimuli.

In this revised article, the authors commanding knowledge of the distinction between thermotaxis navigation (especially negative thermotaxis) and thermonociceptive behaviors is communicated with an admirable depth and clarity to the readers; this study's new findings are helpfully contextualized within the previous literature.

This study represents an important addition to the growing evidence that C. elegans sensory behaviors are strongly impacted by internal states, and that neuropeptigergic signaling plays a key role in mediating behavioral plasticity. To that end, the authors have provided compelling evidence of their claims.

Reviewer #3 (Public review):

Thapliyal, Gopinath, and Glauser show that starvation alters how C. elegans respond to noxious thermal stimuli. Using targeted neural ablation, mutant analysis, and live-cell functional imaging the authors demonstrate that hunger changes the properties of AWC sensory neurons, which sense noxious heat. The authors further show that effects of hunger on nociception require ASI neurons, which are known to respond to hunger and mediate effects of food deprivation on behavior. Finally, the study uses mutant analysis to implicate glutamate and specific neuropeptides in thermal nociception and in modulation of nociceptors by hunger-responsive neurons.

The study clearly shows a strong effect of hunger on nociception and documents a striking effect of hunger on the intrinsic properties of AWC sensory neurons, which respond to noxious heat. The study also clearly and compellingly demonstrates that ablation of hunger-responsive ASI neurons blocks effects of hunger on nociceptive AWCs. These data, which constitute the kernel of the manuscript, are striking and exciting. This revised manuscript analyzes effects of starvation on AWC physiology and clearly shows that starvation alters the way AWCs respond to thermal stimuli by decreases the probability that AWCs will be activated and increases the probability that they will be inhibited. New data also identify ASI-derived neuropeptides that are required for modulation of AWCs by starvation. This study reveals a mechanistic link between an animal's metabolic state and sensory processing and establishes modulation of AWC function as a powerful model to study the molecular basis of this link.

Author response:

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

We believe that the manuscript has been substantially strengthened through the revision process. The main changes are summarized below:

We substantially revised the Introduction and Discussion sections to better position our work relative to previous studies on starvation-dependent thermotaxis plasticity, neuropeptidergic modulation, and AWC function.

We clarified throughout the manuscript the distinction between negative thermotaxis in innocuous thermal ranges and thermonociceptive responses to noxious heat. We now discuss more explicitly that these behaviors involve at least partly distinct molecular, cellular, and circuit-level mechanisms.

We performed new experiments in ins-1 mutants. Unlike what was previously reported for thermotaxis plasticity, ins-1 does not appear required for starvation-dependent thermonociceptive plasticity in our paradigm (new Figure 6—figure supplement 1).

We revised the analysis and terminology used for AWC calcium imaging data. We no longer use the “deterministic/stochastic” terminology and instead describe a starvation-induced shift from predominantly excitatory responses to a mixed distribution of excitatory and inhibitory responses. We also added new quantitative analyses and histogram representations of response distributions, as directly suggested by reviewers, to better illustrate this point.

We performed new genetic interaction experiments using eat-4; flp-6 double mutants. These analyses revealed that glutamatergic and FLP-6 signaling act largely in parallel to mediate heat-evoked reversals after early food deprivation, while prolonged starvation reveals a hierarchical interaction between these pathways.

We revised and clarified the mechanistic model figures accordingly, particularly regarding the proposed ASI → AWC signaling pathway and the role of ASI-derived neuropeptides.

We improved the presentation and statistical rigor throughout the manuscript, including:

- Replacement of heating power values by corresponding temperature increases,

- Clarification of the rationale for using the 1-hour off-food condition as reference,

- Expanded statistical reporting and multiple-comparison procedures,

- Additional methodological details for calcium imaging, rescue validation, and cell ablation approaches,

- Clarification of replotted datasets in figure legends.

We also simplified the manuscript by removing experiments whose interpretation remained ambiguous (notably the nsy-1 and nsy-7 analyses).

Below, we provide a detailed point-by-point response to all reviewer comments.

Public Reviews:

Reviewer #1 (Public review):

This study by Thapliyal and Glauser investigates the neural mechanisms that contribute to the progressive suppression of thermonociceptive behavior that is induced under conditions of starvation. Several previous studies have demonstrated that when starved, C. elegans alters its preferences for a variety of sensory cues, including CO2, temperature, and odors, in order to prioritize food seeking over other behavioral drives. The varied mechanisms that underlie the ability of internal states to alter behavioral responses are not fully understood; however, there is growing evidence for a role of neuropeptidergic signaling as well as the capacity for functionally distinct microcircuits, formed by distinct internal states, to trigger similar behavior outcomes.

Within the physiological range of C. elegans (~15-25{degree sign}C), starvation triggers a profound reduction in temperature-driven thermotaxis behaviors. This reduction involves the recruitment of the amphid sensory neuron pair AWC. The AWC neurons primarily act to sense appetitive chemosensory cues; however, under starvation conditions begin to display temperature responses that previous studies have linked to the reduction in thermotaxis navigation. Here, Thapliyal and Glauser investigate the impact of starvation on thermonociceptive responses, innate escape behaviors that are triggered by exposure to noxious temperatures above 26{degree sign}C or rapid thermal stimuli below 26{degree sign}C. They compare the strength of thermonociceptive behaviors, specifically heat-triggered reversals, in worms experiencing either early food deprivation (1 hour off food) or prolonged starvation (6 hours off food). Their experiments demonstrate a progressive loss of heattriggered reversals that is mediated by AWC and ASI neurons, as well as both glutamatergic and neuropeptidergic signaling.

At the level of neural activity, this study reports that the transition from early food deprivation to prolonged starvation reconfigures the temperature-driven activity of AWC neurons from largely deterministic to stochastic. This finding is interesting in light of previous work that reported the opposite transition (from stochastic to deterministic) in temperature-driven AWC responses when comparing well-fed worms to those kept from food for 3 hours. This study also identifies neural and genetic mechanisms that contribute to differences in thermonociceptive responses at +1 versus +6 hours of starvation; confusingly, these mechanisms are partially distinct from those that contribute to differences in negative thermotaxis behaviors in well-fed and +3 hours of starvation worms (Takeishi et al, 2020). A limitation of this manuscript is that these differences are not particularly acknowledged or addressed, other than the hypothesis that independent mechanisms underlie negative thermotaxis versus thermonociceptive stimuli. However, this suggestion is not experimentally verified.

We thank this reviewer for pointing to the interest of our work. The difference between previous work focusing on negative thermotaxis in the range of innocuous temperatures and our work focusing on thermo-nociceptive response is important and indeed deserves further clarification and a deeper discussion in the manuscript.

Two major empirical evidence for a distinction between negative thermotaxis (as assessed in previous studies) and thermonociceptive plasticity (as assessed in our paradigm) were already included in the initial article version. First, we reported a decrease in average response in AWC neurons due to a shift in the distribution of response polarities from mostly up-response to a mix of ‘up-response’ and ‘downresponse’ after starvation, while previous results showed an increase in response probability of AWCs after starvation (Takeishi et al, 2020). Second, contrary to starvation-evoked thermotaxis adaptation, ASI neurons are required to orchestrate starvation-evoked plasticity in thermonociception. These observations already indicate differences at the circuit and cellular level. For the revision, we conducted further experiments to address the molecular level. We tested ins-1 mutants (see also specific point 3 by reviewer 2, below) and deepened this aspect in the discussion section of the revised manuscript. Previous study found that INS-1 signaling from the intestine is a major mediator of negative thermotaxis plasticity. In contrast, our new data show that INS-1 peptide does not seem critical in regulating starvation-dependent thermonociceptive plasticity (see new Figure 6-supplement 1). Taken together these three lines of empirical evidence support the notion that negative thermotaxis and thermonociceptive starvation-evoked plasticity involves at least partially distinct mechanisms and it seems therefore inappropriate to qualify this notion as purely hypothetical.

Modification in the revised manuscript include extended introduction about the known thermotaxis regulation mechanisms (Introduction section), new Figure 6-supplement 1 about ins-1 and accompanying text in the result section as well as extended discussion about these differences (Discussion section)

Multiple additional aspects of this study make the results difficult to synthesize with existing knowledge, including

(1) Differences in - and insufficient discussion of - the magnitude and kinetics of thermal stimuli;

We have included a better description of the stimuli characteristics in the revised methods section. The discussion section was deepened to better emphasize that different types of thermal stimuli have been used in different studies.

(2) This study's use of "heating power" rather than temperature values when presenting behavioral results;

Thanks for noting this point, which was indeed an unnecessary complication in the result display of the initial manuscript. We have changed ‘power values’ to corresponding ‘temperature increase’ in the revised figures.

(3) The use of +1 hours starvation as a baseline instead of well-fed worms. Indeed, this last point reflects a noticeable experimental result that differs from previous studies, namely that at room temperature, the basal movements of well-fed and starved worms are not different. Such a surprising result warrants further quantification of worm mobility in general and could have prompted a set of experiments directly testing previously published thermal conditions to demonstrate that the new effects reported arise specifically from the use of thermonociceptive stimuli, as hypothesized.

The consideration of on-food and off-food behavioral state is an important point indeed. We found that, at room temp, worms shift from dwelling on food (a state with high spontaneous reversal rate) to global search off-food (a state with low spontaneous reversal, after 1hr starvation) (see Figure 1B). Therefore, unlike the reviewer’s statement, the reported data highlighted key differences in the basal locomotion of worms in fed and 1hr starved conditions. Furthermore, these behavioral states have been characterized very deeply using high-content worm behavioural tracking in our recent publication: Thapliyal et al. 2023 (PMID: 37236963). Our choice of using 1hr as a baseline is primarily driven by the fact that an elevated baseline of spontaneous reversals on food decreased the dynamic range to monitor changes in heat-evoked reversals. 1-hour early food deprivation reduced spontaneous reversals and led to a mild attenuation of heat-evoked responses at low stimulus intensities, while responses to stronger stimuli remained comparable to those of fed animals. Additionally, we observed a clear progressive decrease in heat-evoked reversals with increased duration of food-deprivation, which we further used to dissect the mechanism underlying this plasticity. The choice of the 1hr food deprivation timepoint as a reference is further justified below.

Finally, a previous report (Yeon et al, 2021) demonstrated differences in the impact of chronic versus acute neural silencing on starvation-dependent plasticity in the context of negative thermotaxis. We therefore wonder whether similar developmental compensation impacts the neural circuits that contribute to starvation-dependent plasticity in the thermonociceptive responses.

Indeed, this is an interesting question. Our conclusions are so far based on ablation (with chronic effects). In order to gain insight on this question, future studies could address the impact of chronic vs acute silencing approach in starvation-dependent thermonociceptive responses. We have added an opening on this question in the discussion, as follows:

“Another open question is whether ASI action takes place during development (prior to starvation), or more acutely with active signaling after starvation.”

A weakness of this manuscript is that the introduction is insufficiently scholarly in terms of citations and the description of current knowledge surrounding the impact of internal state on sensory behavior, particularly given previous work on the impact of feeding state on thermosensory behavioral plasticity (Takeshi et al 2020, Yeon et al 2021) and chemosensory valence (Banerjee et al 2023, Rengarajan et al 2019, etc).

To address this weakness, we have revised the introduction section of the manuscript and cited previous relevant research on the impact of internal states on animal behavior, including the papers suggested by the reviewer. We note that 2 out of 4 suggested citations were already present in the initial manuscript (though in the discussion section).

Similarly, the authors' commanding knowledge of the distinction between thermotaxis navigation (especially negative thermotaxis) and thermonociceptive behaviors could be communicated in more depth and clarity to the readers, in order to contextualize this study's new findings within the previous literature.

As mentioned above, we have deepened this aspect in the discussion section of the manuscript (with a dedicated paragraph). It is quite clear that starvationinduced plasticity in negative thermotaxis and thermonociceptive behaviors engage distinct mechanisms (at least in part). These differences include distinct alterations in AWC calcium activity, role of ASI neurons and INS-1 neuropeptide.

Nevertheless, this study represents a solid addition to the growing evidence that C. elegans sensory behaviors are strongly impacted by internal states, and that neuropeptidergic signaling plays a key role in mediating behavioral plasticity. To that end, the authors have provided solid evidence of their claims.

We thank this reviewer for the efforts in evaluating our manuscript, for the positive assessment of our work, and for highlighting some weaknesses which, we believe, have been addressed through the revision.

Reviewer #2 (Public review):

In this work, Thapliyal and Glauser tried to provide a mechanistic understanding by which animals modulate their neural circuit responses to control nociceptive behavior on the basis of the dynamic internal feeding state. It is an important study that adds to the growing body of evidence coming from multiple model systems. They have used elegant genetics, behavioral, and Ca-imaging experiments to demonstrate how the auxiliary thermosensory neuron pair, AWC, and one of the internal state-sensing interneuron pairs, ASI, respond to dynamic internal starvation state to modulate behavioral response to noxious heat. Interestingly, these neuron pairs use distinct molecular mechanisms along with some other unidentified neurons to suppress heat-induced reversal response under short-term and prolonged starvation. The experiments are well performed, supporting most of the claims and providing an important framework for future studies.

I have some queries that, if answered, will certainly enhance the study.

(1) The results suggest that ASI is one of the primary drivers for the starvation-evoked behavioral plasticity, which regulates AWC activity under prolonged starvation. It raises many important questions, including: (a) how starvation modulates ASI response to heat?, and (b) under prolonged starvation, whether ASI also promotes other, non-AWC, glutamatergic inhibitory neurons to suppress heat-induced reversal, and how?

We agree with this reviewer that the mechanisms by which ASI detects and mediates starvation-evoked changes in our model is a very interesting (unsolved) question. However, addressing these questions empirically represents a substantial body of work that would go beyond the scope of the present report. E.g., is temperature-dependent activity in ASI even relevant? At present, we envision that ASI could either work acutely (during heat stimuli) or be modulated over much longer time frames (hours of starvation) as an internal state sensor. Therefore, there will be quite some exploration needed before we figure out the ASI-level regulation more fully (including the critical temporal aspect regarding cell activity, as well as quantitative and qualitative transmission aspects). It will be very interesting in future work to address these questions.

(2) How does ASI regulate AWC activity? In the proposed model (Figure 8) authors suggested an independent, unknown signal, other than INS-32 and NLP-18, from ASI to regulate AWC activity. However, from the results, the existence of another signal is not very clear.

Thanks for raising this point, which reveals a weakness in our graphical representation (in Fig. 8) that was not properly conveying our point. Our current work shows INS-32 and NLP-18 to be important in modulating heat-evoked reversals upon starvation. However, at the moment, we don't know if INS-32, NLP-18, both, and/or other neuropeptides from ASI modulate AWC activity patterns. The calcium imaging experiments in single, double and potentially triple mutants would answer these questions but are not within our current reach, given the time needed to carry out these experiments. However, we acknowledge this point and have changed the figure and its legend to state that the arrow connecting ASI to AWC activity pattern could potentially reflect the action of these neuropeptides.

(3) Previously, Takeishi et. al. showed that ins-1 dynamically modulates AWC-AIAmediated thermotaxis behavior based on the feeding state of the animal. It raises questions whether ins-1 also contributes to noxious heat-induced reversal behavior.

We thank the reviewer for this question. We have now quantified the phenotype of ins-1 mutant in our paradigm. Our data shows that INS-1 neuropeptide is not critical in mediating starvation-evoked thermonociceptive plasticity, unlike plasticity in thermotaxis behavior (See Figure 6- Supplement 1). Together with the differential activity patterns in AWC and the differential need for ASI neurons, these new data further consolidate the notion that starvation-evoked thermotaxis adaptation and noxious-heat avoidance engage separable molecular, cellular and circuit-level modulatory mechanisms. A specific discussion paragraph was added too.

(4) Experiments with AWC fate conversion mutants (nsy-1 and nsy-7) were very good ideas; however, the results obtained were confusing. flp-6 mutant data suggest AWCoff would be essential for heat-induced reversal, especially at the low intensity stimulus level. However, the nsy-1 mutant-forming two AWCon neurons showed complete rescue at the low heat level, which is quite opposite. Similarly, although less prominent, eat-4 rescue experiments suggested both nsy-1 and nsy-7 should behave normally at high heat conditions, which was not the result observed.

We appreciate this comment and the legit attempt to infer what we should expect from a worm with two AWCon or two AWCoff, respectively. From previous studies so far, it's not quite clear if cellular properties of newly formed AWCs in nsy-1 and nsy-7 mutants, including response to sensory cues, formed synapses and their partners, expression of neuromodulator and gap junctions, synaptic output are similar or different. We think further studies are required to first establish if FLP-6 and glutamate signaling (expression, release and action) from altered AWCs in nsy-1 and nsy-7 mutants are the same or different. Therefore, direct comparison between cell fate conversion mutants with flp-6 and glutamate would rely on too many assumptions at this stage. Considering this comment, the limited additional value of the data with nsy1 and nsy-7 mutants (in the absence of additional analyses) and the confusion it could trigger, we have decided to remove these non-essential data of the manuscript.

Reviewer #3 (Public review):

Summary:

Thapliyal and Glauser show that hunger alters how C. elegans responds to noxious thermal stimuli. Using targeted neural ablation, mutant analysis, and live-cell functional imaging, the authors demonstrate that hunger changes the properties of AWC sensory neurons, which sense noxious heat. The authors further show that the effects of hunger on nociception require ASI neurons, which are known to respond to hunger and mediate the effects of food deprivation on behavior. Finally, the study uses mutant analysis to implicate glutamate and specific neuropeptides in thermal nociception and in the modulation of nociceptors by hungerresponsive neurons.

Strengths:

The study clearly shows a strong effect of hunger on nociception and documents a striking effect of hunger on the intrinsic properties of AWC sensory neurons, which respond to noxious heat. The study also clearly and compellingly demonstrates that ablation of hunger-responsive ASI neurons blocks the effects of hunger on nociceptive AWCs. These data, which constitute the kernel of the manuscript, are striking and exciting.

Weaknesses:

The study has some weaknesses that the authors should address.

(1) Ablation of AWC neurons alters the basal sensitivity to noxious heat stimuli. This should be clearly noted in the description of the result and warrants some discussion.

We thank this reviewer for raising this legitimate point. We have clarified this aspect in the results section of the revised manuscript, reading as follows:

“Removal of AWC nearly abolished heat-evoked reversal behavior across all stimulus intensities and timepoints (Figure 2B and E). While one should keep in mind that potential indirect developmental effects might take place in neuro-ablation lines, this observation suggests that AWC plays an essential role in mediating the thermonociceptive response under both early food deprivation and prolonged starvation. Notably, in AWC-ablated animals, the residual response level was unaffected by starvation, suggesting that AWC might also be required for the expression of starvation-dependent plasticity.”

The contrast with known function of the best-characterized sensory neurons mediating thermal nociception (AFD and FLP) is discussed as follows:

“...Therefore, noxious heat-evoked activity in AWC varies widely according to context, which is in line with previous literature [18, 20, 41]. Interestingly, the role of AWC is distinct from that of AFD and FLP neurons, which are canonically linked to thermosensation and nociception [5, 14, 42, 43], but contribute only modestly to heat-evoked behavior in our assay conditions with between 1 and 6 hrs of food deprivation.”

(2) Throughout the study, it seems that data are replotted in multiple figure panels. The authors should clearly indicate in the figure legends when this occurs. Also, the authors should ensure that statistical tests requiring multiple comparisons are correctly implemented and reflect the number of times experimental data are compared to a single set of control data.

Thanks for raising this important point. We have clarified this aspect in the revised figure legends of the manuscript, and in the method section. In some instances, we reconducted some analyses to be perfectly rigorous in multiple comparison accounting. This did not lead to significantly different conclusions. The one exception was that the small effect of eat-4 mutation on spontaneous reversal went below significance threshold. We therefore removed this aspect of the result reporting and of the corresponding interpretation scheme, which became slightly simpler (Figure 3). Globally, this makes the story more focused.

(3) How ASIs modulate AWCs remains unclear. The authors find that loss of INS-6, an insulin-like peptide provided by ASIs, partially recapitulates the effect of ASI ablation. This observation is not further developed, and instead, the authors characterize other secreted factors that seem to mediate sensitization of animals to noxious heat stimuli. While it is interesting that there are multiple opposing inputs into the nociceptor circuit, the essential connection between ASIs and AWCs that underlies the foundational observations in Figures 1 and 2 is not sufficiently characterized.

Whereas we agree that how ASI modulates AWCs is only partially solved by our study, we should emphasize that our work identified two ASI-expressed neuropeptides that function to decrease reversal response after starvation: INS-32 and NLP-18. We initially set a lower priority on INS-6 because the reversal response level in starved mutants appeared lower than that in nlp-18 and ins-32. It is important to note that ins-32 and nlp-18 are not ‘generally potentiated’ mutants, but display reversal upregulation selectively following starvation, which placed them as strong candidates to selectively mediate ASI regulation. This said, it is also true that these two mutants (and ins-6 too) display reduced responsiveness at the early food deprivation time point. Therefore, none of the neuropeptide mutants was strictly identical to ASI ablated line, suggesting that the peptides might also work via non-ASI cells at the early food deprivation timepoint.

Following this reviewer’s comment, we have attempted to complement our story with the idea of using a similar approach and rescue ins-6 with its endogenous promoter or ASI-specific promoter. Unfortunately, we failed to obtain rescue effects, and therefore these data (with a negative result) remain inconclusive (as we cannot guarantee that the rescue constructs were functional). We decided to keep these data aside in the revised manuscript. Globally, our point made graphically in Figure 7F remains valid. We have complemented the figure legend to mention that INS-6 could also potentially work from ASI, but it is not depicted as no ASI-specific data are available. In summary, our data suggests that the connection between ASI and AWC(s) might be established by the integrated action of multiple peptides and their receptors. Further calcium imaging experiments in single, double and potentially triple mutant(s) of peptides and receptors would be required go deeper in this question, which could be performed in future work.

“...Additional neuropeptides (such as INS-6) may also be involved, but in the absence of direct evidence for their origin from ASI, they were not included in this scheme.”

(4) The assertion that 'starvation reshapes AWC responses from deterministic to stochastic' is not clearly supported by the data. AWC neurons seem capable of showing different responses to thermal stimuli, and the probabilities associated with these responses change after fasting. The different kinds of responses are seen under basal and fasted conditions.

We thank this reviewer for the comment. There is an activity response shift that is quite solidly described, including with new quantitative analyses of distributions (histograms in new Fig. 4CD and new Fig. 5C-D, accompanied by Kruskal-Wallis tests). Yet, we totally agree that the wording choice was inappropriate. We have furthermore changed our terminology to avoid using the terms “stochastic” or “deterministic” that were indeed a cause of confusion. We now use the terms “stimulus-locked responses” and describe the shift as “shift from mostly excitatory responses to a mix of both excitatory and inhibitory responses”. We have also included detailed methodology for characterization of traces and statistical analysis in the revised method section of the manuscript, together with the new analyses on peak polarity distribution.

Recommendations for the authors:

Reviewing Editor Comments:

The reviewers agree that the study is clearly presented and makes good use of behavioral, genetic, and imaging approaches to link starvation state with changes in AWC and ASI function. To strengthen the manuscript and ensure clarity for readers, we ask you to address the following points in revision:

(1) Positioning and citations.

Clarify how your findings relate to Takeishi 2020, where the opposite trend in AWC activity was reported, and make a clear distinction between thermonociception and thermotaxis. The introduction should also include additional citations in two specific areas: prior work on AWC and noxious thermal stimuli, and studies demonstrating starvation-dependent behavioral changes via altered neuropeptide release (e.g., Banerjee 2023; Rengarajan 2019).

We have clarified this aspect with extension of the work cited in the introduction and extensive rewriting of the discussion sections.

Our data shows that mechanisms underlying starvation dependent changes in thermonociception and thermotaxis show differences at the molecular, cellular and circuit levels. First, we see a decrease in average response in AWC neurons due to shift from mostly excitatory to a mix of excitatory and inhibitory responses in response to noxious heat after starvation, while previous study found an increase in response probability of AWCs after starvation (Takeishi et al, 2020). Second, contrary to thermotaxis behavior ASI neurons are required to orchestrate starvation evoked plasticity in thermonociception. And, finally, previous study found that INS-1 signaling from the intestine regulates thermotaxis behavioral plasticity while INS-1 peptide does not seem critical in regulating starvation-dependent thermonociceptive plasticity (new data in Figure 6 supplement 1).

We have revised the introduction section of the manuscript and cited previous relevant research on AWC and noxious thermal stimuli and studies demonstrating starvationdependent behavioral changes via altered neuropeptide release including the papers suggested by reviewers. The extended discussion section reads as follows:

“Starvation regulates thermonociceptive and negative thermotaxis plasticity via at least partly different mechanisms

Previous studies showed that AWC plays an important role in starvationdependent plasticity in the negative thermotaxis behavior in an innocuous thermal range between 15 and 25°C [26, 33]. Negative thermotaxis involves the detection of thermal changes created by animal movement in spatial thermogradient (0.5°C/cm), the magnitude of the expected thermal changes approximating 0.01°C/s [26]. The starvation impact on negative thermotaxis was shown to (i) involve an up-regulation of AWC cell activity, (ii) rely on INS1 neuropeptide produced in the intestine and (iii) to occur independently of ASI neurons. In contrast, our study used thermo-nociceptive stimuli, with faster raising thermal slopes (~0.5-2°C/s, hence 50-200 times faster than those occurring for thermotaxis) and covering noxious temperatures (up to 28°C). Our results indicate that the regulation of thermo-nociceptive response by starvation (i) is linked to a shift in the distribution of AWC activity response polarities from mostly excitatory to a mix of excitatory and inhibitory response, (ii) relies on ASI and specific neuropeptide produced in ASI, and (iii) works independently of INS-1 neuropeptide. Therefore, our study complements our understanding of the modulation of temperature-dependent behavior in C. elegans with previously undocumented mechanisms at the circuit, cellular and molecular levels.”

(2) ASI → AWC mechanism.

Because ASI is central to your conclusions, please expand on how ASI is thought to act on AWC and/or other neurons. If an additional ASI signal is proposed beyond INS-32/NLP-18, mark this as speculative unless further rationale can be provided, and adjust the model figure accordingly.

We have revised Figure 8 and its legend to clarify what is still hypothetical in the way ASI could affect AWC activity patterns and reversals. Note that the figure was also modified to integrate the conclusions made from epistasis analysis of eat-4 and flp-6.

(3) AWC response description.

The data support a shift in response distributions rather than a categorical switch from "deterministic to stochastic." Please adjust the language accordingly and provide a clear description of how traces were classified as "up, variable, or down," ideally with a quantification of the distributional shift.

We agree that the term “stochastic” can convey different things, and because it was used for something different for AWC in the past, we should have avoided it. We have revised the nomenclature. What we observe can indeed be better described as a shift in the response polarity distribution. The article was revised accordingly. We also included the quantitative analysis and histogram representation, suggested in one of the specific comments, and added detailed methodology on the categorization of traces.

When ASI is intact, we see a shift from mostly excitatory responses to an ~equal mix of excitatory and inhibitory responses (new Figure 4C-D). This effect is lost when ASI is ablated (new Figure 5C-D).

(4) Presentation and statistics.

In figure legends, indicate where datasets are replotted across panels and confirm that multiple-comparison corrections take account of repeated comparisons to the same controls.

We have included these details in the revised figure legends, and a statement in the method section.

(5) Methods clarity.

Provide justification for using 1-h off-food as the baseline, with quantification of baseline mobility/reversal rates. Expand the calcium-imaging methods to describe the processing pipeline (ΔR calculation, baseline period, drift correction), and add a brief rationale if the approach deviates from common normalization procedures. Clarify how cell ablations were performed and verified for specificity, and how cell-specific rescues were confirmed. Please also acknowledge the potential for developmental compensation with chronic ablation.

The justification of using 1hr off-food as baseline was made more prominent in the revised manuscript.

Revised result section:

“Starvation downregulates thermonociceptive responses in C. elegans

To assess how the feeding state modulates thermonociceptive behavior in C. elegans, we compared responses across different durations of food deprivation (Figure 1A). Synchronized first-day adult animals were stimulated with a series of 4-s infrared pulses of increasing heating power (100, 200, 300, 400 W), causing temperature increase of +2°C, +4°C +6°C and 8°C at the surface of the plate (Figure 1A). Fed animals on food produced robust heat-evoked reversal response to heat, but they also displayed a very elevated baseline of spontaneous reversals (~38%). A 1-hour off-food condition reduced spontaneous reversals (from ~38% to ~10%) and led to an attenuation of heat-evoked responses at low stimulus intensities, while responses to stronger stimuli remained comparable to those of fed animals. More prolonged food deprivation led to a striking progressive reduction in thermonociceptive responses at every heating level, with responses after 6 hours of starvation approaching baseline spontaneous reversal rates (Figure 1B and C). This suggests a robust inhibition of nociceptive behavior caused by prolonged starvation. To determine whether this attenuation was due to the absence of nutrients or chemosensory cues, we conducted similar starvation experiments in the presence of food odor, with OP50 bacteria present on the petri dish lid (Figure 1D). The reduction in thermonociceptive response persisted, indicating that the effect is driven by the internal starvation state rather than external olfactory input.

Although fed animals showed high sensitivity to noxious heat, they also displayed an elevated baseline of spontaneous reversals, which limited their utility as a control group by strongly reducing the dynamic range of heat-evoked reversal quantification and by complicating the quantitative comparison with food-deprivation conditions with much-reduced reversal baseline (Figure 1B). In addition, technical limitations in our calcium imaging setup would have prevented the intended follow-up analyses in fed animals. Based on these observations and technical considerations, we focused subsequent analyses, aiming at dissecting the circuit and molecular underpinnings of starvation-dependent plasticity, to the comparison of two off-food conditions with similar spontaneous reversal baseline: the early food deprivation condition (1-hour off-food, with high responsiveness to noxious heat) and the prolonged starvation (6-hour off-food with almost abolished noxious heat responsiveness).”

In addition, the method section was modified as follows:

- Calcium imaging details were added regarding ΔR calculation, baseline period, drift correction.

- We now explicitly refer to the original respective articles describing the neuroablation lines.

- We clarify that cell-specific transgene expression for rescue was confirmed using SL2::mCherry co-marker

In the result section, we now explicitly address potential developmental compensation in genetic ablation backgrounds in the result section as follows: “…one should keep in mind that potential indirect developmental effects might take place in neuro-ablation lines”.

Reviewer #1 (Recommendations for the authors):

(1) The data availability statement is missing from the reviewed manuscript and should be included.

Thanks, we have included the data availability statement in the revised manuscript.

(2) We request additional information on how n's were determined for individual experiments, as well as the inclusion of post-hoc power measurements for all quantification.

n were determined in agreement with previous studies using similar measures. No a priori power analyses were performed. A posteriori power analyses are not informative beyond the reported effect sizes and p-values (now reported in File S2). We clarified this in the statistical subsection of the method section.

(3) In many cases, the specific statistical tests used are not clear or justified; more details should be provided, including the non-post-hoc test used. Are all tests one-way ANOVAs? For comparisons across genotype and starvation duration, two-way ANOVAs would likely be more appropriate. Also, the authors switch between Bonferroni post-hoc tests and Holm-Bonferroni post-hoc tests. What determined the use of one versus another?

We have now clarified the statistical analyses used and provided full details in File S2. We have now more systematically applied two-way ANOVAs across all relevant analyses (with detailed parameters reported in File S2). When particularly relevant (e.g epistasis analysis between eat-4 and flp-6 mutations) the results of the two-way ANOVAs, is also explicitly stated in the result section.

We also note that all multiple-comparison corrections were performed using the Bonferroni method. Previous mentions of Holm-Bonferroni correction were inaccuracies, and we apologize for this confusion; these mentions have now been corrected throughout the manuscript.

(4) The use of heating power instead of the temperature experienced by the worms is an unwelcome abstraction. We strongly recommend revisiting that choice.

We do agree. We have revised the figures to label the axis with temperature increase.

(5) For calcium imaging, how are the traces categorized into "calcium up", "calcium down", or "no change"? Were those determined blindly - i.e., by individuals unaware of the experimental condition? Did the response direction need to be consistent across different temperatures? Did the change from baseline need to hit a specific threshold, consistent with previous studies in the field (i.e., +/- 3xSD for a minimum amount of time)? We encourage the authors to include these details in their methods section.

We have complemented the method section to clarify the criteria for the qualitative classification of traces. More importantly, new quantitative peak polarities comparisons were added (see specific points below and above, about histograms).

(6) For the experiments showing that exposure to food odor does not prevent response reduction, we suggest that feeding worms heat-killed bacteria would be a helpful control for the importance of bacterial nutritional status. In addition, showing that the impact of starvation was reversible with re-feeding would have been a useful experiment in line with standard experimental design in the starvation field.

Thanks, indeed with our current work we cannot pinpoint the role of additional sensory cues (except food odor) to be mediating starvation-evoked plasticity. Together with refeeding, these are all extremely interesting questions that we aim to answer and potentially link with ASI and AWC activity in our future work.

(7) For the various AWC rescue experiments, we found it curious that there wasn't an AWCon+off rescue, only each neuron individually.

Previous studies have identified similar or opposite responses of both AWCs for distinct sensory cues. Though our calcium imaging experiments point to both AWC on and off having similar response patterns to heat, we cannot rule out the possibility that their output (ability of control reversals) is distinct possibly due to recruited neuromodulators. Therefore, in the present work, we examined where these cell types act via the same or distinct combinations of neuromodulators to control reversals.

Reviewer #2 (Recommendations for the authors):

Experiments suggested:

(1) The authors should look into the Ca-dynamics in ASI.

How does the spontaneous and heat-evoked activity of ASI differ in fed, early food-deprivation and prolong starvation and its link to releases of neuromodulators, modified AWC activity to alter output of thermal nociception are very interesting questions. However, these questions are extremely exploratory (see argumentation above in response to the public review) and addressing them goes beyond the scope of our current manuscript.

(2) The authors should check AWC activity in ins-32 and nlp-18 mutant animals.

In this study, we focused on the roles of INS-32 and NLP-18 released from ASI in modulating heat-evoked reversals, as these mutants exhibit relatively strong behavioral phenotypes. However, these effects remain less pronounced than those observed following ASI ablation. In addition, we cannot exclude the contribution of additional signaling molecules, including INS-6 and other neuropeptides.

A comprehensive analysis of AWC activity in this context would require calcium imaging across multiple genetic backgrounds, including single, double, and potentially higher order peptide and receptor mutants, combined with cell-specific rescue experiments. While we appreciate the suggestion, such an approach would represent a substantial extension of the present work and will be important to pursue in future studies to further elucidate the underlying mechanisms.

(3) Short-term food deprivation completely eliminated heat heat-induced reversal response to 100W stimulus, while the response to 400W stimulus remained unaffected. This suggests fed, short-term starvation, and prolonged starvation are three distinct states, and authors should also test the response of AWC and ASI ablated animals in the fed conditions.

We agree that analyzing thermal nociception in fed states, in addition to short-term and prolonged food deprivation states is an important and interesting question, as these 3 conditions likely represent 3 distinct internal states that may recruit different neural pathways.

Several reasons led us to set the fed condition aside for this study, and we realize we insufficiently explain them in the initial manuscript. There are 2 main reasons.

(1) It is of paramount importance to consider the ‘baseline’ reversal rate (spontaneous reversals not triggered by heat, but visible in our dataset as the first point in the ‘dose-response’ curve). In Fed animals spontaneous reversal rate is very high (~38%) compared to the 1hr and 6hr food-deprivation conditions (>10%). This has two consequences: first a decreased dynamic range for quantify heat-evoked reversal, and, second, the difficulty in judging quantitative differences in heat-evoked reversals with such major differences in baseline reversals.

(2) Experimentally, assessing calcium responses in truly fed animals presents technical challenges. With our current setup, animals must be removed from food for at least ~5 minutes prior to recording (followed by ~5 minutes of imaging), which effectively corresponds to a “freshly starved” condition rather than a fully fed state. While previous studies have used serotonin to mimic aspects of the fed state, such manipulations can be difficult to interpret in this context.

A systematic comparison including fully fed animals, as well as AWC- and ASI-ablated conditions across these states, would be a valuable direction for future work, in particular once the methodological barriers associated with point 2, have been overcome.

We have clarified these choices in the result section as follows:

“Starvation downregulates thermonociceptive responses in C. elegans

To assess how the feeding state modulates thermonociceptive behavior in C. elegans, we compared responses across different durations of food deprivation (Figure 1A). Synchronized first-day adult animals were stimulated with a series of 4-s infrared pulses of increasing heating power (100, 200, 300, 400 W), causing temperature increase of +2°C, +4°C +6°C and 8°C at the surface of the plate (Figure 1A). Fed animals on food produced robust heat-evoked reversal response to heat, but they also displayed a very elevated baseline of spontaneous reversals (~38%). A 1-hour off-food condition reduced spontaneous reversals (from ~38% to ~10%) and led to an attenuation of heat-evoked responses at low stimulus intensities, while responses to stronger stimuli remained comparable to those of fed animals. More prolonged food deprivation led to a striking progressive reduction in thermonociceptive responses at every heating level, with responses after 6 hours of starvation approaching baseline spontaneous reversal rates (Figure 1B and C). This suggests a robust inhibition of nociceptive behavior caused by prolonged starvation. To determine whether this attenuation was due to the absence of nutrients or chemosensory cues, we conducted similar starvation experiments in the presence of food odor, with OP50 bacteria present on the petri dish lid (Figure 1D). The reduction in thermonociceptive response persisted, indicating that the effect is driven by the internal starvation state rather than external olfactory input.

Although fed animals showed high sensitivity to noxious heat, they also displayed an elevated baseline of spontaneous reversals, which limited their utility as a control group by strongly reducing the dynamic range of heat-evoked reversal quantification and by complicating the quantitative comparison with food-deprivation conditions with muchreduced reversal baseline (Figure 1B). In addition, technical limitations in our calcium imaging setup would have prevented the intended follow-up analyses in fed animals. Based on these observations and technical considerations, we focused subsequent analyses, aiming at dissecting the circuit and molecular underpinnings of starvationdependent plasticity, to the comparison of two off-food conditions with similar spontaneous reversal baseline: the early food deprivation condition (1-hour off-food, with high responsiveness to noxious heat) and the prolonged starvation (6-hour off-food with almost abolished noxious heat responsiveness).”

(4) The authors should test the effect of ins-1 in noxious heat-mediated dynamic reversal behavior.

We thank the reviewer for this valuable suggestion. We have now tested the phenotype of ins-1 mutants in our paradigm. Our data shows that INS-1 neuropeptide is not critical in mediating starvation-evoked thermonociceptive plasticity unlike plasticity in thermotaxis behavior (New Figure 6 Sup1). This molecular aspect adds to our initially presented evidence at the cell activity and circuit levels, that noxiousevoked reversal and thermotaxis behaviors are regulated in a clearly separable manner.

(5) Whether Glutamate and flp-6 work in parallel or in the same pathway to regulate reversals?

We thank the reviewer for this question. We tested eat-4; flp-6 double mutants and found:

(1) After short-term food deprivation (1hr), flp-6 and eat-4 separately contribute to heat-evoked reversal at high & low heat and they act in parallel pathways to explain ~90% of animal responsiveness (new version of Fig 3)

Corresponding new text:

“Next, we focused on eat-4 and flp-6 mutants, showing the strongest phenotype. We addressed whether glutamate and FLP-6 signaling act dependently of each other in controlling heat-evoked reversal, by testing eat-4; flp-6 double mutants. The residual response seen in each single mutant (Figure 3 A and B) was almost entirely abolished in the double mutant (Figure 3C). A two-way ANOVA for the highest heat stimuli with eat-4 and flp-6 genotypes as factors (two levels each: mutant or wild type) showed significant main effects of eat-4 (F_(1,67) =48.70, p<.001, η²p=0.421) and flp-6 (F_(1,67) =58.50, p<.001, ηp=0.466), respectively, but no interaction effects (F_(1,67) =0.093, p=.761, η²p=0.001). The significant cumulative effect of the two mutations indicates that the two signaling pathways act mostly independently of each other to mediate heat-evoked reversals.”

(2) After prolong starvation (6hr), flp-6 mutation has a dominant impact on plasticity and eat-4 mutation cannot cause loss of plasticity, pointing to a hierarchy in this context (new version of Fig. 6, including revised hierarchy in the model in panel G, and also revised model in Fig.

8).

Corresponding revised text:

“Second, we tested whether starvation-dependent plasticity was preserved in eat-4 and flp6 mutant backgrounds, which we had suggested to represent the main AWC transmitters controlling heat-evoked reversals under the early food deprivation condition (Figure 3). Even if the heat-evoked response upon early food-deprivation was reduced relative to wild type in flp-6 mutants, a significant further decline was seen after prolonged starvation (Figure 6B). These results indicate that starvation-dependent plasticity can operate independently of FLP-6. In contrast, eat-4 mutants displayed markedly elevated heat-evoked responses after prolonged starvation, even exceeding the response level seen in the early food deprivation condition for low heat stimuli (Figure 6C). This potentiated response in eat-4 mutants was entirely dependent of an intact FLP-6 signaling, since reversal responses in eat-4; flp-6 mutants were entirely abolished, like in flp-6 single mutant (Figure 6C-E, a two-way ANOVA indicating a significant interaction effect between the two mutations: F_(1,70) =0.093, p<.001, η²p=0.247). Moreover, the potentiated response in eat-4 single mutant could not be rescued by expressing eat-4 rescue transgene selectively in either AWC^OFF or AWC^ON neurons (Figure 6F). Interestingly, AWC^OFF-specific rescue produced a further potentiation of heat-evoked reversal response to high heat stimuli (Figure 6F, 6 and 8°C thermal increases), aggravating the phenotype of eat-4 mutants. These results are consistent with a model in which glutamatergic signaling regulates heat-evoked reversals in starved animals via two bidirectional drives (Figure 6G). On the one hand, glutamatergic signaling—originating from AWC^OFF—up-regulates reversals in response to high heat stimuli, thus contributing to prevent starvation-induced thermonociceptive plasticity. On the other hand, glutamatergic signaling—originating from neurons other than AWC— down-regulates reversals over a broad range of heat intensities, thus promoting starvation-induced thermonociceptive plasticity. The latter glutamatergic signaling inhibitory effect seems to be more dominant and to depend on intact FLP-6 signaling.”

(6) The authors should perform flp-6 and eat-4 mutant/rescue experiments in the nsy-1 and nsy-7 background to clarify the results.

Our results indicate that glutamate release via EAT-4 from both AWC^ON and AWC^OFF, as well as FLP-6 from AWC^OFF, contributes to heat-evoked reversals regulation. The experiments suggested by the reviewer would, in principle, provide further insight into the interaction between AWC subtype identity and the respective roles of glutamatergic and peptidergic signaling.

However, as discussed in more details above in the public review, the extent and nature of AWC^ON/OFF remodeling in nsy-1 and nsy-7 mutant backgrounds remain incompletely understood. This introduces significant uncertainty in interpreting results obtained from combining these mutations with eat-4 and flp-6 manipulations. As a result, such experiments would be difficult to interpret in a definitive manner at this stage.

We therefore consider this an important direction for future work, once the roles of nsy1 and nsy-7 in AWC subtype specification and function are more clearly established. As our preliminary results with nsy-1 and nsy-7 mutants added more confusion than clarity, we have chosen to set them aside (former Fig. 3-figure supplement 2 has been removed).

Minor comments:

(1) What is food odor? The experiment should be clearly mentioned.

Thanks for spotting this unintended omission. Food odor experiments were performed by adding OP50 bacteria on the inward side of the petri dish lid instead of the NGM surface. We have added this description in the method section of the revised manuscript.

(2) Panel 3C is coming before 3B. This should be rearranged.

Thanks for pointing this out. Panel arrangement was entirely reorganized in revise Fig. 3, with the addition of eat-4 x flp-6 genetic interaction analysis.

(3) In Figure 6, if the panels are arranged horizontally, it would be easier to follow.

Thanks for pointing this out. Panel arrangement was entirely reorganized in revise Fig. 6, with the addition of eat-4 flp-6 genetic interaction analysis.

Reviewer #3 (Recommendations for the authors):

(1) The authors should consider moving measurements of AFD-ablated animals into the main Figure 1. AFD is a well-known thermosensor, and it is worth showing that responses to noxious thermal stimuli persist in animals lacking AFD.

Thank you for this suggestion. We have moved measurements of AFDablated animals to the main figure (revised Fig. 2).

(2) AFD ablation does affect responses to noxious heat. The authors could consider ablating/silencing AFDs and AWCs simultaneously to determine whether these two neurontypes account for the behavior.

We agree that investigating the combinatorial contributions of thermosensory neurons, including AFD and AWC, to thermal nociception is an important and interesting question. In principle, simultaneous ablation or silencing of these neuron types could indeed reveal unexpected interactions.

In our experimental paradigm, however, we observe only a minimal contribution of AFD neurons to heat-evoked reversals, whereas ablation of AWC nearly abolishes the response. Based on these observations, we chose to focus the present study on AWC, which appears to play a more prominent role in this behavior.

A more detailed dissection of the potential interactions between AFD and AWC, including combinatorial manipulations, would be a valuable direction for future work. In particular, the possibility that AFD exerts a modulatory influence remains an interesting hypothesis to explore.

(3) Given that EAT-4/VGLUT and FLP-6 neuropeptides each contribute to nociception, the authors should consider testing an eat-4; flp-6 double mutant to determine whether this combination of neurochemical signals accounts for AWC signaling to downstream circuits.

We thank the reviewer for this suggestion, which is similar to point 5 of Reviewer 2 (above).

We tested eat-4; flp-6 double mutants and found:

(1) After short-term food deprivation (1hr), flp-6 and eat-4 separately contribute to heat evoked reversal at high & low heat and they act in parallel pathways to explain ~90% of animal responsiveness (new version of Fig 3)

Corresponding new text:

“Next, we focused on eat-4 and flp-6 mutants, showing the strongest phenotype. We addressed whether glutamate and FLP-6 signaling act dependently of each other in controlling heat-evoked reversal, by testing eat-4;flp-6 double mutants. The residual response seen in each single mutant (Figure 3 A and B) was almost entirely abolished in the double mutant (Figure 3C). A two-way ANOVA for the highest heat stimuli with eat-4 and flp-6 genotypes as factors (two levels each: mutant or wild type) showed significant main effects of eat-4 (F_(1,67) =48.70, p<.001, η²p=0.421) and flp-6 (F_(1,67) =58.50, p<.001, η²p=0.466), respectively, but no interaction effects (F_(1,67) =0.093, p=.761, η²p=0.001). The significant cumulative effect of the two mutations indicates that the two signaling pathways act mostly independently of each other to mediate heat-evoked reversals.”

(2) After prolong starvation (6hr), flp-6 mutation has a dominant impact on plasticity and eat-4 mutation cannot cause loss of plasticity, pointing to a hierarchy in this context (new version of Fig. 6, including revised hierarchy in the model in panel G, and also revised model in Fig. 8).

Corresponding revised text:

“Second, we tested whether starvation-dependent plasticity was preserved in eat-4 and flp6 mutant backgrounds, which we had suggested to represent the main AWC transmitters controlling heat-evoked reversals under the early food deprivation condition (Figure 3). Even if the heat-evoked response upon early food deprivation was reduced relative to wild type in flp-6 mutants, a significant further decline was seen after prolonged starvation (Figure 6B). These results indicate that starvation-dependent plasticity can operate independently of FLP-6. In contrast, eat-4 mutants displayed markedly elevated heat-evoked responses after prolonged starvation, even exceeding the response level seen in the early food deprivation condition for low heat stimuli (Figure 6C). This potentiated response in eat-4 mutants was entirely dependent of an intact FLP-6 signaling, since reversal responses in eat-4; flp-6 mutants were entirely abolished, like in flp-6 single mutant (Figure 6C-E, a two-way ANOVA indicating a significant interaction effect between the two mutations: F_(1,70) =0.093, p<.001, η²p=0.247). Moreover, the potentiated response in eat-4 single mutant could not be rescued by expressing eat-4 rescue transgene selectively in either AWC^OFF or AWC^ON neurons (Figure 6F). Interestingly, AWC^OFF-specific rescue produced a further potentiation of heat-evoked reversal response to high heat stimuli (Figure 6F, 6 and 8°C thermal increases), aggravating the phenotype of eat-4 mutants. These results are consistent with a model in which glutamatergic signaling regulates heat-evoked reversals in starved animals via two bidirectional drives (Figure 6G). On the one hand, glutamatergic signaling—originating from AWC^OFF—up-regulates reversals in response to high heat stimuli, thus contributing to prevent starvation-induced thermonociceptive plasticity. On the other hand, glutamatergic signaling—originating from neurons other than AWC— down-regulates reversals over a broad range of heat intensities, thus promoting starvation-induced thermonociceptive plasticity. The latter glutamatergic signaling inhibitory effect seems to be more dominant and to depend on intact FLP-6 signaling.”

(4) The authors should consider representing AWC responses to thermal stimuli as histograms to illustrate how fasting increases the probability of some responses and decreases the probability of others.

We thank this reviewer for the suggestion. The proposed histograms nicely convey the concept of “shift in response polarity distribution” that we observed (using the new terminology we now use, instead of using the term “stochastic”). To create such histograms, we computed the magnitude of the peaks on a trial-by-trial basis. When ASI is intact, we see a shift from mostly excitatory responses to an ~equal mix of excitatory and inhibitory responses (new Figure 4C-D). This effect is lost when ASI is ablated (new Figure 5C-D).

(5) It seems important to better understand the ins-6 mutant phenotype and determine whether ASI-to-AWC signaling involves this insulin-like peptide (ILP). The authors should consider using some of the tools available for disrupting ILP signaling to more clearly demonstrate that a specific neurochemical signal mediates modulation of AWCs by ASIs.

Our work identified two ASI-expressed neuropeptides that function to decrease reversal response after starvation: INS-32 and NLP-18. We initially set a lower priority on INS-6 because the reversal response level in starved mutants appeared lower than that in nlp-18 and ins-32. It is important to note that ins-32 and nlp-18 are not ‘generally potentiated’ mutants, but display reversal up-regulation selectively following starvation, which placed them as strong candidates to selectively mediate ASI regulation. This said, it is also true that these two mutants (and ins-6 too) display reduced responsiveness at the early food deprivation time point. Therefore, none of the neuropeptide mutants was strictly identical to ASI, suggesting that the peptides might also work via non-ASI cells at the early food deprivation timepoint (as follow up data indicated at least for nlp-18).

Following this reviewer’s comment (and the similar one in the public review), we have attempted to complement our story with the idea of using a similar approach and rescue ins-6 with its endogenous promoter or ASI-specific promoter. Unfortunately, we failed to obtain rescue effects, and therefore these data remain inconclusive (as we cannot guarantee that the rescue constructs were functional). We decided to keep these data aside in the revised manuscript. Globally, our point made graphically in Figure 7F remains valid. We have complemented the figure legend to mention that INS6 could also potentially work from ASI, but it is not depicted as no ASI-specific data are available. In summary, our data suggests that the connection between ASI and AWC(s) might be established by the integrated action of multiple peptides and their receptors. Further calcium imaging experiments in single, double and potentially triple mutant(s) of peptides and receptors would be required to go deeper in this question, which could be performed in future work.

eLife Assessment

This important study shows that an odorant that is typically thought of as a repellant actually activates both attractant and repellant olfactory neurons in C. elegans. Convincing evidence is provided that nematode worms can integrate signals in different sensory pathways to drive different behavioral responses to the same cue. These findings will be of interest to scientists interested in combinatorial coding in sensory systems.

Reviewer #1 (Public review):

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

The authors investigated the response of worms to the odorant 1-octanol (1-oct) using a combination of microfluidics-based behavioral analysis and whole-network calcium imaging. They hypothesized that 1-oct may be encoded through two simultaneous, opposing afferent pathways: a repulsive pathway driven by ASH, and an attractive pathway driven by AWC. And the ultimate chemotactic outcome is likely determined by the balance between these two pathways.

It is not surprising that 1-octanol is encoded as attractive at low concentrations and repulsive at higher concentrations. However, the novel aspect of this study is the discovery of the combinatorial coding of 1-oct in the periphery, where it serves as both an attractant and a repellent. Furthermore, the study uses this dual encoding as a model to explore the neural basis of sensory-driven behaviors at a whole-network scale in this organism. The basic conclusions of this study are well supported by the behavioral and imaging experiments, though there are certain aspects of the manuscript that would benefit from further clarification.

A key issue is that several previous studies have demonstrated a combinatorial and concentration-dependent coding of odorant sensing in the nematode peripheral nervous system. Specifically, ASH and AWC are the primary receptors for repellent and attractive responses, respectively. However, other neurons such as AWB, AWA, and ADL are also involved in the coding process. These neurons likely communicate with different interneurons to contribute to 1-oct-induced outputs. The authors' conclusion that loss of tax-4 reduces attractive responses and that osm-9 mutants reduce repulsive responses is not entirely convincing. TAX-4 is required for both AWC (an attractive neuron) and AWB (a repulsive neuron), and osm-9 is essential for ASH, ADL, and AWA (attraction-associated). Therefore, the observed effects on the attractive and repulsive responses could be more complex. Additionally, the interpretation of results involving the use of IAA to reduce the contribution of AWC at lower concentrations lacks clarity.

The authors did not observe any increased correlation between motor command interneurons and sensory neurons, which is consistent with the absence of a consistent relationship between state transitions and 1-oct application. Furthermore, they did not observe significant entrainment of AIB activity with the 2.2 mM 1-oct application. This might be due to the animals being anesthetized with 1 mM tetramisole hydrochloride, which could affect neural activity and/or feedback from locomotion.

Reviewer #2 (Public review):

Summary:

The authors used whole-network imaging to identify sensory neurons that responded to the repellant 1-octanol. While several olfactory neurons responded to the initial onset of odor pulses, two neurons consistently responded to all the pulses, ASH and AWC. ASH typically activates in response to repellants, and AWC typically activates in response to the removal of attractants. However, in this case, AWC activated in response to the removal of 1-octanol, which was unexpected because 1-octanol is a harmful repellant to the worm. The authors further investigated this phenomenon by testing different concentrations of 1-octanol in a chemotaxis assay and found that at lower (less harmful) concentrations the odor is actually an attractant, but becomes repulsive at higher concentrations. The amplitude of the ASH response appeared to be modulated by concentration, but this was not true for AWC. The authors propose a model where the behavioral response of the worm is the result of integrating these two opposing drives, where repulsion is a result of the increased ASH activity over-riding the positive drive from AWC. The authors further tested this theory by testing mutants that ablated the AWC response (tax-4 or AWC::HisCl) or ASH response (osm-9 or ASH::HisCl). The chemo-silencing (HisCl) and tax-4 experiments were consistent with their hypothesis, while the osm-9 mutation had a limited impact on chemotaxis behavior, highlighting the potential role of osm-9-independent signaling in ASH in response to 1-octanol. While the interneuron(s) that integrate these signals to influence behavior were not identified, the authors did find that increasing concentrations of 1-octanol did increase the likelihood of AVA activity, a neuron which drives reversals (and hence, behavioral repulsion).

Strengths:

This was simple and elegant work that identified specific neurons of interest which generated a hypothesis, which was further tested with mutants that altered neuronal activity. The authors performed both neuronal imaging and behavioral experiments to verify their claims.

Weaknesses:

The authors note that other sensory neurons likely contribute to 1-octanol chemotaxis. Given the NeuroPAL data, it would have been nice to identify these other neurons as well. However, the reviewer is aware that this is tangential to the primary focus of this study.

Reviewer #3 (Public review):

Summary:

This work describes how two chemosensory neurons in C. elegans drive opposite behaviors in response to a volatile cue. Because they have different concentration dependencies, this leads to different behavioral responses (attraction at low concentration and repulsion at high concentration). It has been known that many odorants that are attractive at low concentrations are aversive at high concentrations, and the implicated neurons (at least AWC for attraction and ASH for repulsion) have been well established. Nonetheless, by studying behavior and neural responses in a common context (odor pulses, as opposed to gradients) this provides a clear picture of how these sensory neurons may guide the dose dependent response by separately modulating odor entry and odor exit behaviors.

Strengths:

(1) This work provides good evidence that worms are attracted to low concentrations and repelled by high concentrations of 1-oct. Calcium imaging also makes it clear that dose-dependence of this response is stronger for ASH than AWC.

(2) This work presents calcium imaging and behavior with the same stimulus (sudden pulses in volatile odor concentration), while previous studies often focus on using neuronal responses to pulses to understand navigation of gentle gradients.

Weaknesses:

(1) As a whole it is not clear precisely how important AWC is (compared to other cells) for the attractive response (as the authors correctly acknowledge).

(2) The evidence that AIB minus AVA contains relevant information is weak. It appears the entrainment index in Fig. 6H for AIB-AVA could easily be explained by the negative entrainment between AVA and the stimulus (along with no effect or role for AIB). This is suggested by the similar p-values and similar distribution of random EIs (stretched and mirrored) between the first and last rows of this figure.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1 (Public review):

The authors investigated the response of worms to the odorant 1-octanol (1-oct) using a combination of microfluidics-based behavioral analysis and whole-network calcium imaging. They hypothesized that 1-oct may be encoded through two simultaneous, opposing afferent pathways: a repulsive pathway driven by ASH, and an attractive pathway driven by AWC. And the ultimate chemotactic outcome is likely determined by the balance between these two pathways.

It is not surprising that 1-octanol is encoded as attractive at low concentrations and repulsive at higher concentrations. However, the novel aspect of this study is the discovery of the combinatorial coding of 1-oct in the periphery, where it serves as both an attractant and a repellent. Furthermore, the study uses this dual encoding as a model to explore the neural basis of sensory-driven behaviors at a whole-network scale in this organism. The basic conclusions of this study are well supported by the behavioral and imaging experiments, though there are certain aspects of the manuscript that would benefit from further clarification.

A key issue is that several previous studies have demonstrated a combinatorial and concentration-dependent coding of odorant sensing in the nematode peripheral nervous system. Specifically, ASH and AWC are the primary receptors for repellent and attractive responses, respectively. However, other neurons such as AWB, AWA, and ADL are also involved in the coding process. These neurons likely communicate with different interneurons to contribute to 1-oct-induced outputs. The authors' conclusion that loss of tax-4 reduces attractive responses and that osm-9 mutants reduce repulsive responses is not entirely convincing. TAX-4 is required for both AWC (an attractive neuron) and AWB (a repulsive neuron), and osm-9 is essential for ASH, ADL, and AWA (attraction-associated). Therefore, the observed effects on the attractive and repulsive responses could be more complex. Additionally, the interpretation of results involving the use of IAA to reduce the contribution of AWC at lower concentrations lacks clarity.

The authors did not observe any increased correlation between motor command interneurons and sensory neurons, which is consistent with the absence of a consistent relationship between state transitions and 1-oct application. Furthermore, they did not observe significant entrainment of AIB activity with the 2.2 mM 1-oct application. This might be due to the animals being anesthetized with 1 mM tetramisole hydrochloride, which could affect neural activity and/or feedback from locomotion.

Comments on revisions:

The authors have addressed all my previously raised concerns.

Reviewer #2 (Public review):

Summary:

The authors used whole-network imaging to identify sensory neurons that responded to the repellant 1-octanol. While several olfactory neurons responded to the initial onset of odor pulses, two neurons consistently responded to all the pulses, ASH and AWC. ASH typically activates in response to repellants, and AWC typically activates in response to the removal of attractants. However in this case, AWC activated in response to the removal of 1-octanol, which was unexpected because 1-octanol is a harmful repellant to the worm. The authors further investigated this phenomenon by testing different concentrations of 1-octanol in a chemotaxis assay, and found that at lower (less harmful) concentrations the odor is actually an attractant, but becomes repulsive at higher concentrations. The amplitude of the ASH response appeared to be modulated by concentration, but this was not true for AWC. The authors propose a model where the behavioral response of the worm is the result of integrating these two opposing drives, where repulsion is a result of the increased ASH activity over-riding the positive drive from AWC. The authors further tested this theory by testing mutants that ablated the AWC response (tax-4 or AWC::HisCl) or ASH response (osm-9 or ASH::HisCl). The chemo-silencing (HisCl) and tax-4 experiments were consistent with their hypothesis, while the osm-9 mutation had a limited impact on chemotaxis behavior, highlighting the potential role of osm-9-independent signaling in ASH in response to 1-octanol. While the interneuron(s) that integrate these signals to influence behavior were not identified, the authors did find that increasing concentrations of 1-octanol did increase the likelihood of AVA activity, a neuron which drives reversals (and hence, behavioral repulsion).

Strengths:

This was simple and elegant work that identified specific neurons of interest which generated a hypothesis, which was further tested with mutants that altered neuronal activity. The authors performed both neuronal imaging and behavioral experiments to verify their claims.

Weaknesses:

The authors note that other sensory neurons likely contribute to 1-octanol chemotaxis. Given the NeuroPAL data, it would have been nice to identify these other neurons as well. However, the reviewer is aware that this is tangential to the primary focus of this study.

Reviewer #3 (Public review):

Summary:

This work describes how two chemosensory neurons in C. elegans drive opposite behaviors in response to a volatile cue. Because they have different concentration dependencies, this leads to different behavioral responses (attraction at low concentration and repulsion at high concentration). It has been known that many odorants that are attractive at low concentrations are aversive at high concentrations, and the implicated neurons (at least AWC for attraction and ASH for repulsion) have been well established. None the less, by studying behavior and neural responses in a common context (odor pulses, as opposed to gradients) this provides a clear picture of how these sensory neurons may guide the dose dependent response by separately modulating odor entry and odor exit behaviors.

Strengths:

(1) This work provides good evidence that worms are attracted to low concentrations and repelled by high concentrations of 1-oct. Calcium imaging also makes it clear that dose-dependence of this response is stronger for ASH than AWC.

(2) This work presents calcium imaging and behavior with the same stimulus (sudden pulses in volatile odor concentration), while previous studies often focus on using neuronal responses to pulses to understand navigation of gentle gradients.

Weaknesses:

(1) As a whole it is not clear precisely how important AWC is (compared to other cells) for the attractive response (as the authors correctly acknowledge).

(2) The evidence that AIB minus AVA contains relevant information is weak. It appears the entrainment index in Fig. 6H for AIB-AVA could easily be explained by the negative entrainment between AVA and the stimulus (along with no effect or role for AIB). This is suggested by the similar p-values and similar distribution of random EIs (stretched and mirrored) between the first and last rows of this figure.

(3) The model in Figure 7 would be strengthened if it was demonstrated that IAA is attractive when worms are saturated in a 1/10^4 concentration. Panel 7G (and ref. 39) indicate that 10^-4 IAA activates ASH, which would suggest a different explanation for the change from attraction to repulsion in 7C.

It was previously published that 1x10^-4 IAA is attractive in a similar microfluidics context (Albrecht and Bargmann), and we have confirmed its attractiveness in our experimental setup (not shown). Specifically, worms accumulate in zones of the arena where 1x10^-4 IAA is present: they are attracted upon first encounter and maintain attraction after prolonged exposure. A sentence to this effect has been added to Results (line 402).

Recommendations for the authors:

Reviewer #2 (Recommendations for the authors):

This looks great! My only small recommendation is to include the 1-octanol concentration used (2.2 mM) in Fig 4 F,G so that the reader can easily compare these data to Fig 4 D,E. You can do this in the figure or legend, though in the figure would be preferable (as you did for Fig 5 and Fig 6).

[1-oct ] added to Fig. 4D, E

Reviewer #3 (Recommendations for the authors):

(1) The AWC traces in Fig. 3C do not match, despite them being from the same animal. There appear to be too many ASH traces. NeuroPAL data/traces from other animals were not available on Zenodo when this review was prepared. This should be fixed.

Single-animal traces added to figure so reader can see the relationship in the single animal shown and across the full dataset. The averaged ASH/AWC traces are retained in the updated figure because this observation is central to the main hypothesis, so it is important to demonstrate it is a consistent phenomenon. Legend modified (lines 1016-17), ‘consistent’ added to Results (line 147). Data will be uploaded to Zenodo upon finalization of the Version of Record

(2) The normalization of traces in Fig. 4 is not clear (panel A is not min-max, as panel B appears to be). This normalization may be critical to the interpretation that ASH responds dose-dependently. The code linked on Zenodo was not available when this review was prepared. This should be fixed.

The reviewer correctly points out a slight error in normalization of the individual worm traces for the ascending concentration series (4A, left panel; the maximal value shown was slightly under 1.0). This is now corrected, interpretation unchanged.

(3) Chemogenetic data (4F,G and 5D) should specify the promoters used, as they are not ASH and AWC-specific.

Figures 4F, G and 5D altered to specify the promoters used, as requested. As before, the full range of neurons in which these promoters are active is stated in the Discussion

eLife Assessment

Feeding, the circadian rhythm, and the gut microbiota are all intimately linked, motivating new approaches to identify causal relationships while minimizing confounding factors. The authors employ an innovative combination of the osmotic laxative lactulose and a defined 3-member gut microbiota to acutely induce gut bacterial metabolism in mice during the daytime, resulting in changes in the ileal expression of clock genes and altered feeding behavior. Together, this study utilizes convincing methods to provide important new insights into the role of gut microbiota in the circadian rhythm, setting the stage for follow-on studies aimed at better understanding the mechanisms responsible.

Reviewer #2 (Public review):

Summary

The authors aimed to investigate how microbial metabolites, including hydrogen and short-chain fatty acids, influence feeding behavior and clock-gene expression in mice. Specifically, they examined these effects across different microbial environments, including a reduced-community model, germ-free mice, and specific-pathogen-free mice. The study addresses an important and poorly understood question concerning how microbial metabolism may influence host daily rhythms and feeding patterns.

Strengths

The manuscript presents a thoughtful and innovative investigation into the relationship between microbial metabolism, feeding behavior, and host clock-gene expression. A major strength is the use of a reduced microbial community together with real-time measurements of hydrogen production, which provides a useful experimental framework for separating microbial metabolic activity from direct host nutrient intake. The inclusion of germ-free and specific-pathogen-free mice also allows the authors to examine how these effects depend on microbial complexity.

The revised manuscript has addressed several concerns raised in the original review. The authors have clarified aspects of stool collection, provided additional methodological information, refined their use of the terms "circadian" and "diurnal," and added experiments examining the osmotic effects of lactulose across different microbial environments. These revisions improve the clarity and interpretation of the work. The finding that lactulose-induced microbial activity is associated with altered feeding behavior and clock-gene expression in the reduced-community model, but not consistently in germ-free or specific-pathogen-free mice, remains intriguing and highlights the complexity of microbial-host interactions.

Weaknesses

Despite these improvements, several important concerns remain unresolved. Most notably, the response regarding excluded food-intake measurements is insufficient. The authors report that the probability of excluding measurements differed significantly between treatment and control groups in the key feeding experiment shown in Figure 4A/S5A. However, they do not provide the underlying number or percentage of excluded observations, the direction of the imbalance, how many animals were affected, whether the exclusions occurred before or after treatment, or which exclusion criteria accounted for the removed values. Because food-intake rate was calculated from repeated measurements of cumulative intake, differential exclusion of observations could influence both the estimated slope and the reported treatment effect. The authors should provide these details and demonstrate that the principal feeding result is robust to the exclusion procedure.

There is also ambiguity surrounding the analysis and presentation of the quantitative polymerase chain reaction data. Figure 3B-C and Supplementary Figure 4A-C use conflicting mathematical labels, the raw cycle-threshold and replicate-level delta cycle-threshold values are not provided, and the rebuttal appears to conflate logarithmic transformation with exponentiation. I suspect that this may primarily reflect terminology or figure-labeling errors rather than an incorrect underlying analysis. The negative values shown in Figure 3 suggest that the authors may have plotted negative delta-delta cycle-threshold values, corresponding to log2 fold change. Nevertheless, the current description makes the workflow difficult to verify.

Several mechanistic and interpretive issues were acknowledged but only partially addressed. The authors appropriately softened their interpretation of the clock-gene findings and clarified that hormone concentrations were measured at only one time point. However, they did not provide baseline evidence that the positive and negative limbs of the clock-gene network are normally in counterphase, did not discuss the possible involvement of AMP-activated protein kinase, and did not address the limitation of measuring total rather than active glucagon-like peptide 1. These issues should be explicitly discussed as limitations of the current study and as priorities for future work.

Overall assessment

The authors have mostly achieved their aims by providing novel evidence that acute changes in microbial metabolism may influence feeding behavior and clock-gene expression in a simplified microbial environment. The experimental approach is creative, and the study has the potential to contribute meaningfully to the fields of microbiome research and circadian biology. However, unresolved concerns regarding differential data exclusion and the transparency of the gene-expression analysis currently limit confidence in the strength of the evidence supporting the principal conclusions.

Addressing these reporting and analytical issues would substantially strengthen the manuscript and improve its value to researchers studying microbial metabolism, feeding behavior, and host biological rhythms.

Reviewer #3 (Public review):

Summary:

In the manuscript by Greter, et al., entitled "Targeted induction of gut-microbial metabolism acutely affects feeding patterns and clock gene expression in the host" the authors investigate whether acute exposure to a non-nutritive disaccharide (lactulose) promotes microbial metabolism that feeds back onto the host to impact circadian networks. The premise of the study is interesting, and the experiments are thoughtfully designed to dissect these relationships. The evidence presented generally supports the authors' conclusions regarding the impact of lactulose administration during the fasting period, which is intended to mimic a feeding-associated perturbation of the gut microbiota, and its comparison with lactulose administration during the fed state. The studies employ complementary model systems, including germ-free mice, mice colonized with a simplified three-member microbial community, and conventionally colonized animals. These approaches support the authors' conclusions regarding the relationship between diurnal rhythms of microbial fermentation and host circadian clock gene networks. Overall, the work provides a useful experimental framework for developing a deeper mechanistic understanding of how microbial fermentation products contribute to diurnal host-microbe interactions.

Strengths:

Attempting to disentangle nutrient acquisition from microbial fermentation and its impact on diurnal dynamics of gut microbes on host circadian rhythms is an important step for providing insights into these host-microbe interactions.

The authors utilize a novel approach in leveraging lactulose coupled with germ-free animals and metabolic cages fitted with detectors that can measure microbial byproducts of fermentation, particularly hydrogen, in real time.

The authors consider several interesting aspects of lactulose delivery, including how it shifts osmotic balance as well as providing calculations that attempt to explain the caloric contribution of fermentation to the animal in the context of reduced food intake. This provides interesting fundamental insights into the role of microbial outputs on host metabolism.

The authors employ complementary systems, including a simplified three-member microbial community, providing insight into the minimal set of functionally distinct community members necessary to promote the rhythmic production of fermentation products that can affect host physiology.

Residual limitations:<br /> Hypothesis and study framing: The manuscript still does not clearly articulate a specific, testable hypothesis. While the Introduction provides motivation and objectives (e.g., line 53 onward), it remains unclear what precise hypothesis was being evaluated. A more explicit statement would strengthen the conceptual framework of the study.

Interpretation of circadian gene expression changes: The authors have not fully reconciled the differing effects of lactulose treatment on circadian gene expression in the 3MM and SPF settings. In particular, it remains unclear how the increased expression of certain circadian genes observed in lactulose-treated 3MM mice, particularly Cry1, relates to the decreased expression seen in SPF mice, and how the reduction in Arntl expression observed in lactulose-treated SPF mice fits within the proposed model. The authors acknowledge that resolving these mechanistic differences is beyond the scope of the current study, but the limitation should be discussed more explicitly.

Author response:

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

We thank the three reviewers for their encouraging and constructive comments. We have addressed them by increasing clarity of the writing and adding details to the methods section that were previously missing or not stated clearly enough. We have included additional experimental data. Specifically, we tested how the osmotic effect of lactulose depends on lactulose dosage and colonization state, added data on cecum sizes of mice with different microbiomes, and tested how lactulose treatment in the active phase affects feeding behavior.

Public Reviews:

Reviewer #1 (Public Review):

Greter et al. provide an interesting and creative use of lactulose as a "microbial metabolism" inducer, combined with tracking of H2 and other fermentation end products. The topic is timely and will likely be of broad interest to researchers studying nutrition, circadian rhythm, and gut microbiota. However, a couple of moderate to major concerns were noted that may impact the interpretation of the current data:

(1) Much of the data relies on housing gnotobiotic mice in metabolic cages, but I couldn't find any details of methods to assess contamination during multiple days of housing outside of gnotobiotic isolators/cages. Given the complexity of the metabolic cage system used, sterility would likely be incredibly challenging to achieve. More details needed to be included about how potential contamination of the mice was assessed, ideally with 16S rRNA gene sequencing data of the endpoint samples and/or qPCR for total colonization levels relative to the more targeted data shown.

We thank the reviewer for pointing out that we have not made the experimental setup clear in the text. One of the unique features of our metabolic cage setup is that the mice do not need to be housed outside gnotobiotic isolators, but that the whole system is placed inside an isolator. We have developed and published this system recently (Hoces et al, PLOS Biol 2022), including extensive testing for sterility/gnotobiosis. We have now adapted the main text to increase clarity on this issue (lines 76ff).

Given that 16S sequencing of germ-free mice will typically produce false-positive reads, we used Blautia pseudococcoides as an indicator strain for contamination. This strain is present in our SPF mouse colony, forms spores that are highly resilient to decontamination measures, and has been the most likely contaminant in our gnotobiotic system. We have checked for presence of this strain in the cecum content of all our animals at the end of each experiment, and only included experiments which had a B. pseudococcoides signal below threshold level. We have now added this information to the methods section (lines 389ff).

(2) The language could be softened to provide a more nuanced discussion of the results. While lactulose does seem to induce microbial metabolism it also could have direct effects on the host due to its osmotic activity or other off-target effects. Thus, it seems more precise to just refer to lactulose specifically in the figure titles and relevant text.

We have adapted all figure legends to not contain interpretations, but rather state what was done in the experiments shown in the figures. We have also adapted the text in multiple places to soften the language and avoid overinterpretation of our experimental results.

Additionally, the degree to which lactulose "disrupts the diurnal rhythm" isn't clear from the data shown, especially given that the markers of circadian rhythm rapidly recover from the perturbation. It is probably more precise to instead state that lactulose transiently induces fermentation during the light phase or something to that effect.

We tried to make the argument that what we call disruption of the diurnal rhythm is acute, meaning that it is not disrupting the rhythm "chronically" (i.e., for longer), but that it recovers rapidly from this transient disruption. Given the confusion this wording is causing we are introducing this conceptually in the new version of the manuscript (lines 56ff).

The discussion could also be expanded to address what methods are available or could be developed to build upon the concepts here; for example, the use of genetic inducers of metabolism which may avoid the more complex responses to lactulose.

We also appreciate the mention of concepts from our study that can be built on in future studies, and we added a paragraph on potential further research. (lines 301ff).

Despite these concerns, this was still an intriguing and valuable addition to the growing literature on the interface of the microbiome and circadian fields.

We thank the reviewer for all their encouraging and constructive remarks!

Reviewer #2 (Public Review):

Summary:

The authors aimed to investigate how microbial metabolites, such as hydrogen and short-chain fatty acids (SCFAs), influence feeding behavior and circadian gene expression in mice. Specifically, they sought to understand these effects in different microbial environments, including a reduced community model (EAM), germ-free mice, and SPF mice. The study was designed to explore the broader relationship between the gut microbiome and host circadian rhythms, an area that is not well understood. Through their experiments, the authors hoped to elucidate how microbial metabolism could impact circadian clock genes and feeding patterns, potentially revealing new mechanisms of gut microbiome-host interactions.

Strengths: 

The manuscript presents a well-executed investigation into the complex relationship between microbial metabolites and circadian rhythms, with a particular focus on feeding behavior and gene expression in different mouse models. One of the major strengths of the work lies in its innovative use of a reduced community model (EAM) to isolate and examine the effects of specific microbial metabolites, which provides valuable insights into how these metabolites might influence host behavior and circadian regulation. The study also contributes to the broader understanding of the gut microbiome's role in circadian biology, an area that remains poorly understood. The experiments are thoughtfully designed, with a clear rationale that ties together the gut microbiome, metabolic products, and host physiological responses. The authors successfully highlight an intriguing paradox: the significant influence of microbial metabolites in the EAM model versus the lack of effect in germ-free and SPF mice, which adds depth to the ongoing exploration of microbial-host interactions. Despite some methodological concerns, the manuscript offers compelling data and opens up new avenues for research in the field of microbiome and circadian biology.

We thank the reviewer for their encouraging remarks, specifically on the surprising findings that microbial metabolism seems to affect circadian clock gene expression and behavior differently in EAM and SPF mice.

Weaknesses:

The manuscript, while providing valuable insights, has several methodological weaknesses that impact the overall strength of the findings. First, the process for stool collection lacks clarity, raising concerns about potential biases, such as the risk of coprophagia, which could affect the dry-to-wet weight ratio analysis and compromise the validity of these measurements.

We thank the reviewer for pointing out that our description of the specific methods used for collecting feces were presented in a somewhat confusing manner. In short, dry and wet faecal weights were determined based on fecal pellets that were freshly produced and directly collected from restrained mice. To determine total fecal output over time, we collected all fecal pellets produced in a 5-hour window in a cage, determined their dry weight, and then used the water content determined for fresh faeces to calculate wet weight. Using this method, we cannot account for potential differences in coprophagia between the groups. However, this is not likely to affect the dry-to-wet ratio of faecal output in our results. We have now adapted the section in the methods to increase clarity (lines 440ff), and changed the quantity shown in Figure S2C and E to "water content", which is a more intuitive measure for the same thing.

Additionally, the use of the term "circadian" in some contexts appears inaccurate, as "diurnal" might be more appropriate, especially given the uncertainty regarding whether the observed microbiome fluctuations are truly circadian.

Similarly to our answer to reviewer 1 above, we appreciate this remark about imprecise language and have addressed this issue in the text and the figure legends. Indeed, we do not think the fluctuations in microbiota activity are truly circadian, but likely a result of the entrainment through the host's food intake.

Another significant issue is the unexpected absence of an osmotic effect of lactulose in EAM mice, which contradicts the known properties of lactulose as an osmotic laxative. This finding requires further verification, including the use of a positive control, to ensure it is not artifactual.

This is a good point. We have used this lactulose dosage specifically to induce microbial metabolism without causing osmotic diarrhoea and went to some lengths do demonstrate this (FigureS2C-E). In response to this comment (and one by reviewer 3 below about transit time), we have now performed additional experiments using higher lactulose dosage (new FigureS3). Our results indicate that the effect of lactulose on water content and transit time depends on microbiota complexity, with no change in fecal water content in 3MM mice even when treated with higher lactulose doses, and a stronger change in SPF mice. We now address this in the main text (lines 127ff).

The presentation of qRT-PCR data as log2-fold changes, with a mean denominator, could introduce bias by artificially reducing variability, potentially leading to spurious findings or increased risk of Type I error. This approach may explain the unexpected activation of both the positive and negative limbs of the circadian clock.

While we agree that our description of the qPCR method used for measuring circadian clock gene expression was lacking detail, we do not see how our analysis would lead to an increased risk of Type 1 error.

Briefly, we use the standard ΔΔCt method to analyze our RT-PCR results. We first normalize gene expression values for each gene of interest to an internal housekeeping gene. Then, we use these normalized values to compare gene expression values in treatment vs control groups (or to the control group at time point 0 in the case of Figure 3C). We then use a log2 transformation to convert the logarithmic RT-PCR values to fold changes. We apologize for the confusing labeling in the figures, where we called the values "log2(fold changes)", which we have now changed to "log2(ΔΔCt) of expression" in Figures 3B, C and S4.

The simultaneous activation of both limbs of the circadian clock is indeed a surprising result and somewhat complicates interpreting the effect of lactulose treatment on clock gene expression. We take it as evidence that it generally interferes with clock gene expression, while a clearer understanding of the effects would require further research.

Moreover, the lack of detailed information on the primers and housekeeping genes used in the experiments is concerning, particularly given the importance of using non-circadian housekeeping genes for accurate normalization.

It seems like the resource table describing these important experimental details was omitted in the original submission. We have now included it in the revised version (Table S1).

The methods for measuring metabolic hormones, such as GLP-1 and GIP, are also not adequately described. If DPP-IV/protease inhibitor tubes were not used, the data could be unreliable due to the rapid degradation of these hormones by circulating proteases.

We thank the reviewer for pointing out this omission. We have now added details of how we measured the metabolic hormones to the methods section, including the fact that we have added a DPP-IV inhibitor to the tubes at sampling (lines 469ff).

Finally, the manuscript does not address the collection of hormone levels during both fasting and fed phases, a critical aspect for interpreting the metabolic impact of microbial metabolites.

While we agree that it would be interesting to measure hormone levels also in the fed phase, a more thorough examination of hormone levels over the diurnal cycle, as suggested by reviewer 3, would be relevant for a full-scale follow-up. Given our data, we of course cannot exclude that there may be time-point-specific differences and therefore have softened the language around this conclusion to state that hormone levels are not acutely changed after a lactulose intervention at the time-points examined. (lines 318ff).

These methodological concerns collectively weaken the robustness of the study's results and warrant careful reconsideration and clarification by the authors.

Because of these weaknesses, the authors have partially achieved their aims by providing novel insights into the relationship between microbial metabolites and host circadian rhythms. The data do suggest that microbial metabolites can significantly influence feeding behavior and circadian gene expression in specific contexts. However, the unexpected absence of an osmotic effect of lactulose, the potential biases introduced by the log2-fold change normalization in qRT-PCR data, and the lack of clarity in critical methodological details weaken the overall conclusions. While the study provides valuable contributions to understanding the gut microbiome's role in circadian biology, the methodological weaknesses prevent a full endorsement of the authors' conclusions. Addressing these issues would be necessary to strengthen the support for their findings and fully achieve the study's aims.

We thank the reviewer again for their careful and critical reading of our work, and for their constructive input. In the revised version of our manuscript, we address the reviewer's concerns by providing more methodological detail and additional experimental data.

Despite the methodological concerns raised, this work has the potential to make a significant impact on the field of circadian biology and microbiome research. The study's exploration of the interaction between microbial metabolites and host circadian rhythms in different microbial environments opens new avenues for understanding the complex interplay between the gut microbiome and host physiology. This research contributes to the growing body of evidence that microbial metabolites play a crucial role in regulating host behaviors and physiological processes, including feeding and circadian gene expression.

We thank the reviewer for their encouraging remarks!

Reviewer #3 (Public Review):

Summary:

In the manuscript by Greter, et al., entitled "Acute targeted induction of gut-microbial metabolism affects host clock genes and nocturnal feeding" the authors are attempting to demonstrate that an acute exposure to a non-nutritive disaccharide (lactulose) promotes microbial metabolism that feeds back onto the host to impact circadian networks. The premise of the study is interesting and the authors have performed several thoughtful experiments to dissect these relationships, providing valuable insights for the field. However, the work presented does not necessarily support some of the conclusions that are drawn. For instance, lactulose is administered during the fasting period to mimic the impact of a feeding bout on the gut microbiota, but it would be important to perform this treatment during the fed state as well to show that the effects on food intake, etc. do not occur.

We thank the reviewer for this important point. In the revised version, we include an experiment where we administer lactulose during the fed state and do not observe a significant change in food intake. We describe this in the text (lines 189ff) and in the new Figure S5C and D.

To truly draw the conclusion that the current outcomes are directly connected to and mediated via an impact on the host circadian clock, it would be ideal to perform these studies in a circadian gene knock-out animal (i.e., Cry1 or Cry2 KO mice, or perhaps Bmal-VilCre tissue-specific KO mice). If the effects are lost in these animals, this would more concretely connect the current findings to the circadian clock gene network.

We agree that these would be interesting experiments to follow up on the question how the observed effects are actuated by host functions. However, they would require a large amount of preparatory work (including rederiving the KO mice to get them germ-free in our gnotobiotic facility), we argue that they are beyond the scope of this study.

Despite these reservations, the work is promising.

We thank the reviewer for their encouraging assessment.

Strengths:

Attempting to disentangle nutrient acquisition from microbial fermentation and its impact on diurnal dynamics of gut microbes on host circadian rhythms is an important step for providing insights into these host-microbe interactions.

The authors utilize a novel approach in leveraging lactulose coupled with germ-free animals and metabolic cages fitted with detectors that can measure microbial byproducts of fermentation, particularly hydrogen, in real-time.

The authors consider several interesting aspects of lactulose delivery, including how it shifts osmotic balance as well as provides calculations that attempt to explain the caloric contribution of fermentation to the animal in the context of reduced food intake. This provides interesting fundamental insights into the role of microbial outputs on host metabolism.

Thank you!

Weaknesses:

While the authors have done a large amount of work to examine the osmotic vs. metabolic influence of lactulose delivery, the authors have not accounted for the enlarged cecum and increased cecal surface area in germ-free mice. The authors could consider an additional control of cecectomy in germ-free mice.

We thank the reviewer for pointing out the potential effect of the anatomical differences of germ-free and conventionally colonized mice. We agree that when comparing germ-free mice to SPF mice, the enlarged cecum area in germ-free animals could lead to differences in water release or uptake. However, this difference is smaller in gnotobiotic mice colonized with our minimal microbiota, even though their ceca are still slightly smaller than those of germ-free mice (new Figure S2F). While we agree that including control of cecectomy in germ-free mice could be interesting, we do not have the option of doing surgery on germ-free mice given our current experimental setup. We have now added information on cecum weight, a good proxy for cecum size, in the new Figure S2F done.

The authors have examined GI hormones as one possible mechanism for how food intake is altered by microbial fermentation of lactulose. However, the authors measure PYY and GLP-1 only at a single time point, stating that there are no differences between groups. Given the goal of the studies is to tie these findings back into circadian rhythms, it would be important to show if the diurnal patterns of these GI hormones are altered.

We fully agree that a deeper investigation of the diurnal fluctuations of hormone levels would be an interesting next step in studying whether perturbations in food intake can disturb these rhythms. Doing this for the whole rhythm would really require a full second study.

In response to the reviewer's comments, we have changed the statements made around these data to point out just that hormone level fluctuations could not be detected during specific time points after lactulose treatment and therefore do not seem to explain the imminent behavioral changes (lines 318ff).

Considerations of other factors, such as conjugated vs. deconjugated bile acids, microbial bile salt hydrolase activity, and bile acid resorption, might be an important consideration for how lactulose elicits more influence on ileal circadian clock genes relative to cecum and colon.

We absolutely agree that investigation of microbial bile acid modification and their metabolism by the host would be an interesting topic for a follow-up study.

Measurements of GI transit time (both whole gut and regional) would be an important for consideration for how lactulose might be impacting the ileum vs. cecum vs. colon.

This is also an interesting point. While we did not add an experiment in which we specifically measure transit time to the revised version, we now measure total faecal output in a 5 h time period after PBS or lactulose treatment (Figure S3C). Faecal output is known to be a good proxy for transit time, and we see no significant difference between lactulose treatment (even with a two-fold higher dose than used before, new Figure S3) and PBS treatment.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) Line 126 - see the point in the public review, this data argues against disrupting the rhythm.

See our response above

(2) Line 156 - the metric used for water content is confusing. Why not just subtract dry weight from wet weight to get water content? The ratio is much harder to think about for me. Perhaps more importantly, this data is very confusing given that colonization seems to impact the activity of lactulose, which complicates the interpretation. Could be an interesting area for future study that you might highlight more in the discussion.

We thank the reviewer for pointing out that our presentation of water content could be clearer. We have changed the dry/wet ratio we have used in the previous version to the "water fraction" (new Figure S2C, E; new Figure S3A,B), i.e., the per cent of weight of the wet sample that is made up by water. We would argue that this is a measurement that is easier to interpret than the difference suggested by the reviewer, because it is independent of the absolute sample weight.

We also agree that it is surprising that colonization state changes the osmotic state of the gut environment and have now added text discussing that (lines 127ff).

(3) Line 188 - the lack of expression changes in the distal gut (cecum/colon) potentially conflicts with the model, warranting additional discussion/qualifications. Is lactulose getting metabolized in the small intestine? Alternatively, does lactulose have a direct effect on the host? The current text seems to imply that lactulose is fermented in the colon, the fermentation products are absorbed, and then they only impact the ileum through circulation, which doesn't seem physiologically possible.

It is possible that lactulose affects small intestinal tissue directly, but we show that the effect depends on microbial activity. Microbial activity is much larger in the large intestine than in the small intestine, which is why we hypothesize that it acts through systemic signals rather than locally. These systemic signals might be fermentation products impacting the ileum through circulation. We would argue that this is not implausible, given that there are well-documented systemic effects of fermentation products in circulation (e.g., den Besten et al, 2013). It is, however, also possible that, e.g., metabolism of fermentation products in the liver triggers a secondary signal that acts systemically.

(4) Line 241 - need to weaken this sub-heading. The experimental design shows that fermentation products impact feeding behavior, but this does not necessarily imply that fermentation products are responsible for the lactulose effect.

Done.

(5) Line 256 - The lack of an effect in SPF mice is surprising and potentially conflicts with the model proposed. Given this and other confusing results (gene expression site specificity, osmolarity effects, etc) it seems prudent to be more cautious as to the potential mechanisms through which lactulose supplementation impacts host gene expression and feeding behavior, which would likely require a lot more experiments to provide a definitive answer.

While the lack of an effect in SPF mice is surprising, we would argue that this is rather points towards the need for a better understanding of host-microbiota-diet interactions than a conflict with our interpretation. We do agree with the reviewer that more work is necessary for providing definitive proof of the underlying mechanisms of the observed effects and have adapted the language throughout the text.

(6) Line 260 - A lot of text is devoted to the potential caloric effects of the fermentation products and the lactulose itself. Was this in response to a prior reviewer? Either way, it seems too speculative to me and I would recommend trimming it down and moving it to the discussion.

We have rewritten this section to make it more concise and increase readability (lines 224ff).

(7) Line 305 - Not fasting, just lower caloric intake, as shown by Figure 1c.

We have changed the text to reflect that.

(8) Line 371 - Too definitive given the current data. Need to qualify the interpretation here.

We have adapted the text and qualified the interpretation.

(9) Figure 2 - Need to revise the title, no data showing that the rhythm is disrupted.

Done.

(10) Figure 3b - Clarify what timepoint is shown in the legend.

Done.

(11) Figure 3c - label when the treatment started. Consider changing to 2-way ANOVA which is probably more appropriate than t-tests.

We have added information on treatment start to the figure legend and have changed the statistical analysis to a two-way ANOVA (time, treatment).

(12) Figure 4c - move to supplement as this negative data isn't sufficient to rule out an impact on the hypothalamus or liver. Even when only considering transcript levels it's possible that the timepoint is just not ideal.

We agree that this data only represents a snapshot of gene expression at one timepoint after treatment and does not rule out involvement of hypothalamus or liver in this process. We have now moved the previous Figure 4C to the supplementary information (new Figure S6A).

(13) Figure 5 - defined the "fermentation products" and their concentrations in the legend. Consider moving panels d, and e to the supplement - negative results with unclear interpretation. Clarify in the legend how many calories/g were assumed for the fermentation products and provide a scientific rationale for this decision. Modify the title to be more cautious - as discussed above.

We thank the reviewer for pointing out that this was not clear. We have now added the formulation of the fermentation products to the figure legend and moved panels D and E to the supplement. We have also combined the former Figures 4AB and 5ABC into the new Figure 4.

(14) Figure S1 - The patterns in panel a are really intriguing and could be discussed more. For example, what do you think is driving the rapid oscillations in E. rectale?

We agree that the patterns are potentially interesting, but the fact that the patterns we observe do not replicate well between the two light-dark cycles we monitor suggest a large contribution of experimental noise. We therefore refrain from interpreting too much into this dataset.

(15) Figure S2 - Could changing the metric for water content be more easily interpretable? Modify the title to better match the data shown.

We thank the reviewer for pointing that out. We have now changed the metric in Figure S2 (and the new Figure S3) to % water in feces/cecum content, which is easier to interpret.

(16) Figure S3 - need to specify the multiple testing correction used.

Done.

(17) Figure S4 title - replace "inducing microbial metabolism" with "lactulose".

Done.

(18) Figure S5 title - modify to weaken the causal claim.

We have adapted all figure titles to conform to this comment.

Reviewer #2 (Recommendations For The Authors):

Greter and colleagues present an insightful manuscript investigating the effects of microbial metabolites, such as hydrogen and SCFAs, on feeding behavior and circadian gene expression at a single time point. Notably, they observed that these metabolites exert a significant influence on behavior in a reduced community model (EAM). However, this effect was not evident in germ-free or SPF mice, highlighting an intriguing paradox. The manuscript is well-written, with experiments that are thoughtfully designed and clearly rationalized. The study contributes valuable data to the poorly understood relationship between the gut microbiome and host circadian rhythms. While I find the manuscript compelling, I believe that providing additional experimental and analytical details would enhance clarity and rigor.

Major Comments:

(1) Additional clarity is needed regarding the stool collection process. Were the samples collected as fresh specimens, or was there a possibility of coprophagia before collection? Clarification on this point is important, as it could impact the results, particularly the dry-to-wet weight ratio analysis. Ensuring the collection process did not introduce this bias is crucial for the validity of these measurements.

Thank you for pointing out that this was not stated clearly. Wherever we assessed water content in faeces, we used fresh samples that were directly collected from a live animal and immediately frozen in a closed container or analyzed. When we assessed total faecal output, we collected the total bedding from a cage, sorted out the faecal pellets, and only measured dry weight. Total wet weight output was then assessed by using the water content of fresh faeces of animals with the same microbiota and the same treatment as a correction factor. We have now made this clear in the methods section (lines 435ff).

In cases where we performed total output measurements by collecting bedding, there was the possibility for coprophagia. We did not control for this but assumed that coprophagia will have a small effect that is likely similar between groups and should thus not affect the comparison.

(2) Caution is advised in the use of the term 'circadian,' which is sometimes used when 'diurnal' might be more appropriate. For example, the title of the first results section could be revised to 'Host Feeding [or Diurnal] Rhythms Influence Microbial Metabolic Fluctuations.' Additionally, line 357 should likely use 'diurnal' instead of 'circadian.' It's important to note that 'circadian' implies that cyclical fluctuations would persist without environmental cues (e.g., feeding). Since it's not clear whether most microbiome compositional or functional fluctuations are truly circadian, 'diurnal' is likely the more accurate term.

We thank the reviewer for pointing out our imprecise use of the term circadian. We have now adapted this throughout the manuscript.

(3) The lack of an osmotic effect of lactulose in EAM mice is quite surprising, given that lactulose is known to be an osmotic laxative. Was this finding specific to EAM mice, or was a similar lack of osmotic effect observed in SPF mice? A positive control is necessary to verify that this unexpected result is not artifactual. If there is no osmotic laxative effect in SPF mice, an explanation is needed as to why this medication is not functioning as expected in these mice.

This is a good point. While we initially thought that the lack of an osmotic effect was due to the specific lactulose dose we were using, we also did not observe an osmotic effect (measured by the water content of fresh faecal pellets produced after treatment) when we used twice the amount of lactulose in EAM mice (new Figure S3). However, in SPF mice, we did see an increase in faecal water content after treatment. This intriguing result suggests that the effect of lactulose as an osmotic laxative depends on the presence of a complex microbiota. We now show this data in the new Figure S3A and discuss it in the text (lines 127ff).

(4) The authors state that 'To account for faulty measurements due to disruptive events and for measurement noise, some datapoints were excluded from the raw datasets.' It would be important for the authors to confirm that this data exclusion was unbiased, meaning it did not disproportionately affect one group over another, and that any exclusions affected groups randomly.

This is a good point, and we analyzed this for the experiments we show in Figs 4A/S5A, 4B/S5E, 4C, and S5F, and discuss this in the methods part (lines 493ff). The resulting statistics is not fully conclusive: using a Chi-square test to check whether the probability of excluding values differs between experiments, we get significant differences (p=3.9 x 10^-18). We would, however, argue that this is not surprising, as different experiments sometimes different in the number of times we needed to do maintenance work on the isolators, which could lead to actuation of the scales measuring feed values, and thus faulty measurements.

We face the same problem within experiments: we found a significant difference in the probability to exclude values between treatment and control in the experiment shown in Figs 4A/S5A (p=1.7 x 10^-6), but no significant differences in the experiments in Figs 4B/S5E and S5F (p=0.71, and p=0.10, respectively). While it is hard to strictly exclude an influence of our data exclusion strategy, these findings speak against a systematic effect of treatment.

(5) The authors should provide details on the primers and housekeeping genes used in their experiments. It's crucial that the housekeeping gene is not circadian and is stable at all time points (PMID 17878933).

In the previous submission, the main resources table was omitted by accident. We have now added it (Table S1), including details on the primers and reagents used for all experimental work.

(6) The presentation of qRT-PCR data as log2-fold change is confusing. It's unclear what the numerator and denominator represent for this ratio or why such normalization was deemed necessary. Ideally, transcripts should be normalized to a housekeeping gene (as noted in a previous comment), not to a baseline measure of other same genes acquired from other mice. Log2-fold change is typically appropriate when comparing two measures from the same mice; however, in this study, the mice were euthanized, and the denominator is a mean of genes measured from other samples. This approach could introduce bias and might explain why both the positive and negative limbs appear to be activated by the microbial metabolites. It would be more rigorous to present these values as absolute gene expression levels.

We thank the reviewer for pointing out that this was not described clearly in the previous version of our manuscript. We have now adapted the methods part to explain that all data showing RT-PCR data is normalized to a housekeeping gene. Only after that, we compare the gene expression levels of the treatment group to the control group (or to gene expression of the control group at timepoint 0 in the case of Figure 3C).

(7) The use of log-ratios, with a mean as the denominator, could artificially reduce the variability in the data, potentially leading to spurious findings or an increased risk of Type I error.

As pointed out above, we have used internal normalization to a housekeeping gene before comparing the resulting values of the treatment and control groups. This method (commonly known as ΔΔC_t method) is a standard way of comparing gene expression values obtained by RT-PCR. The use of a log2 transformation is commonly used to convert the logarithmic data resulting from the RT-PCR measurement to a linear fold-change measurement. We do not see that this data analysis strategy should lead to an increased risk for producing false positives. We now explain this better in the methods section of the manuscript (lines 541ff), and have adapted the labels in Figures 3B, C and S4 to avoid confusion.

(8) It is unusual that both the positive and negative limbs of the circadian clock are overexpressed following lactulose administration. The authors should provide data confirming that these genes are in counter phase to each other at baseline. This clarification would help readers better understand the effects of the experimental interventions. As it stands, this critical part of the results is quite confusing.

We agree that this result does not allow a clear interpretation of the effect of lactulose on the diurnal rhythm of the host. While we agree that this would be interesting to understand in detail, we are merely taking this as a first indication that actuation of microbial metabolism during the inactive phase of the diurnal rhythm can lead to changes in clock genes. This claim is supported by our data.

(9) Could the effects of the microbial metabolites be mediated by AMPK, a known nutrient sensor that can influence the post-translational modification of Cry proteins? It would be beneficial for the authors to explore whether these metabolites have a more direct, previously unknown mechanism of affecting the circadian clock, or if their effects are mediated through known signaling pathways such as AMPK.

We agree that this would be a valuable path to continue investigating the effect of microbial metabolism on clock gene activity.

(10) The methods section does not specify how the metabolic hormones (e.g., GLP-1, GIP, leptin, ghrelin) were measured in the experiments. It is important for the authors to confirm that DPP-IV/protease inhibitor tubes were used for hormone measurement, as these proteins can be rapidly degraded by circulating proteases. Without the use of appropriate tubes, this data cannot be reliably interpreted. Additionally, it would have been ideal to collect these hormone levels during both the fasting and fed phases, but it appears this was not done. This represents a significant limitation of the study and should be addressed in the discussion.

We thank the reviewer for pointing out this omission, we have now added a description of our protocol to measure metabolic hormones to the methods section (lines 469ff).

(11) If the samples were appropriately collected in DPP-IV/protease inhibitor tubes, the authors should consider measuring active GLP-1, as this would likely provide a more accurate assessment of GLP-1 activity.

We agree that this would be a valuable next step, in addition to testing the effect of changes in microbial metabolism on the time traces of hormone levels.

Minor Comments:

(12) Line 350 appears to have an incomplete sentence, as it seems part of the first sentence in the paragraph has been inadvertently deleted. This should be reviewed and corrected for clarity.

Done.

Reviewer #3 (Recommendations For The Authors):

Major comments:

(1) Could the authors provide a deeper description about what they are referring to in the following statement? "...higher order interactions and microbial metabolism are variable..." it is difficult to interpret as written. Do the authors mean cross-feeding interactions?

We have changed this sentence to clarify the meaning.

(2) Could the authors explicitly state their hypothesis in the introduction and provide a brief, but deeper explanation of the intervention prior to the results section?

This is a good point, we have adapted the text accordingly (lines 53ff).

(3) Could the authors include a bit more information regarding the diet provided to the mice? If grain-based chow, please provide insights into the fiber source, etc.

While we agree that it would be interesting to know what part of the mouse diet is available to the microbes, this is hard for the standard mouse chow that we (and most others doing experiments with mice) feed the experimental animals. We have now added more detailed information on the specific type of chow the mice were fed (lines 347f). We would argue that, because the control and treatment groups were always fed the same chow, the effect of the fiber source and other specifics are controlled for, even if we do not know them.

(4) In figure 1A - cells/g does not seem to be the correct unit - # of copies/g feces perhaps?

Cells/g is the appropriate unit, but it seems like we have not explained the way we arrive at this unit in sufficient detail. In short, we use a qPCR run on a known standard curve of bacterial counts (known cells/g values) to estimate these numbers from qPCR results from faeces. We have now explained this better in the methods section (lines 385f).

(5) Figure 1B/C and Figure S1B/C are confusing - the legend states these measurements were taken over two days, however, the plot shows a single 12:12 LD period. Was the data averaged? It might be best to show each day separately (i.e., over a 48-hour period) rather than in one 24-hour plot. Then, the authors could also show the averages in the light period vs. the dark period in a separate, complementary graph.

We thank the reviewer for pointing this out. The previous figures were indeed averaged over the two days of measurement and projected onto one 24h period for plotting. We have now changed Figures 1B,C and S1B,C to show the full 48h time windows.

(6) Line 109 - The reviewer concurs that lactulose is a non-nutritive, synthetic disaccharide, however, in theory, lactulose may have a high heat increment, which could cause the animal to undergo metabolic responses to defend core body temperature (which also exhibits diurnal rhythmicity). Have the authors considered core body temperature rhythms, their connection to microbial metabolism, and the core circadian clock gene network in their model?

This is an interesting thought. We have not measured body temperature in our experiments. As the heat increment from food is typically associated with metabolic activity, and lactulose is not metabolized, we do not expect its heat increment to be high, at least in GF mice. In mice with a microbiota, we agree that metabolic heat will be produced upon lactulose metabolism by the microbes, which could be a contributor the observed effect.

(7) Figure 2A and corresponding text in lines 121 - 123 - indicate at what time the bar and whisker plots were taken (assuming at ZT8, but please be explicit in the figure and corresponding text). Could the authors also include statistics for these waveforms?

Done.

(8) In Figure 2B, the authors state that microbial metabolism had been restored to normal levels by ZT15, however, did this persist into the next light phase? It would be ideal if the authors could present these data in a similar manner to that shown in Figure S2A for SPF mice. Further, what are the statistical considerations here to describe changes in phase, amplitude, periodicity, etc.

This is a good point. Unfortunately, we do not have H2 measurements for EAM and GF mice over comparable time periods as shown for SPF mice in Figure S2A. However, the food intake data shown in Figure S5BCD is a indicates that the food intake normalized in the second dark phase after treatment.

(9) In lines 147 - 149 and in Figure S2B figure legend - assuming these measurements are from individual bacteria? Could this be stated clearly in the text or legend? Also, what are the statistical considerations? Were there significant differences in SCFA production between bacteria?

We thank the reviewer for pointing out that this was unclear. We have now adapted the legend to explain the way these data were collected and added a statistical analysis.

(10) The authors have done a large amount of work to examine the osmotic vs. metabolic influence of lactulose delivery - however, have the authors accounted for the enlarged cecum and increased cecal surface area in germ-free mice? Would an additional control be cecectomy in germ-free mice to be more in-line w/ SPF animals? Further, could the authors tie in these findings more explicitly and state how they pertain to the overall goal of the study? Is this simply to draw the conclusion that microbial biomass is increased w/ lactulose?

We wanted to make sure that the effect we see with lactulose is due to microbial metabolism and not due to the induction of osmotic diarrhoea or other host-dependent effects, and we have added a statement to that effect (lines 127ff). We have not corrected for the change in cecum size between GF and EAM mice, as EAM size (and many other gnotobiotic mouse models) also have enlarged ceca relative to conventionally colonized mice, but have added a dataset showing how cecum size of EAM and GF mice compare and discuss this in the text (new Figure S2F, lines 135ff).

(11) Is Figure 3A necessary?

It might not be strictly necessary, but it can help with understanding the relation of the genes tested in B and C, and we would therefore like to keep it in.

(12) Line 155 - 157 - the authors make the statement that dry/wet feces weight ratio is decreased in GF mice, but this does not appear to be statistically significant. Please adjust to state numerical differences were observed or provide statistics.

We have changed the statement in the text.

(13) Could Figure S3A be moved to the main Figure 3 as this may provide a more logical flow? qRT-PCR data is expressed as - log2 (fold expression), but relative to what? Could the authors provide further info about the control?

The former Figure S3A (new S4A) and Figure 3B show the same data in slightly different ways. We therefore opted to keep only one of those illustrations in a main figure. The fold changes are always relative to the average of the PBS control group, which we now state explicitly in the figure legend.

(14) The authors state that the qRT-PCR data shows that microbial metabolism of lactulose impacts peripheral circadian gene expression, but this conclusion seems simplified. Lactulose treatment only impacted the ileum circadian gene expression. Additional peripheral tissues (liver, adipose tissue, etc.) could be moved to this figure, i.e., move Figure 4C data. Why do the authors think the ileum was most impacted beyond GI hormones as discussed later in the manuscript? Could changes in bile acid deconjugation (i.e., BSH activity?) and/or bile acid resorption by the host in the distal ileum due to lactulose delivery be involved? Or is it simply due to differences in GI transit time (which was not measured in the current study)? Further, lactulose had minimal impact in SPF ileum, and in fact, shifted Cry1 in the opposite direction relative to EAM mice. Could the authors provide more insight into these disparate observations (line 196 - 200)?

We agree that the statement "lactulose impacts peripheral gene expression" is oversimplified, and we have now adapted the text to avoid the impression that this is our conclusion. No tested tissues other than the ileum showed significant differences in gene expression at the time point tested, which does not rule out that other tissues would react to the treatment at that time point, or the tested tissues would do so at the tested time point. As we don't have a good enough understanding of what mechanism causes the gene expression changes in the ileum, we refrain from speculating in the text, even though we agree that this is an intriguing question.

(15) Could the authors provide more insight into the statistical approaches used to assess amplitude, peak, nadir, etc. in Figure 3C? Was the co-sinor waveform tested?

This is a good question. Even though this was a highly work-intensive experiment using many animals, we would argue that the noise level is too high and the coverage of the time analyzed too sparse to infer meaningful statistics on the fluctuations of gene expression over time. In the new version, we have changed our statistical analysis of this dataset to a two-way ANOVA (treatment, time) to better analyze this dataset.

(16) The food intake decrease and interpretation following treatment (Figure 4A and S4A) is curious - all animals were gavaged and in EAM mice, many animals, regardless of PBS or lactulose are trending down in food intake rate/total intake. It seems to be more of an impact of gavage and not of treatment, which the authors somewhat acknowledge in Lines 228 - 230.

We agree that gavage is a possible factor in future food intake of experimental animals, which is precisely why we used the PBS gavage as a control. Even though the difference is not large, we see a significant change in food intake when lactulose is given, but not when PBS is given (Figure 4A, S4A).

(17) Could the authors provide a deeper rationale for their line of thinking for lines 234 - 240? What is the evidence that systemic effects are likely to occur 3 hours after lactulose delivery? Further, as stated in comment 13, could brain and liver data be moved to Figure 3/Figure S3 as an additional example of peripheral tissue clocks?

We have added an explanation for the rationale we use to justify the 8h time point (it is 3h after the peak of H2 production, as shown in Fig2AB, which happens 5h after lactulose delivery). While we agree that the brain and liver data would also fit into the Figure 3, we have now moved all negative data to the supplementary information (in response to a comment by reviewer 1, and in a general effort to clean up the data in the manuscript).

(18) The authors measure PYY and GLP-1 at a single time point and state there are no differences, yet, the goal of the studies is to tie this back to circadian networks. Would it be possible to measure these GI hormones over a 24-hour period to show that the diurnal patterns are altered?

We fully agree that measuring the metabolic hormones over time would be very interesting. It is possible but would represent a major effort using many animals and a large amount of work. We would therefore argue that it is beyond the scope of this revision, but a good starting point for a follow-up study.

(19) The authors state that the administration of fermentation products acutely altered circadian food intake, but the studies do not support that this change is connected to the circadian network. Suggest softening the interpretation of the findings.

We have changed the language there to soften the interpretation.

Minor comments:

(1) The authors should consider when it is appropriate to refer to rhythms as diurnal vs. circadian, as each has a distinct meaning. Diurnal follows entrainment cues while circadian is endogenously driven (i.e., line 39, line 58).

We thank the reviewer for pointing out this important difference, we have adapted this in the whole text accordingly.

(2) Circadian rhythm should be plural throughout the manuscript (circadian rhythms).

Thank you, we have changed that where we refer to host circadian rhythms generally.

(3) Lines 54 - 63. Fermentation should be capitalized when used at the beginning of a sentence.

Done.

(4) Line 289 - This should be Figure 5D and 5E.

Done.

(5) Line 290 - heart should be cardiac.

Done.

eLife Assessment

This paper introduces a valuable optical method for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photostimulation in different axial planes. The evidence for effective optical separation of excitation and imaging is convincing, although the in vivo data in the olfactory bulb suggest potential confounding factors arising if stimulation not only affects cell bodies but also neuronal processes. The work will be of broad interest to neurobiologists working in circuit and systems neuroscience, as well as to specialists in optical microscopy.

Reviewer #1 (Public review):

In this methods paper, the authors introduce a novel and innovative imaging approach for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photo-stimulation in different axial planes. This is a highly exciting technique that enables the axial decoupling of optical imaging of deep neural circuits from surface photo-stimulation of spatially precise (tens of micrometres) brain spots. This method builds on previous developments from the same laboratory, combining DMD-based patterned photo-stimulation with in vivo electrophysiological recordings. To my knowledge, this is the first instance in which patterned photo-stimulation has been combined and axially decoupled from two-photon (2P) imaging.

Beginning with a thorough characterisation of the optical resolution of the photo-stimulation system, the authors applied this method to the olfactory bulb (OB) network, in which sensory inputs are topographically organised at the surface of the OB and thus ideally suited to demonstrate the relevance of this approach. They first showed that this technique can be used to rapidly reveal connectivity patterns of OB output neurons and to identify sister mitral cells. In addition, they manipulated a specific glomerular inhibitory population and demonstrated that these neurons provide spatially heterogeneous long-range inhibition of OB output neurons, with differential effects on mitral and tufted cells (a result previously observed in a paper from the same lab: Banerjee et al., 2015, Neuron). Altogether, the data demonstrate that this technique is well-suited for high-throughput functional mapping of neural circuit properties. The results are compelling and illustrate both the significant advance represented by this method and its feasibility.

Despite my initial enthusiasm, there are several concerns in the present study that must be addressed in order to rule out confounding observations and to resolve remaining uncertainties regarding photo-stimulation resolution. These include the following:

(1) Spatial resolution: Although the authors provide convincing data on spatial resolution in vitro, several observations throughout the paper suggest that the effective photo-stimulation precision may be lower than initially reported. For instance, in Figure 2, the authors observe repeated responses in neighbouring glomeruli (e.g., glomeruli #3 & #5, #4 & #6). To what extent could light scattering along the X/Y/Z-axis above the targeted glomerulus recruit en passage axons, resulting in the inadvertent activation of multiple glomeruli?

A further observation concerns the presence of "inhibited" sister mitral cells (Figure 3). The authors claim this is reminiscent of the differential spike-timing reported between sister cells (Dwawale et al., 2010, Nat Neuro). However, observing both excitatory and inhibitory responses following stimulation of glutamatergic inputs is an altogether different matter, particularly given that sister mitral cells are reciprocally connected via gap junctions. This observation requires further clarification and raises serious questions about the effective resolution of the stimulation. Could the inhibited cell simply correspond to a non-sister mitral cell receiving disynaptic feed-forward inhibition? To verify sister cell identity, the authors could confirm that the predicted sister cells share a similar odour receptive field compared to randomly selected mitral cell pairs. In their previous study employing analogous DMD-based photo-stimulation (Dhawale et al., 2010, Nat. Neurosci.), sister mitral cells did not exhibit such opposite response profiles (firing rate correlation of ∼0.7 between sister cells). Could the authors verify that a comparable activity correlation is also observed among the sister cells identified using ADePT in the present study? In Figure S6, the authors show recordings and stimulation of the same neurons co-expressing GCaMP and ChR2. Applying this experimental design to the mitral/tufted cell population (using a Tbet-Cre mouse transduced in the OB with both GCaMP and Chrimson virus) would constitute a valuable control to clarify the nature of these "inhibited" sister cells.

An additional concern relates to the 21 out of 162 mitral cells that were activated by two distinct glomeruli - a finding that is incompatible with the established OB wiring diagram and that further challenges the claimed stimulation resolution.

A critical control experiment is also absent: in a Thy1-GCaMP6 mouse lacking any light-sensitive opsin, do the authors observe any unintended side effects of photo-stimulation?

Regarding sister cells (Figure 3), tufted cells are not analysed alongside mitral cells in this dataset, whereas this is elegantly performed in Figure 5 using the DAT+ model. Could the authors also demonstrate how the technique can reveal the complete family portrait of sister mitral and tufted cells?

(2) The authors have explored only a limited set of photo-stimulation parameters, primarily varying light intensity. They should present additional tests, such as varying the spot size (which appears to be arbitrarily fixed at 30-50 µm) and the z plane of stimulation. The level of activation can vary considerably: for example, in Figure 3a(iii), identical stimulations elicit responses of markedly different amplitudes (see glom#3 and #4). In Figure 2, 5 out of 15 glomeruli failed to respond - could the choice of z-plane account for this variability? The stimulation duration (50-150 ms) also appears somewhat arbitrary: can the authors demonstrate that the technique is compatible with finer temporal patterns (e.g., 10 Hz stimulation for 500 ms using 20 ms light pulses)? What are the spatiotemporal and axial scanning limits of this approach, and can two or three glomeruli be targeted simultaneously with temporally patterned stimulation?

(3) One particularly relevant application of this method would be to guide photo-stimulation based on prior functional measurements - for instance, by generating a photo-stimulation mask specifically targeting odour-responsive glomeruli. In the DAT-Cre × Thy1-GCaMP6 experiment shown in Figure 5e, which glomeruli are activated by a given odour, and how does this odor responsiveness influence the efficiency of DAT+ cell-mediated inhibition?

Reviewer #2 (Public review):

Summary:

In this manuscript, Koh and colleagues describe ADePT (Axially Decoupled Photo-stimulation and Two-photon Readout), a modular approach for combining patterned one-photon optogenetic stimulation with two-photon calcium imaging in independently controlled axial planes. The method relies on a digital micromirror device together with a motorized holographic diffuser to generate spatially confined stimulation patterns while imaging deeper neuronal populations. As proof-of-principle applications, the authors use the system to map excitatory and inhibitory functional connectivity in the mouse olfactory bulb by stimulating superficial glomerular circuits and recording responses from mitral and tufted cells in deeper layers.

This is a well-executed Tools and Resources manuscript. The technical implementation is described in considerable detail, the optical performance is systematically characterized, and the biological experiments provide convincing demonstrations of the types of circuit questions that can be addressed using the method.

Strengths:

The greatest strength of the manuscript is the comprehensive technical characterization of the optical system. The authors carefully benchmark the spatial resolution, axial confinement, registration accuracy, calibration procedure, and practical operating limits of the setup. I found the extensive optical benchmarking particularly helpful, as it gives readers a realistic sense of the operating regime and practical limitations of the approach.

Another strength is the high level of methodological transparency. The optical design, calibration procedures, stimulation strategies, and analysis pipeline are described in sufficient detail that an experienced laboratory could realistically evaluate whether the system is suitable for its own applications. This level of documentation is particularly appropriate for a Tools and Resources article.

A further strength is the clear positioning of ADePT relative to existing approaches. The authors are transparent about the trade-off between spatial resolution and implementation complexity: ADePT does not provide single-cell photostimulation, but offers flexible axial separation, a large stimulation field, and cellular-resolution two-photon readout in deeper planes without requiring a full holographic stimulation system. This defines a credible and potentially useful experimental niche.

The biological applications convincingly demonstrate the utility of ADePT. The experiments identifying sister mitral/tufted cells through selective glomerular stimulation and the mapping of heterogeneous inhibitory influences from DAT-positive interneurons illustrate the types of functional connectivity questions that become experimentally accessible with this approach. Importantly, the authors generally avoid overstating these biological findings and appropriately present them as proof-of-principle demonstrations of the technology.

Weaknesses:

The primary limitation is inherent to the method itself rather than the execution of the study. Because ADePT relies on one-photon patterned illumination, photo-stimulation remains restricted to relatively superficial structures and does not achieve single-cell spatial resolution. The authors appropriately acknowledge these constraints and clearly position the method within this operating regime. Consequently, ADePT occupies a useful niche for interrogating spatially organized functional units such as olfactory glomeruli or cortical barrels, rather than applications requiring single-cell precision or deeper tissue penetration.

Although the manuscript describes the approach as relatively simple and cost-effective, implementation still requires careful optical alignment, registration, calibration, and optimization. This does not diminish the value of the approach, but terms such as modular or accessible may better reflect the practical implementation than simple. Likewise, a brief bill of materials, approximate add-on cost, and indication of which components are essential versus substitutable would help prospective users assess the accessibility of the system.

Finally, the manuscript provides an impressive level of technical characterization, but much of the practical guidance for adopting the system is distributed across the Results and Discussion. Bringing together the principal limitations, recommended operating regime, expected calibration workflow, evidence for long-term alignment stability, and the circumstances in which ADePT is preferable to alternative approaches would further strengthen the manuscript as a community resource.

Author response:

eLife Assessment

This paper introduces a valuable optical method for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photostimulation in different axial planes. The evidence for effective optical separation of excitation and imaging is convincing, although the in vivo data in the olfactory bulb suggest potential confounding factors arising if stimulation not only affects cell bodies but also neuronal processes. The work will be of broad interest to neurobiologists working in circuit and systems neuroscience, as well as to specialists in optical microscopy.

We thank the reviewers for their constructive feedback. We are planning to address their concerns as detailed below. Specifically, in the revised manuscript, we will streamline and consolidate the text to include:

(1) An in-depth discussion of the awake recording results in the main text.

(2) Further discussion of light scattering, photo-stimulation specificity of targeting individual glomeruli and resolution.

(3) A summary (including also a table) of the operating regime and comparisons with alternative techniques for patterned photo-stimulation and imaging of the ensuing responses. We will highlight the advantages and constraints of the current implementation of ADePT. In particular, here we explored a small set of spatiotemporal parameters to understand the limits of our technique and provide a proof of principle of the strategy. These parameters can be varied further depending on the exact research question.

(4) Implementation considerations and technical guidelines for calibration and long-term stability of the rig for ADePT (i.e. ‘a how-to guide’).

(5) A bill of materials and estimated hardware costs.

Furthermore, we will provide additional controls and rephrase some of the statements in the text as suggested (e.g. replace ‘accessible’ with ‘simple’, etc.). We will correct the unfortunate grammatical errors, improve clarity of text, and update the references accordingly.

Public Reviews:

Reviewer #1 (Public review):

In this methods paper, the authors introduce a novel and innovative imaging approach for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photo-stimulation in different axial planes. This is a highly exciting technique that enables the axial decoupling of optical imaging of deep neural circuits from surface photo-stimulation of spatially precise (tens of micrometres) brain spots. This method builds on previous developments from the same laboratory, combining DMD-based patterned photo-stimulation with in vivo electrophysiological recordings. To my knowledge, this is the first instance in which patterned photo-stimulation has been combined and axially decoupled from two-photon (2P) imaging.

Beginning with a thorough characterisation of the optical resolution of the photo-stimulation system, the authors applied this method to the olfactory bulb (OB) network, in which sensory inputs are topographically organised at the surface of the OB and thus ideally suited to demonstrate the relevance of this approach. They first showed that this technique can be used to rapidly reveal connectivity patterns of OB output neurons and to identify sister mitral cells. In addition, they manipulated a specific glomerular inhibitory population and demonstrated that these neurons provide spatially heterogeneous long-range inhibition of OB output neurons, with differential effects on mitral and tufted cells (a result previously observed in a paper from the same lab: Banerjee et al., 2015, Neuron). Altogether, the data demonstrate that this technique is well-suited for high-throughput functional mapping of neural circuit properties. The results are compelling and illustrate both the significant advance represented by this method and its feasibility.

We thank the Reviewer for their constructive input.

Despite my initial enthusiasm, there are several concerns in the present study that must be addressed in order to rule out confounding observations and to resolve remaining uncertainties regarding photo-stimulation resolution. These include the following:

(1) Spatial resolution: Although the authors provide convincing data on spatial resolution in vitro, several observations throughout the paper suggest that the effective photo-stimulation precision may be lower than initially reported. For instance, in Figure 2, the authors observe repeated responses in neighbouring glomeruli (e.g., glomeruli #3 & #5, #4 & #6). To what extent could light scattering along the X/Y/Z-axis above the targeted glomerulus recruit en passage axons, resulting in the inadvertent activation of multiple glomeruli?

Indeed, we cannot rule out this possibility. Fibers of passage are a potential concern. This is why we systematically sample different light intensities and assess their impact on specificity of dendritic mitral and tufted cell responses within the glomerular layer (same axial-plane optical stimulation and imaging experiments, Fig. 2). We identify a range of intensities that on average result mostly in activation of the targeted glomeruli. Within the range of intensities used for identifying sister cells, >90% of responses were on the diagonal (targeted glomeruli) and ~5% pixels that cleared the signal significance criterion used were in off-target glomeruli, as stated in the text and quantified in Fig. 2f. In the revised manuscript, we will further clarify and expand on these points.

A further observation concerns the presence of "inhibited" sister mitral cells (Figure 3). The authors claim this is reminiscent of the differential spike-timing reported between sister cells (Dwawale et al., 2010, Nat Neuro). However, observing both excitatory and inhibitory responses following stimulation of glutamatergic inputs is an altogether different matter, particularly given that sister mitral cells are reciprocally connected via gap junctions. This observation requires further clarification and raises serious questions about the effective resolution of the stimulation. Could the inhibited cell simply correspond to a non-sister mitral cell receiving disynaptic feed-forward inhibition?

This is indeed what we think it is happening (i.e. disynaptic feed-forward inhibition as the Reviewer points out). We observe inhibition in some of the mitral cells in the field of imaging when we stimulate not their parent glomerulus, but other glomeruli in the neighbourhood. As the Reviewer points out, sister cells are connected via gap junctions, but they also receive inhibitory chemical synaptic inputs via their secondary (and primary dendrites) from other (not-their-parent) glomeruli mediated by numerous types of interneurons including the DAT+/GABAergic (a.k.a. superficial short axon cells) and granule cells. Our data is consistent with differential inhibitory input from other glomeruli on sister cells getting input from the same parent glomerulus. As it appears that we failed to present this point clearly in the initial submission, we will further expand along these lines in the revised manuscript. Briefly:

First, we identify a photo-stimulation regime that results mostly in the activation of a given targeted glomerulus (and not of other glomeruli in the field of stimulation). To this end, we photo-stimulate and image ensuing neuronal responses in the same axial optical plane. We strobe (alternate) between monitoring dendritic mitral and tufted cell (enhanced) GCaMP responses within the targeted glomerulus and other glomeruli in the field of imaging, while varying systematically the light intensity (Figs. 2,3a; Suppl. Fig. 4b, Suppl. Fig, 5b-d;h-j). We use as criterion for specificity a condition when >95% of significantly responding pixels (above a statistically defined signal response threshold) lie within the anatomical boundaries of the targeted glomerulus. For each glomerulus (or pixel within a glomerulus) we compared the average light response across trials with the baseline reference distribution in the absence of light stimulation. If this value crossed the 99th percentile of the baseline distribution, the glomerulus/pixel within glomerulus was classified as responsive to the photo-stimulation.

Second, using the minimal light intensity regime experimentally identified as ‘specific’ for targeting individual glomeruli in the field of photo-stimulation (< 5% significant activation of off-target pixels), we decouple photo-stimulation in the glomerular layer from monitoring responses of mitral and tufted cells in the deeper layers of the olfactory bulb (100-250 µm axial displacement). This approach enables us to map cohorts of sister (daughter) cells associated with any specific target glomerulus in the field of view (1,2,3…n) by monitoring excitatory (enhanced) responses of mitral and tufted cell bodies. A cohort of sister cells associated with glomerulus x_i (daughters of glomerulus x_i) is defined by those cells which show statistically significant excitatory responses (4 SD - standard deviations - above their baseline fluctuations) specifically in response to photo-stimulation of glomerulus x_i.

Third, in the process, as we photo-stimulate different glomeruli in the field of stimulation, we also observe at times suppressed (inhibitory) responses in a subset of the mitral and tufted cells (exceeding 3 SD in the negative direction their baseline fluctuations). These suppressed responses occur in response to photo-stimulating not the parent glomerulus of a given cell, but other glomeruli in the field. These experiments revealed that within a cohort of sister cells (daughters of glomerulus x_i), only a subset of cells are suppressed by activation of glomerulus x_j, and, in a few example cases, different cells are suppressed by activation of different glomeruli (e.g. x_j vs. x_k), presumably through disynaptic feed-forward inhibition (Fig. 3b iii; 3d; Suppl. Figs. 5f,g; l,m). These preliminary observations suggest that sister cells receive differential inhibitory inputs from glomeruli in the neighborhood. In the revised manuscript, we will expand to further clarify these points.

To verify sister cell identity, the authors could confirm that the predicted sister cells share a similar odour receptive field compared to randomly selected mitral cell pairs. In their previous study employing analogous DMD-based photo-stimulation (Dhawale et al., 2010, Nat. Neurosci.), sister mitral cells did not exhibit such opposite response profiles (firing rate correlation of ∼0.7 between sister cells). Could the authors verify that a comparable activity correlation is also observed among the sister cells identified using ADePT in the present study? In Figure S6, the authors show recordings and stimulation of the same neurons co-expressing GCaMP and ChR2. Applying this experimental design to the mitral/tufted cell population (using a Tbet-Cre mouse transduced in the OB with both GCaMP and Chrimson virus) would constitute a valuable control to clarify the nature of these "inhibited" sister cells.

We thank the Reviewer for the suggestion. We consider that the experiments shown here are proof-of-principle in nature, highlighting the potential of ADePT for mapping functional neural circuit connectivity. In our opinion, further investigating the logic of similarities and differences in the odor responses of sister mitral cells and the nature of inhibitory glomerular interactions forms the focus of future studies. We also note that firing rate correlations can be notoriously difficult to compare and interpret across experimental regimes (spikes vs. calcium imaging).

An additional concern relates to the 21 out of 162 mitral cells that were activated by two distinct glomeruli - a finding that is incompatible with the established OB wiring diagram and that further challenges the claimed stimulation resolution.

Indeed, this reflects some degree of non-specific activation of the targeted glomeruli as discussed above. In the revised manuscript, we will further highlight this issue.

A critical control experiment is also absent: in a Thy1-GCaMP6 mouse lacking any light-sensitive opsin, do the authors observe any unintended side effects of photo-stimulation?

In the revised manuscript, we will include an additional control as suggested by the reviewer. Within the range of intensities used, we did not observe significant modulation of GCaMP6s activity in mitral and tufted cells in mice lacking light-sensitive opsins.

Regarding sister cells (Figure 3), tufted cells are not analysed alongside mitral cells in this dataset, whereas this is elegantly performed in Figure 5 using the DAT+ model. Could the authors also demonstrate how the technique can reveal the complete family portrait of sister mitral and tufted cells?

We thank the Reviewer for the suggestion. We think that the differences between mitral and tufted cells are indeed very interesting to investigate, but in our opinion form the subject of future studies.

(2) The authors have explored only a limited set of photo-stimulation parameters, primarily varying light intensity. They should present additional tests, such as varying the spot size (which appears to be arbitrarily fixed at 30-50 µm) and the z plane of stimulation. The level of activation can vary considerably: for example, in Figure 3a(iii), identical stimulations elicit responses of markedly different amplitudes (see glom#3 and #4). In Figure 2, 5 out of 15 glomeruli failed to respond - could the choice of z-plane account for this variability? The stimulation duration (50-150 ms) also appears somewhat arbitrary: can the authors demonstrate that the technique is compatible with finer temporal patterns (e.g., 10 Hz stimulation for 500 ms using 20 ms light pulses)? What are the spatiotemporal and axial scanning limits of this approach, and can two or three glomeruli be targeted simultaneously with temporally patterned stimulation?

Indeed, here we explored a limited set of spatiotemporal parameters to understand the limits of our technique and provide proof of principle. These can be varied depending on the exact question. In the revised manuscript, we will clearly state what the constraints of the current implementation are and provide context for further optimizations. Briefly, in the current version, individual as well as multiple glomeruli can be photo-stimulated together and 20 ms per pulse regime in trains of pulses is feasible.

(3) One particularly relevant application of this method would be to guide photo-stimulation based on prior functional measurements - for instance, by generating a photo-stimulation mask specifically targeting odour-responsive glomeruli. In the DAT-Cre × Thy1-GCaMP6 experiment shown in Figure 5e, which glomeruli are activated by a given odour, and how does this odor responsiveness influence the efficiency of DAT+ cell-mediated inhibition?

We thank the Reviewer for the suggestion. We are thinking along exactly the same lines. In particular, we would like to investigate the relationship between the degree of overlap in odor responses of individual glomeruli and the strength and specificity of their inhibitory interactions mediated by DAT+ interneurons. In the revised manuscript, we will further expand on discussing this venue of study. We feel however that this investigation is beyond the scope of this technical report.

Reviewer #2 (Public review):

Summary:

In this manuscript, Koh and colleagues describe ADePT (Axially Decoupled Photo-stimulation and Two-photon Readout), a modular approach for combining patterned one-photon optogenetic stimulation with two-photon calcium imaging in independently controlled axial planes. The method relies on a digital micromirror device together with a motorized holographic diffuser to generate spatially confined stimulation patterns while imaging deeper neuronal populations. As proof-of-principle applications, the authors use the system to map excitatory and inhibitory functional connectivity in the mouse olfactory bulb by stimulating superficial glomerular circuits and recording responses from mitral and tufted cells in deeper layers.

This is a well-executed Tools and Resources manuscript. The technical implementation is described in considerable detail, the optical performance is systematically characterized, and the biological experiments provide convincing demonstrations of the types of circuit questions that can be addressed using the method.

Strengths:

The greatest strength of the manuscript is the comprehensive technical characterization of the optical system. The authors carefully benchmark the spatial resolution, axial confinement, registration accuracy, calibration procedure, and practical operating limits of the setup. I found the extensive optical benchmarking particularly helpful, as it gives readers a realistic sense of the operating regime and practical limitations of the approach.

Another strength is the high level of methodological transparency. The optical design, calibration procedures, stimulation strategies, and analysis pipeline are described in sufficient detail that an experienced laboratory could realistically evaluate whether the system is suitable for its own applications. This level of documentation is particularly appropriate for a Tools and Resources article.

A further strength is the clear positioning of ADePT relative to existing approaches. The authors are transparent about the trade-off between spatial resolution and implementation complexity: ADePT does not provide single-cell photostimulation, but offers flexible axial separation, a large stimulation field, and cellular-resolution two-photon readout in deeper planes without requiring a full holographic stimulation system. This defines a credible and potentially useful experimental niche.

The biological applications convincingly demonstrate the utility of ADePT. The experiments identifying sister mitral/tufted cells through selective glomerular stimulation and the mapping of heterogeneous inhibitory influences from DAT-positive interneurons illustrate the types of functional connectivity questions that become experimentally accessible with this approach. Importantly, the authors generally avoid overstating these biological findings and appropriately present them as proof-of-principle demonstrations of the technology.

We thank the Reviewer for their constructive input.

Weaknesses:

The primary limitation is inherent to the method itself rather than the execution of the study. Because ADePT relies on one-photon patterned illumination, photo-stimulation remains restricted to relatively superficial structures and does not achieve single-cell spatial resolution. The authors appropriately acknowledge these constraints and clearly position the method within this operating regime. Consequently, ADePT occupies a useful niche for interrogating spatially organized functional units such as olfactory glomeruli or cortical barrels, rather than applications requiring single-cell precision or deeper tissue penetration.

We agree. In the revised manuscript, we will further expand on these points, highlighting the limitations and advantages of ADePT compared to other techniques in a table discussing various operating regimes.

Although the manuscript describes the approach as relatively simple and cost-effective, implementation still requires careful optical alignment, registration, calibration, and optimization. This does not diminish the value of the approach, but terms such as modular or accessible may better reflect the practical implementation than simple. Likewise, a brief bill of materials, approximate add-on cost, and indication of which components are essential versus substitutable would help prospective users assess the accessibility of the system.

We agree. We will change the text accordingly and provide the additional information as suggested.

Finally, the manuscript provides an impressive level of technical characterization, but much of the practical guidance for adopting the system is distributed across the Results and Discussion. Bringing together the principal limitations, recommended operating regime, expected calibration workflow, evidence for long-term alignment stability, and the circumstances in which ADePT is preferable to alternative approaches would further strengthen the manuscript as a community resource.

We agree. We will proceed accordingly in the revised manuscript.

eLife Assessment

This study presents a valuable finding on the spatio-temporal neural dynamics of language processing modulated by story context and presentation speed. The presentation of evidence in the version of the original submission is solid, as providing further clarifications on methodological details and addressing potential confounds would strengthen the study. The work will be of interest to cognitive neuroscientists working on reading and language comprehension.

Reviewer #1 (Public review):

Summary:

This paper reports an important MEG study that is interesting from many different angles. By manipulating presentation rate (Fast vs. Slow) and "Structure" (word list vs. sentence list vs. story), the authors revealed the spatiotemporal dynamics of processing constituents at different speeds and semantic-conceptual scales. The methods are solid, and the results would be interesting to both researchers interested in the neural basis of structure building and researchers interested in the consequences of presentation rate, which, as the authors note, is understudied for visual sentence presentation. Overall, I enjoyed reading this paper and agree that it reports valuable findings for language processing, but there are a few points where clarity can be improved for readers to better evaluate and appreciate this work.

Strengths:

This paper studies the effects of presentation rate and linguistic structure building at different scales with MEG, which is perhaps one of the first MEG studies approaching these questions, and MEG is an appropriate technology to investigate spatiotemporal dynamics in the brain. The authors conducted a breadth of analysis, enabling us to fully understand what is shown by their data.

Weaknesses:

Here I would like to point out a few points where clarity can be improved (e.g., analytical details) for readers to better appreciate this paper.