Reviewer #2 (Public review):

Summary:

The authors decomposed response times into component processes and manipulated the duration of these processes in opposing directions by varying contrast, and overall by manipulating speed-accuracy tradeoffs. They identify different processes and their durations by identifying neural states in time and validate their functional significance by showing that their properties vary selectively as expected with predicted effects of the contrast manipulation. They identify 4 processes: stimulus encoding, attention orienting, decision and motor execution. These map onto 5 classical event related potentials. The decision-making component matched the CPP and its properties varied with contrast and predicted decision-accuracy.

Strengths:

The design of the experiment is remarkable and offers crucial insights. The analyses techniques are beyond-state-of-the art and the analyses are well motivated and offer clear insights.

Weaknesses:

The number of identified events depends on the parameter setting of the analysis. While the authors discuss weaknesses of the approach this needs to be made explicit as well. It is also unclear to what extent topographies map onto processes since e.g., different combinations of sources can lead to the same scalp topography.

Reviewer #3 (Public review):

Summary:

In this manuscript the authors examine the processing stages involved in perceptual decision-making using a new approach to analysing EEG data, combined with a critical stimulus manipulation. This new EEG analysis method enables single-trial estimates of the timing and amplitude of transient changes in EEG time-series recurrent across trials in a behavioural task. The authors find evidence for five events between stimulus onset and the response in a two-spatial-interval visual discrimination task. By analysing the timing and amplitude of these events in relation to behaviour and the stimulus manipulation, the authors interpret these events as related to separable processing stages for stimulus encoding (first two events), attention orientation (second event), motor planning (fourth event) and decision (deliberation, final event). This is largely consistent with previous findings from both event-related potentials (across trials) and single-trial estimates using decoding techniques and neural network approaches. However, by taking a data-driven approach (as opposed to theory-driven decoding analyses) a more nuanced picture emerges: there are several stimulus encoding steps which may contribute differently to behaviour, and decision processes extend beyond the planning of the motor response.

Strengths:

This work is not only important for the conceptual advance, but also in promoting this new analysis technique, which will likely prove useful in future research. For the broader picture, this work is an excellent example of the utility of neural measures for mental chronometry.

Weaknesses:

Though beyond the scope of this manuscript, these results should be considered within the broader decision-making literature, where task or domain-specific processes may not generalise (for example, in value-based decision-making).

Author response:

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

Reviewer #1 (Public review):  

From my reading, this study aimed to achieve two things:  

(1) A neurally-informed account of how Pieron's and Fechner's laws can apply in concert at distinct processing levels.  

(2) A comprehensive map in time and space of all neural events intervening between stimulus and response in an immediately-reported perceptual decision.  

I believe that the authors achieved the first point, mainly owing to a clever contrast comparison paradigm, but with good help also from a new topographic parsing algorithm they created. With this, they found that the time intervening between an early initial sensory evoked potential and an "N2" type process associated with launching the decision process varies inversely with contrast according to Pieron's law. Meanwhile, the interval from that second event up to a neural event peaking just before response increases with contrast, fitting Fechner's law, and a very nice finding is that a diffusion model whose drift rates are scaled by Fechner's law, fit to RT, predicts the observed proportion of correct responses very well. These are all strengths of the study.   

We thank the reviewer for their comments that added context to the events we detected in relation to previous findings. We also believe that the change in the HMP algorithm suggested by the reviewer improved the precision of our analyses and the manuscript. We respond to the reviewer’s specific comments below.

(1) The second, generally stated aim above is, in the opinion of this reviewer, unconvincing and ill-defined. Presumably, the full sequence of neural events is massively task-dependent, and surely it is more in number than just three. Even the sensory evoked potential typically observed for average ERPs, even for passive viewing, would include a series of 3 or more components - C1, P1, N1, etc. So are some events being missed? Perhaps the authors are identifying key events that impressively demarcate Pieron- and Fechner-adherent sections of the RT, but they might want to temper the claim that they are finding ALL events. In addition, the propensity for topographic parsing algorithms to potentially lump together distinct processes that partially co-evolve should be acknowledged.  

We agree with the reviewer that the topographical solutions found by HMP will be dependent on the task and the quality and type of data. We address this point in the last section of the discussion (see also response to R3.5). We would also like to add that the events detected by HMP are, by construction, those that contribute to the RT and not necessarily all ERPs elicited by a stimulus.

In addition to the new last section of the discussion we also make these points clear in the revised manuscript at the discussion start: 

“By modeling the recorded single-trial EEG signal between stimulus onset and response as a sequence of multivariate events with varying by-trial peak times, we  aimed to detect recurrent events that contribute to the duration of the reaction time in the present perceptual decision-making task”.

Regarding the typical visual ERPs, in response to this comment but also comments R1.2, R1.3 and R2.1, we aimed for a more precise description of the topographies and thus reduced the width of the HMP expected events to 25ms. This ensures that we do not miss events shorter than the initial expectations of 50ms (see Appendix B of Weindel et al., 2024 and also response to  R1.3). This new estimation provides evidence for at least two of the visual ERPs that, based on their timings and topographies (in relation with the spatial frequency of the stimulus), we interpret as the N40 and the P100 (see response to R1.5 for the justification of this categorization). We provide a description and justification of the interpretations in the result section “Five trial-recurrent sequential events occur in the EEG during decisions” and the discussion section “Visual encoding time”.

(2) To take a salient example, the last neural event seems to blend the centroparietal positivity with a more frontal midline negativity, some of which would capture the CNV and some motor-execution related components that are more tightly time-locked to, of course, the response. If the authors plotted the traditional single-electrode ERP at the frontal focus and centroparietal focus separately, they are likely to see very different dynamics and contrast- and SAT-dependency. What does this mean for the validity of the multivariate method? If two or more components are being lumped into one neural event, wouldn't it mean that properties of one (e.g., frontal burstiness at response) are being misattributed to the other (centroparietal signal that also peaks but less sharply at response)?

Using the new HMP parameterization described above we show that the reviewer's intuition was correct. Using an expected pattern duration of 25ms the last event in the original manuscript splits in two events. The before-last event, now referred to the lateralized readiness potential (LRP) presents a strong lateralization (Figure 3) with an increased negativity over the motor cortex contralateral to the right hand. The effect of contrast is mostly on the last event that we interpret as the CPP (Figure 5). Despite the improved precision of the topographies of the identified events, it is however to be noted that some components will overlap. If the LRP is generated when a certain amount of evidence is accumulated (e.g. that the CPP crosses a certain value) then a time-based topography will necessarily include that CPP activity in addition to the lateralized potential. We discuss this in the section “Motor execution” of the discussion:

“Adding the abrupt onset of this potential, we believe that this event is the start of motor execution, engaged after a certain amount of evidence. The evidence for this interpretation is manifest in the fact that the event's topography shares some activity with the CPP event that follows, an expected result if the LRP is triggered at a certain amount of evidence, indexed by the CPP”.

(3) Also related to the method, why must the neural events all be 50 ms wide, and what happens if that is changed? Is it realistic that these neural events would be the same duration on every trial, even if their duration was a free parameter? This might be reasonable for sensory and motor components, but unlikely for cognitive.  

The HMP method is sensitive to the event's duration as shown in the manuscript about the method (Appendix B of Weindel et al., 2024). Nevertheless as long as the topography in the real data is longer than the expected one it shouldn't be missed (i.e. same goes for by-trial variations in the event width). For this reason we halved the expected event width of 50ms (introduced by the original HsMM-MVPA paper by Anderson and colleagues) in the revision. This new estimation with 25ms thus is much less likely to miss events as evidenced by the new visual and motor events. In the revised manuscript this is addressed at the start of the Results section:

“Contrary to previous applications (Anderson et al.,2016; Berberyan et al., 2021; Zhang et al., 2018; Krause et al., 2024) we assumed that the multivariate pattern was represented by a 25ms half-sine as our previous research showed that a shorter expected pattern width increases the likelihood of detecting cognitive events (see Appendix B of Weindel et al., 2024)”.

Regarding the event width as a free parameter this is both technically and statistically difficult to implement as the amount of computing capacity, flexibility and trade-offs among the HMP parameters would, given the current implementation, render the model unfit for most computers and statistically unidentifiable.

(4) In general, I wonder about the analytic advantage of the parsing method - the paradigm itself is so well-designed that the story may be clear from standard average event-related potential analysis, and this might sidestep the doubts around whether the algorithm is correctly parsing all neural events.  

Average ERP analysis suffers from an impossibility to differentiate between an effect of an experimental factor on the amplitude vs. on the timing of the underlying components (Luck, 2005). Furthermore the overlap of components across trials bluries the distinction between them. For both reasons we would not be able to reach the same level of certainty and precision using ERP analyses. Furthermore the relatively low number of trials per experimental cell (contrast level X SAT X participant = 6 trials) makes the analyses hard to perform on ERP which typically require more trials per modality. From the reviewer’s comment we understand that this point was not clear. We therefore discuss this in the revision, Section “Functional interpretation of the events” of the results:

“Nevertheless identifying neural dynamics on these ERPs centered on stimulus is complicated by the time variation of the underlying single-trial events (see probabilities displayed in Figure 3 for an illustration and Burle et al., 2008, for a discussion). The likely impact of contrast on both amplitude and time on the underlying single-trial event does not allow one to interpret the average ERP traces as showing an effect in one or the other dimension without strong assumptions (Luck, 2005)”.

(5) In particular, would the authors consider plotting CPP waveforms in the traditional way, across contrast levels? The elegant design is such that the C1 component (which has similar topography) will show up negative and early, giving way to the CPP, and these two components will show opposite amplitude variations (not just temporal intervals as is this paper's main focus), because the brighter the two gratings, the stronger the aggregate early sensory response but the weaker the decision evidence due to Fechner. I believe this would provide a simple, helpful corroborating analysis to back up the main functional interpretation in the paper.  

We agree with the suggestion and have introduced the representation on top of Figure 5 for sets of three electrodes in the occipital, posterior and frontal regions. The new panels clearly show an inversion of the contrast effect dependent on the time and locus of the electrodes. We discuss this in Section “Functional interpretation of the events” of the results:

“This representation shows that there is an inversion of the contrast effect with higher contrasts having a higher amplitude on the electrodes associated with visual potentials in the first couple of deciseconds (left panel of Figure 5A) while parietal and frontal electrodes shows a higher amplitude for lower contrasts in later portions of the ERPs (middle and right panel of Figure 5A)”.

To us, this crucially shows that we cannot achieve the same decomposition using traditional ERP analyses. In these plots it appears that while, as described by the reviewer, there is an inversion, the timing and amplitude of the changes due to contrast can hardly be interpreted.

(6) The first component is picking up on the C1 component (which is negative for these stimulus locations), not a "P100". Please consult any visual evoked potential study (e.g., Luck, Hillyard, etc). It is unexpected that this does not vary in latency with contrast - see, for example. Gebodh et al (2017, Brain Topography) - and there is little discussion of this. Could it be that nonlinear trends were not correctly tested for?  

We disagree with the reviewer on the interpretation of the ERP. The timing of the detected component is later than the one usually associated with a C1. Furthermore the central display does not create optimal conditions to detect a C1

We do agree that the topography raises the confusion but we believe that this is due to the spatial frequency of the stimulus that generates a high posterior positivity (see references in the following extract). The new HMP solution also now happens to show an effect of contrast on the P100 latencies, we believe this is due to the increased precision in the time location of the component. We discuss this in the “Visual encoding time” section of the discussion:

“The following event, the P100, is expressed around 70ms after the N40, its topography is congruent with reports for stimuli with low spatial frequencies as used in the current study (Kenemans et al., 2002, 2000; Proverbio et al., 1996). The timing of this P100 component is changed by the contrast of the stimulus in the direction expected by the Piéron law (Figure 4A)”. 

(7) There is very little analysis or discussion of the second stage linked to attention orientation - what would the role of attention orientation be in this task? Is it spatial attention directed to the higher contrast grating (and if so, should it lateralise accordingly?), or is it more of an alerting function the authors have in mind here?  

We agree that we were not specific enough on the interpretation of this attention stage. We now discuss our hypothesis in the section “Attention orientation” of the discussion:  

“We do however observe an asymmetry in the topographical map Figure 3. This asymmetry might point to an attentional bias with participants (or at least some participants) allocating attention to one side over the other in the same way as the N2pc component (Luck and Hillyard, 1994, Luck et al., 1997). Based on this collection of observations, we conclude that this third event represents an attention orientation process. In line with the finding of Philiastides et al. (2006), this attention orientation event might also relate to the allocation of resources. Other designs varying the expected cognitive load or spatial attention could help in further interpreting the functional role of this third event”.

We would like to add that it is unlikely that the asymmetry we mention in the discussion cannot stem from the redirection towards higher contrast as the experimental design balanced the side of presentation. We therefore believe that this is a behavioral bias rather than a bias toward the highest contrast stimulus as suggested by the reviewer. We hope that, while more could be tested and discussed, this discussion is sufficient given the current manuscript's goal.

Reviewer #2 (Public review):  

Summary:  

The authors decomposed response times into component processes and manipulated the duration of these processes in opposing directions by varying contrast, and overall by manipulating speed-accuracy tradeoffs. They identify different processes and their durations by identifying neural states in time and validate their functional significance by showing that their properties vary selectively as expected with the predicted effects of the contrast manipulation. They identify 3 processes: stimulus encoding, attention orienting, and decision. These map onto classical event-related potentials. The decision-making component matched the CPP, and its properties varied with contrast and predicted decision-accuracy, while also exhibiting a burst not characteristic of evidence accumulation.  

Strengths:  

The design of the experiment is remarkable and offers crucial insights. The analysis techniques are beyond state-of-the-art, and the analyses are well motivated and offer clear insights.  

Weaknesses:  

It is not clear to me that the results confirm that there are only 3 processes, since e.g., motor preparation and execution were not captured. While the authors discuss this, this is a clear weakness of the approach, as other components may also have been missed. It is also unclear to what extent topographies map onto processes, since, e.g., different combinations of sources can lead to the same scalp topography.  

We thank the reviewer for their kind words and for the attention they brought on the question of the missing motor preparation event. In light of this comment (and also R1.1, R3.3) the revised manuscript uses a finer grained approach for the multivariate event detection. This preciser estimation comes from the use of a shorter expected pattern in which the initial expectation of a 50ms half-sine was halved, therefore ensuring that we do not miss events shorter than the initial expectations (see Appendix B of Weindel et al., 2024 and also response to  R1.3). In the new solution the motor component that the reviewer expected is found as evidenced by the topography of the event, its lateralization and a time-to-response congruent with a response execution event. This is now described in the section “Motor execution” of the revised manuscript: 

“The before last event, identified as the LRP, shows a strong hemispheric asymmetry congruent with a right hand response. The peak of this event is approximately 100 ms before the response which is congruent with reports that the LRP peaks at the onset of electromyographical activity in the effector muscle (Burle et al., 2004), typically happening 100ms before the response in such decision-making tasks (Weindel et al., 2021). Furthermore, while its peak time is dependent on contrast, its expression in the EEG is less clearly related to the contrast manipulation than the following CPP event”.

Reviewer #3 (Public review):  

Summary:  

In this manuscript, the authors examine the processing stages involved in perceptual decision-making using a new approach to analysing EEG data, combined with a critical stimulus manipulation. This new EEG analysis method enables single-trial estimates of the timing and amplitude of transient changes in EEG time-series, recurrent across trials in a behavioural task. The authors find evidence for three events between stimulus onset and the response in a two-spatial-interval visual discrimination task. By analysing the timing and amplitude of these events in relation to behaviour and the stimulus manipulation, the authors interpret these events as related to separable processing stages for stimulus encoding, attention orientation, and decision (deliberation). This is largely consistent with previous findings from both event-related potentials (across trials) and single-trial estimates using decoding techniques and neural network approaches.  

Strengths:  

This work is not only important for the conceptual advance, but also in promoting this new analysis technique, which will likely prove useful in future research. For the broader picture, this work is an excellent example of the utility of neural measures for mental chronometry.  

We appreciate the very positive review and thank the reviewer for pointing out important weaknesses in our original manuscript and also providing resources to address them in the recommendations to authors. Below we comment on each identified weakness and how we addressed them.   

Weaknesses:  

(1) The manuscript would benefit from some conceptual clarifications, which are important for readers to understand this manuscript as a stand-alone work. This includes clearer definitions of Piéron's and Fechner's laws, and a fuller description of the EEG analysis technique.

We agree that the description of both laws were insufficient, we therefore added the following text in the last paragraph of the introduction:

“Piéron’s law predicts that the time to perceive the two stimuli (and thus the choice situation) should follow a negative power law with the stimulus intensity (Figure 1, green curve). In contradistinction, Fechner’s law states that the perceived difference between the two patches follows the logarithm of the absolute contrast of the two patches (Figure 1, yellow curve). As the task of our participants is to judge the contrast difference, Piéron’s law should predict the time at which the comparison starts (i.e. the stimuli become perceptible), while Fechner’s law should implement the comparison, and thus decision, difficulty”.

Regarding the EEG analysis technique we added a few elements at the start of the result:

“The hidden multivariate pattern model (HMP) implemented assumed that a task-related multivariate pattern event is represented by a half-sine whose timing varies from trial to trial based on a gamma distribution with a shape parameter of 2 and a scale, controlling the average latency of the event, free-to-vary per event (Weindel et al., 2024)”.

We also made the technique clearer at the start of the discussion:

“By modeling the recorded single-trial EEG signal between stimulus onset and response as a sequence of multivariate events with varying by-trial peak times, we aimed to detect recurrent events that contribute to the duration of the reaction time in the present perceptual decision-making task. In addition to the number of events, using this hidden multivariate pattern approach (Weindel et al., 2024) we estimated the trial-by-trial probability of each event’s peak, therefore accessing at which time sample each event was the most likely to occur”.

Additionally, we added a proper description in the method section (see the new first paragraph of the “Hidden multivariate pattern” subsection). 

(2) The manuscript, broadly, but the introduction especially, may be improved by clearly delineating the multiple aims of this project: examining the processes for decision-making, obtaining single-trial estimates of meaningful EEG-events, and whether central parietal positivity reflects ramping activity or steps averaged across trials.

For the sake of clarity we removed the question of the ramping activity vs steps in the introduction and focused on the processes in decision-making and their single-trial measurement as this is the main topic of the paper. Furthermore the references provided by the reviewer allowed us to write a more comprehensive review of previous studies and how the current study is in line with those. These changes are mainly manifested in these new sentences:

“As an example Philiastides et al. (2006) used a classifier on the EEG activity of several conditions to show that the strength of an early EEG component was proportional to the strength of the stimulus while a later component was related to decision difficulty and behavioral performance (see also Salvador et al., 2022; Philiastides and Sajda, 2006). Furthermore the authors interpreted that a third EEG component was indicative of the resource allocated to the upcoming decision given the perceived decision difficulty. In their study, they showed that it is possible to use single-trial information to separate cognitive processes within decision-making. Nevertheless, their method requires a decoding approach, which requires separate classifiers for each component of interest and restrains the detection of the components to those with decodable discriminating features (e.g. stimuli with strong neural generators such as face stimuli, see Philiastides et al., 2006)”.

(3) A fuller discussion of the limitations of the work, in particular, the absence of motor contributions to reaction time, would also be appreciated. 

As laid out in responses to comments R1.1 and R2 the new estimates now include evidence for a motor preparation component. We discuss this in the new “motor execution” paragraph in the discussion section. Additionally we discuss the limitation of the study and the method in the two last paragraphs of the discussion (in the new Section “Generalization and limitation”).

(4) At times, the novelty of the work is perhaps overstated. Rather, readers may appreciate a more comprehensive discussion of the distinctions between the current work and previous techniques to gauge single-trial estimates of decision-related activity, as well as previous findings concerning distinct processing stages in decision-making. Moreover, a discussion of how the events described in this study might generalise to different decision-making tasks in different contexts (for example, in auditory perception, or even value-based decision-making) would also be appreciated.  

We agree that the original text could be read as overstating. In addition to the changes linked to R3.2 we also now discuss the link with the previous studies in the before-last paragraph of the discussion before the conclusion in the new “Generalization and limitations” section:

“The present study showed what cognitive processes are contributing to the reaction time and estimated single-trial times of these processes for this specific perceptual decision-making task. The identified processes and topographies ought to be dependent on the task and even the stimuli (e.g. sensory events will change with the sensory modality). More complex designs might generate a higher number of cognitive processes (e.g. memory retrieval from a cue, Anderson et al., 2016) and so could more natural stimuli which might trigger other processes in the EEG (e.g. appraisal vs. choice as shown by Frömer et al., 2024). Nevertheless, the observation of early sensory vs. late decision EEG components is likely to generalize across many stimuli and tasks as it has been observed in other designs and methods (Philiastides et al., 2006; Salvador et al., 2022). To these studies we add that we can evaluate the trial-level contribution, as already done for specific processes (e.g. Si et al., 2020; Sturm et al., 2016), for the collection of events detected in the current study”.

Reviewing Editor Comments:  

As you will see, all three reviewers agree that the paper makes a valuable contribution and has many strengths. You will also see that they have provided a range of constructive comments highlighting potential issues with the interpretation of the outcomes of your signal decomposition method. In particular, all three reviewers point out that your results do not identify separate motor preparation signals, which we know must be operating on this type of task. The reviewers suggest further discussion of this issue and the potential limitations of your analysis approach, as well as suggesting some additional analyses that could be run to explore this further. While making these changes would undoubtedly enhance the paper and the final public reviews, I should note that my sense is that they are unlikely to change the reviewers' ratings of the significance of the findings and the strength of evidence in the final eLife assessment  

Reviewer #1 (Recommendations for the authors):  

(1) Abstract: "choice onset" is ill-defined and not the label most would give the start of the RT interval. Do you mean stimulus onset?  

We replaced with "choice onset" with "stimulus onset" in the abstract

(2) Similarly "choice elements" in the introduction seem to refer to sensory attributes/objects being decided about?  

We replaced "choice-elements" with "choice-relevant features of the stimuli"

(3) "how the RT emerges from these putative components" - it would be helpful to specify more what level of answer you're looking for, as one could simply answer "when they're done."  

We replaced with "how the variability in RTs emerges from these putative components"

(4) Line 61-62: I'm not sure this is a fully correct characterisation of Frömer et al. It was not similar in invoking a step function - it did not invoke any particular mechanism or function, and in that respect does not compare well to Latimer et al. Also, I believe it was the overlap of stimulus-locked components, not response-locked, that they argued could falsely generate accumulator-like buildup in the response-locked ERP.  

We indeed wrongly described Frömer et al. The sentence is now "In human EEG data, the classical observation of a slowly evolving centro-parietal positivity, scaling with evidence accumulation, was suggested to result from the overlap of time-varying stimulus-related activity in the response-locked event related potential"

(5) Line 78: Should this be single-trial *latency*?  

This referred to location in time but we agree that the term is confusing and thus replaced it with latencies.

(6) The caption of Figure 1 should state what is meant by the y-axis "time"  

We added the sentence "The y-axis refers the time predicted by each law given a contrast value (x-axis) and the chosen set of parameters." in the caption of Figure 1

(7) Line 107: Is this the correct description of Fechner's law? If the perceived difference follows the log of the physical difference, then a constant physical difference should mean a constant perceived difference. Perhaps a typo here.  

This was indeed a typo we replaced the corresponding part of the sentence with "the perceived difference between the two patches follows the logarithm of the absolute contrast of the two patches"

(8) Line 128: By scale, do you mean magnitude/amplitude?  

No, this refers to the parameter of a gamma distribution. To clarify we edited the sentence:  "based on a gamma distribution with a shape parameter of 2 and a scale parameter, controlling the average latency of the event, free-to-vary per event"

(9) The caption of Figure 3 is insufficient to make sense of the top panel. What does the inter-event interval mean, and why is it important to show? What is the "response" event?  

We agree that the top panel was insufficiently described. To keep the length of the paper short and because of the relatively low amount of information provided by these panels we replaced them for a figure only showing the average topographies as well as the asymmetry tests for each event.

(10) Figure 4: caption should say what the top vs bottom row represents (presumably, accuracy vs speed emphasis?), and what the individual dots represent, given the caption says these are "trial and participant averaged". A legend should be provided for the rightmost panels.  

We agree and therefore edited Figure 4. The beginning of the caption mentioned by the reviewer now reads: “A) The panels represent the average duration between events for each contrast level, averaged across participants and trials (stimulus and response respectively as first and last events) for accuracy (top) and speed instructions (bottom).”. Additionally we added legends for the SAT instructions and the model fits.

(11) Line 189: argued for a decision-making role of what?  

Stafford and Gurney (2004) proposed that Pieron’s law could reflect a non-linear transformation from sensory input to action outcomes, which they argued reflected a response mechanism. We (Van Maanen et al., 2012) specified this result by showing that a Bayesian Observer Model in which evidence for two alternative options was accumulated following Bayes Rule indeed predicted a power relation between the difference in sensory input of the two alternatives, and mean RT. However, the current data suggest that such an explanation cannot be the full story, as also noted by R3. To clarify this point we replaced the comment by the following sentence:

“Note that this observation is not necessarily incongruent with theoretical work that argued that Piéron’s law could also be a result of a response selection mechanism (Stafford and Gurney, 2004; Van Maanen et al., 2012; Palmer et al., 2005). It could be that differences in stimulus intensity between the two options also contribute to a Piéron-like relationship in the later intervals, that is convoluted with Fechner’s law (see Donkin and Van Maanen, 2014 for a similar argument). Unfortunately, our data do not allow us to discriminate between a pure logarithmic growth function and one that is mediated by a decreasing power function”.

(12) Table 2: There is an SAT effect even on the first interval, which is quite remarkable and could be discussed more - does this mean that the C1 component occurs earlier under speed pressure? This would be the first such finding.  

The original event we qualified as a P100 was sensitive to SAT but the earliest event is now the N40 and isn’t statistically sensitive to speed pressure in this data. We believe that the fact that the P100 is still sensitive to SAT is not a surprise and therefore do not outline it.

(13) Line 221: "decrease of activation when contrast (and thus difficulty) increases" - is this shown somewhere in the paper?  

The whole section for this analysis was rewritten (see comment below)

(14) I find the analysis of Figure 5 interesting, but the interpretation odd. What is found is that the peak of the decision signal aligns with the response, consistent with previous work, but the authors choose to interpret this as the decision signal "occurring as a short-lived burst." Where is the quantitative analysis of its duration across trials? It can at least be visually appraised in the surface plot, and this shows that the signal has a stimulus-locked onset and, apart from the slowest RTs, remains present and for the most part building, until response. What about this is burst-like? A peak is not a burst.  

This was the residue of a previous version of the paper where an analysis reported that no evidence accumulation trace was found. But after proper simulations this analysis turned out to be false because of a poor statistical test. Thus we removed this paragraph in the revised manuscript and Figure 5 has now been extended to include surface plots for all the events.

Reviewer #2 (Recommendations for the authors):  

Overall, I really enjoyed reading this paper. However, in some places the approach is a bit opaque or the results are difficult to follow. As I read the paper, I noted:  

Did you do a simple DDM, or did you do a collapsing bound for speed?  

The fitted DDM was an adaptation of the proportional rate diffusion model. We make this clearer at the end of the introduction: "Given that Fechner’s law is expected to capture decision difficulty we connected this law to the classical diffusion decision models by replacing the rate of accumulation with Fechner’s law in the proportional rate diffusion model of Palmer et al.(2005).”

It is confusing that the order of intervals in the text doesn't match the order in the table. It might be better to say what events the interval is between rather than assuming that the reader reconstructs.  

We agree and adapted the order in both the text and the table. The table is now also more explicit (e.g. RT instead of S-R)

Otherwise, I do wonder to what extent the method is able to differentiate processes that yield similar scalp topographies and find it a bit concerning that no motor component was identified.  

We believe that the new version with the LRP/CPP is a demonstration that the method can handle similar topographies. The method can handle events with close topographies as long as they are separate in time, however if they are not sequential to one another the method cannot capture both events. We now discuss this, in relation with the C1/P100 overlap, in the discussion section “Visual encoding time”:

“Nevertheless this event, seemingly overlapping with the P100 even at the trial level (Figure 5C), cannot be recovered by the method we applied. The fact that the P100 was recovered instead of the C1 could indicate that only the timing of the P100 contributes to the RT (see Section 3 of Weindel et al., 2024)”.

And we more generally address the question of overlap in the new section “Generalization and limitation”.

Reviewer #3 (Recommendations for the authors):  

Major Comments:  

(1) If we agree on one thing, it is that motor processes contribute to response time. Line 364: "In the case of decision-making, these discrete neural events are visual encoding, attention-orientation, and decision commitment, and their latency make up the reaction time." Does the third event, "decision commitment", capture both central parietal positivity (decision deliberation) and motor components? If so, how can the authors attribute the effects to decision deliberation as opposed to motor preparation?  

Thanks to the suggestions also in the public part. This main problem is now addressed as we do capture both a motor component and a decision commitment.

Line 351 suggests that the third event may contain two components.  

This was indeed our initial, badly written, hypothesis. Nevertheless the new solution again addresses this problem.

The time series in Figure 6 shows an additional peak that is not evident in the simulated ramp of Appendix 1.  

This was probably due to the overlap of both the CPP and the LRP. It is now much clearer that the CPP looks mostly like a ramp while the LRP looks much more like a burst-like/peaked activity. We make this clear in the “Decision event” paragraph of the discussion section:

“Regarding the build-up of this component, the CPP is seen as originating from single-trial ramping EEG activities but other work (Latimer et al., 2015; Zoltowski et al., 2019) have found support for a discrete event at the trial-level. The ERPs on the trial-by-trial centered event in Figure 5 show support for both accounts. As outlined above, the LRP is indeed a short burst-like activity but the build-up of the CPP between high vs low contrast diverges much earlier than its peak”.

Previous analyses (Weindel et al., 2024) found motor-related activity from central parietal topographies close to the response by comparing the difference in single-trial events on left- vs right-hand response trials. The authors suggest at line 315 that the use of only the right hand for responding prevented them from identifying a motor event.  

The use of only the right hand should have made the event more identifiable because the topography would be consistent across trials (rather than inverting on left vs right hand response trials).  

The reviewer is correct, in the original manuscript we didn’t test for lateralization, but the comment of the reviewer gave us the idea to explicitly test for the asymmetry (Figure 3). This test now clearly shows what would be expected for a motor event with a strong negativity over the left motor cortex.

The authors state on line 422 that the EEG data were truncated at the time of the response.  

Could this have prevented the authors from identifying a motor event that might overlap with the timing of the response?  

We thank the reviewer for this suggestion. This would have been a possibility but the problem is that adding samples after the response also adds the post-response processes (error monitoring, button release, stimulus disappearance, etc.). While increasing the samples after the response is definitely something that we need to inspect, we think that the separation we achieved in this revision doesn’t call for this supplementary analysis.

The largest effects of contrast on the third event amplitude appear around the peak as opposed to the ramp. If the peak is caused by the motor component, how does this affect the conclusions that this third event shows a decision-deliberation parietal processes as opposed to a motor process (a number of studies suggest a causal role for motor processes in decision-making e.g. Purcell et al., 2010 Psych Rev; Jun et al., 2021 Nat Neuro; Donner et al., 2009 Curr Bio).  

This result now changed and it does look like the peak capturing most of the effect is no longer true. We do however think that there might be some link to theories of motor-related accumulation. We therefore added this to the discussion in the Motor execution section:

“Based on all these observations, it is therefore very likely that this LRP event signs the first passage of a two-step decision process as suggested by recent decision-making models (Servant et al., 2021; Verdonck et al., 2021; Balsdon et al., 2023)”.

I would suggest further investigation into the motor component (perhaps by extending the time window of analysed EEG to a few hundred ms after the response) and at least some discussion of the potential contribution of motor processes, in relation to the previous literature.  

We believe that the absence of a motor component is sufficiently addressed in the revised manuscript and in the responses to the other comments.    

(2) What do we learn from this work? Readers would appreciate more attention to previous findings and a clearer outline of how this work differs. Two points stand out, outlined below. I believe the authors can address these potential complaints in the introduction and discussion, and perhaps provide some clarification in the presentation of the results.  

In the introduction, the authors state that "... to date, no study has been able to provide single-trial evidence of multiple EEG components involved in decision-making..." (line 64). Many readers would disagree with this. For example, Philiastides, Ratcliff, & Sadja (2006) use a single-trial analysis to unravel early and late EEG components relating to decision difficulty and accuracy (across different perceptual decisions), which could be related to the components in the current work. Other, network-based single-trial EEG analyses (e.g., Si et al., 2020, NeuroImage, Sturn et al., 2016 J Neurosci Methods) could also be related to the current component approach. Yet other approaches have used inverse encoding models to examine EEG components related to separable decision processes within trials (e.g., Salvador et al., 2022, Nat Comms). The results of the current work are consistent with this previous work - the two components from Philiastides et al., 2006 can be mapped onto the components in the current work, and Salvador et al., 2022 also uncover stimulus- and decision-deliberation related components.  

We completely agree with the reviewer that the link to previous work was insufficient. We now include all references that the reviewer points out both in the introduction (see response R3.2) and in the discussion (see response R3.4). We wish to thank the reviewer for bringing these papers to our attention as they are important for the manuscript.

The authors relate their components to ERPs. This prompts the question of whether we would get the same results with ERP analyses (and, on the whole, the results of the current work are consistent with conclusions based on ERP analyses, with the exception of the missing motor component). It's nice that this analysis is single-trial, but many of the follow-up analyses are based on grouping by condition anyway. Even the single-trial analysis presented in Figure 4 could be obtained by median splits (given the hypotheses propose opposite directions of effects, except for the linear model). 

We do not agree with the reviewer in the sense that classical ERP analyses would require much more data-points. The performance of the method is here to use the information shared across all contrast levels to be able to model the processing time of a single contrast level (6 trials per participant). Furthermore, as stated in the response to R1.4 and R1.5, the aim of the paper is to have the time of information processing components which cannot be achieved with classical ERPs without strong, and likely false, assumptions.

Medium Comments:  

(1) The presentation of Piéron's law for the behavioural analysis is confusing. First, both laws should be clearly defined for readers who may be unfamiliar with this work. I found the proposal that Piéron's law predicts decreasing RT for increasing pedestal contrast in a contrast discrimination paradigm task surprising, especially given the last author's previous work. For example, Donkin and van Maanen (2014) write "However, the commonality ofPiéron's Law across so many paradigms has lead researchers (e.g., Stafford & Gurney, 2004; Van Maanen et al., 2012) to propose that Piéron's Law is unrelated to stimulus scaling, but is a result of the architecture of the response selection (or decision making) process." The pedestal contrast is unrelated to the difficulty of the contrast discrimination task (except for the consideration of Fechner's law). Instead, Piéron's law would apply to the subjective difference in contrast in this task, as opposed to the pedestal contrast. The EEG results are consistent with these intuitions about Piéron's law (or more generally, that contrast is accumulated over time, so a later EEG component for lower pedestal contrast makes sense): pedestal contrast should lead to faster detection, but not necessarily faster discrimination. Perhaps, given the complexity of the manuscript as a whole, the predictions for the behavioural results could be simplified?  

We agree that the initial version was confusing. We now clarified the presentation of Piéron's law at the end of the introduction (see also response to R2).

Once Fechner's law is applied, decision difficulty increases with increasing contrast, so Piéron's law on the decision-relevant intensity (perceived difference in contrast) would also predict increasing RT with increasing pedestal contrast. It is unlikely that the data are of sufficient resolution to distinguish a log function from a power of a log function, but perhaps the claim on line 189 could be weakened (the EEG results demonstrate Piéron's law for detection, but do not provide evidence against Piéron's law in discrimination decisions).  

This is an excellent observation, thank you for bringing it to our attention. Indeed, the data support the notion that Pieron’s law is related to detection, but do not rule out that it is also related to decision or discrimination. In earlier work, we (Donkin & Van Maanen, 2014) addressed this question as well, and reached a similar conclusion. After fitting evidence accumulation models to data, we found no linear relationship between drift rates and stimulus difficulty, as would have been the case if Pieron's law could be fully explained by the decision process (as -indirectly- argued by Stafford & Gurney, 2004; Van Maanen et al., 2012). The fact that we observed evidence for a non-linear relationship between drift rates and stimulus difficulty led us to the same conclusion, that Pieron’s law could be reflected in both discrimination and decision processes. We added the following comment to the discussion about the functional locus of Pieron's law to clarify this point:

“Note that this observation is not necessarily incongruent with theoretical work that argued that Piéron’s law could also be a result of a response selection mechanism (Stafford and Gurney, 2004; Van Maanen et al., 2012; Palmer et al., 2005). It could be that differences in stimulus intensity between the two options also contribute to a Piéron like relationship in the later intervals, that is convoluted with Fechner’s law (see Donkin and Van Maanen, 2014, for a similar argument). Unfortunately, our data do not allow us to discriminate between a pure logarithmic growth function and one that is mediated by a decreasing power function”.

(2) Appendix 1 shows that the event detection of the HMP method will also pick up on ramping activity. The description of the problem in the introduction is that event-like activity could look like ramping when averaged across trials. To address this problem, the authors should simulate events (with some reasonable dispersion in timing such that they look like ramping when averaged) and show that the HMP method would not pull out something that looked like ramping. In other words, the evidence for ramping in this work is not affected by the previously identified confounds.  

We agree that this demonstration was necessary and thus added the suggested simulation to Appendix 1. As can be seen in the Figure 1 of the appendix, when we simulate a half-sine the average ERP based on the timing of the event looks like a half-sine.

(3) Some readers may be interested in a fuller discussion of the failure of the Fechner diffusion model in the speed condition.  

We are unsure which failure the reviewer refers to but assumed it was in relation to the behavioral results and thus added: 

It is unlikely that neither Piéron nor Fechner law impact the RT in the speed condition. Instead this result is likely due to the composite nature of the RT where both laws co-exist in the RT but cancel each other out due to their opposite prediction.

Minor Comments:  

(1) "By-trial" is used throughout. Normally, it is "trial-by-trial" or "single-trial" or "trial-wise".

We replaced all occurrences of “by-trial”  with the three terms suggested were appropriate.

(2) Line 22: "The sum of the times required for the completion of each of these precessing steps is the reaction time (RT)." The total time required. Processing.  

Corrected for both.

(3) Line 26/27: "Despite being an almost two century old problem (von Helmholtz, 2021)." Perhaps the citation with the original year would make this point clearer.  

We agree and replaced the citation.

(4) Line 73: "accounted by estimating". Accounted for by estimating.  

Corrected.

(5) Line 77 "provides an estimation on the." Of the.  

Corrected.

(6) Line 86: "The task of the participants was to answer which of two sinusoidal gratings." The picture looks like Gabor's? Is there a 2d Gaussian filter on top of the grating? Clarify in the methods, too.  

We incorrectly described the stimuli as those were indeed just Gabor’s. This is now corrected both in the main text and the method section.

(7) Figure 1 legend: "The Fechner diffusion law" Fechner's law or your Fechner diffusion model?  

Law was incorrect so we changed to model as suggested.

(8) Line 115: "further allows to connects the..." Allows connecting the.  

Corrected.

(9) Line 123: "lower than 100 ms or higher than..." Faster/slower.  

Corrected.

(10) Line 131: "To test what law." Which law.?  

Corrected to model.

(11) Figure 2 legend: "Left: Mean RT (dot) and average fit (line) over trials and participants for each contrast level used." The fit is over trials and participants? Each dot is? Average trials for each contrast level in each participant?  

This sentence was corrected to “Mean RT (dot) for each contrast level and averaged predictions of the individual fits (line) with Accuracy (Top) and Speed (Bottom) instructions.”.

(12) Line 231: "A comprehensive analysis of contrast effect on". The effect of contrast on.  

This title was changed to “functional interpretation of the events”.

(13) Line 23: "the three HMP event with". Three HMP events.

The sentence no longer exists in the revised manuscript.

(14) Line 270: "Secondly, we computed the Pearson correlation coefficient between the contrast averaged proportion of correct." Pearson is for continuous variables. Proportion correct is not continuous. Use Spearman, Kendall, or compute d'.  

The reviewer rightly pointed out our error, we corrected this by computing Spearman correlation.

(15)  Line 377: "trial 𝑛 + 1 was randomly sampled from a uniform distribution between 0.5 and 1.25 seconds." It's just confusing why post-response activity in Figure 5 does look so consistent. Throughout methods: "model was fitted" should be "was fit", and line 448, "were split".  

We do not have a specific hypothesis of why the post-response activity in the previous Figure 5 was so consistent. Maybe the Gaussian window (same as in other manuscripts with a similar figure, e.g. O’Connell et al. 2012) generated this consistency. We also corrected the errors mentioned in the methods.

(16) The linear mixed models paragraph is a bit confusing. Can it clearly state which data/ table is being referred to and then explain the model? "The general linear mixed model on proportion of correct responses was performed using a logit link. The linear mixed models were performed on the raw milliseconds scale for the interval durations and on the standardized values for the electrode match." We go directly from proportion correct to raw milliseconds...  

The confusion was indeed due to the initial inclusion of a general linear mixed model on proportion correct which was removed as it was not very informative. The new revision should be clearer on the linear mixed models (see first sentence of subsection ‘linear mixed models' in the method section).

(17) A fuller description of the HMP model would be appreciated.  

We agree that this was necessary and added the description of the HMP model in the corresponding method section “Hidden multivariate pattern” in addition to a more comprehensive presentation of HMP in the first paragraph of the Result and Discussion sections.

(18) Line 458: "Fechner's law (Fechner, 1860) states that the perceived difference (𝑝) between the two patches follows the logarithm of the difference in physical intensity between..." ratio of physical intensity.  

Corrected.

(19) P is defined in equations 2 and 4. I would include the beta in equation 4, like in equation 2, then remove the beta from equations 3 and 5 (makes it more readable). I would also just include the delta in equation 2, state that in this case, c1 = c+delta/2 or whatever.  

This indeed makes the equation more readable so we applied the suggestions for equations 2, 3, 4 and 5. The delta was not added in equation 2 but instead in the text that follows:

“Where 𝐶1 = 𝐶0 + 𝛿, again with a modality and individual specific adjustment slope (𝛽).” 

(20) The appendix suggests comparing the amplitudes with those in Figure 3, but the colour bar legend is missing, so the reader can only assume the same scale is used?  

We added the color bar as it was indeed missing. Note though that the previous version displayed the estimation for the simulated data while this plot in the revised manuscript shows the solution on real data obtained after downsampling the data (and therefore look for a larger pattern as in the main text). We believe that this representation is more useful given that the solution for the downsampled data is no longer the same as the one in the main text (due to the difference in pattern width).

eLife Assessment

This Review Article provides a thorough overview of whole-brain activity changes induced by brain stimulation and summarizes the current state of the field. However, it lacks integration across spatial and mechanistic scales, which limits the reader's ability to understand how the different findings relate to one another. In addition, several key concepts are not explained in sufficient depth for non-expert readers. The manuscript would benefit from the development of a cohesive conceptual framework to more clearly synthesize the existing literature.

Reviewer #1 (Public review):

Summary:

This paper is a comprehensive review of perturbation studies and the state-dependence of the brain's response to perturbation at the circuit, mesoscale, and macroscale levels.

Strengths:

The strengths of the paper are the thorough description of many perturbation studies at different levels of organization, and the integration of both experimental and modeling studies. The review clearly communicates the need to consider (1) brain or local-population state, and (2) multiple levels of organization, in order to understand perturbation responses. Another major strength is the ability for the reader to reproduce figures using the EBRAINS platform.

Weaknesses:

Two major points of improvement should be resolved with the review, in order to make it useful for a broad audience.

The first is that the review does not include a significant integration across scales, and as a result, reads like three separate (though comprehensive) reviews. Currently, the only integration across the scales is in the brief conclusion paragraph. I would recommend adding an additional section, in which the overarching picture is discussed. (i.e. a unifying view of state dependence, and what is learned by considering across scales). This need not be too long, but it should be longer than a single conclusion paragraph.

The second major weakness is that there is a lack of clarity on many points throughout, which is needed for the reader to fully understand the results described.

Reviewer #2 (Public review):

Summary:

In this review article, the authors discuss the whole-brain activity changes induced by brain stimulation. They review the literature on how these activity changes depend on the cognitive state of the brain and divide the results by the scale of the change being induced, from microscale changes across small groups of neurons, up to macroscale changes across the entire brain. Finally, they describe attempts to model these changes using computational models.

Strengths:

The review provides an overview of the results within this subfield of neuroscience, and the authors are able to discuss a lot of prior results. The framing of the changes in neuronal activity in terms of computational changes is also a helpful approach.

Weaknesses:

However, the authors are not able to contextualize these results within a single framework, i.e. explaining from first principles how different aspects of stimulus-induced changes interact to generate functional changes in the brain, and how different changes - at distinct spatiotemporal scales - combine to form larger effects. This is a significant weakness in generating a review of the literature, since the authors do not provide a cohesive conceptual framework on which to frame the results. Similarly, the authors do not explain how their different computational models fit together, and how one can get a singular computational understanding of the distinct mechanisms of brain activity changes due to stimulation under different brain states, by combining the results derived from each separate model.

Major Comments:

(1) The authors have written this review as if it were intended for an audience who is already familiar with the topics. For example, they introduce concepts like complexity, spiral vs planar waves, without much explanation.

(2) Regarding complexity, the authors present a quantification termed PCI. However, in the associated box, they state that PCI could be implemented in a number of different ways, using analogous metrics (which are, nonetheless, not identical). Yet the authors simply claim that all these metrics are sufficiently similar to be grouped together as "PCI". The authors do not provide much intuition about this, and they also don't present any other potential quantifications. This makes any interpretation of their results strongly dependent on your understanding of the concept of PCI. It would be helpful to present some other, analogous metric to demonstrate that the results that the authors are focusing on are not somehow tied to the specific computational structure of the PCI metric.

(3) The authors divide the review into sections organized by the spatial extent of the effects that they are exploring (e.g. from microscale to macroscale). However, they don't bring together these insights into a cohesive structure - for example, by providing potential explanations of the macroscale effects by using the microscale changes.

(4) The authors completely ignore any aspect of cell-type specificity in their review, despite the known importance of specific cell types at the microcircuit scale. This makes it difficult to map their results onto the true biological system.

(5) The authors introduce several different computational models, such as the Hopf model, the AdEx model, and the MPR model. However, they do not provide the reader with a conceptual understanding of the structure of each of these models (except through potentially more complex terminology, e.g. the Hopf model is a "phenomenological Stuart-Landau nonlinear oscillator"). Additionally, though they present the results of each simulation, they don't provide the reader with intuition about how these models compare against each other, and how best to interpret results derived from each model.

(6) In several cases, the authors make statements that they appear to believe to be completely straightforward (and require no justification), but that do not appear so to the reader. For example, they mention: "In wakefulness and REM sleep, ..., the membrane potential is depolarized and close to the spike threshold, which explains why neurons respond more reliably and with less response variability compared with slow-wave sleep". However, this statement is not obvious to the reader and requires explanation (for example, in a system that is close to balance, bringing cells closer to the firing threshold can result in increased response jitter).

Author response:

We would like to thank the editors and reviewers for excellent suggestions to improve the paper. We will carefully address these suggestions in a revised version.

eLife Assessment

This potentially valuable cross-sectional longitudinal study leverages high-definition transcranial direct current stimulation to the left dorsolateral prefrontal cortex to examine its effect on procrastination behavior over an extended time span. Support for the conclusions is incomplete owing to missing information about the analyses, the nature of the procrastination tasks, and the derived dependent measures.

Reviewer #1 (Public review):

Summary:

The authors report the results of a tDCS brain stimulation study (verum vs sham stimulation of left DLPFC; between-subjects) in 46 participants, using an intense stimulation protocol over 2 weeks, combined with an experience-sampling approach, plus follow-up measures after 6 months.

Strengths:

The authors are studying a relevant and interesting research question using an intriguing design, following participants quite intensely over time and even at a follow-up time point. The use of an experience-sampling approach is another strength of the work.

Weaknesses:

There are quite a few weaknesses, some related to the actual study and some more strongly related to the reporting about the study in the manuscript. The concerns are listed roughly in the order in which they appear in the manuscript.

(1) In the introduction, the authors present procrastination nearly as if it were the most relevant and problematic issue there is in psychology. Surely, procrastination is a relevant and study-worthy topic, but that is also true if it is presented in more modest (and appropriate) terms. The manuscript mentions that procrastination is a main cause of psychopathology and bodily disease. These claims could possibly be described as 'sensationalized'. Also, the studies to support these claims seem to report associations, not causal mechanisms, as is implied in the manuscript.

(2) It is laudable that the study was pre-registered; however, the cited OSF repository cannot be accessed and therefore, the OSF materials cannot be used to (a) check the preregistration or to (b) fill in the gaps and uncertainties about the exact analyses the authors conducted (this is important because the description of the analyses is insufficiently detailed and it is often unclear how they analyzed the data).

(3) Related to the previous point: I find it impossible to check the analyses with respect to their appropriateness because too little detail and/or explanation is given. Therefore, I find it impossible to evaluate whether the conclusions are valid and warranted.

(4) Why is a medium effect size chosen for the a priori power analysis? Is it reasonable to assume a medium effect size? This should be discussed/motivated. Related: 18 participants for a medium effect size in a between-subjects design strikes me as implausibly low; even for a within-subjects design, it would appear low (but perhaps I am just not fully understanding the details of the power analysis).

(5) It remains somewhat ambiguous whether the sham group had the same number of stimulation sessions as the verum stimulation group; please clarify: Did both groups come in the same number of times into the lab? I.e., were all procedures identical except whether the stimulation was verum or sham?

(6) The TDM analysis and hyperbolic discounting approach were unclear to me; this needs to be described in more detail, otherwise it cannot be evaluated.

(7) Coming back to the point about the statistical analyses not being described in enough detail: One important example of this is the inclusion of random slopes in their mixed-effects model which is unclear. This is highly relevant as omission of random slopes has been repeatedly shown that it can lead to extremely inflated Type 1 errors (e.g., inflating Type 1 errors by a factor of then, e.g., a significant p value of .05 might be obtained when the true p value is .5). Thus, if indeed random slopes have been omitted, then it is possible that significant effects are significant only due to inflated Type 1 error. Without more information about the models, this cannot be ruled out.

(8) Related to the previous point: The authors report, for example, on the first results page, line 420, an F-test as F(1, 269). This means the test has 269 residual degrees of freedom despite a sample size of about 50 participants. This likely suggests that relevant random slopes for this test were omitted, meaning that this statistical test likely suffers from inflated Type 1 error, and the reported p-value < .001 might be severely inflated. If that is the case, each observation was treated as independent instead of accounting for the nestedness of data within participants. The authors should check this carefully for this and all other statistical tests using mixed-effects models.

(9) Many of the statistical procedures seem quite complex and hard to follow. If the results are indeed so robust as they are presented to be, would it make sense to use simpler analysis approaches (perhaps in addition to the complex ones) that are easier for the average reader to understand and comprehend?

(10) As was noted by an earlier reviewer, the paper reports nearly exclusively about the role of the left DLPFC, while there is also work that demonstrates the role of the right DLPFC in self-control. A more balanced presentation of the relevant scientific literature would be desirable.

(11) Active stimulation reduced procrastination, reduced task aversiveness, and increased the outcome value. If I am not mistaken, the authors claim based on these results that the brain stimulation effect operates via self-control, but - unless I missed it - the authors do not have any direct evidence (such as measures or specific task measures) that actually capture self-control. Thus, that self-control is involved seems speculation, but there is no empirical evidence for this; or am I mistaken about this? If that is indeed correct, I think it needs to be made explicit that it is an untested assumption (which might be very plausible, but it is still in the current study not empirically tested) that self-control plays any role in the reported results.

(12) Figures 3F and 3H show that procrastination rates in the active modulation group go to 0 in all participants by sessions 6 and 7. This seems surprising and, to be honest, rather unlikely that there is absolutely no individual variation in this group anymore. In any case, this is quite extraordinary and should be explicitly discussed, if this is indeed correct: What might be the reasons that this is such an extreme pattern? Just a random fluctuation? Are the results robust if these extreme cells are ignored? The authors remove other cells in their design due to unusual patterns, so perhaps the same should be done here, at least as a robustness check.

(13) The supplemental materials, unfortunately, do not give more information, which would be needed to understand the analyses the authors actually conducted. I had hoped I would find the missing information there, but it's not there.

In sum, the reported/cited/discussed literature gives the impression of being incomplete/selectively reported; the analyses are not reported sufficiently transparently/fully to evaluate whether they are appropriate and thus whether the results are trustworthy or not. At least some of the patterns in the results seem highly unlikely (0 procrastination in the verum group in the last 2 observation periods), and the sample size seems very small for a between-subjects design.

Reviewer #2 (Public review):

Summary:

Chen and colleagues conducted a cross-sectional longitudinal study, administering high-definition transcranial direct stimulation targeting the left DLPFC to examine the effect of HD-tDCS on real-world procrastination behavior. They find that seven sessions of active neuromodulation to the left DLPFC elicited greater modulation of procrastination measures (e.g., task-execution willingness, procrastination rates, task aversiveness, outcome value) relative to sham. They report that tDCS effects on task-execution willingness and procrastination are mediated by task outcome value and claim that this neuromodulatory intervention reduces procrastination rates quantified by their task. Although the study addresses an interesting question regarding the role of DLPFC on procrastination, concerns about the validity of the procrastination moderate enthusiasm for the study and limit the interpretability of the mechanism underlying the reported findings.

Strengths:

(1) This is a well-designed protocol with rigorous administration of high-definition transcranial direct current stimulation across multiple sessions. The approach is solid and aims to address an important question regarding the putative role of DLPFC in modulating chronic procrastination behavior.

(2) The quantification of task aversiveness through AUC metrics is a clever approach to account for the temporal dynamics of task aversiveness, which is notoriously difficult to quantify.

Weaknesses:

(1) The lack of specificity surrounding the "real-world measures" of procrastination is problematic and undermines the strength of the evidence surrounding the DLPFC effects on procrastination behavior. It would be helpful to detail what "real-world tasks" individuals reported, which would inform the efficacy of the intervention on procrastination performance across the diversity of tasks. It is also unclear when and how tasks were reported using the ESM procedure. Providing greater detail of these measures overall would enhance the paper's impact.

(2) Additionally, it is unclear whether the reported effects could be due to differential reporting of tasks (e.g., it could be that participants learned across sessions to report more achievable or less aversive task goals, rather than stimulation of DLPFC reducing procrastination per se). It would be helpful to demonstrate whether these self-reported tasks are consistent across sessions and similar in difficulty within each participant, which would strengthen the claims regarding the intervention.

(3) It would be helpful to show evidence that the procrastination measures are valid and consistent, and detail how each of these measures was quantified and differed across sessions and by intervention. For instance, while the AUC metric is an innovative way to quantify the temporal dynamics of task-aversiveness, it was unclear how the timepoints were collected relative to the task deadline. It would be helpful to include greater detail on how these self-reported tasks and deadlines were determined and collected, which would clarify how these procrastination measures were quantified and varied across time.

(4) There are strong claims about the multi-session neuromodulation alleviating chronic procrastination, which should be moderated, given the concerns regarding how procrastination was quantified. It would also be helpful to clarify whether DLPFC stimulation modulates subjective measures of procrastination, or alternatively, whether these effects could be driven by improved working memory or attention to the reported tasks. In general, more work is needed to clarify whether the targeted mechanisms are specific to procrastination and/or to rule out alternative explanations.

Reviewer #3 (Public review):

This manuscript explores whether high-definition transcranial direct current stimulation (HD-tDCS) of the left DLPFC can reduce real-world procrastination, as predicted by the Temporal Decision Model (TDM). The research question is interesting, and the topic - neuromodulation of self-regulatory behavior - is timely.

However, the study also suffers from a limited sample size, and sometimes it was difficult to follow the statistics.

The preregistration and ecological design (ESM) are commendable, but I was not able the find the preregistration, as reported in the paper.

Overall, the paper requires substantial clarification and tightening.

Author response:

Reviewer #1:

(1) We fully thank you to point out the risks of sensationalizing ramification of procrastination on psychopathology, and would rewrite the Introduction section by adding balanced evidence and overall toning down such inappropriate claims meanwhile.

(2) Thank you to raise this crucial question. We are sorry for this fundamental technical issue to preregistration. This occurs from a seriously technical hurdle. The OSF has banned my OSF account, as it claimed to detect “suspicious user’s activities” in my account. This causes no accesses to all materials that already deposited in this OSF account, including this preregistration. We have contacted OSF team, but received no valid technical solution. We reckon that this may be mistaken by my affiliation changes to Third Military Medical University of People’s Liberation Army (PLA). To tackle with this technical issue, we shall upload preregistration in a new repository soon.

(3) This is a back-to-back study to conceptually probe into whether strengthening left DLPFC can mitigate procrastination via reducing task aversiveness or weighting outcome value. Thus, the current study selected a medium effect size in aprior by following the previous one (Xu et al., 2023). This effect size is calculated by the new tool called “Power Contours” (Baker et al., 2021), which weights statistical power by increasing within-subject repeated measures. As you kindly pointed out, we shall clarify effect size calculation in the revised manuscript.

(4) Yes, both groups come in the same number of times into the lab for tDCS stimulation, except to the type (active vs sham).

(5) We shall add full details for clarifying TDM and hyperbolic discounting modeling.

(6) Thank you to raise this very crucial statistical question. We shall double-check whether multiple sessions are modeled as random slopes, and would like to reanalysis it in case which those random slopes are omitted.

(7) Thank you. We have no intentions of confusing you by adding those complicated statistics, but indeed enrich understanding of how we can interpret those findings.

(8) Yes, as mentioned above, we shall add balanced evidence to clarify both left and right DLPFC may function to self-control capability in the Introduction section.

(9) Yes, this is a conceptual hypothesis --- actively stimulating left DLPFC could improve self-control functions. Thank you for this very nuanced but crucial insight, and we could explicitly clarify the nature of our conclusions.

(10) Yes, we ensure that all the participants successfully completed their tasks before deadline at session 6 and 7, and the procrastination rates have been all decreased to 0. Personally speaking, this is somewhat surprise to us as well, but we affirmed this case. For a portion of participants included in the active group, we have received written letters of thanks from them. Thus, this is surprise but exciting finding. Furthermore, thank you for this helpful suggestion, and we would like to do this robustness check by iteratively removing each session, to obviate the statistical biases from an extreme pattern.

(11) Yep, we fully agree with you to add full details in the main text rather in Supplemental materials, and would like to do so in the first round of revision.

Reviewer #2:

(1) Thank you for this very crucial suggestion. We are sorry for this case that much details are omitted to comply with editorial requirement at Nature Human Behaviour (last submission). We do apologize to confuse you as those ambiguous descriptions, and would like to clearly clarify how we measure participants’ procrastination in the real-world tasks. In brief, we asked participant to report a real task that would really happen in the tomorrow and its deadline is also no more than tomorrow. When tomorrow comes, we used ESM to require participant reporting real task completion rate (0-100%) at five time points before the deadline. The five time points are determined by a hyperbolic discounting model (see how and why we set those five time points in the full author’s response letter later). When participant reports the real task completion rate (0-100%) at a given time point, she/he is required to provide a photo to prove its authenticity. The dependent variable --- real-world procrastination rates --- is thus calculated as 100% subtracts the task completion rate (0-100%) when the deadline meets. That is to say, if participant reports task has been fully completed before or when deadline meets, his/her real-world procrastination rate is 100% - 100% = 0%; if reporting task has been completed 60% when deadline meets, the real-world procrastination rate is determined as 100% - 60% = 40%. Do not worry for spurious reporting, we asked all the participants to provide photo verifying the real task completion rate. This is merely a short instance. We shall show the full details in the formal author response letter later.

(2) This is a very meaningful point. We agree with you for this case that participants may learn how to complete this experiment task swiftly rather benefit from neuromodulation. This speculation makes sense, but is compromised by experimental control and empirical observations. Firstly, we do not say “You must complete this task” or “The task completion is associated with bonus/rewards you may get” for participants, which indicates no motivations to do so. Then, the measures to task completion rate are not yet fully based on self-reporting, and we mandate them to provide photos for verification. Thus, this controls the marked risks of spurious reporting. Lastly, all the participants, including ones in either active or sham group, received all the same treatments, excepting “real simulation” and “sham simulation” protocol. Results demonstrated the significant amelioration in the active group rather sham one, indicating no significant “placebo” or “task learning” side effect.

(3) Thank you. As you kindly suggested, we would like to add huge details for those measures in the revised manuscript. While this is a great idea, we did not collect procrastination scores from scales after neuromodulation, and would like to warrant this point into the Limitation section.

(4) Yep, this is a conceptual hypothesis --- actively stimulating left DLPFC could improve self-control functions. We cannot rule out possibilities of amplifying working memory, attention or other cognitive components from this neuromodulation protocol. We fully agree with you for this helpful recommendation --- we would like tone down those claims regarding the roles of DLPFC on self-control, and explicitly warrant that this mechanism may be specialized to the procrastination.

Reviewer #3:

(1) Thank you for taking valuable time to review our manuscript. Yep, limited sample size should warrant cautions to draw a solid conclusion. We would like to claim it into the limitation section. Also, we have streamlined and tightened statistic section by removing complicated and redundancy statistical models.

(2) As mentioned above, we are sorry for this fundamental technical issue to preregistration. This occurs from a seriously technical hurdle. The OSF has banned my OSF account, as it claimed to detect “suspicious user’s activities” in my account. This causes no accesses to all materials that already deposited in this OSF account, including this preregistration. We have contacted OSF team, but received no valid technical solution. We reckon that this may be mistaken by my affiliation changes to Third Military Medical University of People’s Liberation Army (PLA). To tackle with this technical issue, we shall upload preregistration in a new repository soon.

(3) Yep, thank you for this very helpful suggestion. As you kindly indicated, we would like to clarify measures, analyses, methods, and protocols, as well as tighten the whole manuscript.

References

Baker, D. H., Vilidaite, G., Lygo, F. A., Smith, A. K., Flack, T. R., Gouws, A. D., & Andrews, T. J. (2021). Power contours: Optimising sample size and precision in experimental psychology and human neuroscience. Psychological methods, 26(3), 295–314. https://doi.org/10.1037/met0000337

Xu, T., Zhang, S., Zhou, F., & Feng, T. (2023). Stimulation of left dorsolateral prefrontal cortex enhances willingness for task completion by amplifying task outcome value. Journal of experimental psychology. General, 152(4), 1122-1133. https://doi.org/10.1037/xge0001312

Again, we wholeheartedly appreciate all of those very helpful and insightful comments, with each one to contribute substantially for the quality of this manuscript. Notably, those response we presented above are merely provisional and initial. We shall revise our manuscript following those suggestions, one-by-one, along with a full-length response letter.

eLife Assessment

In this Review Article, the authors survey the literature describing how correlated dynamical states relate to various cognitive states, including anesthesia and sleep. While the topic is significant and the coverage broad, the manuscript does not yet provide a synthesis that connects the many available findings or highlights converging themes across studies. Additionally, many of the disparate concepts are not introduced at the level of first principles. As a result, the Review remains difficult to access for readers outside the immediate subfield. Developing a clearer integrative perspective would help make the article informative to a wider audience.

Reviewer #1 (Public review):

Summary:

In the paper, the authors review literature on synchronous activity, its relationship to brain state, and the multi-scale mechanisms underlying it.

Strengths:

The overall strength of the paper is the wide range of information reviewed, and the diversity of perspectives/approaches it brings together.

Weaknesses:

However, this strength is also the source of its major weaknesses - namely, that the overall structure lacks clarity, and there are inconsistencies throughout. Overall, in the opinion of this reviewer, the manuscript reads as disorganized and incomplete. Major and minor points are delineated below.

Major points:

(1) Most of the text in many figures was too small to read.

(2) Terminology is inconsistent throughout the manuscript. What is the difference between slow oscillations and delta waves? Sometimes the term slow waves is used instead. For sleep state, sometimes the term SWS is used, sometimes non-REM. Similarly, "spindle activity" is not defined, but simply stated as if the reader knows. This brings up two issues: (a) the manuscript should be clearer and more consistent about its terminology, and (b) it's unclear who is the intended readership of the review - is it a pedagogical review for people outside the field of sleep and slow oscillations, or is it meant to be a consensus statement for readers who are already in the field in which a pressing concern has been addressed? It seems part way between these two, and as a result, is ineffective at either goal.

(3) I suggest the authors look again at the overall structure and flow of the review... many sections feel redundant, and it's unclear how they fit together into a single review.

(4) There are many speculative statements in the review that are not justified or explained sufficiently for the reader. For example: "While highly regular slow waves in vivo suggest a single mechanism of generation, namely local cortical circuits, irregular cycles are compatible with a larger role of subcortical nuclei, ..."; "The involvement of different cortical areas and subcortical nuclei can form the basis of these different roles in memory.". For these statements, I assume the relationship between slow wave statistics, subcortical nuclei, and memory either has been written about before, and then should be cited and summarized, or is a novel claim of the authors, which then should be explained and defended rather than stated. There are other similar examples, and I suggest the authors go through the manuscript and make sure that it's clear what is a novel claim of the authors vs a cited claim, and make sure that both are sufficiently justified for the reader.

(5) An especially notable example can be found in the section on the role of the thalamus, where the authors state that they "hold that slow oscillations are fundamentally cortical". However, this section is far too short, and very little evidence is provided to back up this claim. Please review the ways in which the thalamus modulates, and, e.g., ways in which up-down is similar/different without the thalamus.

Reviewer #2 (Public review):

Summary:

In this review article, the authors discuss the correlated dynamical states associated with distinct cognitive states, including those associated with anesthesia and sleep. They present evidence that these states are primarily cortically generated, and demonstrate the properties of these dynamical states at different levels, from the microscale dynamics in individual neurons to the macroscale dynamics across the brain.

Strengths:

Multiple groups have been adding to this field over the past decades, and therefore, a review of this literature is very helpful. This review collates a large amount of the literature within this field into a single document, which should make it a valuable resource within this area of neuroscience.

Weaknesses:

Unfortunately, this review does not seem to be a balanced viewpoint of the field in question. Although there are a lot of authors in the review, it feels as if they are from a common school of thought. The authors provide only a single perspective on these dynamical states, focusing on the perspective of wave-like electrical dynamics across the cortex. Their perspective is embedded in methods such as EEG and LFP recordings. This makes the work hard to interpret outside of the field in which the authors reside. Indeed, the review seems intended for a more specialized audience.

In addition, the article reads more like a catalog of prior studies as opposed to a true synthesis across the large volume of data in this field that highlights links across multiple sources. Hence, it does not seem to provide a novel way of understanding the dynamics involved in cognitive state transitions.

We have included more details on these general comments below:

Major Comments:

(1) The authors have written this review as if it were intended for an audience who is already familiar with these topics. They do not define many of the terms that they introduce within the review, including concepts like complexity, metastability, and oscillations that are fundamental to the concepts that the authors are introducing. Though these may seem like first principles concepts to the authors, they often introduce assumptions that may be unfamiliar to the general reader. For example, are slow wave oscillations periodic? A naïve reader may assume that oscillations - characterized by their frequency - should be somewhat periodic, but that is often not the case. For a journal with a general biological science readership, it would be particularly helpful for each of these terms to be formally defined and characterized.

(2) It would be helpful for the authors to reframe their work in different perspectives and to incorporate all the literature on the dynamics of cortical brain states, and not simply the work that is most familiar to them. As one example, the authors do not discuss cell-type-specific changes in brain state during anesthesia and in altered states of consciousness (including dissociative states and hallucinatory states). There is recent work in this vein (Suzuki and Larkum, 2020; Vesuna et al, 2020; Bharioke, Munz et al, 2023), and yet the authors do not discuss these papers.

(3) Given the authors' clear, extensive knowledge of their field, it would also be extremely helpful for the authors to reframe fundamental concepts in terms of neuronal population activity, trajectory analyses, etc. This would enable a more general audience to better understand their work.

(4) The authors have one section focused on thalamic contributions to cortical wave-like activity. This is a cursory treatment of a subject that is quite controversial in the field. It would be helpful if the authors could provide a more balanced consideration of all the evidence regarding potential thalamocortical interactions and their role in wave-like activity.

(5) The authors present many computational models and describe the results of simulations with these different models. However, this doesn't provide the reader with intuition about what each model adds or removes from the true biological picture. It would be helpful for the authors to provide some intuition about the assumptions and constraints that underlie each model.

(6) The authors state that "The main mechanism [of slow oscillatory dynamics] consists of a combination of two ingredients: the recurrent connectivity, which maintains the excitability in the network, and adaptation, an activity-dependent fatigue variable that provides inhibitory feedback". They make this statement as a fact, yet they don't provide much justification for it. Additionally, it's not clear that any other possible combination of ingredients would be able to produce slow oscillatory dynamics.

(7) The authors often define one concept in terms of other equally complex concepts. For example: "EIA (excitatory-inhibitory with adaptation) cortical circuits then display the typical slow-fast dynamics of relaxation oscillators". The reader would need an explanation of slow-fast dynamics and relaxation oscillators to understand this line, neither of which is provided in the text.

(8) When discussing sleep, the authors do not discuss REM sleep, focusing on slow-wave non-REM sleep. It would be helpful if the authors could at least frame the full sleep cycle and discuss why they are focusing on one part of it.

(9) The authors introduce the concept of sleep spindles without any explanation.

eLife Assessment

This important work combines theoretical analysis with precise experimental perturbation to demonstrate a previously unappreciated quantitative characteristic of the Wnt signaling pathway, which is anti-resonance, or a suppression of pathway output at intermediate activation frequencies. This effect is demonstrated experimentally with compelling evidence from optogenetic stimulation in multiple cell types, alongside modeling results that corroborate the phenomenon. While the demonstration of this phenomenon has yet to be extended to fully physiological situations, its clear existence within optogenetically stimulated systems shows that it is likely a significant factor that contributes to the behavior of this central signaling pathway.

Reviewer #1 (Public review):

Summary:

This report demonstrates that the gene expression output of the Wnt pathway, when controlled precisely by a synthetic light-based input, depends substantially on the frequency of stimulation. The particular frequency-dependent trend that is observed - anti-resonance, a suppression of target gene expression at intermediate frequencies given a constant duty cycle - is a novel aspect that has not been clearly shown before for this or other signaling pathways. The paper provides both clear experimental evidence of the phenomenon with engineered cellular systems and a model-based analysis of how the pairing of rate constants in pathway activation/deactivation could result in such a trend.

Strengths:

This report couples in vitro experimental data with an abstracted mathematical model. Both of these approaches appear to be technically sound and to provide consistent and strong support for the main conclusion. The experimental data are particularly clear, and the demonstration that Brachyury expression is subject to anti-resonance in ESCs is particularly compelling. The modeling approach is reasonably scaled for the system at the level of detail that is needed in this case, and the hidden variable analysis provides some insight into how the anti-resonance works.

In this revised manuscript, the authors have addressed issues in presentation and in discussing the broader relevance of their study to other pathways. Other limitations of the paper, including the fact that the anti-resonance phenomenon has not yet been demonstrated using physiological Wnt ligands and that the model has not been validated using experimental manipulations to establish that the mechanisms of the cell system and the model are the same, were deemed out of the scope of this initial demonstration by both the reviewers and authors. These questions will provide an interesting basis for further studies.

Reviewer #2 (Public review):

Summary:

By combining optogenetics with theoretical modelling the authors identify an anti-resonance behavior in the WnT signaling pathway. This behavior is manifested as a minimal response at a certain stimulation frequency. Using an abstracted hidden variable model, the authors explain their findings by a competition of timescales. Furthermore, they experimentally show that this anti-resonance influences the cell fate decision involved in human gastrulation.

Strengths:

- This interdisciplinary study combines precise optogenetic manipulation with advanced modelling.<br /> - The results are directly tested in two different systems: HEK293T cells and H9 human embryonic stem cells.<br /> - The model is implemented based on previous literature and has two levels of detail: i) a detailed biochemical model and ii) an abstract model with a hidden parameter

Weaknesses:

- While the experiments provide both single-cell data and population data, the model only considers population data.<br /> - Although the model captures the experimental data for TopFlash very well, the beta-Cat curves (Fig 2B) are only described qualitatively. This discrepancy is not discussed.

Overall Assessment:

The authors convincingly identified an anti-resonance behavior in a signaling pathway that is involved in cell fate decisions. The focus on a dynamic signal and the identification of such a behavior is important. I believe that the model approach of abstracting a complicated pathway with a hidden variable is an important tool to obtain an intuitive understanding of complicated dependencies in biology. Such a combination of precise ontogenetical manipulation with effective models will provide a new perspective on causal dependencies in signaling pathways and should not be limited only to the system that the authors study.

Comments on revisions:

I don't have any more comments for the authors and would like to congratulate them for the nice piece of work!

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

This report demonstrates that the gene expression output of the Wnt pathway, when controlled precisely by a synthetic light-based input, depends substantially on the frequency of stimulation. The particular frequency-dependent trend that is observed - anti-resonance, a suppression of target gene expression at intermediate frequencies given a constant duty cycle - is a novel aspect that has not been clearly shown before for this or other signaling pathways. The paper provides both clear experimental evidence of the phenomenon with engineered cellular systems and a model-based analysis of how the pairing of rate constants in pathway activation/deactivation could result in such a trend.

Strengths:

This report couples in vitro experimental data with an abstracted mathematical model. Both of these approaches appear to be technically sound and to provide consistent and strong support for the main conclusion. The experimental data are particularly clear, and the demonstration that Brachyury expression is subject to anti-resonance in ESCs is particularly compelling. The modeling approach is reasonably scaled for the system at the level of detail that is needed in this case, and the hidden variable analysis provides some insight into how the anti-resonance works.

Weaknesses:

(1) The anti-resonance phenomenon has not been demonstrated using physiological Wnt ligands; however, I view this as only a minor weakness for an initial report of the phenomenon. The potential significance of the phenomenon for Wnt outweighs the amount of effort it would take to carry the demonstration further - testing different frequencies/duty cycles at the level of ligand stimulus using microfluidics could get quite involved, and would likely take quite some time. Adding some more discussion about how the time scales of ligand-receptor binding could play into the reduced model would further ameliorate this issue.

We thank the reviewer for this comment and the interesting suggestion to test the anti-resonance phenomenon with microfluidics. We agree that combining physiological Wnt ligands with microfluidic stimulation would go beyond the scope of this current study, though it is an interesting extension. One advantage of the optogenetic setup, as mentioned in the discussion, is that the Wnt stimulus can be turned off sharply. This allows us to test the output from perfectly square wave input profiles; in microfluidics, washing the sticky ligand off the cells might “smear” the effective input profile cells respond to.

We show in Supplement Fig. 6, that our reduced model matches the experimental data and that we would expect the antiresonance phenomenon as long as (see Fig. 4). Practically, a smeared input profile implies an effective reduction of 𝑘_off, which means that the phenomenon would be visible with microfluidics (provided the minimum is deep enough, see Fig. 4). However, this should still be considered with caution, as the antiresonance would then appear because the cells essentially receive a smeared out or continuous pulse in the high frequency limit, rather than cells responding to a square wave in a specific way.

(2) While the model is fully consistent with the data, it has not been validated using experimental manipulations to establish that the mechanisms of the cell system and the model are the same. There may be some ways to make such modifications, for example, using a proteasome inhibitor. An alternative would be to more explicitly mention the need to validate the model's mechanism with experiments.

We thank the reviewer for this valuable and constructive comment. We agree that future experimental perturbations that directly modulate pathway activation and reset kinetics—such as proteasome inhibition, targeted degradation of pathway components, or engineered changes in receptor turnover—would provide an important validation of the model’s mechanistic interpretation. In the present study, our primary goal was to establish the existence and quantitative features of anti-resonance in the Wnt pathway and to identify the minimal set of timescale relationships that can explain it. We view the proposed experimental validations as exciting next steps that extend beyond the scope of the current work, and we are grateful to the reviewer for emphasizing their importance. We now mention this explicitly in the discussion of our manuscript.

(3) I think the manuscript misses an opportunity to discuss the potential of the phenomenon in other pathways. The hedgehog pathway, for example, involves GSK3-mediated partial proteolysis of a transcription factor, which could conceivably be subject to similar behaviors, and there are certainly other examples as well.

We thank the reviewer for pointing out an opportunity to emphasize the possibility of this phenomenon in other pathways. The minimal model indicates that anti-resonance emerges whenever a rapid activating process is paired with a slower deactivating/reset process. Beyond Hedgehog/Gli processing, candidate circuits include: NF-κB (rapid IκBα phosphorylation/degradation vs slower IκBα resynthesis), ERK (fast phosphorylation bursts vs slower transcriptional negative feedback such as DUSPs), Notch (fast γ-secretase NICD release vs slower NICD turnover and feedback), BMP/TGF-β–SMAD (fast R-SMAD phosphorylation vs slower receptor trafficking/SMAD7 feedback), and Hippo/YAP (rapid cytoplasmic sequestration vs slower transcriptional feedback). Each contains the same timescale separation that should create a frequency ‘stop-band,’ predicting suppressed gene expression or fate transitions at intermediate stimulation frequencies. We have updated the manuscript’s discussion to mention the Hedgehog connection with the following added sentence in the discussion: Analogous band-stop filtering should arise in other developmental circuits that couple a fast ‘ON’ step to slower deactivation or negative feedback. In Hedgehog, for example, PKA/CK1/GSK3-mediated partial proteolysis of Gli with slower recovery of full-length Gli creates the same fast-activation/slow-reset motif our hidden-variable model predicts will yield anti-resonance, and Wnt–Hedgehog crosstalk through the shared kinase GSK3 suggests such frequency selectivity could occur in other developmental signaling pathways.

We also added an additional sentence regarding different activation and deactivation timescales in other pathways.

(4) Some aspects of the modeling and hidden variable analysis are not optimally presented in the main text, although when considered together with the Supplemental Data, there are no significant deficiencies.

We have addressed the model choices and analysis now more clearly in the main manuscript and also referred to the Supplemental Data more directly.

Reviewer #2 (Public review):

Summary:

By combining optogenetics with theoretical modelling, the authors identify an anti-resonance behavior in the WnT signaling pathway. This behavior is manifested as a minimal response at a certain stimulation frequency. Using an abstracted hidden variable model, the authors explain their findings by a competition of timescales. Furthermore, they experimentally show that this anti-resonance influences the cell fate decision involved in human gastrulation.

Strengths:

(1) This interdisciplinary study combines precise optogenetic manipulation with advanced modelling.

(2) The results are directly tested in two different systems: HEK293T cells and H9 human embryonic stem cells.

(3) The model is implemented based on previous literature and has two levels of detail: i) a detailed biochemical model and ii) an abstract model with a hidden parameter.

Weaknesses:

(1) While the experiments provide both single-cell data and population data, the model only considers population data.

We thank the reviewer for correctly pointing out that the single-cell measurements would in principle allow us to incorporate the cell-to-cell heterogeneity into the model. In this study, we sought to identify a minimal quantitative model of the Wnt pathway that could explain anti-resonance through competing time scales. We believe that, for our purposes, focusing on population data allowed us to keep the complexity of the model to a minimum to increase its explanatory value. We agree with the reviewer that considering single-cell trajectories is an interesting direction for further work.

(2) Although the model captures the experimental data for TopFlash very well, the beta-Cat curves (Figure 2B) are only described qualitatively. This discrepancy is not discussed.

Indeed, our model fits to mean β-catenin expressions are more qualitative than for TopFlash. The fit for β-catenin was tricky, as expression of β-catenin is typically low and closer to the detectable limits than TopFlash. These experimental constraints mean that the variation between individual signal trajectories is higher for β-catenin compared to the light-off condition than for TopFlash. Therefore, we strove to obtain a qualitative rather than a quantitative fit to the mean expression profile in β-catenin.  The current model fit is well within the standard deviation of variation. Given the observed heterogeneity and the fact that we take the parameters from literature (which ensures that the order of magnitude of parameters is in a sensible range), we believe that the model fits are reasonable. We now mention this explicitly in the text.

Overall Assessment:

The authors convincingly identified an anti-resonance behavior in a signaling pathway that is involved in cell fate decisions. The focus on a dynamic signal and the identification of such a behavior is important. I believe that the model approach of abstracting a complicated pathway with a hidden variable is an important tool to obtain an intuitive understanding of complicated dependencies in biology. Such a combination of precise ontogenetic manipulation with effective models will provide a new perspective on causal dependencies in signaling pathways and should not be limited only to the system that the authors study.

We thank both reviewers for the positive assessment of our manuscript.

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

There are several points that deserve more discussion, as noted above in the review.

(1) It would be worthwhile to consider whether a relatively simple experiment with a proteasome inhibitor or similar pharmacological manipulation could provide useful validation data for the model.

We address this point above in the weaknesses section from reviewer 1.

(2) The figure legend for S5C should clarify whether the values plotted are at a particular fixed time point, or (more likely) at a certain time following the second pulse, which would be variable.

We have modified the figure caption to clarify that the values plotted are at a fixed time point in the simulation (t\=48 hrs). We chose this timepoint sufficiently long after the second pulse to ensure that there are no residual dynamical effects. We thank the reviewer for noting this.

(3) As noted in the Sci Score document, various aspects of the resource reporter should be improved, such as including RRIDs, etc.

We are sending out our plasmids to AddGene; versions for Python and Matlab are listed in our methods section.

Reviewer #2 (Recommendations for the authors):

I mostly have suggestions to improve the clarity of the presentation.

(1) Not all symbols in the equations given in the main text are explained. This is rather annoying, because either you present them and explain what they are or you don't show them and refer to the supplements. For example, d_0 or c_o or \bar{b} or n or K are not explained.

We have now more clearly presented the parameters in the main text and added signposts to the Methods section.

(2) Overall, it is often not clear what data in the figures are redundant, although the authors referred to them in the text. For example, in Figure 2c, a curve for 24 hours is shown and referred back to Figure 1D. However, in Figure 1D there is no curve for 24 hours. Is the data from Supplementary Figure 1 H and K also in the main text?

We thank the referee for pointing out these redundancies. We have now included the 24hr line in Figure 1D and are now only showing the unsmoothed data, also in the main text of the manuscript. To clarify supplemental figures, we have now removed S1H and S1K since all they showed was the unsmoothed version of the data. The remaining plots in Supplementary Figure 1 are normalized differently from what we show in Figure 1 to demonstrate our choice of normalization is not the reason for the observed optogenetic response.

eLife Assessment

Following retinal injury, zebrafish Müller glia reenter the cell cycle and generate replacement cells; this potentially valuable study proposes that injury induces a cxcl18b+ transitional state in Müller cells, which then express nitric oxide, inhibiting Notch signaling and allowing Müller glial cells to reenter the cell cycle. However, the evidence supporting the claims is incomplete, and the authors have made interpretations and conclusions that are not supported by the data. Questions of the temporal expression and function of cxcl18b, as well as the source of potential inflammatory cues before cxcl18b expression, remain unanswered and technical limitations and data inconsistencies raise concerns. Using larval animals complicates the analysis since the retina is still forming, and distinguishing between injury-induced regeneration and ongoing development is complex. With more rigorous testing of the signaling pathways proposed and a clear demonstration of their interdependence, the link between nitric oxide signaling and Notch activity, particularly, would interest those investigating retinal regeneration.

Reviewer #1 (Public review):

Summary:

This study presents a valuable contribution of NO signaling in zebrafish retinal regeneration in larval animals. The data on NO signaling are solid. There are multiple limitations to the study, but these are largely acknowledged by the authors in the revised text.

Strengths:

New data on NO signaling is valuable to the field but may be limited to larval "regeneration".

Weaknesses:

A weakness of the approach is testing cone ablation and regeneration in early larval animals. A near identical study was already done by Hoang et al 2020 in the adult zebrafish, a more relevant biological timepoint.

Reviewer #2 (Public review):

Summary:

In this manuscript Ye at al. examine the sequence of events that occur in the damaged zebrafish Muller glia (MG) in states between quiescence and the onset of proliferation. Using an inducible metronidazole (MTZ) and nitroreductase system to ablate red/green cones in larval zebrafish, they identify a novel transitional MG state that is characterized by the expression of cxcl18b. Using trajectory analysis from single-cell RNA-seq datasets, they find that cxcl18b is expressed before MG expression PCNA and become proliferative. They find that cxcl18b expression peaks in MG at approximately 24 hours post injury (hpi) and rapidly declines as MG proliferate following injury. In a most interesting finding, the authors find a link between nos2b-dependent nitric oxide signaling and cxcl18b-mediated proliferation. Mutagenesis of nos2b decreases MG proliferation. The mechanism linking NO signaling to proliferation was suggested to function via notch signaling as pharmacological inhibition of nitric oxide signaling resulted in elevated Notch activity, thus preventing MG proliferation. The authors suggest a model whereby cxcl18b induces autocrine NO signaling in MG to reduce activity of Notch3, thereby promoting MG proliferation.

Strengths:

The authors utilize a number of sophisticated transgenic approaches and generate novel lines that will have value to the field. The identification of a novel cxcl18b transition state is exciting and the putative link between NO signaling and Notch activity would provide new insight into the drivers of Muller glia proliferation.

Weaknesses:

While the overall model is appealing and may serve as a foundation for future studies, some information gaps remain and certain conclusions rely on correlational data. The cellular expression of nos2b remains unclear as the single-cell RNA-seq data cannot provide expression data that matches RT-PCR results. The temporal sequence of events are based on transgene expression in the Tg(cxcl18b:GFP) lines, where persistence of the GFP fluorescence may not reflect endogenous cxcl18b. The identity of putative cxcl18b receptors on MG to support an autocrine signaling pathway remains unclear. Nevertheless, this is an interesting study that should open new avenues of exploration.

Author response:

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

Reviewer #1:

(1) The authors state that more is known about glial reactivation than cell-cycle re-entry. They are confusing many points here. More gene networks that require cell-cycle re-entry are known. Some of the genes listed for "reactivation" are, in fact, required for cell cycle re-entry/proliferation. And the authors confuse gliosis vs glial reactivation.

We thank the reviewer for this important and constructive comment. We fully agree that clearly distinguishing between the concepts of glial reactivation, glial proliferation, gliosis, and neurogenesis is essential to avoid conceptual confusion in our study.

Injury-induced retinal regeneration in zebrafish:

Glial reactivation refers to the initial response of quiescent Müller glia (MG) to injury, characterized by morphological changes and upregulation of reactive markers (e.g., gfap, ascl1a, lin28a) and activation of signaling pathways such as Notch, Jak/Stat, and Wnt (Lahne et al., 2020; Pollak et al., 2013; Sifuentes et al., 2016; Yao et al., 2016).

Glial proliferation refers to the clonal expansion of these MG-derived progenitor cells, which undergo rapid cell-cycle re-entry and amplify to generate sufficient progenitors for regeneration (Iribarne and Hyde, 2022; Lee et al., 2024; Wan and Goldman, 2016)

Gliosis vs neurogenesis represents a divergent fate decision following proliferation. In zebrafish, MG-derived progenitor cells differentiate into retinal neurons that can replace those damaged or lost due to retinal injury. In contrast, mammalian MG tend to undergo an initial gliotic surge and rapidly revert to a quiescent state, exhibiting gliosis and glial scarring (Thomas et al., 2016; Yin et al., 2024). Thus, we totally agreed that gliosis cannot be confused with glial reactivation because glial reactivation is the very first step of glial injury responses, whereas gliogensis is the very last glial response to the injury.

We agree with the reviewer that many genes typically described as “reactivation markers” (e.g., ascl1a, lin28a, sox2, mycb, mych) are also essential regulators of cell-cycle re-entry (Gorsuch et al., 2017; Hamon et al., 2019; Lee et al., 2024; Lourenço et al., 2021; Pollak et al., 2013; Thomas et al., 2016). Because the glial reactivation is a leading event for glial proliferation, the regulators of glial reactivation are expected to be responsible for glial proliferation as well.

In our study, we focused on the states preceding glial proliferation to understand the mechanism underlying injury-induced glial cell-cycle re-entry. We defined these transitional states and the subsequent proliferative MG states based on single-cell RNA-seq trajectory analysis. (revised lines: 41-58)

(2) A major weakness of the approach is testing cone ablation and regeneration in early larval animals. For example, cones are ablated starting the day that they are born. MG that are responding are also very young, less than 48 hrs old. It is also unclear whether the immune response of microglia is a mature response. All of these assays would be of higher significance if they were performed in the context of a mature, fully differentiated, adult retina. All analysis in the paper is negatively affected by this biological variable.

We thank the reviewer for raising this important point regarding the developmental stage of the retina in our model system. We have carefully considered this concern and now provide additional clarification and justification, as follows:

(1) The glial responses in larval and adult retina:

Previous studies have demonstrated that injury-induced glial responses are largely conserved in larval and adult zebrafish retina, including reactive gliosis marked by gfap upregulation and proliferation(Meyers et al., 2012; Sarich et al., 2025). In our study, G/R cones were ablated beginning at 5 dpf using metronidazole (MTZ), and we observed robust induction of PCNA⁺ MG in the inner nuclear layer, consistent with injury-induced proliferation (Figure 1E). These findings align with previous studies showing that key features of MG regenerative responses are conserved across larval and adult stages.

(2) The microglial responses in larval and adult retina:

Retinal microglia functionally mature at 5 dpf in the zebrafish retina (Mazzolini et al., 2020; Svahn et al., 2013), and prior studies have demonstrated that microglia in larval and adult zebrafish exhibit similar responses to injury, including migration, morphological activation, and phagocytosis(Nagashima and Hitchcock, 2021; White et al., 2017). In our experiments using Tg(mpeg1: GFP) larvae, we observed clear microglial recruitment to the outer nuclear layer (ONL) following cone ablation (Figure 1E and Figure 1-figure supplement 1A), supporting the functional competence of larval microglia in injury-induced immune responses

(3) The contribution using larval animals to study the regeneration program:

We agree that regeneration studies in the adult retina can provide important biological insights, particularly in a fully differentiated tissue environment. Accordingly, we have acknowledged this limitation in our revised manuscript “limitations of this study” section (revised lines 534-540: “1. Our study focuses on larval zebrafish, in which the core features of MG and immune responses are conserved compared to the adult. However, we acknowledge that the adult retina—with its fully matured differentiated retina and immune response—provides irreplaceable biological insight. Nevertheless, larval models offer a powerful platform to uncover conserved regenerative mechanisms and serve as a valuable complement for identifying age-dependent differences in MG-mediated regeneration.”) and have stated our intention to extend future analyses to adult zebrafish, especially to explore age-dependent differences in redox signaling and MG proliferation. At the same time, we believe that the larval model offers unique advantages for uncovering fundamental, conserved mechanisms of regeneration and enables characterization of age-dependent regulatory differences. Thus, our study in larval animals serves as a complementary and informative platform for understanding both the conserved and developmental stage-specific features of injury-induced regeneration.

(4) Related to the above point, the clonal analysis of cxcl18b+ MG is complicated by the fact that new MG are still being born in the CMZ (as are new cones for that matter).

We thank the reviewer for raising this important point regarding potential contributions from CMZ-derived progenitors to the lineage-traced cxcl18b⁺ MG clones. To address this concern, we have implemented evidence to rule out a CMZ origin for the clones analyzed:

Spatial restriction of clones: All clones included in our analysis were located exclusively within the central and dorsal retina, as shown in Figure 2H. From the spatial distribution of reactive MG populations across the retina, we observed a patterned organization in which the vast majority of proliferating MG arose from local mature MG–derived progenitors, rather than from peripheral CMZ-derived progenitors. However, we acknowledge that we cannot entirely exclude the possibility that CMZ-derived progenitors contribute to injury-induced MG proliferation, particularly in the peripheral retina.

We have clarified this point in the revised Methods section (revised lines 756–762: “Clone analysis of cxcl18b⁺ lineage-traced MG was restricted to cells located in the central and dorsal region of the zebrafish retina after G/R cone ablation in Figure 2, Figure 6, and their figure supplement. This spatial restriction strongly suggests that the proliferative MG originate from local mature MG, although we cannot completely rule out the possibility that CMZ-derived progenitors contribute to the generation of proliferative MG in the peripheral retina.”) and updated the corresponding figure legends.

(4) A near identical study was already done by Hoang et al., 2020, in adult zebrafish, a more relevant biological timepoint. Did the authors check this published RNA-seq database for their gene(s) of interest?

We thank the reviewer for pointing out the relevance of the study by Hoang et al., 2020, which characterized the transcriptional dynamics of MG reactivation in the adult zebrafish retina. We agree that comparisons with their single-cell RNA-seq dataset are important to confirm the conservation of our findings in larval vs adult zebrafish.

To this end, we examined the adult zebrafish MG dataset reported by Hoang et al., and confirmed that cxcl18b is also present and enriched in their analysis, particularly in activated MG populations under various injury paradigms:

(1) cxcl18b is listed as a differentially expressed gene (DEG) in Supplementary Table ST2, enriched in GFP⁺ MG following injury. It is also significantly upregulated in both NMDA-induced and light damage conditions, as shown in Supplementary Table ST3.

(2) In Supplementary Table ST5, cxcl18b is identified as a classifier of activated MG, with classification power scores of 0.552 (NMDA), 0.632 (light damage), and 0.574 (TNFα + γ-secretase inhibitor treatment), indicating its consistent expression across multiple injury models.

(3). In their pseudotime analysis (Figure 4C and Supplementary Table ST8), cxcl18b is specifically expressed in Module 5, which is expressed earlier along the trajectory than ascl1a. This temporal pattern of cxcl18b preceding ascl1a expression is consistent with our trajectory analysis in larval MG (Figure 1H), further supporting its role as an early marker of the transitional state before proliferation.

These findings underscore the robustness and biological relevance of cxcl18b as a conserved marker of injury-responsive MG in both larval and adult zebrafish. Our data expand upon the prior work by specifically characterizing a cxcl18b-defined transitional MG state preceding cell-cycle re-entry, thereby offering additional insights into the temporal staging of MG activation during regeneration.

(5) KD of cxcl18b did not affect MG proliferation or any other defined outcome. And yet the authors continually state such phrases as "microglia-mediated inflammation is critical for activating the cxcl18b-defined transitional states that drive MG proliferation." This is false. Cxcl18b does not drive MG proliferation at all.

We thank the reviewer for raising this concern. We agree with the reviewer and have revised this statement as "These findings suggest that microglia-mediated inflammation may contribute to the activation of cxcl18b-defined transitional states that precede MG proliferation, although a causal relationship remains to be established." (revised lines 251-253).

(6) A technical concern is that intravitreal injections are not routinely performed in larval fish.

We appreciate the reviewer’s technical concern regarding the use of intravitreal injections in larval zebrafish. In our study, we performed intraocular injection according to previously established methods (Alvarez et al., 2009; Giannaccini et al., 2018; Rosa et al., 2023). This approach involves carefully delivering a small volume of viral suspension into the intraocular space by a glass micropipette. To address this concern, we will revise the Materials and Methods section to clearly describe the injection procedure and will cite the relevant references accordingly.

Reviewer #2:

(1) The authors note a peak of PCNA+ Muller glia at 72 hours post injury. This is somewhat surprising as the MG would be expected to generate progenitor cells that would continue proliferating and stain with PCNA. Indeed, only a handful of PCNA+ cells are seen in the INL/ONL layer in Figure 1E2 with few clusters of progenitors present. It would be helpful to stain with a Muller glia marker to confirm these PCNA+ cells are Muller glia. It's also curious that almost all the PCNA+ cells are in the dorsal retina, even though G/R cone loss extends across both dorsal and ventral retina. Is this typical for cone ablation models in larval zebrafish?

We thank the reviewer for their insightful comment regarding the spatial distribution and identity of PCNA⁺ cells following injury.

In our study, we observed that the injury-induced proliferating cells (PCNA⁺) were predominantly located in the central and dorsal regions of the retina at 72 hours post-injury (hpi) (Figure 1E). To verify the identity of these proliferating cells, we performed additional immunostaining using BLBP, and confirmed that the majority of PCNA⁺ cells also express BLBP (Figure 1–figure supplement 1B in our revised Data), these results supporting their MG origin.

The regional bias of MG proliferation towards the central and dorsal retina is consistent with previous findings. Notably, (Krylov et al., 2023) demonstrated that MG exhibit region-specific heterogeneity in their regenerative responses to photoreceptor ablation. Their study identified proliferative MG subpopulations predominantly in the central (fgf24-expressing) and dorsal (efnb2a-expressing) domains, whereas ventral MG showed limited proliferative capacity (Krylov et al., 2023). These observations provide a plausible explanation for the spatially restricted PCNA⁺ MG population observed in our model following cone ablation.

(2) In Line 148: What is meant by "most original MG states" in this context? Original meaning novel? Or original meaning the earliest state MG adopted following injury? The language here is confusing.

We thank the reviewer for pointing out the ambiguous phrasing in our original manuscript. The term “most original MG states” was imprecise and misleading, as it could be interpreted as referring to the quiescent state of MG. In our context, we intended to describe the earliest transitional states in MG respond to injury, as they begin to exit quiescence and enter reactive characteristics. These early transitional MG populations co-express quiescent markers such as cx43 and early reactive markers gfap, as shown in Figure 1H.

To avoid confusion and improve conceptual clarity, we have revised the manuscript by replacing “most original MG states” with “early transitional MG state” (revised line 154) and have added a clearer explanation in the corresponding Results section to define this population more accurately.

(3) Perhaps provide a better image in Figure 2A of the cxcl18b at 48 hpi and 72 hpi. The current images appear virtually identical, with very little cxcl18b expression observed, especially compared to the 24 hpi. This is in contrast to the Tg(cxcl18b:GFP) transgenic line shown in Figure 2D, which indicates either much higher expression in proliferating cells at 48 hpi or the stability of GFP protein. Can the authors provide guidance on the accurate temporal expression of cxcl18b? Does expression peak rapidly at 24 hpi and then rapidly decline or is there persistence of expression to 48-72 hpi?

We appreciate the reviewer’s careful observation regarding the apparent similarity of cxcl18b expression at 48 hpi and 72 hpi in the in situ hybridization (ISH) images (Figure 2A), and the differences compared to the Tg(cxcl18b: GFP) reporter line shown in Figure 2D.

(1) The similarity of ISH images at the 48 hpi and 72 hpi (Figure 2A):

The cxcl18b mRNA signal peaked at 24 hpi, suggesting a rapid transcriptional response after retina injury. By 48 hpi, cxcl18b expression had already declined substantially, and by 72 hpi, the signal was further reduced to near-background levels. This temporal expression pattern explains why the ISH images at 48 hpi and 72 hpi appear nearly identical and much weaker compared to 24 hpi.

(2) The discrepancy between ISH and GFP reporter signal (Figure 2D):

The Tg(cxcl18b: GFP) reporter line shows persistent GFP expression beyond the transcriptional window of cxcl18b mRNA. This may be due to the prolonged stay of GFP protein, which remains detectable even after the endogenous transcription of cxcl18b has diminished. This explanation is also noted in the manuscript (revised lines 198–200). As a result, GFP⁺ MG cells are still visible at 48–72 hpi, and some of them co-label with PCNA.

These findings are consistent with our Pseudotime analysis based on scRNA-seq data (Figure 1H), which shows that cxcl18b expression precedes the induction of proliferative markers such as pcna and ascl1a.

(4) Line 198: The establishment of the Tg(cxcl18b:Cre-vhmc:mcherry::ef1a:loxP-dsRed-loxP-EGFP::lws2:nfsb-mCherry) is considerable but the nomenclature doesn't properly fit. Is the mCherry fused with Cre and driven by the cxcl18b promoter? What is the vhmc component? Finally, while this may provide the ability to clonally track cxcl18b-expressing MG, it does not address the prior question of what is the actual temporal expression of cxcl18b? If anything, this only addresses whether proliferating MG expressed cxcl18b at some point in their history, but does not indicate that cxcl18b expression co-exists in proliferating cells. The most convincing evidence is in Supplemental Figure 2B.

The "vmhc" component refers to the ventricular myosin heavy chain promoter, commonly used to label atrial cardiomyocytes (Jin et al., 2009). We cloned the vmhc upstream region containing its promoter and fusing with mCherry for selection during transgenic fish line construction.

Clone analysis using the Tg(cxcl18b: Cre-vmhc: mCherry::ef1a: loxP-DsRed-loxP-EGFP::lws2: nfsb-mCherry) further indicates that cxcl18b-defined the transitional state is the essential routing for MG proliferation. We have clarified in the revised text that this lineage tracing indicates a “history of injury-induced cxcl18b expression” rather than its ongoing expression during proliferation (revised line 205).

(5) Line 203: The data shown in Figure 2F do not indicate that these MG are cxcl18b+. Rather, the data are consistent with the interpretation that these MG expressed Cre at some prior stage and now express GFP from the ef1a promoter rather than DsRed. Whether these MG continue to express cxcl18b at the time these fish were collected is not addressed by these data. It is not accurate to conclude that these cells are cxcl18b+.

We thank the reviewer for pointing out this important issue. We agreed that the GFP⁺ MG shown in Figure 2F represents cells that have previously expressed cxcl18b and thus belong to the cxcl18b-expressing cell lineage, but this does not indicate that they continue to express cxcl18b at the time of sample collection. Performing clonal analysis using the Cre-loxp system, the GFP signal reflects historical cxcl18b promoter activity rather than ongoing transcription. We have revised the relevant sentence in our manuscript to clarify this point and now refer to these GFP⁺ cells as "cxcl18b lineage-traced MG" rather than "cxcl18b⁺ MG" to avoid any misinterpretation (revised line 207).

(6) Line 213: The statement that proliferative MG mostly originated from cxcl18b+ MG transitional states is a conclusion that does appear fully supported by the data. Whether those MG continue to express cxcl18b remains unanswered by the data in Figure 2 and would likely be inconsistent with the single-cell data in Figure 1.

We thank the reviewer for this valuable comment. We agree that the original statement on Line 213 regarding the lineage relationship between cxcl18b⁺ transitional MG and proliferative MG required clarification.

(1) The cxcl18b expression dynamics:

Our single-cell RNA-seq and ISH analyses consistently show that cxcl18b expression peaks as early as 24 hpi and declines rapidly, with significantly reduced expression by 48 and 72 hpi. These findings suggest that cxcl18b marks an early transitional MG state, rather than being maintained in proliferative MG. Indeed, in our scRNA-seq pseudotime trajectory analysis (Figure 1H), cxcl18b expression is highest in early transitional MG clusters (Clusters 1) and downregulated as cells progress toward proliferative states (Clusters 3/6), supporting a model in which cxcl18b is downregulated before cell-cycle re-entry.

(2) Prolonged stability of GFP protein:

The GFP signal observed in Tg(cxcl18b: GFP) retinas at 72 hpi may be because of the prolonged stability of GFP protein, rather than sustained cxcl18b transcription. The actual expression dynamics of cxcl18b are more directly reflected by our in situ hybridization and single-cell RNA-seq data, both showing a rapid decline after its early peak at 24 hpi. This explanation is also noted in the manuscript (revised lines 196–197).

(7) Line 246: The use of Dexamethasone to block inflammation is a widely used approach. However, dexamethasone is a broad-spectrum anti-inflammatory molecule that works through glucocorticoid signaling that may involve more than microglia. The observation that microglia recruitment and cxcl18a expression are both reduced is correlative but does not prove causation. Thus, the data are not sufficient to conclude that microglia-mediated inflammation is critical for activating cxcl18b expression. Indeed, data in Figure 1 indicate that cxcl18b expression occurs prior to microglia migration to the ONL.

We thank the reviewer for this thoughtful and important comment. We fully acknowledge that dexamethasone is a broad-spectrum anti-inflammatory agent that acts via glucocorticoid receptor signaling and may influence multiple immune and non-immune pathways beyond microglia.

In our study, dexamethasone treatment led to a reduction in both microglial recruitment and the number of cxcl18b⁺ MG at 72 hpi, suggesting a potential association between inflammation and cxcl18b activation. However, we agree that this observation remains correlative and is not sufficient to establish a direct link between microglia activity and cxcl18b induction. Our time-course analysis indicates that cxcl18b expression peaks at 24 hpi, preceding robust microglial accumulation in the ONL, further highlighting the need to clarify the temporal dynamics and cellular sources of inflammatory cues.

To address this question more conclusively, selective ablation of microglia during cone injury would be necessary. However, implementing such an approach would require a complex intersection of three transgenic lines—Tg(mpeg1: nfsB-mCherry) for microglia ablation, Tg(lws2: nfsB-mCherry) for cone ablation, and Tg(cxcl18b: GFP) for reporting—posing substantial genetic and experimental challenges.

We have revised the Results section accordingly to state: “These findings suggest that microglia-mediated inflammation may contribute to the activation of cxcl18b-defined transitional states that precede MG proliferation, although a causal relationship remains to be established.” (revised lines 251–253). We also added a new paragraph in the “Result: Clonal analysis reveals injury-induced MG proliferation via cxcl18b-defined transitional states associated with inflammation” as “While dexamethasone suppressed both microglial recruitment and cxcl18b⁺ MG generation, its broad anti-inflammatory action precludes definitive conclusions about microglial causality. Dissecting this relationship would require concurrent ablation of microglia and cone photoreceptors using a triple-transgenic strategy, which is beyond the scope of the current study. Targeted approaches will be necessary to resolve the specific role of microglia in initiating cxcl18b expression.” (revised lines 251–258) to explicitly acknowledge this limitation and the need for future studies using microglia-specific ablation models to resolve the mechanism.

(8) Could the authors clarify the basis of investigating NO signaling, given the relative expression of the genes by either cxcl18b+ MG or uninjured MG? Based on the expression illustrated in Supplemental Figure 4A, there is almost no expression of nos1 or nos2b in any MG. The authors are encouraged to revisit the earlier single-cell data sets to identify those cells that express components of NO signaling to determine the source(s) of NO that could be impacting the Muller glia.

We thank the reviewer for raising these important points.

Nitric oxide (NO) signaling has been implicated in the regeneration of multiple zebrafish tissues, including the heart (Rochon et al., 2020; Yu et al., 2024), spinal cord (Bradley et al., 2010), and fin (Matrone et al., 2021). Based on these findings, we hypothesized that NO signaling might also contribute to retinal regeneration.

As described in the manuscript, we compiled a redox-related gene list and systematically screened their roles in injury-induced MG proliferation using CRISPR-Cas9-mediated gene disruption. Among the candidates, disruption of nos genes significantly reduced the number of PCNA⁺ MG cells following G/R cone ablation (Figure 4), prompting us to further investigate the role of NO signaling.

(9) Line 319-320: this sentence appears to be missing text as "while no influenced across the nos mutants and gsnor mutants" does not make sense.

We appreciate the reviewer’s observation and agree that the original sentence was unclear. We have revised the sentence in the manuscript as follows:

“In contrast, no significant change in MG proliferation was observed in nos1, nos2a, or gsnor mutants compared to wild type (Figures 4F–4I)” (revised lines 326-328).

(10) Line 326-328: The text should be rewritten as the current meaning would suggest there was no significant loss of photoreceptors in the nos2b mutants. That is incorrect. Rather, there was no significant difference between WT and the nos2b mutants in the number of photoreceptors lost at 72 hpi following MTZ treatment. Both groups lost photoreceptors, but the number lost in nos2b hets and homozygotes was the same as WT.

We agree with the suggestion and have revised our manuscript. We have revised the sentence in the manuscript as follows:

“We observed no significant difference in the loss of cone photoreceptor at 72 hpi between nos2b mutants and WT, indicating that the reduced MG proliferation observed in nos2b mutants is independent of the injury (WT: 45 ± 8 remaining cones, n = 24; nos2b⁺/⁻: 49 ± 12, n = 20; nos2b⁻/⁻: 46 ± 9, n = 20; mean ± SEM) (Figure 4K).” (revised lines 331-335).

(11) There is concern over the inconsistencies with some of the data. In Figure 4, Supplement 1A, the single-cell data found virtually no expression of nos2b in either uninjured MG or cxcl18b+ MG. In contrast, the authors find nos2b expression by RT-PCR in the cxcl18b:GFP+ MG. The in situ expression of nos2b in Figure 5 - Supplement 1 is not persuasive. The red puncta are seen in a single cxcl18b:GFP+ cell but also in the plexiform layer and is other non cxcl18b:GFP+ cells.

We appreciate the concern regarding the apparent inconsistencies in nos2b expression across different datasets. We provide the following explanations:

(1) Low expression of nos2b in scRNA-seq data:

We propose a potential explanation: Nitric oxide (NO) signaling is known to exert its biological functions in a dose-dependent manner and is tightly regulated post-transcriptionally, especially in inducible nitric oxide synthase (iNOS) (Bogdan, 2001; Nathan and Xie, 1994; Thomas et al., 2008). Thus, even modest changes in nos2b expression may exert meaningful biological effects without producing strong transcriptional signals detectable by scRNA-seq, which could fall below the detection threshold of scRNA-seq methods. Supporting this idea, our functional assay (Figure 4J) reveals a clear concentration-dependent effect of NO on MG proliferation, consistent with the biological relevance of Nos2b activity despite its low transcript abundance.

(2) Regarding the in situ hybridization data:

We used both commercially available in situ hybridization probes from (HCR^TM) and RNAscope^TM (data not shown) to detect nos2b transcripts. While the nos2b signal was observed in other retinal cell types, including cells in the plexiform layer, our primary study was focused on examining its expression within the cxcl18b⁺ MG lineage.

(3) Regarding RT-PCR detection of nos2b in cxcl18b: GFP⁺ MG:

To enhance detection sensitivity, we enriched cxcl18b: GFP⁺ MG by FACS at 72 hpi and performed cDNA amplification before RT-PCR. This approach allowed the detection of low-abundance transcripts such as nos2b. It is also important to note that RT-PCR reflects fold changes in expression compared to MG to other retina cell type. The subtle but biologically upregulated of nos2b expression may not be readily captured by in situ hybridization or scRNA-seq.

(12) Line 356 - there is a disagreement over the interpretation of the current data. The statement that nos2b was specifically expressed in cxcl18b+ transitional MG states is not entirely accurate. This conclusion is based on expression of GFP from a cxcl18b promoter, which may reflect persistence of the GFP protein and not evidence of cxcl18b expression. Even assuming that the nos2b in situ hybridization and RT-PCR data are correct, the data would indicate that nos2b is expressed in proliferating MG that are derived from the cxcl18b+ transitional states. The single-cell trajectory analysis in Figure 2 indicates that cxcl18b is not co-expressed with PCNA. Furthermore, the single-cell data in Figure 4, Supplement 1, indicates no expression of nos2b in cxcl18b+ MG. The authors need to reconcile these seemingly contradictory pieces of data.

We thank the reviewer for this thoughtful and important comment. We agree that clarification is needed to accurately interpret the relationship between cxcl18b, nos2b, and MG proliferation, particularly considering the different temporal and technical contexts of our datasets.

(1) Lineage labeling and interpretation of GFP expression:

We acknowledge that in the Tg(cxcl18b: Cre-vhmc: mcherry::ef1a: loxP-dsRed-loxP-EGFP::lws2: nfsb-mCherry) line, GFP expression reflects historical activity of the cxcl18b promoter, rather than ongoing transcription. This GFP signal, due to its prolonged stay, may persist beyond the time window of endogenous cxcl18b expression. Accordingly, we have revised the manuscript to replace “cxcl18b⁺ MG” with “cxcl18b⁺ lineage-traced MG” throughout the relevant sections to prevent potential misinterpretation.

(2) Functional experiments support a lineage relationship between cxcl18b⁺ states and nos2b activity:

To further investigate the regulatory relationship between cxcl18b and nos2b, we conducted NO scavenging experiments using C-PTIO in the Tg(cxcl18b: GFP) background. We observed that the generation of cxcl18b: GFP⁺ MG after injury was not affected by NO depletion, indicating that cxcl18b activation precedes NO signaling (data not shown). However, PCNA⁺ MG was significantly reduced under the same treatment, suggesting that NO signaling is not required for cxcl18b⁺ transitional state formation, but is necessary for proliferation. Together with our MG-specific nos2b knockout data, these results support a model in which nos2b-derived NO acts downstream of the cxcl18b⁺ transitional state to promote MG cell-cycle re-entry.

(3) The scRNA-seq data with nos2b expression:

We agree with the reviewer that our scRNA-seq dataset shows minimal overlap between cxcl18b and pcna expression, which is consistent with our interpretation that cxcl18b expression marks a transitional phase before cell-cycle entry. Furthermore, nos2b transcripts were not robustly detected in cxcl18b⁺ MG clusters in our scRNA dataset. This discrepancy may be caused by technical limitations of scRNA-seq in capturing low-abundance or transient transcripts such as nos2b, as discussed in response to comment #11.

(13) The data in Figure 7 are interesting and suggest a link between NO signaling and notch activity. The use of the C-PTIO NO scavenger is not specific to MG, which limits the conclusions related to autocrine NO signaling in cxcl18b+ MG.

We acknowledge that the use of C-PTIO cannot distinguish between NO signaling within MG and paracrine effects from other retinal cells. Currently, technical limitations prevent MG-specific NO depletion. We have discussed this limitation accordingly in our revised “Limitations of this study” section (revised lines 540-545: “2. While our data suggest that injury-induced NO suppresses Notch signaling activation and promotes MG proliferation, the use of a general NO scavenger (C-PTIO) does not allow us to determine whether this regulation occurs in an autocrine or paracrine manner. The specific role of NO signaling within cxcl18b⁺ MG requires further validation using MG-specific NO depletion.”)

(14) Line 446-448. As mentioned before, the data do not support a causative link between microglia recruitment and cxcl18b induction. More specifically, dexamethasone is a broad-spectrum anti-inflammatory drug that blocks microglia activation and recruitment. Critically, the authors demonstrate that expression of cxcl18b occurs prior to microglia recruitment (see Figure 1, Supplement 1). Thus, the statement that cxcl18b induction depends on microglia recruitment is not accurate.

We thank the reviewer for reiterating this important point. We fully agree that the current data do not support a direct causal relationship between microglia recruitment and cxcl18b induction. As also addressed in our response to Comment 7, dexamethasone, as a broad-spectrum anti-inflammatory agent, cannot distinguish microglia-specific effects from those of other immune components. We have revised the text in revised lines 251–258 to clarify that microglia-mediated inflammation is associated with—but not required for—activation of cxcl18b-defined transitional MG states.

Reference:

Bogdan, C. (2001). Nitric oxide and the immune response. Nature immunology 2, 907-916.

Bradley, S., Tossell, K., Lockley, R., and McDearmid, J.R. (2010). Nitric oxide synthase regulates morphogenesis of zebrafish spinal cord motoneurons. The Journal of neuroscience : the official journal of the Society for Neuroscience 30, 16818-16831.

Gorsuch, R.A., Lahne, M., Yarka, C.E., Petravick, M.E., Li, J., and Hyde, D.R. (2017). Sox2 regulates Müller glia reprogramming and proliferation in the regenerating zebrafish retina via Lin28 and Ascl1a. Experimental eye research 161, 174-192.

Hamon, A., García-García, D., Ail, D., Bitard, J., Chesneau, A., Dalkara, D., Locker, M., Roger, J.E., and Perron, M. (2019). Linking YAP to Müller Glia Quiescence Exit in the Degenerative Retina. Cell reports 27, 1712-1725.e1716.

Iribarne, M., and Hyde, D.R. (2022). Different inflammation responses modulate Müller glia proliferation in the acute or chronically damaged zebrafish retina. Frontiers in cell and developmental biology 10, 892271.

Jin, D., Ni, T.T., Hou, J., Rellinger, E., and Zhong, T.P. (2009). Promoter analysis of ventricular myosin heavy chain (vmhc) in zebrafish embryos. Developmental dynamics : an official publication of the American Association of Anatomists 238, 1760-1767.

Krylov, A., Yu, S., Veen, K., Newton, A., Ye, A., Qin, H., He, J., and Jusuf, P.R. (2023). Heterogeneity in quiescent Müller glia in the uninjured zebrafish retina drive differential responses following photoreceptor ablation. Frontiers in molecular neuroscience 16, 1087136.

Lahne, M., Nagashima, M., Hyde, D.R., and Hitchcock, P.F. (2020). Reprogramming Müller Glia to Regenerate Retinal Neurons. Annual review of vision science 6, 171-193.

Lee, M.S., Jui, J., Sahu, A., and Goldman, D. (2024). Mycb and Mych stimulate Müller glial cell reprogramming and proliferation in the uninjured and injured zebrafish retina. Development (Cambridge, England) 151.

Lourenço, R., Brandão, A.S., Borbinha, J., Gorgulho, R., and Jacinto, A. (2021). Yap Regulates Müller Glia Reprogramming in Damaged Zebrafish Retinas. Frontiers in cell and developmental biology 9, 667796.

Matrone, G., Jung, S.Y., Choi, J.M., Jain, A., Leung, H.E., Rajapakshe, K., Coarfa, C., Rodor, J., Denvir, M.A., Baker, A.H., et al. (2021). Nuclear S-nitrosylation impacts tissue regeneration in zebrafish. Nat Commun 12, 6282.

Mazzolini, J., Le Clerc, S., Morisse, G., Coulonges, C., Kuil, L.E., van Ham, T.J., Zagury, J.F., and Sieger, D. (2020). Gene expression profiling reveals a conserved microglia signature in larval zebrafish. Glia 68, 298-315.

Meyers, J.R., Hu, L., Moses, A., Kaboli, K., Papandrea, A., and Raymond, P.A. (2012). β-catenin/Wnt signaling controls progenitor fate in the developing and regenerating zebrafish retina. Neural development 7, 30.

Nagashima, M., and Hitchcock, P.F. (2021). Inflammation Regulates the Multi-Step Process of Retinal Regeneration in Zebrafish. Cells 10.

Nathan, C., and Xie, Q.W. (1994). Nitric oxide synthases: roles, tolls, and controls. Cell 78, 915-918.

Pollak, J., Wilken, M.S., Ueki, Y., Cox, K.E., Sullivan, J.M., Taylor, R.J., Levine, E.M., and Reh, T.A. (2013). ASCL1 reprograms mouse Muller glia into neurogenic retinal progenitors. Development (Cambridge, England) 140, 2619-2631.

Rochon, E.R., Missinato, M.A., Xue, J., Tejero, J., Tsang, M., Gladwin, M.T., and Corti, P. (2020). Nitrite Improves Heart Regeneration in Zebrafish. Antioxidants & redox signaling 32, 363-377.

Sarich, S.C., Sreevidya, V.S., Udvadia, A.J., Svoboda, K.R., and Gutzman, J.H. (2025). The transcription factor Jun is necessary for optic nerve regeneration in larval zebrafish. PloS one 20, e0313534.

Sifuentes, C.J., Kim, J.W., Swaroop, A., and Raymond, P.A. (2016). Rapid, Dynamic Activation of Müller Glial Stem Cell Responses in Zebrafish. Investigative ophthalmology & visual science 57, 5148-5160.

Svahn, A.J., Graeber, M.B., Ellett, F., Lieschke, G.J., Rinkwitz, S., Bennett, M.R., and Becker, T.S. (2013). Development of ramified microglia from early macrophages in the zebrafish optic tectum. Developmental neurobiology 73, 60-71.

Thomas, D.D., Ridnour, L.A., Isenberg, J.S., Flores-Santana, W., Switzer, C.H., Donzelli, S., Hussain, P., Vecoli, C., Paolocci, N., Ambs, S., et al. (2008). The chemical biology of nitric oxide: implications in cellular signaling. Free radical biology & medicine 45, 18-31.

Thomas, J.L., Ranski, A.H., Morgan, G.W., and Thummel, R. (2016). Reactive gliosis in the adult zebrafish retina. Experimental eye research 143, 98-109.

Wan, J., and Goldman, D. (2016). Retina regeneration in zebrafish. Current opinion in genetics & development 40, 41-47.

White, D.T., Sengupta, S., Saxena, M.T., Xu, Q., Hanes, J., Ding, D., Ji, H., and Mumm, J.S. (2017). Immunomodulation-accelerated neuronal regeneration following selective rod photoreceptor cell ablation in the zebrafish retina. Proceedings of the National Academy of Sciences of the United States of America 114, E3719-e3728.

Yao, K., Qiu, S., Tian, L., Snider, W.D., Flannery, J.G., Schaffer, D.V., and Chen, B. (2016). Wnt Regulates Proliferation and Neurogenic Potential of Müller Glial Cells via a Lin28/let-7 miRNA-Dependent Pathway in Adult Mammalian Retinas. Cell reports 17, 165-178.

Yin, Z., Kang, J., Xu, H., Huo, S., and Xu, H. (2024). Recent progress of principal techniques used in the study of Müller glia reprogramming in mice. Cell regeneration (London, England) 13, 30.

Yu, C., Li, X., Ma, J., Liang, S., Zhao, Y., Li, Q., and Zhang, R. (2024). Spatiotemporal modulation of nitric oxide and Notch signaling by hemodynamic-responsive Trpv4 is essential for ventricle regeneration. Cellular and molecular life sciences : CMLS 81, 60.

eLife Assessment

This important study focuses on the molecular mechanisms underlying the generation of neuronal diversity. Taking advantage of a well-defined neuroblast lineage in Drosophila, the authors provide convincing evidence that two transcription factors of the conserved forkhead box (FOX) family provide a mechanistic link between transient spatial cues that initially specify neuroblast identity and terminal selector genes that define post-mitotic neuron identity. The findings will be of interest to developmental neurobiologists.

Reviewer #1 (Public review):

Summary:

Lai and Doe address the integration of spatial information with temporal patterning and genes that specify cell fate. They identify the Forkhead transcription factor Fd4 as a lineage-restricted cell fate regulator that bridges transient spatial transcription factors to terminal selector genes in the developing Drosophila ventral nerve cord. The experimental evidence convincingly demonstrates that Fd4 is both necessary for late-born NB7-1 neurons, but also sufficient to transform other neural stem cell lineages toward the NB7-1 identity. This work addresses an important question that will be of interest to developmental neurobiologists: How can cell identities defined by initial transient developmental cues be maintained in the progeny cells, even if the molecular mechanism remains to be investigated? In addition, the study proposes a broader concept of lineage identity genes that could be utilized in other lineages and regions in the Drosophila nervous system and in other species.

Strengths:

While the spatial factors patterning the neuroepithelium to define the neuroblast lineages in the Drosophila ventral nerve cord are known, these factors are sometimes absent or not required during neurogenesis. In the current work, Lai and Doe identified Fd4 in the NB7-1 lineage that bridges this gap and explains how NB7-1 neurons are specified after Engrailed (En) and Vnd cease their expression. They show that Fd4 is transiently co-expressed with En and Vnd and is present in all nascent NB7-1 progenies. They further demonstrate that Fd4 is required for later-born NB7-1 progenies and sufficient for the induction of NB7-1 markers (Eve and Dbx) while repressing markers of other lineages when force-expressed in neural progenitors, e.g., in the NB5-6 lineage and in the NB7-3 lineage. They also demonstrate that, when Fd4 is ectopically expressed in NB7-3 and NB5-6 lineages, this leads to the ectopic generation of dorsal muscle-innervating neurons. The inclusion of functional validation using axon projections demonstrates that the transformed neurons acquire appropriate NB7-1 characteristics beyond just molecular markers. Quantitative analyses are thorough and well-presented for all experiments.

Weaknesses:

(1) While Fd4 is required and sufficient for several later-born NB7-1 progeny features, a comparison between early-born (Hb/Eve) and later-born (Run/Eve) appears missing for pan-progenitor gain of Fd4 (with sca-Gal4; Figure 4) and for the NB7-3 lineage (Figure 6). Having a quantification for both could make it clearer whether Fd4 preferentially induces later-born neurons or is sufficient for NB7-1 features without temporal restriction.

(2) Fd4 and Fd5 are shown to be partially redundant, as Fd4 loss of function alone does not alter the number of Eve+ and Dbx+ neurons. This information is critical and should be included in Figure 3.

(3) Several observations suggest that lineage identity maintenance involves both Fd4-dependent and Fd4-independent mechanisms. In particular, the fact that fd4-Gal4 reporter remains active in fd4/fd5 mutants even after Vnd and En disappear indicates that Fd4's own expression, a key feature of NB7-1 identity, is maintained independently of Fd4 protein. This raises questions about what proportion of lineage identity features require Fd4 versus other maintenance mechanisms, which deserves discussion.

(4) Similarly, while gain of Fd4 induces NB7-1 lineage markers and dorsal muscle innervation in NB5-6 and NB7-3 lineages, drivers for the two lineages remain active despite the loss of molecular markers, indicating some regulatory elements retain activity consistent with their original lineage identity. It is therefore important to understand the degree of functional conversion in the gain-of-function experiments. Sparse labeling of Fd4 overexpressing NB5-6 and NB7-3 progenies, as was done in Seroka and Doe (2019), would be an option.

(5) The less-penetrant induction of Dbx+ neurons in NB5-6 with Fd4-overexpression is interesting. It might be worth the authors discussing whether it is an Fd4 feature or an NB5-6 feature by examining Dbx+ neuron number in NB7-3 with Fd4-overexpression.

(6) It is logical to hypothesize that spatial factors specify early-born neurons directly, so only late-born neurons require Fd4, but it was not tested. The model would be strengthened by examining whether Fd4-Gal4-driven Vnd rescues the generation of later-born neurons in fd4/fd5 mutants.

(7) It is mentioned that Fd5 is not sufficient for the NB7-1 lineage identity. The observation is intriguing in how similar regulators serve distinct roles, but the data are not shown. The analysis in Figure 4 should be performed for Fd5 as supplemental information.

Reviewer #2 (Public review):

Summary:

Via a detailed expression analysis, they find that Fd4 is selectively expressed in embryonic NB7-1 and newly born neurons within this lineage. They also undertake a comprehensive genetic analysis to provide evidence that fd4 is necessary and sufficient for the identity of NB7-1 progeny.

Strengths:

The analysis is both careful and rigorous, and the findings are of interest to developmental neurobiologists interested in molecular mechanisms underlying the generation of neuronal diversity. Great care was taken to make the figures clear and accessible. This work takes great advantage of years of painstaking descriptive work that has mapped embryonic neuroblast lineages in Drosophila.

Weaknesses:

The argument that Fd4 is necessary for NB7-1 lineage identity is based on a Fd4/Fd5 double mutant. Loss of fd4 alone did not alter the number of NB7-1-derived Eve+ or Dbx+ neurons. The authors clearly demonstrate redundancy between fd4 and fd5, and the fact that the LOF analysis is based on a double mutant should be better woven through the text. The authors generated an Fd5 mutant. I assume that Fd5 single mutants do not display NB7-1 lineage defects, but this is not stated. The focus on Fd4 over Fd5 is based on its highly specific expression profile and the dramatic misexpression phenotypes. But the LOF analysis demonstrates redundancy, and the conclusions in the abstract and through the results should reflect the existence of Fd5 in the conclusions of this manuscript.

It is notable that Fd4 overexpression can rewire motor circuits. This analysis adds another dimension to the changes in transcription factor expression and, importantly, demonstrates functional consequences. Could the authors test whether U4 and U5 motor axon targeting changes in the fd4/fd5 double mutant? To strengthen claims regarding the importance of fd4/fd5 for lineage identity, it would help to address terminal features of U motorneuron identity in the LOF condition.

Reviewer #3 (Public review):

The goal of the work is to establish the linkage between the spatial transcription factors (STFs) that function transiently to establish the identities of the individual NBs and the terminal selector genes (typically homeodomain genes) that appear in the newborn post-mitotic neurons. How is the identity of the NB maintained and carried forward after the spatial genes have faded away? Focusing on a single neuroblast (NB 7-1), the authors present evidence that the fork-head transcription factor, fd4, provides a bridge linking the transient spatial cues that initially specified neuroblast identity with the terminal selector genes that establish and maintain the identity of the stem cell's progeny.

The study is systematic, concise, and takes full advantage of 40+ years of work on the molecular players that establish neuronal identities in the Drosophila CNS. In the embryonic VNC, fd4 is expressed only in the NB 7-1 and its lineage. They show that Fd4 appears in the NB while the latter is still expressing the Spatial Transcription Factors and continues after the expression of the latter fades out. Fd4 is maintained through the early life of the neuronal progeny but then declines as the neurons turn on their terminal selector genes. Hence, fd4 expression is compatible with it being a bridging factor between the two sets of genes.

Experimental support for the "bridging" role of Fd4 comes from a set of loss-of-function and gain-of-function manipulations. The loss of function of Fd4, and the partially redundant gene Fd5, from lineage 7-1 does not affect the size of the lineage, but terminal markers of late-born neuronal phenotypes, like Eve and Dbx, are reduced or missing. By contrast, ectopic expression of fd4, but not fd5, results in ectopic expression of the terminal markers eve and Dbx throughout diverse VNC lineages.

A detailed test of fd4's expression was then carried out using lineages 7-3 and 5-6, two well-characterized lineages in Drosophila. Lineage 7-3 is much smaller than 7-1 and continues to be so when subjected to fd4 misexpression. However, under the influence of ectopic Fd4 expression, the lineage 7-3 neurons lost their expected serotonin and corazonin expression and showed Eve expression as well as motoneuron phenotypes that partially mimic the U motoneurons of lineage 7-1.

Ectopic expression of Fd4 also produced changes in the 5-6 lineage. Expression of apterous, a feature of lineage 5-6, was suppressed, and expression of the 7-1 marker, Eve, was evident. Dbx expression was also evident in the transformed 5-6 lineages, but extremely restricted as compared to a normal 7-1 lineage. Considering the partial redundancy of fd4 and fd5, it would have been interesting to express both genes in the 5-6 lineage. The anatomical changes that are exhibited by motoneurons in response to Fd4 expression confirm that these cells do, indeed, show a shift in their cellular identity.

Author response:

Reviewer #1 (Public Review):

Lai and Doe address the integration of spatial information with temporal patterning and genes that specify cell fate. They identify the Forkhead transcription factor Fd4 as a lineage-restricted cell fate regulator that bridges transient spatial transcription factors to terminal selector genes in the developing Drosophila ventral nerve cord. The experimental evidence convincingly demonstrates that Fd4 is both necessary for lateborn NB7-1 neurons, but also sufficient to transform other neural stem cell lineages toward the NB7-1 identity. This work addresses an important question that will be of interest to developmental neurobiologists: How can cell identities defined by initial transient developmental cues be maintained in the progeny cells, even if the molecular mechanism remains to be investigated? In addition, the study proposes a broader concept of lineage identity genes that could be utilized in other lineages and regions in the Drosophila nervous system and in other species. 

Thanks for the accurate summary and positive comments!

While the spatial factors patterning the neuroepithelium to define the neuroblast lineages in the Drosophila ventral nerve cord are known, these factors are sometimes absent or not required during neurogenesis. In the current work, Lai and Doe identified Fd4 in the NB7-1 lineage that bridges this gap and explains how NB7-1 neurons are specified after Engrailed (En) and Vnd cease their expression. They show that Fd4 is transiently co-expressed with En and Vnd and is present in all nascent NB7-1 progenies. They further demonstrate that Fd4 is required for later-born NB7-1 progenies and sufficient for the induction of NB7-1 markers (Eve and Dbx) while repressing markers of other lineages when force-expressed in neural progenitors, e.g., in the NB56 lineage and in the NB7-3 lineage. They also demonstrate that, when Fd4 is ectopically expressed in NB7-3 and NB5-6 lineages, this leads to the ectopic generation of dorsal muscle-innervating neurons. The inclusion of functional validation using axon projections demonstrates that the transformed neurons acquire appropriate NB7-1 characteristics beyond just molecular markers. Quantitative analyses are thorough and well-presented for all experiments.

Thanks for the positive comments!

(1) While Fd4 is required and sufficient for several later-born NB7-1 progeny features, a comparison between early-born (Hb/Eve) and later-born (Run/Eve) appears missing for pan-progenitor gain of Fd4 (with sca-Gal4; Figure 4) and for the NB7-3 lineage (Figure 6). Having a quantification for both could make it clearer whether Fd4 preferentially induces later-born neurons or is sufficient for NB7-1 features without temporal restriction.

We quantified the percentage of Hb+ and Runt+ cells among Eve+ cells with sca-gal4, and the results are shown in Figure 4-figure supplement 1. We found that the proportion of early-born cells is slightly reduced but the proportion of later-born cells remain similar. Interestingly, we also found a subset of Eve+ cells with a mixed fate (Hb+Runt+) but the reason remains unclear.

(2) Fd4 and Fd5 are shown to be partially redundant, as Fd4 loss of function alone does not alter the number of Eve+ and Dbx+ neurons. This information is critical and should be included in Figure 3.

Because every hemisegment in an fd4 single mutant is normal, we just added it as the following text: “In fd4 mutants, we observe no change in the number of Eve+ neurons or Dbx+ neurons (n=40 hemisegments).”

(3) Several observations suggest that lineage identity maintenance involves both Fd4dependent and Fd4-independent mechanisms. In particular, the fact that fd4-Gal4 reporter remains active in fd4/fd5 mutants even after Vnd and En disappear indicates that Fd4's own expression, a key feature of NB7-1 identity, is maintained independently of Fd4 protein. This raises questions about what proportion of lineage identity features require Fd4 versus other maintenance mechanisms, which deserves discussion.

We agree, thanks for raising this point. We add the following text to the Discussion. “Interestingly, the fd4 fd5 mutant maintains expression of fd4:gal4, suggesting that the fd4/fd5 locus may have established a chromatin state that allows “permanent” expression in the absence of Vnd, En, and Fd4/Fd5 proteins.”

(4) Similarly, while gain of Fd4 induces NB7-1 lineage markers and dorsal muscle innervation in NB5-6 and NB7-3 lineages, drivers for the two lineages remain active despite the loss of molecular markers, indicating some regulatory elements retain activity consistent with their original lineage identity. It is therefore important to understand the degree of functional conversion in the gain-of-function experiments. Sparse labeling of Fd4 overexpressing NB5-6 and NB7-3 progenies, as was done in Seroka and Doe (2019), would be an option.

We agree it is interesting that the NB7-3 and NB5-6 drivers remain on following Fd4 misexpression. To explore this, we used sca-gal4 to overexpress Fd4 and observed that Lbe expression persisted while Eg was largely repressed (see Author response image 1 below). The results show that Lbe and Eg respond differently to Fd4. A non-mutually exclusive possibility is that the continued expression of lbe-Gal4 UAS-GFP or eg-Gal4 UAS-GFP may be due to the lengthy perdurance of both Gal4 and GFP.

Author response image 1.

(5) The less-penetrant induction of Dbx+ neurons in NB5-6 with Fd4-overexpression is interesting. It might be worth the authors discussing whether it is an Fd4 feature or an NB56 feature by examining Dbx+ neuron number in NB7-3 with Fd4-overexpression.

In the NB7-3 lineages misexpressing Fd4, only 5 lineages generated Dbx+ cells (0.1±0.4, n=64 hemisegments), suggesting that the low penetrance of Dbx+ induction is an intrinsic feature of Fd4 rather than lineage context. We have added this information in the results section. 

(6) It is logical to hypothesize that spatial factors specify early-born neurons directly, so only late-born neurons require Fd4, but it was not tested. The model would be strengthened by examining whether Fd4-Gal4-driven Vnd rescues the generation of laterborn neurons in fd4/fd5 mutants.

When we used en-gal4 driver to express UAS-vnd in the fd4/fd5 mutant background, we found an average 7.4±2.2 Eve+ cells per hemisegment (n=36), significantly higher than fd4/fd5 mutant alone (3.9±0.8 cells, n=52, p=2.6x10<sup.-11</sup>) (Figure 3J). In addition, 0.2±0.5 Eve+ cells were ectopic Hb+ (excluding U1/U2), indicating that Vnd-En integration is sufficient to generate both early-born and late-born Eve+ cells in the fd4/fd5 mutants. We have added the results to the text.

(7) It is mentioned that Fd5 is not sufficient for the NB7-1 lineage identity. The observation is intriguing in how similar regulators serve distinct roles, but the data are not shown. The analysis in Figure 4 should be performed for Fd5 as supplemental information.

Thanks for the suggestion. Because the results are exactly the same as the wild type, we don’t think it is necessary to provide an additional images or analysis as supplemental information.

Reviewer #2 (Public review):

Via a detailed expression analysis, they find that Fd4 is selectively expressed in embryonic NB7-1 and newly born neurons within this lineage. They also undertake a comprehensive genetic analysis to provide evidence that fd4 is necessary and sufficient for the identity of NB7-1 progeny. 

Thanks for the accurate summary!

The analysis is both careful and rigorous, and the findings are of interest to developmental neurobiologists interested in molecular mechanisms underlying the generation of neuronal diversity. Great care was taken to make the figures clear and accessible. This work takes great advantage of years of painstaking descriptive work that has mapped embryonic neuroblast lineages in Drosophila. 

Thanks for the positive comments!

The argument that Fd4 is necessary for NB7-1 lineage identity is based on a Fd4/Fd5 double mutant. Loss of fd4 alone did not alter the number of NB7-1-derived Eve+ or Dbx+ neurons. The authors clearly demonstrate redundancy between fd4 and fd5, and the fact that the LOF analysis is based on a double mutant should be better woven through the text.

The authors generated an Fd5 mutant. I assume that Fd5 single mutants do not display NB7-1 lineage defects, but this is not stated. The focus on Fd4 over Fd5 is based on its highly specific expression profile and the dramatic misexpression phenotypes. But the LOF analysis demonstrates redundancy, and the conclusions in the abstract and through the results should reflect the existence of Fd5 in the conclusions of this manuscript.

We agree, and have added new text to clarify the single mutant phenotypes (there are none) and the double mutant phenotype (loss of NB7-1 molecular and morphological features. The following text is added to the manuscript: “Not surprisingly, we found that fd4 single mutants or fd5 single mutants had no phenotype (Eve+ neurons were all normal). Thus, to assess their roles, we generated a fd4 and fd5 double mutant. Because many Eve+ and Dbx+ cells are generated outside of NB7-1 lineage, it was also essential to identify the Eve+ or Dbx+ cells within NB7-1 lineage in wild type and fd4 mutant embryos. To achieve this, we replaced the open reading frame of fd4 with gal4 (called fd4-gal4) (see Methods); this stock simultaneously knocked out both fd4 and fd5 (called fd4/fd5 mutant hereafter) while specifically labeling the NB7-1 lineage. For the remainder of this paper we use the fd4/fd5 double mutant to assay for loss of function phenotypes.”

It is notable that Fd4 overexpression can rewire motor circuits. This analysis adds another dimension to the changes in transcription factor expression and, importantly, demonstrates functional consequences. Could the authors test whether U4 and U5 motor axon targeting changes in the fd4/fd5 double mutant? To strengthen claims regarding the importance of fd4/fd5 for lineage identity, it would help to address terminal features of U motorneuron identity in the LOF condition.

Thanks for raising this important point. We examined the axon targeting on body wall muscles in both wild type and in fd4/fd5 mutant background and added the results in Figure 3-figure supplement 2. We found that the axon targeting in the late-born neuron region (LL1) is significantly reduced, suggesting that the loss of late-born neurons in fd4/fd5 mutant leads to the absence of innervation of corresponding muscle targets.

Reviewer #3 (Public review):

The goal of the work is to establish the linkage between the spatial transcription factors (STFs) that function transiently to establish the identities of the individual NBs and the terminal selector genes (typically homeodomain genes) that appear in the newborn postmitotic neurons. How is the identity of the NB maintained and carried forward after the spatial genes have faded away? Focusing on a single neuroblast (NB 7-1), the authors present evidence that the fork-head transcription factor, fd4, provides a bridge linking the transient spatial cues that initially specified neuroblast identity with the terminal selector genes that establish and maintain the identity of the stem cell's progeny. 

Thanks for the positive comments!

The study is systematic, concise, and takes full advantage of 40+ years of work on the molecular players that establish neuronal identities in the Drosophila CNS. In the embryonic VNC, fd4 is expressed only in the NB 7-1 and its lineage. They show that Fd4 appears in the NB while the latter is still expressing the Spatial Transcription Factors and continues after the expression of the latter fades out. Fd4 is maintained through the early life of the neuronal progeny but then declines as the neurons turn on their terminal selector genes. Hence, fd4 expression is compatible with it being a bridging factor between the two sets of genes. 

Thanks for the accurate summary!

Experimental support for the "bridging" role of Fd4 comes from a set of loss-of-function and gain-of-function manipulations. The loss of function of Fd4, and the partially redundant gene Fd5, from lineage 7-1 does not aoect the size of the lineage, but terminal markers of late-born neuronal phenotypes, like Eve and Dbx, are reduced or missing. By contrast, ectopic expression of fd4, but not fd5, results in ectopic expression of the terminal markers eve and Dbx throughout diverse VNC lineages. 

Thanks for the accurate summary!

A detailed test of fd4's expression was then carried out using lineages 7-3 and 5-6, two well-characterized lineages in Drosophila. Lineage 7-3 is much smaller than 7-1 and continues to be so when subjected to fd4 misexpression. However, under the influence of ectopic Fd4 expression, the lineage 7-3 neurons lost their expected serotonin and corazonin expression and showed Eve expression as well as motoneuron phenotypes that partially mimic the U motoneurons of lineage 7-1.

Thanks for the positive comments!

Ectopic expression of Fd4 also produced changes in the 5-6 lineage. Expression of apterous, a feature of lineage 5-6, was suppressed, and expression of the 7-1 marker, Eve, was evident. Dbx expression was also evident in the transformed 5-6 lineages, but extremely restricted as compared to a normal 7-1 lineage. Considering the partial redundancy of fd4 and fd5, it would have been interesting to express both genes in the 5-6 lineage. The anatomical changes that are exhibited by motoneurons in response to Fd4 expression confirm that these cells do, indeed, show a shift in their cellular identity.

We appreciate the positive comments. We agree double misexpression of Fd4 and Fd5 might give a stronger phenotype (as the reviewer says) but the lack of this experiment does not change the conclusions that Fd4 can promote NB7-1 molecular and morphological aspects at the expense of NB5-6 molecular markers.

eLife Assessment

This study presents a valuable open-source and cost-effective method for automating the quantification of male aggression and courtship in Drosophila melanogaster. The work as presented provides solid evidence that the use of the behavioral setup that the authors designed - using readily available laboratory equipment and standardised high-performing classifiers they developed using existing software packages - accurately and reliably characterises social behavior in Drosophila. The work will be of interest to Drosophila neurobiologists and particularly to those working on male social behaviors.

Reviewer #1 (Public review):

The study introduces an open-source, cost-effective method for automating the quantification of male social behaviors in Drosophila melanogaster. It combines machine-learning based behavioral classifiers developed using JAABA (Janelia Automatic Animal Behavior Annotator) with inexpensive hardware constructed from off-the-shelf components. This approach addresses the limitations of existing methods, which often require expensive hardware and specialized setups. The authors demonstrate that their new "DANCE" classifiers accurately identify aggression (lunges) and courtship behaviors (wing extension, following, circling, attempted copulation, and copulation), closely matching manually annotated ground-truth data. Furthermore, DANCE classifiers outperform existing rule-based methods in accuracy. Finally, the study shows that DANCE classifiers perform as well when used with low-cost experimental hardware as with standard experimental setups across multiple paradigms, including RNAi knockdown of the neuropeptide Dsk and optogenetic silencing of dopaminergic neurons.

The authors make creative use of existing resources and technology to develop an inexpensive, flexible, and robust experimental tool for the quantitative analysis of Drosophila behavior. A key strength of this work is the thorough benchmarking of both the behavioral classifiers and the experimental hardware against existing methods. In particular, the direct comparison of their low-cost experimental system with established systems across different experimental paradigms is compelling. A weakness of the study is that the use of JAABA-based classifiers to analyze aggression and courtship is not novel (Tao et al., J. Neurosci., 2024; Sten et al., Cell, 2023; Chiu et al., Cell, 2021; Isshi et al., eLife, 2020; Duistermars et al., Neuron, 2018). However, the demonstration the JAABA classifiers they developed work as well without expensive experimental hardware opens the door to more low-cost systems for quantitative behavior analysis.

In summary, this work provides a practical and accessible approach to quantifying Drosophila behavior, reducing the economic barriers to the study of the neural and molecular mechanisms underlying social behavior.

Reviewer #2 (Public review):

Summary:

This manuscript addresses the development of a low-cost behavioural setup and standardised open-source high performing classifiers for aggression and courtship behaviour. It does so by using readily available laboratory equipment and previously developed software packages. By comparing the performance of the setup and the classifiers to previously developed ones, this study shows the classifier's overperformance and the reliability of the low-cost setup in recapitulating previously described effects of different manipulations on aggression and courtship.

Strengths:

The newly developed classifiers for lunges, wing extension, attempted copulation, copulation, following, circling, perform better than previously available developed ones. The behavioural setup developed is low cost and reliably allows analysis of both aggression and courtship behaviour, validated through social experience manipulation (social isolation), gene knock (Dsk in Dilp2 neurons) and neuronal inactivation (dopaminergic neurons) know to affect courtship and aggression.

Weaknesses:

This framework only encompasses analysis of lunges, while aggression encompasses multiple behaviours. Even though DANCE can serve as a template allowing future development of additional classifiers, the current study compares performance to CADABRA which analyses further aggression behaviours, making the comparisons incomplete.

Reviewer #3 (Public review):

The study by Yadav et al. describes a new setup to quantify a number of aggression and mating behaviors in Drosophila melanogaster. The investigation of these behaviors requires the analysis of large number of videos to identify each kind of behavior displayed by a fly. Several approaches to automatize this process have been published before, but each of them has their limitations. The authors set out to develop a new setup that includes a very low-cost, easy to acquire hardware and open-source machine-learning classifiers to identify and quantify the behavior.

Strengths:

(1) The study demonstrates that their cheap, simple, and easy to obtain hardware works just as well as custom-made, specialized hardware for analyzing aggression and mating behavior. This enables the setup to be used in a wide range of settings, from research with limited resources to classroom teaching.

(2) The authors used previously published software to train new classifiers for detecting a range of behaviors related to aggression and mating and make them freely available. The classifiers are very positively benchmarked against a manually acquired ground-truth as well as existing algorithms.

(3) The study demonstrates the applicability of the setup (hardware and classifiers) to common methods in the field by confirming a number of expected phenotypes with their setup.

Taken together, this work can greatly facilitate research of aggression and mating in Drosophila. The combination of low-cost, off-the-shelf hardware and open-source, robust software enables researchers with very little funding or technical expertise to contribute to the scientific process, and also allows large-scale experiments, for example, in classroom teaching with many students, or for systematic screenings.

Author response:

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

Reviewer #1 (Public review):

The study introduces an open-source, cost-effective method for automating the quantification of male social behaviors in Drosophila melanogaster. It combines machine-learning-based behavioral classifiers developed using JAABA (Janelia Automatic Animal Behavior Annotator) with inexpensive hardware constructed from off-the-shelf components. This approach addresses the limitations of existing methods, which often require expensive hardware and specialized setups. The authors demonstrate that their new "DANCE" classifiers accurately identify aggression (lunges) and courtship behaviors (wing extension, following, circling, attempted copulation, and copulation), closely matching manually annotated groundtruth data. Furthermore, DANCE classifiers outperform existing rule-based methods in accuracy. Finally, the study shows that DANCE classifiers perform as well when used with low-cost experimental hardware as with standard experimental setups across multiple paradigms, including RNAi knockdown of the neuropeptide Dsk and optogenetic silencing of dopaminergic neurons.

The authors make creative use of existing resources and technology to develop an inexpensive, flexible, and robust experimental tool for the quantitative analysis of Drosophila behavior. A key strength of this work is the thorough benchmarking of both the behavioral classifiers and the experimental hardware against existing methods. In particular, the direct comparison of their low-cost experimental system with established systems across different experimental paradigms is compelling.

While JAABA-based classifiers have been previously used to analyze aggression and courtship (Tao et al., J. Neurosci., 2024; Sten et al., Cell, 2023; Chiu et al., Cell, 2021; Isshi et al., eLife, 2020; Duistermars et al., Neuron, 2018), the demonstration that they work as well without expensive experimental hardware opens the door to more low-cost systems for quantitative behavior analysis.

We thank the reviewer for their positive assessment and constructive suggestions. We have cited these additional JAABA studies in the Introduction. We clarified that several prior JAABA-based classifiers were developed using specialized machinevision cameras or custom setups, and that in some cases the original code and classifiers were not made publicly available, which limits reproducibility and wider adoption. To address this, we explicitly note in the revised manuscript that DANCE was developed with accessibility in mind.

Although the study provides a detailed evaluation of DANCE classifier performance, its conclusions would be strengthened by a more comprehensive analysis. The authors assess classifier accuracy using a bout-level comparison rather than a frame-level analysis, as employed in previous studies (Kabra et al., Nat Methods, 2013). They define a true positive as any instance where a DANCE-detected bout overlaps with a manually annotated ground-truth bout by at least one frame. This criterion may inflate true positive rates and underestimate false positives, particularly for longer-duration courtship behaviors. For example, a 15-second DANCE-classified wing extension bout that overlaps with ground truth for only one frame would still be considered a true positive. A frame-level analysis performance would help address this possibility.

We thank the reviewer for raising this important point. Our original use of bout-level analysis followed existing literature (Duistermars et al., 2018; Ishii et al., 2020; Chiu et al., 2021; Tao et al., 2024; Hindmarsh Sten et al., 2025). While our lunge classifier already operates at the frame level, we have now performed additional frame-level evaluations for the duration based courtship classifiers. These analyses revealed only minor differences in precision, recall, and F1 scores compared with the original bout-level approach (see new Figure 5—Figure Supplement 3). Details of this analysis are now included in the Materials and Methods.

In summary, this work provides a practical and accessible approach to quantifying Drosophila behavior, reducing the economic barriers to the study of the neural and molecular mechanisms underlying social behavior.

We thank the reviewer for their encouraging comments and for recognizing the accessibility and practical value of our approach. We appreciate the constructive suggestions, which have helped strengthen the manuscript.

Reviewer #2 (Public review):

Summary:

This manuscript addresses the development of a low-cost behavioural setup and standardised open-source high-performing classifiers for aggression and courtship behaviour. It does so by using readily available laboratory equipment and previously developed software packages. By comparing the performance of the setup and the classifiers to previously developed ones, this study shows the classifier's overperformance and the reliability of the low-cost setup in recapitulating previously described effects of different manipulations on aggression and courtship.

Strengths:

The newly developed classifiers for lunges, wing extension, attempted copulation, copulation, following, and circling, perform better than available previously developed ones. The behavioural setup developed is low cost and reliably allows analysis of both aggression and courtship behaviour, validated through social experience manipulation (social isolation), gene knock (Dsk in Dilp2 neurons) and neuronal inactivation (dopaminergic neurons) known to affect courtship and aggression.

We thank the reviewer for the clear summary of our work and for highlighting its strengths. We appreciate these positive comments and suggestions, which have helped improve the clarity of the manuscript.

Weaknesses:

Aggression encompasses multiple defined behaviours, yet only lunges were analysed. Moreover, the CADABRA software to which DANCE was compared analyses further aggression behaviours, making their comparisons incomplete. In addition, though DANCE performs better than CADABRA and Divider in classifying lunges in the behavioural setup tested, it did not yield very high recall and F1 scores.

We thank the reviewer for raising this important point. We focused on lunges because they are widely used as a standard proxy for male aggression across multiple laboratories (Agrawal et al., 2020; Asahina et al., 2014; Chiu et al., 2021; Chowdhury et al., 2021; Dierick et al., 2007; Hoyer et al., 2008; Jung et al., 2020; Nilsen et al., 2004; Watanabe et al., 2017). As noted in the Discussion, our study also provides a template for the future development of additional aggression classifiers (fencing, wing flick, tussle, chase, female headbutt) and courtship classifiers (tapping, licking, rejection), which can be trained and shared through the same DANCE framework. Developing and validating these was beyond the scope of the present work.

To address the concern regarding precision, recall, and F1 scores, we performed additional analyses across all training videos and compiled these results in the new Figure 2—Figure Supplement 2. Our earlier lunge classifier had performance metrics obtained after training on a total of 11 videos. Our analysis shows performance metrics for classifiers trained on four independent datasets (Videos 8– 11). We found that the classifier trained on nine videos provided the best balance of precision, recall, and F1 (78.73%, 73.07%, and 75.79%, respectively), which was slightly better than the earlier classifier. We therefore updated the main figure, text, and Materials and Methods to use this version and uploaded the corresponding classifier and training details to the GitHub repository. 

DANCE is of limited use for neuronal circuit-level enquiries, since mechanisms for intensity and temporally controlled optogenetic manipulations, which are nowadays possible with open-source software and low-cost hardware, were not embedded in its development.

We thank the reviewer for this valuable point. The primary aim of DANCE is to provide an accessible, modular, and low-cost behavioural recording and analysis platform. It was designed so that users can readily integrate additional components such as optogenetic control when needed. As a proof of concept, we implemented optogenetic silencing of dopaminergic neurons using the DANCE hardware and confirmed that this manipulation increased aggression (Figure 7R). 

To facilitate adoption, we now provide schematic diagrams, LED control code, and instructions on our GitHub page and setup photographs in the manuscript (see new Figure 7—Figure Supplement 1). The released code allows programmable timing and intensity control, enabling users to reproduce temporally precise optogenetic protocols or extend the system for other stimulation paradigms.

Reviewer #3 (Public review):

The preprint by Yadav et al. describes a new setup to quantify a number of aggression and mating behaviors in Drosophila melanogaster. The investigation of these behaviors requires the analysis of a large number of videos to identify each kind of behavior displayed by a fly. Several approaches to automatize this process have been published before, but each of them has its limitations. The authors set out to develop a new setup that includes very low-cost, easy-to-acquire hardware and open-source machine-learning classifiers to identify and quantify the behavior.

Strengths:

(1) The study demonstrates that their cheap, simple, and easy-to-obtain hardware works just as well as custom-made, specialized hardware for analyzing aggression and mating behavior. This enables the setup to be used in a wide range of settings, from research with limited resources to classroom teaching.

(2) The authors used previously published software to train new classifiers for detecting a range of behaviors related to aggression and mating and to make them freely available. The classifiers are very positively benchmarked against a manually acquired ground truth as well as existing algorithms.

(3) The study demonstrates the applicability of the setup (hardware and classifiers) to common methods in the field by confirming a number of expected phenotypes with their setup.

We thank the reviewer for the positive assessment of our work and for highlighting its strengths. We appreciate these encouraging comments and suggestions, which have helped improve the clarity and presentation of the manuscript.

Weaknesses:

(1) When measuring the performance of the duration-based classifiers, the authors count any bout of behavior as true positive if it overlaps with a ground-truth positive for only 1 frame - despite the minimal duration of a bout is 10 frames, and most bouts are much longer. That way, true positives could contain cases that are almost totally wrong as long there was an overlap of a single frame. For the mating behaviors that are classified in ongoing bouts, I think performance should be evaluated based on the % of correctly classified frames, not bouts.

We thank the reviewer for raising this concern. In response to this point, and to Reviewer #1’s similar comment, we performed a frame-level evaluation of all duration-based courtship classifiers. The analysis revealed only minor differences compared with the original bout-level metrics (see new Figure 5—Figure Supplement 3), confirming the robustness of our classifiers. We have also added a description of this analysis in the Materials and Methods section.

(2) In the methods part, only one of the pre-existing algorithms (MateBook), is described. Given that the comparison with those algorithms is a so central part of the manuscript, each of them should be briefly explained and the settings used in this study should be described.

We thank the reviewer for this helpful suggestion. In the revised manuscript, we expanded the Materials and Methods to include concise descriptions and parameter settings for all pre-existing algorithms used for comparison. This includes dedicated subsections for CADABRA and the Divider assay, with explicit reference to their rulebased or geometric features. For MateBook, we specified the persistence filters used and the adjustments made for fair benchmarking. These changes ensure transparency and reproducibility.

Taken together, this work can greatly facilitate research on aggression and mating in Drosophila. The combination of low-cost, off-the-shelf hardware and open-source, robust software enables researchers with very little funding or technical expertise to contribute to the scientific process and also allows large-scale experiments, for example in classroom teaching with many students, or for systematic screenings.

We thank the reviewer for the encouraging comments and for recognizing the accessibility and broad applicability of DANCE. We believe these revisions have further strengthened the manuscript.

Reviewer #1 (Recommendations for the authors):

The following comments highlight areas where additional context, clarification, or further analysis could strengthen the manuscript. I hope these suggestions will be useful in refining your work.

(1) Lines 71-73: The authors state that Ctrax "leads to frequent identity switches among tracked flies, which is not the case while using FlyTracker." However, Ctrax was specifically designed to minimize identity errors, and Kabra et al. (2013) reported a low frequency of such errors-approximately one per five fly-hours in 10-fly videos. In contrast, Caltech FlyTracker does not correct identity errors automatically, requiring manual corrections, as noted in the Methods section of this study. If this is not an oversight, please provide further context to clarify this distinction.

We thank the reviewer for raising this clarification. As reported by Bentzur et al. (2021), when groups of flies were tracked simultaneously, Ctrax often generated multiple identities for the same individual, sometimes producing more trajectories than the actual number of flies. To prevent ambiguity, we revised the text to read: “While both Ctrax and FlyTracker (Eyjolfsdottir et al., 2014) may produce identity switches, when groups of flies were tracked simultaneously, Ctrax led to inaccuracies that required manual correction using specialized algorithms such as FixTrax (Bentzur et al., 2021).”  We also quantified FlyTracker identity-switch rates in our datasets and report them in new Supplementary File 5, confirming that such events were rare (< 2% of tracked intervals). We believe, this updated version provides the necessary context and ensures accuracy in describing each tracker’s limitations.

(2) Line 85: Providing additional context on how this study builds on previous work using JAABA-based classifiers for fly social behavior and comparing these classifiers to rule-based methods would more accurately situate it within the field. The authors state that "recently, a few JAABA-based classifiers have been developed for measuring aggression and courtship" and cite four related studies. However, this statement seems to underrepresent the use of JAABA-based classifiers for quantifying fly social behavior, which has become common in the field. Several additional studies (as noted in the public review) have developed JAABA-based classifiers for scoring aggression or courtship. Furthermore, other studies have compared the performance of JAABA-based classifiers with rule-based classifiers like CADABRA (e.g., Chowdhury et al., Comm Biology 2021; Leng et al., PlosOne 2020; Kabra et al., Nat Methods 2013). Mentioning the similar findings in those studies and your own helps strengthen the conclusion that machine-learning-based classifiers outperform rule-based classifiers in several experimental contexts.

We thank the reviewer for this helpful suggestion. We have revised the Introduction to include additional references to studies that applied JAABA-based classifiers for aggression and courtship and made textual edits to reflect this. We further noted that, unlike several previous studies, all DANCE classifiers and analysis code are publicly available.

Reviewer #2 (Recommendations for the authors):

(1) Suggestions for improved or additional experiments, data or analyses: As mentioned in the description of the effect of optogenetic inactivation of dopaminergic neurons, in the conclusion and also reported in the literature, there are other important identified aggression behaviours, such as fencing, wing flick, tussle, and chase. Similarly, for courtship, tapping and licking have also been defined. This study, as opposed to proposed future studies, would benefit from creating opensource classifiers for these established behaviours, which are important for the analysis of aggression and courtship.

We thank the reviewer for this valuable suggestion. As clarified in the Discussion, this manuscript intentionally focuses on six core, well-validated aggression and courtship behaviors to demonstrate DANCE’s modularity and reproducibility. Developing additional classifiers such as fencing, wing flick, tussle, chase, tapping, and licking would require extensive annotation and validation beyond the present scope. To address this point, we explicitly note in the revised text that the DANCE pipeline is readily extendable, allowing the community to build new classifiers within the same framework.

In terms of observer bias assessment for ground-truthing in courtship, this was only presented for circling and it would be beneficial to have encompassed all behaviours analysed.

We thank the reviewer for this suggestion. Observer-bias comparisons for all six classifiers are presented in Figure 2—Figure Supplement 1 (panels A–F). We clarified in the Results that annotations from two independent evaluators were compared for all classifiers, with no significant differences observed, confirming their robustness.

Finally, intensity and temporal optogenetic control are important for neuronal circuit analysis of underlying behaviour. The authors could embed this aspect in DANCE by integrating control of the green light LED strip used in this study using, for example, the open-source visual reactive programming software Bonsai (Lopes et al., 2015) and open-source electronics platform Arduino. This is an important and valuable addition in line with maintaining low cost.

We thank the reviewer for this valuable suggestion. DANCE was designed to be modular, allowing integration of temporal optogenetic control. To support immediate adoption, we now provide Arduino LED control code, setup schematics, and photographs (new Figure 7—Figure Supplement 1) along with step-by-step instructions on our GitHub page. We also note that Bonsai and Arduino frameworks are compatible with DANCE, enabling future extensions for closed-loop or behaviortriggered stimulation.

(2) Minor corrections to the text and figures:

Figure Supplement 1 refers only to Figure 2, yet panels D-F refer to the behaviour circling in courtship and therefore should be assigned to the respective figure.

Thanks, we have corrected this.

In lines 315-316, the cumbersome task of fluon coating for aggression assays seems to be ubiquitous across assays which is not the case, and therefore the sentence should include the word 'some'.

Thanks, we have edited this.

The cost of the phone and/or tablet should be included in the DANCE setup costs, as presumably these devices will be dedicated to the behavioural studies, for consistency purposes.

We thank the reviewer for this comment. We intentionally did not include smartphones or tablets in the setup cost because, in our experiments, these devices were not dedicated exclusively to DANCE but were repurposed from routine personal use. Our aim was to leverage readily available consumer electronics so that their cost does not become a barrier to adoption. We confirmed that commonly available Android phones capable of 30 fps at 1080p in H.264 format, as well as tablets or phones running a simple white-screen light app, are sufficient for reliable behavior classification and illumination. Since these devices can be returned to regular use after recordings, including their cost in the setup would not accurately reflect the intended accessibility of DANCE. For consistency, we now clarify in the Materials and Methods that such devices should be placed in airplane mode during recordings.

Reviewer #3 (Recommendations for the authors):

(1) For my taste, the authors put too much emphasis on the point that their method outperforms existing methods. I understand the value in comparing to published methods and it is of course fully justified to state the advantages of the new method. But the whole preprint is set up as a competition with the old algorithms, and the conclusion that the new classifier is better is repeated in each figure caption and after each paragraph of the results. This competitive mindset also extends to the selection of which results are presented as main figures and which as supplements - all cases in which the previous methods actually perform well are only presented in the supplement. I think this is simply unnecessary as the authors' results speak for themselves, and do not need the continuous competitive comparison.

We thank the reviewer for this thoughtful suggestion. Our intention was to benchmark DANCE rigorously against existing methods, not to frame the study competitively. We agree that repeated emphasis on relative performance was unnecessary. In the revised version, we streamlined figure captions and text throughout the manuscript to balance comparisons and removed redundant phrasing. Instances where other methods performed well are now presented with equal clarity to maintain a neutral and informative tone.

(2) When describing the DANCE hardware, as a reader I would find it interesting to also read about potential issues that the authors encountered. For example, how difficult is it to handle the materials without breaking or deforming them, which could affect the behavioral assays? How critical is it to use specific blister packs - the availability of which will likely vary strongly between countries? Did the authors try different sizes, and products? Such information, even as a supplement, could be very helpful for the widespread use of the hardware.

We thank the reviewer for this important point. To address this, we conducted additional tests comparing DANCE arenas of different diameters (new Figure 7— Figure Supplement 3A–C and new Figure 7—Figure Supplement 4A–L). We also consulted colleagues in multiple countries and verified that the blister packs used in our assays are readily available. The Materials and Methods now include practical handling notes: blister foils can be reused ~30–40 times for aggression assays and ~10–15 times for courtship assays before deformation. We also describe how to prevent agar surface damage during assembly and how to wash and dry the arenas for optimal reusability.

(3) I find the arrows pointing to several videos in a number of figures rather distracting and redundant, and suggest omitting them.

Thanks, we have omitted these arrows from all relevant figures and clarified the figure legends to enhance readability.

(4) P8, line 169 ff: this is a very long sentence that should be separated into several sentences.

We have rewritten this as follows: “DANCE scores remained comparable to groundtruth scores across all categories, whereas CADABRA and Divider underestimated the lunge counts (Figure 2B–E). Correlation analysis revealed a strong relationship between DANCE and ground-truth scores (Figure 2F, Supplementary File 2). In comparison, CADABRA and the Divider assay classifier showed a weaker correlation (Figure 2G-H, Supplementary File 2).”

(5) P10, line 216: please explain, here and in the methods, how these behavioral indices are calculated. I did not find this information anywhere in the paper.

We thank the reviewer for pointing this out. We now define the behavioral index explicitly in Materials and Methods: “For each assay, a behavioral index was calculated as the proportion of frames in which the male engaged in the specified behavior. This was obtained by dividing the total number of frames annotated for that behavior by the total number of frames in the recording.”

(6) P11, line 253: I don't understand the modifications to MateBook regarding attempted copulations, neither in the results nor the methods section. I would ask the authors to explain more explicitly what was done.

We thank the reviewer for this helpful suggestion. We have re-written several parts of the Materials and methods to clarify these details and streamline the text. To train the attempted copulation classifier, we combined datasets from assays with mated and decapitated virgin females, using manual annotations as ground truth. We also adapted MateBook’s persistence filters (Ribeiro et al., 2018) and defined thresholds explicitly: mounting lasting >45 s (>1350 frames at 30 fps) was defined as copulation, whereas abdominal curling without mounting, or mounting lasting 0.33– 45 s, was defined as attempted copulation.

(7) Figure 7F: this is the only case with a significant difference between the two setups. What explanations do the authors have for the discrepancy?

We thank the reviewer for raising this point. After repeating the experiments, we no longer found a significant difference between the setups. Figure 7 and its legend have been updated to reflect these results.

(8) Figure 2 - Supplement 1: I do not understand why the boxes for Observer 1 have different colors in different figures. Does this have a meaning?

Thanks for pointing this out. The color differences had no intended meaning, and we have corrected the figure for consistency across panels.

(9) P22, line 517ff: It would be interesting to know how frequently identity switches occurred. For large-scale, automatic behavioral screenings that step could be a crucial bottleneck.

We thank the reviewer for this valuable suggestion. We analyzed identity switches using the FlyTracker “Visualizer” package, which flags frames with possible overlaps or jumps. Flagged intervals were manually verified, and we report these data in new Supplementary File 5. Identity switch rates were very low: 0.66% for high-resolution recordings and 1.9% for smartphone DANCE videos in the most challenging decapitated-virgin dataset. These findings demonstrate robust tracking performance under both setups.

eLife Assessment

This important study presents a compelling theoretical framework for understanding condensation or phase separation of membrane-bound proteins, with a focus on the organization of tight junction components. By incorporating non-dilute binding effects into thermodynamic models and validating the model's predictions with in vitro experiments on the tight junction protein ZO-1, the authors provide a quantitative tool that combines theory and experiments and will help researchers in the field quantitatively interpret their findings. Given that phase separation of membrane bound molecules is becoming key in signaling, spanning from immune signaling to cell-cell adhesion, this work will be of broad interest for cell biologists and biophysicists.

Reviewer #1 (Public review):

Summary:

Biomolecular condensates are essential part of cellular homeostatic regulation. In this manuscript, authors develop a theoretical framework for phase separation of membrane bound proteins. They show the effect of non-dilute surface binding and phase separation on tight junction protein organization.

Strengths:

It is an important study considering the phase separation of membrane bound molecules are taking the center stage of signaling, spanning from immune signaling to cell-cell adhesion. A theoretical framework will help biologists to quantitatively interpret their findings.

Weaknesses:

Understandably, authors used one system to test their theory (ZO-1). However, to establish a theoretical framework, this is sufficient.

Comments on revisions:

I do not recommend new experiments. The manuscript is clear and establishes a new step in understanding the physical chemistry of biomolecular condensates.

Author response:

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

Reviewer #1 (Public review): 

Summary: 

Biomolecular condensates are an essential part of cellular homeostatic regulation. In this manuscript, the authors develop a theoretical framework for the phase separation of membrane-bound proteins. They show the effect of non-dilute surface binding and phase separation on tight junction protein organization. 

Strengths: 

It is an important study, considering that the phase separation of membrane-bound molecules is taking the center stage of signaling, spanning from immune signaling to cell-cell adhesion. A theoretical framework will help biologists to quantitatively interpret their findings. 

Weaknesses: 

Understandably, the authors used one system to test their theory (ZO-1). However, to establish a theoretical framework, this is sufficient. 

We acknowledge this limitation. While we agree that additional systems would strengthen the generality of our theory, we note that the focus of this work is to introduce and validate a theoretical framework. As the reviewer notes, this is sufficient for establishing the framework. Nonetheless, we are open to further collaborations or future studies to test the model with other systems.

Reviewer #2 (Public review): 

Summary: 

The authors present a clear expansion of biophysical (thermodynamic) theory regarding the binding of proteins to membrane-bound receptors, accounting for higher local concentration effects of the protein. To partially test the expanded theory, the authors perform in vitro experiments on the binding of ZO1 proteins to Claudin2 C-terminal receptors anchored to a supported lipid bilayer, and capture the effects that surface phase separation of ZO1 has on its adsorption to the membrane. 

Strengths: 

(1) The derived theoretical framework is consistent and largely well-explained. 

(2) The experimental and numerical methodologies are transparent. 

(3) The comparison between the best parameterized non-dilute theory is in reasonable agreement with experiments. 

Weaknesses: 

(1) In the theoretical section, what has previously been known, compared to which equations are new, should be made more clear. 

We have revised the theory section to clearly distinguish previously established formulations from novel contributions following equation (4), which is .

(2) Some assumptions in the model are made purely for convenience and without sufficient accompanying physical justification. E.g., the authors should justify, on physical grounds, why binding rate effects are/could be larger than the other fluxes. 

For our problem, binding is relevant together with diffusive transport in each phase. Each process is accompanied by kinetic coefficients that we estimate for the experimental system. For the considered biological systems (and related ones), it is difficult to determine whether other fluxes (see, e.g., Eq. 8(e)) have relaxed or not. We note that their effects are, of course, included in the kinetic model applied to the coarsening of ZO1 surface condensates as boundary conditions. But we cannot exclude that the corresponding kinetic coefficient in the actual biological system is large enough such that, e.g., Eq. (9e) does not vanish to zero “quasi-statically”. We have now added a sentence to the outlook highlighting the relevance of testing those flux-force relationships in biological systems. 

(3) I feel that further mechanistic explanation as to why bulk phase separation widens the regime of surface phase separation is warranted.  

We have discussed the mechanistic explanation related to bulk protein interaction strength in the manuscript in the section: “Effects of binding affinity and interactions on surface phase separation”. We explained how the bulk interaction parameter affects the binding equilibrium. 

(4) The major advantage of the non-dilute theory as compared with a best parameterized dilute (or homogenous) theory requires further clarification/evidence with respect to capturing the experimental data. 

We thank reviewer for this helpful question. To address this point, we have added new paragraphs in the conclusion section, which explicitly discuss the necessity of employing the non-dilute theory for interpreting the experimental data.

(5) Discrete (particle-based) molecular modelling could help to delineate the quantitative improvements that the non-dilute theory has over the previous state-of-the-art. Also, this could help test theoretical statements regarding the roles of bulk-phase separation, which were not explored experimentally.  

We appreciate the suggestion and agree that such modeling would be valuable. However, this is beyond the scope of the current study. 

(6) Discussion of the caveats and limitations of the theory and modelling is missing from the text. 

We sincerely appreciate the reviewer’s helpful comment. We have added a discussion in the conclusion section outlining the caveats and limitations of our modeling approach.

Reviewing Editor Comments: 

Upon discussing with the reviewers, we feel that this manuscript could significantly be improved if testing the model with a different model system (beyond ZO1/tight junctions), in which case we foresee that we could enhance the strength of evidence from "compelling" to "exceptional". But of course, this is up to the authors to go for it or not, the paper is already very good. 

Reviewer #2 (Recommendations for the authors): 

(1) Lines 132-134: Re-word, the use of "complex" is confusing.

We have rephrased the sentence for clarity. The revised version reads: ṽ_{_𝑃𝑅}_ are the molecular volume and area of the protein-receptor complex ѵ_𝑃𝑅, respectively”, and the changes have been in the revised manuscript.

(2) Line 154 use of ""\nu"" for volume and area could be avoided for better clarity. 

We thank the reviewer for this helpful suggestion. We have removed the statement involving ""\nu"" as these quantities have already been defined in the preceding context.

(3) Line 158 the total "Helmholtz" free energy F... 

We have added the word "Helmholtz" to the sentence.

(4) Line 160 typo "In specific,..." 

We carefully checked this sentence but could not identify a typo.  

(5) For equation 5 explain the physical origins of each term, or provide a reference if this equation is explained elsewhere. 

Thank you very much for your valuable suggestions. We have carefully rephrased Equation (5) and added a paragraph immediately afterward to provide a detailed explanation of its physical meaning.

(6) Derivation on lines 163-174 is poorly written. Make the logical flow between the equations clearer. 

We greatly appreciate your insightful suggestions. Equation (6) has been carefully revised for clarity, and the explanation has been rewritten to ensure better readability. All modifications are Done.

(7) Define bold "t" in Equation 6. 

The variable “t” has been explicitly defined in the context for clarity.

(8) In equations. 7b-7c the nablas (gradients) should be the 2D versions.  

We have updated the gradient operators in Equations (7b) and (7c) [Eq. (9) in revised manuscript]  to their 2D forms for consistency. 

(9) Line 190, avoid referring to the future Equation 14, and state in words what is meant by "thermodynamic equilibrium". 

We have added the explanation of “thermodynamic equilibrium” and remove the reference to equation accordingly.

(10) In Equation 11 you don't explain what you are doing ( which is a perturbation around the minimum of the free energy). 

We have revised the paragraph before equation (11) [Eq. (13) in revised manuscript] to clarify that the expression represents a perturbation around the minimum of the free energy.

(11)  In Equation 12, doesn't this also depend on how you have written equation 6 (not just equation 5). 

Eq. (12) [Eq. (14) in revised manuscript] is derived directly from the variation of the total free energy F. In contrast, Eq. (6) contains the time derivative of free energies that were not written in their final form. In the revised version, we have now given the conjugate forces and fluxes in Eqs. (7) and (8) for clarity.

(12) Line 206 specify the threshold of local concentration (or provide a reference). 

We have specified the threshold of local concentration in the revised text, and the corresponding statement has been highlighted.

(13) Line 223 is the deviation from ideality captured in a pair-wise fashion? I presume it does not account for N many-body interactions?  

Yes, our model is formulated within a mean-field framework that incorporates pairwise (second order) interaction coefficients. For example, 𝜒_{𝑃𝑅 -𝑅} characterizes the interaction between the complex 𝑃𝑅 and the free receptor 𝑅, 𝜒_{𝑅 -L} the interaction between free receptor 𝑅 and free lipid 𝐿, 𝜒_{𝑃𝑅-𝐿} the interaction between complex 𝑃𝑅and free lipid 𝐿. We have stressed this choice of free energy in the revised manuscript.

(14) Line 274, how do the authors know the secondary effects (of which they should mention a few) do not significantly impact the observed behaviour?  

We sincerely thank the reviewer for the helpful comment. First, the parameters 𝜒_{𝑅 -L} and 𝜒_{𝑃𝑅 -𝑅} are not essential based on the experimental observations. For more information, please see our revised paragraph on the choice of the specific parameter values, which has been in the following Eq. (21).

(15) It's not clear how Figures 3 b and c are generated with reference to which parameters are changed to investigate with/without bulk phase separation. 

To improve clarity, we have revised Figure 3 to display the corresponding parameter values directly in each panel. Figures 3b and 3c were generated by computing the surface binding curves (as shown in Fig. 2) for each binding affinity 𝜔_𝑃𝑅 and membrane-complex interaction strength 𝜒_{𝑃𝑅-𝐿}, under different bulk interaction strengths chi, to compare the cases with and without bulk phase separation. 

(16) The jump between theory and the "Mechanism in ..." section is too much. The authors should include the biological context of tight junctions and ZO1 in the main introduction. 

We appreciate the reviewer’s suggestion. Following this comment, we have added an extended discussion in the main introduction to provide the necessary biological context of tight junctions and ZO1. In addition, we inserted new bridging paragraphs between the theoretical section and the section “Mechanism in tight junction formation” to create a smoother transition from theory to experiments. These revisions help to better connect the theoretical framework with the biological phenomena discussed in the later section.

eLife Assessment

This important study shows that orientation tuning of V1 neurons is suppressed during a continuous flash suppression paradigm, especially when the neurons have a binocular receptive field. However, the evidence presented is incomplete and, in particular, does not distinguish whether this suppression is due to reduced contrast or due to masking.

Reviewer #1 (Public review):

Disclaimer: While I am familiar with the CFS method and the CFS literature, I am not familiar with primate research or two-photon calcium imaging. Additionally, I may be biased regarding unconscious processing under CFS, as I have extensively investigated this area but have found no compelling evidence in favor of unconscious processing under CFS.

This manuscript reports the results of a nonhuman-primate study (N=2 behaving macaque monkeys) investigating V1 responses under continuous flash suppression (CFS). The results show that CFS substantially suppressed V1 orientation responses, albeit slightly differently in the two monkeys. The authors conclude that CFS-suppressed orientation information "may not suffice for high-level visual and cognitive processing" (abstract).

The manuscript is clearly written and well-organized. The conclusions are supported by the data and analyses presented (but see disclaimer). However, I believe that the manuscript would benefit from a more detailed discussion of the different results observed for monkeys A and B (i.e., inter-individual differences), and how exactly the observed results are related to findings of higher-order cognitive processing under CFS, on the one hand, and the "dorsal-ventral CFS hypothesis", on the other hand.

Major Comments:

(1) Some references are imprecise. For example, l.53: "Nevertheless, two fMRI studies reported that V1 activity is either unaffected or only weakly affected (Watanabe et al., 2011; Yuval-Greenberg & Heeger, 2013)". "To the best of my understanding, the second study reaches a conclusion that is entirely opposite to that of the first, specifically that for low-contrast, invisible stimuli, stimulus-evoked fMRI BOLD activity in the early visual cortex (V1-V3) is statistically indistinguishable from activity observed during stimulus-absent (mask-only) trials. Therefore, high-level unconscious processing under CFS should not be possible if Yuval-Greenberg & Heeger are correct. The two studies contradict each other; they do not imply the same thing.

(2) Line 354: "The flashing masker was a circular white noise pattern with a diameter of 1.89{degree sign}{degree sign}, a contrast of 0.5, and a flickering rate of 10 Hz. The white noise consisted of randomly generated black and white blocks (0.07 × 0.07 each)." Why did the authors choose a white noise stimulus as the CFS mask? It has previously been shown that the depth of suppression engendered by CFS depends jointly on the spatiotemporal composition of the CFS and the stimulus it is competing with (Yang & Blake, 2012). For example, Hesselmann et al. (2016) compared Mondrian versus random dot masks using the probe detection technique (see Supplementary Figure S4 in the reference below) and found only a poor masking performance of the random dot masks.

Yang, E., & Blake, R. (2012). Deconstructing continuous flash suppression. Journal of Vision, 12(3), 8. https://doi.org/10.1167/12.3.8

Hesselmann, G., Darcy, N., Ludwig, K., & Sterzer, P. (2016). Priming in a shape task but not in a category task under continuous flash suppression. Journal of Vision, 16, 1-17.

(3) Related to my previous point: I guess we do not know whether the monkeys saw the CF-suppressed grating stimuli or not? Therefore, could it be that the differences between monkey A and B are due to a different individual visibility of the suppressed stimuli? Interocular suppression has been shown to be extremely variable between participants (see reference below). This inter-individual variability may, in fact, be one of the reasons why the CFS literature is so heterogeneous in terms of unconscious cognitive processing: due to the variability in interocular suppression, a significant amount of data is often excluded prior to analysis, leading to statistical inconsistencies. Moreover, the authors' main conclusion (lines 305-307) builds on the assumption that the stimuli were rendered invisible, but isn't this speculation without a measure of awareness?

Yamashiro, H., Yamamoto, H., Mano, H., Umeda, M., Higuchi, T., & Saiki, J. (2014). Activity in early visual areas predicts interindividual differences in binocular rivalry dynamics. Journal of Neurophysiology, 111(6), 1190-1202. https://doi.org/10.1152/jn.00509.2013

(4) The authors refer to the "tool priming" CFS studies by Almeida et al. (l.33, l.280, and elsewhere) and Sakuraba et al. (l.284). A thorough critique of this line of research can be found here:

Hesselmann, G., Darcy, N., Rothkirch, M., & Sterzer, P. (2018). Investigating Masked Priming Along the "Vision-for-Perception" and "Vision-for-Action" Dimensions of Unconscious Processing. Journal of Experimental Psychology. General. https://doi.org/10.1037/xge0000420

This line of research ("dorsal-ventral CFS hypothesis") has inspired a significant body of behavioral and fMRI/EEG studies (see reference for a review below). The manuscript would benefit from a brief paragraph in the discussion section that addresses how the observed results contribute to this area of research.

Ludwig, K., & Hesselmann, G. (2015). Weighing the evidence for a dorsal processing bias under continuous flash suppression. Consciousness and Cognition, 35, 251-259. https://doi.org/10.1016/j.concog.2014.12.010

Reviewer #2 (Public review):

Summary:

The goal of this study was to investigate the degree to which low-level stimulus features (i.e., grating orientation) are processed in V1 when stimuli are not consciously perceived under conditions of continuous flash suppression (CFS). The authors measured the activity of a population of V1 neurons at single neuron resolution in awake fixating monkeys while they viewed dichoptic stimuli that consisted of an oriented grating presented to one eye and a noise stimulus to the other eye. Under such conditions, the mask stimulus can prevent conscious perception of the grating stimulus. By measuring the activity of neurons (with Ca2+ imaging) that preferred one or the other eye, the authors tested the degree of orientation processing that occurs during CFS.

Strengths:

The greatest strength of this study is the spatial resolution of the measurement and the ability to quantify stimulus representations during CSF in populations of neurons, preferring the eye stimulated by either the grating or the mask. There have been a number of prominent fMRI studies of CFS, but all of them have had the limitation of pooling responses across neurons preferring either eye, effectively measuring the summed response across ocular dominance columns. The ability to isolate separate populations offers an exciting opportunity to study the precise neural mechanisms that give rise to CFS, and potentially provide insights into nonconscious stimulus processing.

Weaknesses:

While this is an impressive experimental setup, the major weakness of this study is that the experiments don't advance any theoretical account of why CFS occurs or what CFS implies for conscious visual perception. There are two broad camps of thinking with regard to CFS. On the one hand, Watanabe et al. (2011) reported that V1 activity remained intact during CFS, implying that CFS interrupts stimulus processing downstream of V1. On the other hand, Yuval-Greenberg and Heeger (2013) showed that V1 activity is, in fact, reduced during CFS. By using a parametric experimental design, they measured the impact of the mask on the stimulus response as a function of contrast and concluded that the mask reduces the gain of neural responses to the grating stimulus. They presented a theoretical model in which the mask effectively reduced the SNR of the grating, making it invisible in the same way that reducing contrast makes a stimulus invisible.

An important discussion point of Yuval-Greenberg and Heeger is that null results (such as those presented by Watanabe et al.) are difficult to interpret, as the lack of an effect may be simply due to insufficient data. I am afraid that this critique also applies to the present study. Here, the authors report that CFS effectively 'abolishes' tuning for stimuli in neurons preferring the eye with the grating stimulus. The authors would have been in a much stronger position to make this claim if they had varied the contrast of the stimulus to show that the loss of tuning was not simply due to masking. So, while this is an incredibly impressive set of measurements that in many ways raises the bar for in vivo Ca2+ imaging in behaving macaques, there isn't anything in the results that constitutes a real theoretical advance.

Reviewer #3 (Public review):

Summary:

In this study, Tang, Yu & colleagues investigate the impact of continuous flash suppression (CFS) on the responses of V1 neurons using 2-photon calcium imaging. The report that CFS substantially suppressed V1 orientation responses. This suppression happens in a graded fashion depending on the binocular preference of the neuron: neurons preferring the eye that was presented with the marker stimuli were most suppressed, while the neurons preferring the eye to which the grating stimuli were presented were least suppressed. The binocular neuron exhibited an intermediate level of suppression.

Strengths:

The imaging techniques are cutting-edge, and the imaging results are convincing and consistent across animals.

Weaknesses:

I am not totally convinced by the conclusions that the authors draw based on their machine learning models.

Author response:

Reviewer #2

We respectfully disagree with Reviewer 2’s critiques, upon which the eLife assessment of “incomplete evidence” is primarily based. We believe these critiques do not accurately reflect our study and are rooted in a misinterpretation of the evidence. Consequently, we suggest that the conclusion of “incomplete evidence” is not warranted.

On the basis of Reviewer 2’s critiques, the eLife assessment states: “However, the evidence presented is incomplete and, in particular, does not distinguish whether this suppression is due to reduced contrast or due to masking.” We emphasize that the suppression we observed is a consequence of interocular masking, not contrast reduction. Reviewer 2 cites Yuval-Greenberg and Heeger (2013), which proposes that during CFS, the mask reduces the gain of neural responses in V1 in a manner analogous to reducing stimulus contrast. We agree that both CFS masking and contrast reduction can decrease signal-to-noise ratio and thereby reduce visibility. However, in our paradigm, the physical stimulus contrast was held constant, while suppression was induced by interocular competition under CFS. This is a fundamentally different mechanism from lowering stimulus contrast. Our results therefore reflect genuine masking-induced suppression, rather than the effect of physical contrast reduction.

Furthermore, Reviewer 2 cited Yuval-Greenberg and Heeger’s discussion that null results can arise from insufficient data, and suggested that this applies to our study. This main critique from Reviewer 2 is misplaced for two reasons: First, our main result is not a null effect. A null effect would mean that CFS masking had no impact on population orientation responses. Instead, we observed significant suppression, including abolished tuning in some conditions, which clearly indicates a strong effect of masking. Second, our findings are based on large neural populations recorded using two-photon calcium imaging, providing extensive sampling and high statistical power. Thus, concerns about “insufficient data” do not apply to our study.

Finally, we used machine learning approaches to examine the effects of CFS masking on orientation discrimination and recognition, providing new insight into the long-standing debate over whether the brain can perform high-level cognitive processing under CFS. Although it is, to some extent, a matter of personal judgment whether our work represents a theoretical advance, Reviewer 2 made no comment, positive or negative, on this major component of our study while forming his/her judgment. (In response to Reviewer 3’s main concern about the suitability of SVMs, we now performed a multi-way classification analysis, which yielded results largely consistent with those obtained using the SVM approach in the original manuscript, confirming the robustness of our mechine learning results.

eLife Assessment

In this important paper, Garcia et al seek to determine whether the superior frontal sulcus (SFS), an area previously implicated in evidence accumulation for perceptual decisions, plays a causal role in perceptual and/or value-based decisions. Through a combination of careful paradigm design, computational modelling, transcranial magnetic stimulation and fMRI analyses, the authors provide convincing evidence that the SFS supports perceptual but not value-based decisions and that its disruption leads to a lowering of decision boundaries.

Reviewer #1 (Public review):

Summary:

In this study, participants completed two different tasks. A perceptual choice task in which they compared the sizes of pairs of items and a value-different task in which they identified the higher value option among pairs of items with the two tasks involving the same stimuli. Based on previous fMRI research, the authors sought to determine whether the superior frontal sulcus (SFS) is involved in both perceptual and value-based decisions or just one or the other. Initial fMRI analyses were devised to isolate brain regions that were activated for both types of choices and also regions that were unique to each. Transcranial magnetic stimulation was applied to the SFS in between fMRI sessions and it was found to lead to a significant decrease in accuracy and RT on the perceptual choice task but only a decrease in RT on the value-different task. Hierarchical drift diffusion modelling of the data indicated that the TMS had led to a lowering of decision boundaries in the perceptual task and a lowering of non-decision times on the value-based task. Additional analyses show that SFS covaries with model derived estimates of cumulative evidence, that this relationship is weakened by TMS.

The paper has many strengths including the rigorous multi-pronged approach of causal manipulation, fMRI and computational modelling which offers a fresh perspective on the neural drivers of decision making. Some additional strengths include the careful paradigm design which ensured that the two types of tasks were matched for their perceptual content while orthogonalizing trial-to-trial variations in choice difficulty. The paper also lays out a number of specific hypotheses at the outset regarding the behavioural outcomes that are tied to decision model parameters and well justified.

Reviewer #2 (Public review):

Summary:

The authors set out to test whether a TMS-induced reduction in excitability of the left Superior Frontal Sulcus influenced evidence integration in perceptual and value-based decisions. They directly compared behaviour-including fits to a computational decision process model---and fMRI pre and post TMS in one of each type of decision-making task. Their goal was to test domain-specific theories of the prefrontal cortex by examining whether the proposed role of the SFS in evidence integration was selective for perceptual but not value-based evidence.

Strengths:

The paper presents multiple credible sources of evidence for the role of the left SFS in perceptual decision making, finding similar mechanisms to prior literature and a nuanced discussion of where they diverge from prior findings. The value-based and perceptual decision making tasks were carefully matched in terms of stimulus display and motor response, making their comparison credible.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this study, participants completed two different tasks. A perceptual choice task in which they compared the sizes of pairs of items and a value-different task in which they identified the higher value option among pairs of items with the two tasks involving the same stimuli. Based on previous fMRI research, the authors sought to determine whether the superior frontal sulcus (SFS) is involved in both perceptual and value-based decisions or just one or the other. Initial fMRI analyses were devised to isolate brain regions that were activated for both types of choices and also regions that were unique to each. Transcranial magnetic stimulation was applied to the SFS in between fMRI sessions and it was found to lead to a significant decrease in accuracy and RT on the perceptual choice task but only a decrease in RT on the value-different task. Hierarchical drift diffusion modelling of the data indicated that the TMS had led to a lowering of decision boundaries in the perceptual task and a lower of nondecision times on the value-based task. Additional analyses show that SFS covaries with model derived estimates of cumulative evidence, that this relationship is weakened by TMS.

Strengths:

The paper has many strengths, including the rigorous multi-pronged approach of causal manipulation, fMRI and computational modelling, which offers a fresh perspective on the neural drivers of decision making. Some additional strengths include the careful paradigm design, which ensured that the two types of tasks were matched for their perceptual content while orthogonalizing trial-to-trial variations in choice difficulty. The paper also lays out a number of specific hypotheses at the outset regarding the behavioural outcomes that are tied to decision model parameters and well justified.

We thank the reviewer for their thoughtful summary of the study and for highlighting these strengths. We are pleased that the multi-pronged approach combining causal manipulation, fMRI, and hierarchical drift–diffusion modelling, as well as the careful matching of perceptual content across the two tasks, came across clearly. We also appreciate the reviewer’s positive remarks on the specificity of our a priori hypotheses and their links to decision-model parameters. In revising the manuscript, we have aimed to further streamline the presentation of these hypotheses and to more explicitly connect the behavioural predictions, model parameters, and neural readouts throughout the Results and Discussion sections.

Weaknesses:

In my previous comments (1.3.1 and 1.3.2) I noted that key results could be potentially explained by cTBS leading to faster perceptual decision making in both the perceptual and value-based tasks. The authors responded that if this were the case then we would expect either a reduction in NDT in both tasks or a reduction in decision boundaries in both tasks (whereas they observed a lowering of boundaries in the perceptual task and a shortening of NDT in the value task). I disagree with this statement. First, it is important to note that the perceptual decision that must be completed before the value-based choice process can even be initiated (i.e. the identification of the two stimuli) is no less trivial than that involved in the perceptual choice task (comparison of stimulus size). Given that the perceptual choice must be completed before the value comparison can begin, it would be expected that the model would capture any variations in RT due to the perceptual choice in the NDT parameter and not as the authors suggest in the bound or drift rate parameters since they are designed to account for the strength and final quantity of value evidence specifically. If, in fact, cTBS causes a general lowering of decision boundaries for perceptual decisions (and hence speeding of RTs) then it would be predicted that this would manifest as a short NDT in the value task model, which is what the authors see.

We thank the reviewer for raising these points and for the helpful clarification. We agree that, in principle, the architecture of the value-based task can be conceived as involving an upstream perceptual process that must be completed, to some degree, before value comparison can proceed. Under such a multistage framework, it is indeed possible that cTBS-induced changes in a perceptual decision stage could manifest as a reduction in boundary separation in the pure perceptual task, while the same perturbation appears as a shortening of non-decision time (NDT) when fitting a single-stage DDM to the value task. In this sense, our earlier statement that a “general speeding effect” would necessarily produce identical parameter changes (either NDT or boundaries) in both tasks was too strong, and we are grateful to the reviewer for pointing this out.

At the same time, this alternative explanation remains fully compatible with our central claim that the left SFS plays a perceptual rather than value-based role. We agree with the reviewer that there must be a stimulus-related circuit (in visual and parietal regions) that encodes the physical attributes of the options, and that this upstream processing can influence both tasks. However, a large body of work suggests that left SFS is not part of this primary identification circuitry, but rather contributes specifically to the accumulation and comparison of sensory evidence (e.g., Heekeren et al., 2004, 2006), downstream from areas such as FFA, PPA, or MT/V5 that encode stimulus identity. In other words, stimulus identification (forming a representation of “what is where”) is anatomically and functionally distinct from the accumulation of evidence toward a perceptual decision. Within this framework, the reviewer’s proposal that cTBS speeds “perceptual decisions” across tasks can be understood as targeting precisely the evidence-accumulation stage we ascribe to SFS, with the value-comparison stage proper likely implemented in other regions (e.g., vmPFC and connected valuation circuitry).

We therefore do not rely solely on the dissociation between boundary changes in the perceptual task and NDT changes in the value task as decisive evidence against a “general speeding” account. Instead, our interpretation is based on the convergence of behavioural, model-based, and neural results. First, in the perceptual task, cTBS to left SFS leads to a selective reduction in decision boundary and a concomitant change in trialwise BOLD activity within the stimulated region that covaries with perceptual choice behaviour and with the latent decision variable inferred from the HDDM. Second, in the value task, cTBS does not affect value sensitivity or accuracy, nor does it alter value-related drift or boundary parameters; the only robust HDDM effect is a modest shortening of NDT. Third, critically, left SFS BOLD activity is modulated by perceptual evidence and by cTBS in the perceptual task, but we observe no evidence that SFS activity encodes value evidence or shows value-related cTBS neuronal effects in the value task.

Taken together, these findings indicate that the left SFS serves a causal role in the accumulation of perceptual evidence and in the setting of the choice criterion for perceptual decisions. The reviewer’s suggestion that cTBS may induce a general speeding of perceptual processes that also influences the value task is compatible with this conclusion, in the sense that any contribution of SFS to the value task is best understood as acting via a perceptual component that is upstream of value comparison, rather than via the value accumulation process itself. We have clarified this point in the Discussion of the revised manuscript and now explicitly acknowledge that our DDM dissociation alone does not exclude a general perceptual speeding account, but that the combination of task-specific neural effects in SFS, preserved value-based choice behaviour, and the absence of value-related BOLD changes in SFS strongly support a primarily perceptual role for this region.

Reviewer #2 (Public review):

Summary:

The authors set out to test whether a TMS-induced reduction in excitability of the left Superior Frontal Sulcus influenced evidence integration in perceptual and value-based decisions. They directly compared behaviour-including fits to a computational decision process model---and fMRI pre and post TMS in one of each type of decision-making task. Their goal was to test domain-specific theories of the prefrontal cortex by examining whether the proposed role of the SFS in evidence integration was selective for perceptual but not value-based evidence.

Strengths:

The paper presents multiple credible sources of evidence for the role of the left SFS in perceptual decision making, finding similar mechanisms to prior literature and a nuanced discussion of where they diverge from prior findings. The value-based and perceptual decision-making tasks were carefully matched in terms of stimulus display and motor response, making their comparison credible.

We thank the reviewer for their clear summary of our aims and approach, and for highlighting these strengths. We are pleased that the convergence between causal TMS, fMRI, and hierarchical modelling comes across as providing credible evidence for the role of left SFS in perceptual decision-making, and that our attempt to link these results to the existing literature is seen as appropriately nuanced. We also appreciate the reviewer’s positive assessment of the task design, in particular the close matching of perceptual content and motor output across perceptual and value-based decisions, which was central to our goal of testing domain-specific theories of prefrontal function. In revising the manuscript, we have further clarified these design choices and their rationale, and we have streamlined the exposition of how the hypotheses, model parameters, and neural readouts are connected across the two decision domains.

Weaknesses:

I was confused about the model specification in terms of the relationship between evidence level and drift rate. While the methods (and e.g. supplementary figure 3) specify a linear relationship between evidence level and drift rate, suggesting, unless I misunderstood, that only a single drift rate parameter (kappa) is fit. However, the drift rate parameter estimates in the supplementary tables (and response to reviewers) do not scale linearly with evidence level.

We thank the reviewer for raising this point and appreciate the opportunity to clarify the model specification. In our hierarchical DDM, we did not fit separate, free drift parameters for each evidence level. As shown in Supplementary Fig. 3, the drift on each trial is specified as

where 𝐸_{𝑐,𝑠,𝑖} the trial-wise evidence (difference in size or value) and κ_𝑐,𝑠 is a single drift-scaling parameter per condition and session. Thus, the linear dependence of drift on evidence is implemented at the trial level via 𝜅; we do not estimate independent 𝛿 parameters for each evidence level.

In Supplementary Tables 8 and 9 we report, for descriptive purposes, the posterior means of 𝛿 conditional on each evidence bin (levels 1–4), alongside the corresponding decision boundary and nondecision time summaries. These values are therefore derived quantities that reflect the combination of (i) the single κ_𝑐,𝑠 parameter, (ii) the empirical distribution of continuous evidence values 𝐸 within each bin, and (iii) hierarchical pooling across subjects and sessions. Consequently, they are expected to increase monotonically with evidence level—as they do in our data—but not to lie exactly on a straight line in the discrete level index, because the underlying evidence bins are not equally spaced in physical units and because of between-subject variability and posterior uncertainty.

We will revise the text and table captions to make clear that the evidence-level entries are descriptive summaries of 𝛿 implied by the 𝜅×𝐸 formulation, rather than independently estimated drift parameters, in order to avoid this confusion.

-The fit quality for the value-based decision task is not as good as that for the PDM, and this would be worth commenting on in the paper.

We agree that the HDDM fit for the value-based task is somewhat weaker than for the perceptual task. This is reflected in the somewhat higher DIC values for VDM compared with PDM and in slightly broader posterior-predictive distributions (Supplementary Tables 8–11 and Supplementary Figs. 11–16). We believe this difference primarily reflects the greater intrinsic variability of subjective value-based choices (e.g. trial-to-trial fluctuations in preferences, satiety, or attention), coupled with our decision to use the same relatively simple DDM architecture for both tasks to allow a principled cross-task comparison. Importantly, posterior-predictive checks show that, for VDM as well, the model adequately reproduces both accuracy and full RT distributions at the group and subject level (Supplementary Figs. 11–16), indicating that the fit quality is sufficient for our purposes. In the revised manuscript we now explicitly note that the model captures PDM behaviour more tightly than VDM and that this may reduce sensitivity to very small cTBS effects on value-based decision parameters, even though no systematic effects are evident in our data. Crucially, our central conclusion—that left SFS plays a domain-specific role in setting the decision boundary for perceptual evidence—relies on the robust behavioural, computational, and neural effects observed in PDM and does not depend on assuming a perfect model fit for VDM.

- Supplementary Figure 3 specifies the distribution for kappa hyper-parameter twice.

We thank the reviewer for spotting this typo. We have revised Supplementary Figure 3 legend.

eLife Assessment

Combining state-of-the-art in-situ cell-surface proteomics, functional genetic screening, and single-nucleus RNA sequencing, this fundamental work substantially advances our understanding of glial contributions to organismal lifespan. The evidence supporting the conclusions is compelling, although additional clarification, control experiments, and analysis would further strengthen the study. The work will be of broad interest to researchers studying aging biology, glia-neuron communication, and in vivo proteomic profiling.

Reviewer #1 (Public review):

Summary:

Age-related synaptic dysfunction can have detrimental effects on cognitive and locomotor function. Additionally, aging makes the nervous system vulnerable to late-onset neurodegenerative diseases. This manuscript by Marques et al. seeks to profile the cell surface proteomes of glia to uncover signaling pathways that are implicated in age-related neurodegeneration. They compared the glial cell-surface proteomes in the central brain of young (day 5) and old (day 50) flies, and identified the most up- and down-regulated proteins during the aging process. 48 genes were selected for analysis in a lifespan screen, and interestingly, most sex-specific phenotypes. Among these, adult-specific pan-glial DIP-β overexpression (OE) significantly increased the lifespan of both males and females and improved their motor control ability. To investigate the effect of DIP-β in the aging brain, Marques et al. performed snRNA-seq on 50-day-old Drosophila brains with or without DIP-β OE in glia. Cortex and ensheathing glia showed the most differentially expressed genes. Computational analysis revealed that glial DIP-β OE increased cell-cell communication, particularly with neurons and fat cells.

Strengths:

(1) State-of-the-art methodology to reveal the cell surface proteomes of glia in young and old flies.

(2) Rigorous analyses to identify differentially expressed proteins.

(3) Examination of up- and down-regulated candidates and identification of glial-expressed mediators that impact fly lifespan.

(4) Intriguing sex-specific glial genes that regulate life span.

(5) Follow-up RNA-seq analysis to examine cellular transcriptomes upon overexpression of an identified candidate (DIP-β).

(6) A compelling dataset for the community that should generate extensive interest and spawn many projects.

Weaknesses:

(1) DIP-β OE using flySAM:

a) These flies showed a larger increase in lifespan compared to using UAS-DIP-β (Figure 2 C, D). Do the authors think that flySAM is a more efficient way of OE than UAS? Also, the UAS construct would be specific to one DIP-β isoform, while flySAM would likely express all isoforms. Could this also contribute to the phenotypes observed?

b) The Glial-GS>DIP-β flySAM flies without RU-486 have significantly shorter lifespans (Figure 2C) than their UAS-DIP-β counterparts. flySAM is lethal when expressed under the control of tubulin-GAL4 (Jia et al. 2018), likely due tothe toxicity of such high levels of overexpression. Is it possible that a larger increase in lifespan is due to the already reduced viability of these flies?

c) Statistics: It is stated in the Methods that "statistical methods used are described in the figure legend of each relevant panel." However, there is no description of the statistics or sample sizes used in Figure 2.

(2) Figure 3: The authors use a glial GeneSwitch (GS) to knock down and overexpress candidate genes. In Figure 3A, they look at glial-GS>UAS-GFP with and without RU. Without RU, there is no GFP expression, as expected. With RU, there is GFP expression. It is expected that all cell body GFP signal should colocalize with a glial nuclear marker (Repo). However, there is some signal that does not appear to be glia. Also, many glia do not express GFP, suggesting the glial GS driver does not label all glia. This could impact which glia are being targeted in several experiments.

(3) It is interesting that sex-specific lifespan effects were observed in the candidate screen.

a) The authors should provide a discussion about these sex-specific differences and their thoughts about why these were observed.

b) The authors should also provide information regarding the sex of the flies used in the glial cell surface proteome study.

c) Also, beyond the scope of this study, examining sex-specific glial proteomes could reveal additional insights into age-related pathways affecting males and females differentially.

(4) The behavioral assay used in this study (climbing) tests locomotion driven by motor neurons. The proteomic analysis was performed with the central adult brain, which does not include the nerve cord, where motor neurons reside. While likely beyond the scope of this study, it would be informative to test other behaviors, including learning, circadian rhythms, etc.

(5) It is surprising that overexpressing a CAM in glia has such a broad impact on the transcriptomes of so many different cell types. Could this be due to DIP-β OE maintaining the brain in a "younger" state and indirectly influencing the transcriptomes? Instead of DIP-β OE in glia directly influencing cell-cell interactions? Can the authors comment on this?

Reviewer #2 (Public review):

This manuscript presents an ambitious and technically innovative study that combines in situ cell-surface proteomics, functional genetic screening, and single-nucleus RNA sequencing to uncover glial factors that influence aging in Drosophila. The authors identify DIP-β as a glial protein whose overexpression extends lifespan and report intriguing sex-specific differences in lifespan outcomes. Overall, the study is conceptually compelling and offers a valuable dataset that will be of considerable interest to researchers studying glia-neuron communication, aging biology, and proteomic profiling in vivo.

The in-situ proteomic labeling approach represents a notable methodological advance. If validated more extensively, it has the potential to become a widely used resource for probing glial aging mechanisms. The use of an inducible glial GeneSwitch driver is another strength, enabling the authors to carefully separate aging-relevant effects from developmental confounds. These technical choices meaningfully elevate the rigor of the study and support its central conclusions. The discovery of new candidate genes from the proteomics pipeline, including DIP-β, is intriguing and opens new avenues for understanding glial contributions to organismal lifespan. The observation of sex-specific lifespan effects is particularly interesting and warrants further exploration; the study sets the stage for future work in this direction.

At the same time, several areas would benefit from clarification or additional analysis to fully support the manuscript's claims:

(1) The manuscript frequently refers to "improved" or "increased" cell-cell communication following DIP-β overexpression, but the meaning of this term remains somewhat vague. Because the current analysis relies largely on transcriptomic predictions, it would be helpful to define precisely what metric is being used, e.g., increased numbers of predicted ligand-receptor interactions, enrichment of specific signaling pathways, or altered expression of communication-related components. Strengthening the mechanistic link between DIP-β, cell-cell communication, and lifespan extension, potentially through targeted validation of specific glial interactions, would substantially reinforce the interpretation.

(2) The lifespan screen is central to the paper, and clearer visualization and contextualization of these results would significantly improve the manuscript's impact. For example, Figure 3D is challenging to interpret in its current form. More explicit presentation of which manipulations extend lifespan in each sex, along with effect sizes and significance values, would provide clarity. Including positive controls for lifespan extension would also help contextualize the magnitude of the observed effects. The reported effects of DIP-β, while promising, are modest relative to baseline effects of RU feeding, and a discussion of this would help appropriately calibrate the conclusions.

(3) Several figures would benefit from improved labeling or more detailed legends. For instance, the meaning of "N" and "C" in Figure 1D is unclear; Figure 3A should clarify that Repo is a glial marker; and Figure 5C appears to have truncated labels. Reordering certain panels (e.g., moving control data in Figure 4A-B) may also improve narrative flow. These refinements would greatly aid reader comprehension.

(4) A few claims would be strengthened by more specific references or acknowledgment of alternative interpretations. Examples include the phenoxy-radical labeling radius, the impact of H₂O₂ exposure, and the specificity of neutravidin. Additionally, downregulation of synapse-related GO terms may reflect age-related transcriptional changes rather than impaired glia-neuron communication per se, and this possibility should be recognized. The term "unbiased" to describe the screen may also be reconsidered, given the preselection of candidate genes.

(5) Clarifying the rationale for focusing on central brain glia over optic-lobe glia would be useful.

eLife Assessment

This important study presents novel data on temporal variation in sperm whale communication, contributing to a richer understanding of the social transmission of vocal styles across neighbouring clans. The evidence is solid, although some terminology limits comparisons to other taxa. This research will be of interest to bioacoustics and cetacean communication specialists, particularly those working on social learning and culture.

Reviewer #2 (Public review):

Summary:

The current article adapts standard rhythmic measures to describe the temporal organisation of whale song units.

Strengths:

The detailed description of the internal temporal structure of whale songs is something that has thus far been lacking.

Weaknesses:

Conceptual and terminological bases of the paper are problematical and hamper comparison with other taxa, including humans. According to signal theory, codas are indexical rather than symbolic. They signal an individual's group identity. Borrowing from humans and linguistics, coda inter-group variation represents a case of accents -- phonologically different varieties of the same call -- not dialects, confirming they are an index. Moreover, symbolism is not a feature detectable or confirmed through rhythmic analyses or temporal characterisation. This raises serious doubt whether alleged "dialects," "symbolism" and similarity between whales and humans is factual. The comparative scope and relevance of this paper for the broader field is limited and evolutionary claims are potentially misleading and perilous.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

Reviewer #2 (Public review):

Summary:

The current article presents a new type of analytical approach to the sequential organisation of whale song units.

Strengths:

The detailed description of the internal temporal structure of whale songs is something that has been thus far lacking.

Weaknesses:

The conceptual and terminological bases of the paper are problematical and hamper comparison with other taxa, including humans. According to signal theory, codas are indexical rather than symbolic. They signal an individual's group identity. Borrowing from humans and linguistics, coda inter-group variation represents a case of accents - phonologically different varieties of the same call - not dialects, confirming they are an index. This raises serious doubt about whether alleged "symbolism" and similarity between whale and human vocal behaviour is factual.

We respect that the reviewer does not agree with describing codas as symbolic markers of cultural identity in sperm whales, but ultimately we find the quantitative evidence presented in Hersh et al. (2022) compelling, and stand by the framing of our manuscript, which builds on this foundation.

The same applies to the difference between ICIs (inter-click interval) and IOIs (inter-onset interval). If the two are equivalent, variation in click duration needs to be shown so small that can be considered negligible. This raises serious doubt about whether the alleged variation in whale codas is indeed rhythmic in nature and prevents future efforts for comparison with the vocal capacities of other species. The scope and relevance of this paper for the broader field is limited.

We believe there has been a miscommunication. Coda inter-click intervals are calculated as the time between the onsets of sequential clicks within a coda. This is identical to definitions of inter-onset intervals in many publications, including:

• Burchardt and Knörnschild (2020): “the duration between the beginning of one element and the next”

• Friberg and Battel (2002): “the time interval between the onset of the tone and the onset of the immediately following tone”

• De Gregorio et al. (2021): “the time between the onset of a note and the next one”

In response to a comment from this reviewer in the first round of revisions, we made the point that we do not believe rhythm analyses need be restricted to inter-onset intervals alone. Regardless of that stance, we did analyze inter-onset intervals in this manuscript and accordingly are capturing aspects of rhythm in our analyses. We have removed a poorly worded sentence in our introduction and apologize for any confusion it caused. We have also made this explicit in lines 30–35: “This classification is based on the total number of clicks and their rhythm and tempo extrapolated from the time interval between the onsets of consecutive clicks: the inter-click interval (ICI) [15, 16] (Fig. 1A). This measure is equivalent to the inter-onset intervals (IOIs) often used in rhythm analyses [17, 18, 19] but for the sake of compatibility with studies on sperm whale acoustics, we use ICI terminology throughout this paper.”

In our analyses, inter-click intervals and inter-onset intervals are equivalent measures.

Recommendations for the authors:

Reviewer #2 (Recommendations for the authors):

My concerns regarding interdisciplinary terminology and methods remain unaddressed. The study's inaccurate terminology hinders reliable comparison with other taxa, including humans. Being "symbolic" bears no weight on the new method that the authors present, thus, the unwillingness for compatibility is limiting and perplexing. The authors state that codas have been previously described as being symbolic, but just because poor terminology has been used before doesn't justify perpetuating it, especially when it confounds and conflicts with broader comparative efforts.

We agree that being symbolic bears no weight on the new method we present, but we believe it does bear weight on our interpretation of what our method reveals about patterns in sperm whale communication. For that reason, we have opted to maintain the current framing of our manuscript.

The same applies to the difference between ICIs and IOIs. The authors resist amending terminology, even though they state the two represent the same measure. If so, want prevents the correct use of IOIs?

We have opted to use ICI throughout the paper because it is standard terminology in sperm whale acoustics, but we have now made the ICI/IOI equivalence explicitly clear in the introduction.

References:

Burchardt LS, Knörnschild M. 2020. Comparison of methods for rhythm analysis of complex animals’ acoustic signals. PLoS Computational Biology 16. doi:10.1371/journal.pcbi.1007755

De Gregorio C, Valente D, Raimondi T, Torti V, Miaretsoa L, Friard O, Giacoma C, Ravignani A, Gamba M. 2021. Categorical rhythms in a singing primate. Current Biology 31:R1379–R1380. doi:10.1016/j.cub.2021.09.032

Friberg A, Battel GU. 2002. Structural communication In: Parncutt R, McPherson G, editors. The Science & Psychology of Music Performance: Creative Strategies for Teaching and Learning. Oxford University Press. doi:10.1093/acprof:oso/9780195138108.001.0001

Hersh TA, Gero S, Rendell L, Cantor M, Weilgart L, Amano M, Dawson SM, Slooten E, Johnson CM, Kerr I, Payne R, Rogan A, Andrews O, Ferguson EL, Hom-Weaver CA, Norris TF, Barkley YM, Merkens KP, Oleson EM, Doniol-Valcroze T, Pilkington J, Gordon J, Fernandes M, Guerra M, Hickmott L, Whitehead H. 2022. Evidence from sperm whale clans of symbolic marking in non-human cultures. Proceedings of the National Academy of Sciences 119:e2201692119. doi:10.1073/pnas.2201692119

eLife Assessment

This study makes a valuable contribution by elucidating the genetic determinants of growth and fitness across multiple clinical strains of Mycobacterium intracellulare, an understudied non-tuberculous mycobacterium. Using transposon sequencing (Tn-seq), the authors identify a core set of 131 genes essential for bacterial adaptation to hypoxia, providing a convincing foundation for anti-mycobacterial drug discovery.

Reviewer #1 (Public review):

Summary:

In this descriptive study, Tateishi et al. report a Tn-seq based analysis of genetic requirements for growth and fitness in 8 clinical strains of Mycobacterium intracellulare Mi), and compare the findings with a type strain ATCC13950. The study finds a core set of 131 genes that are essential in all nine strains, and therefore are reasonably argued as potential drug targets. Multiple other genes required for fitness in clinical isolates have been found to be important for hypoxic growth in the type strain.

Strengths:

The study has generated a large volume of Tn-seq datasets of multiple clinical strains of Mi from multiple growth conditions, including from mouse lungs. The dataset can serve as an important resource for future studies on Mi, which despite being clinically significant, remains a relatively understudied species of mycobacteria.

Weaknesses:

The primary claim of the study that the clinical strains are better adapted for hypoxic growth is yet to be comprehensively investigated. However, this reviewer thinks such an investigation would require a complex experimental design and perhaps form an independent study.

Comments on revisions:

The revised paper has satisfactorily addressed my previous concerns, and I have no further issues with this paper.

Author response:

The following is the authors’ response to the previous reviews

Reviewer #1 (Public review) :

Comments on revisions:

The revised manuscript has responded to the previous concerns of the reviewers, albeit modestly. The overemphasis on hypoxic adaptation of the clinical isolates persist as a key concern in the paper. The authors have compared the growth-curve of each of the clinical and ATCC strains under normal and hypoxic conditions (Fig. 8), but don't show how mutations in some of the genes identified in Tn-seq would impact the growth phenotype under hypoxia. They largely base their arguments on previously published results.

As I mentioned previously, the paper will be better without over-interpreting the TnSeq data in the context of hypoxia.

Thank you for the comment on the issue of not determining the impact of individual gene mutations identified in TnSeq on the growth phenotypes under hypoxia.

We agree that the lack of validation of TnSeq results is a limitation of this study. Without evidence of growth pattern of each gene-deletion mutant under hypoxia there might be a risk of over-interpretating the data, even though the data are carefully interpreted based on previous reports. We consider that it is necessary to confirm the phenomenon by using knockout mutants.

We have just recently succeeded in constructing the vector plasmids for making knockout mutants of M intracellulare (Tateishi. Microbiol Immunol. 2024). We will proceed to the validation experiment of TnSeq-hit genes by constructing knockout mutants. We already mentioned this point as a limitation of this study in the Discussion (pages 35-36 lines 630-640 in the revised manuscript).

Reference.

Tateishi, Y., Nishiyama, A., Ozeki, Y. & Matsumoto, S. Construction of knockout mutants in Mycobacterium intracellulare ATCC13950 strain using a thermosensitive plasmid containing negative selection marker rpsL+. Microbiol Immunol 68, 339-347 (2024).

Other points:

The y-axis legends of plots in Fig.8c are illegible.

Following the comment, we have corrected Figure 8c and checked the uploaded PDF

The statements in lines 376-389 are convoluted and need some explanation. If the clinical strains enter the log phase sooner than ATCC strain under hypoxia, then how come their growth rates (fig. 8c) are lower? Aren't they expected to grow faster?

Thank you for the comment on the interpretation of the difference in bacterial growth under hypoxia between MAC-PD strains and the ATCC type strain. The growth curve consists of the onset of logarithmic growth and its growth speed. In this study, we evaluated the former as timing of midpoint and the latter as growth rate at midpoint. Timing of midpoint and growth rate at midpoint are individual parameters. The early entry to log-phase does not mean the fast growth rate at midpoint.

Our results demonstrated that 5 (M.i.198, M.i.27, M003, M019 and M021) out of 8 clinical MAC-PD strains entered log-phase early and continued to grow logarithmically long time (slow growth). This data suggests the capacity for MAC-PD to continue replication long time under hypoxic conditions. By contrast, the ATCC type strain showed delayed onset of logarithmic growth caused by long-term lag phase. The duration of logarithmic growth was short even once after it started. The log phase soon transited to the stationary phase. This data suggests the lower capacity for the ATCC strain to continue replication under hypoxic conditions.

Following the comment, we have added the interpretation of the growth curve pattern as follows (page 22 lines 379-392 in the revised manuscript): “The growth rate at midpoint under hypoxic conditions was significantly lower in these 5 clinical MAC-PD strains than in ATCC13950. The early entry to log phase followed by long-term logarithmic growth (slow growth rate at midpoint) suggests the capacity for these 5 clinical MAC-PD strains to continue replication long time under hypoxic conditions. On the other hand, the rest 3 clinical MAC-PD strains (M018, M001 and MOTT64) did not show significant change in the growth rate between aerobic and hypoxic conditions, suggesting that there are different levels of capacity in maintaining long-term replication under hypoxia among clinical MAC-PD strains. In ATCC13950, the entry to log phase was significantly delayed under 5% oxygen compared to aerobic conditions, and the growth rate at midpoint was significantly increased under hypoxic conditions compared to aerobic conditions in ATCC13950. Such long-term lag phase followed by short-term log phase suggests lower capacity for ATCC13950 to continue replication under hypoxic conditions compared to clinical MAC-PD strains.”

Reviewer #4 (Public review):

Comments on revisions:

The revised version has satisfactorily addressed my initial comments in the discussion section.

The authors thank the Reviewer for understanding our reply.

Reviewer #5 (Public review):

Comments on revisions:

There is quite a lot of data and this could have been a really impactful study if the authors had channelized the Tn mutagenesis by focusing on one pathway or network. It looks scattered. However, from the previous version, the authors have made significant improvements to the manuscript and have provided comments that fairly address my questions.

The authors thank the Reviewer for understanding our reply. And the authors thank the Reviewer for the comments suggesting the future studies of TnSeq that focus on one pathway or network.

eLife Assessment

This is an important study that utilized in vivo optical measurements of the cortical metabolic rate of O2 and blood flow, as well as measurements in isolated mitochondria to assess the uncoupling of the oxidative phosphorylation due to hypoxia-ischemia injury of the neonatal brain, and effects of the hypothermia treatment. The combination of state-of-the-art optical measurements, mitochondrial assays, and the use of various control experiments provides convincing evidence for the derived conclusions. This work will be of interest to those in the mitochrondrial metabolomics, brain injury and hypoxia fields.

Reviewer #1 (Public review):

Summary:

This manuscript addresses the important problem of the uncoupling of oxidative phosphorylation due to hypoxia-ischemia injury in the neonatal brain and provides insight into the neuroprotective mechanisms of hypothermia treatment.

Strengths:

The authors used a combination of in vivo imaging of awake P10 mice and experiments on isolated mitochondria to assess various key parameters of brain metabolism during hypoxia-ischemia with and without hypothermia treatment. This unique approach resulted in a comprehensive data set that provides solid evidence to support the derived conclusions.

Weaknesses:

Several potential weaknesses were identified in the original submission, which the authors subsequently addressed in the revised manuscript. Here is the brief list of the questions:

(1) Is it possible that the observed relatively low baseline OEF and trends of increased OEF and CBF over several hours after the imaging start were partially due to slow recovery from anesthesia?

(2) What was the pain management, and is there a possibility that some of the observations were influenced by the pain-reducing drugs or their absence?

(3) Were P10 mice significantly stressed during imaging in the awake state because they didn't have head-restraint habituation training?

(4) Considering high metabolism and blood flow in the cortex, it could be potentially challenging to predict cortical temperature based on the skull temperature, particularly in the deeper part of the cortex.

(5) The map of estimated CMRO2 looks quite heterogeneous across the brain surface. Could this be partially resulting from the measurement artefact?

(6) It would be beneficial to provide more detailed justification for using P10 mice in the experiments.

Reviewer #3 (Public review):

Sun et al. present a comprehensive study using a novel photoacoustic microscopy setup and mitochondrial analysis to investigate the impact of hypoxia-ischemia (HI) on brain metabolism and the protective role of therapeutic hypothermia. The authors elegantly demonstrate three connected findings: (1) HI initially suppresses brain metabolism, (2) subsequently triggers a metabolic surge linked to oxidative phosphorylation uncoupling and brain damage, and (3) therapeutic hypothermia mitigates HI-induced damage by blocking this surge and reducing mitochondrial stress.

The study's design and execution are great, with a clear presentation of results and methods. Data is nicely presented, and methodological details are thorough.

However, a minor concern is the extensive use of abbreviations, which can hinder readability. As all the abbreviations are introduced in the text, their overuse may render the text hard to read to non-specialist audiences. Additionally, sharing the custom Matlab and other software scripts online, particularly those used for blood vessel segmentation, would be a valuable resource for the scientific community. In addition, while the study focuses on the short-term effects of HI, exploring the long-term consequences and definitively elucidating HI's impact on mitochondria would further strengthen the manuscript's impact.

Despite these minor points, this manuscript is very interesting.

Comments on revisions:

All addressed.

Author response:

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

Reviewer #1 (Public review)

(1) This manuscript addresses an important problem of the uncoupling of oxidative phosphorylation due to hypoxia-ischemia injury of the neonatal brain and provides insight into the neuroprotective mechanisms of hypothermia treatment.

The authors used a combination of in vivo imaging of awake P10 mice and experiments on isolated mitochondria to assess various key parameters of the brain metabolism during hypoxia-ischemia with and without hypothermia treatment. This unique approach resulted in a comprehensive data set that provides solid evidence for the derived conclusions

We thank the reviewer for the positive feedback.

(2) The experiments were performed acutely on the same day when the surgery was performed. There is a possibility that the physiology of mice at the time of imaging was still affected by the previously applied anesthesia. This is particularly of concern since the duration of anesthesia was relatively long. Is it possible that the observed relatively low baseline OEF (~20%) and trends of increased OEF and CBF over several hours after the imaging start were partially due to slow recovery from prolonged anesthesia? The potential effects of long exposure to anesthesia before imaging experiments were not discussed.

We thank the reviewer for this important comment and for pointing out the potential influence of anesthesia on the physiological state of the animals. We apologize for any confusion. To clarify, all PAM imaging experiments were conducted in awake animals. Isoflurane anesthesia was used only during two brief surgical procedures: (1) the installation of the head-restraint plastic head plate and (2) the right common carotid artery (CCA) ligation. Each anesthesia session lasted less than 20 minutes.

We have revised the Methods section to provide additional details:

For the subsection Procedures for PAM Imaging on page 17, we clarified the sequence of procedures during the head plate installation, as well as the corresponding anesthesia duration:

“After the applied glue was solidified (~20 min), the animal was first returned to its cage for full recovery from anesthesia, and then carefully moved to the treadmill and secured to the metal arm-piece with two #4–40 screws for awake PAM imaging. The total duration of anesthesia, including preparation and glue solidification, was approximately 20 minutes.”

For the subsection Neonatal Cerebral HI and Hypothermia Treatment on page 19, we also clarified the CCA ligation procedure:

“Briefly, P10 mice of both sexes anesthetized with 2% isoflurane were subjected to the right CCA-ligation. To manage pain, 0.25% Bupivacaine was administered locally prior to the surgical procedures, which took less than 10 minutes. After a recovery period for one hour, awake mice were exposed to 10% O₂ for 40 minutes in a hypoxic chamber at 37 °C.”

Regarding the reviewer’s concern about the observed trends in OEF and CBF, we agree that residual effects of anesthesia could, in principle, influence physiological parameters. However, we believe this is unlikely in this study for the following reasons. First, all imaging was conducted in awake animals after a clearly defined recovery period. Second, the trend of increasing OEF and CBF over time was consistent across animals and aligned with expected physiological responses following hypoxic-ischemic injury. In particular, the relatively low baseline OEF (0.21 at 37°C) is consistent with our previous study (0.25; (Cao et al., 2018)). The gradual increase in CBF and OEF reflects metabolic compensation and reperfusion following hypoxia-ischemia, as previously described (Lin and Powers, 2018). Therefore, we believe the observed changes are of physiological origin rather than anesthesia-related artifacts.

(3) The Methods Section does not provide information about drugs administered to reduce the pain. If pain was not managed, mice could be experiencing significant pain during experiments in the awake state after the surgery. Since the imaging sessions were long (my impression based on information from the manuscript is that imaging sessions were ~4 hours long or even longer), the level of pain was also likely to change during the experiments. It was not discussed how significant and potentially evolving pain during imaging sessions could have affected the measurements (e.g., blood flow and CMRO₂). If mice received pain management during experiments, then it was not discussed if there are known effects of used drugs on CBF, CMRO₂, and lesion size after 24 hr.

We thank the reviewer for this valuable comment regarding pain management. We confirm that local analgesia was administered to all animals prior to surgical procedures. Specifically, 0.25% Bupivacaine was applied locally before both the head-restraint plate installation and the CCA ligation. These details have now been clarified in the Methods section:

For the subsection Procedures for PAM Imaging on page 16, we added:

“To manage pain, 0.25% Bupivacaine was administered locally prior to the surgical procedures.”

For the subsection Neonatal Cerebral HI and Hypothermia Treatment on page 18, we added:

“To manage pain, 0.25% Bupivacaine was administered locally prior to the surgical procedures, which took less than 10 minutes.”

To our knowledge, Bupivacaine has minimal systemic effects at the dose used and is unlikely to significantly alter CBF, CMRO₂, or lesion development (Greenberg et al., 1998). No other analgesics (e.g., NSAIDs or opioids) were administered unless distress symptoms were observed—which did not occur in this study.

Additionally, although imaging sessions were extended (up to 2 hours), animals remained calm and showed no signs of pain or distress during or after the procedures. Throughout the experimental period (up to 24 hours post-surgery), animals were monitored for signs of discomfort (e.g., abnormal activity, breathing, or weight gain), but no additional analgesia was required. The neonatal HI procedures are considered minimally invasive, and based on our protocol and prior experience, local Bupivacaine provides effective analgesia during and after the brief surgeries. We have added a corresponding note in the Discussion section (newly added subsection: Limitations in this study, the last paragraph) on page 15:

“We observed no signs of distress or pain and did not use stress- or pain-reducing drugs during imaging. However, potential effects of stress or residual pain on CBF and CMRO₂ cannot be fully ruled out. Future studies could incorporate more detailed pain assessment and stress-mitigation strategies to further enhance physiological reliability.”

(4) Animals were imaged in the awake state, but they were not previously trained for the imaging procedure with head restraint. Did animals receive any drugs to reduce stress? Our experience with well-trained young-adult as well as old mice is that they can typically endure 2 and sometimes up to 3 hours of head-restrained awake imaging with intermittent breaks for receiving the rewards before showing signs of anxiety. We do not have experience with imaging P10 mice in the awake state. Is it possible that P10 mice were significantly stressed during imaging and that their stress level changed during the imaging session? This concern about the potential effects of stress on the various measured parameters was not discussed.

We thank the reviewer for this important comment regarding the potential effects of stress during awake imaging. The neonatal mice used in our study were P10, a stage at which animals are still physiologically immature and relatively inactive. Due to their small size and limited mobility, these animals did not struggle or show signs of distress during the imaging sessions. All animals remained calm and stable throughout the procedure, and no stress-reducing drugs were administered.

We agree that, unlike older animals, P10 mice are not amenable to prior behavioral training. However, their underdeveloped motor activity and natural docility at this stage allowed for stable head-restrained imaging without inducing overt stress responses. Although no behavioral signs of stress were observed, we acknowledge that subtle physiological effects cannot be entirely excluded. We have added a brief discussion in the Discussion section (newly added subsection: Limitations in this study, the last paragraph) on page 15:

“Lastly, for awake imaging, the small size of neonatal mice at P10 aids stability during awake PAM imaging, though it limits the feasibility of prior training, which is typically possible in older animals.”

(5) The temperature of the skull was measured during the hypothermia experiment by lowering the water temperature in the water bath above the animal's head. Considering high metabolism and blood flow in the cortex, it could be challenging to predict cortical temperature based on the skull temperature, particularly in the deeper part of the cortex.

We thank the reviewer for this helpful comment and for highlighting an important technical consideration. We acknowledge that we did not directly measure intracortical tissue temperature during the hypothermia experiments. While we recognize that relying on skull temperature may have limitations—particularly in reflecting temperature changes in deeper cortical regions—this approach is consistent with clinical practice, where intracortical temperature is typically not measured. Moreover, prior studies have shown that skull or brain surface temperature generally reflects cortical thermal dynamics to a reasonable extent under controlled conditions (Kiyatkin, 2007). We have added the following note in the Discussion section (newly added subsection: Limitations in this study, the 2^nd paragraph) on page 14:

“A technical limitation is the absence of direct intracortical temperature measurements during hypothermia; we relied on skull temperature, which may not fully capture temperature dynamics in deeper cortical layers. However, this approach aligns with clinical practice, where intracortical temperature is not typically measured. Future studies could benefit from more precise intracortical assessments.”

(6) The map of estimated CMRO₂ (Fig. 4B) looks very heterogeneous across the brain surface. Is it a coincidence that the highest CMRO₂ is observed within the central part of the field of view? Is there previous evidence that CMRO₂ in these parts of the mouse cortex could vary a few folds over a 1-2 mm distance?

We appreciate the reviewer’s insightful observation regarding the spatial heterogeneity observed in the estimated CMRO₂ map (Fig. 4B). This heterogeneity is not a result of scanning bias, as uniform contour scanning was performed across the entire field of view. The higher CMRO₂ values observed in the central region are unlikely to be artifacts and more likely reflect underlying physiological variability.

Our CMRO₂ estimation is based on an algorithm we previously developed and validated in other tissues. Specifically, we have successfully applied this algorithm to assess oxygen metabolism in the mouse kidney (Sun et al., 2021) and to monitor vascular adaptation and tissue oxygen metabolism during cutaneous wound healing (Sun et al., 2022). These studies demonstrated the algorithm's capability to capture spatial variations in oxygen metabolism. Although the current application to the brain is novel, the algorithm has been validated in controlled experimental settings and shown to produce consistent results. We acknowledge that the observed range of CMRO₂ appears relatively broad across a 1–2 mm distance; however, such heterogeneity may arise from local differences in vascular density, metabolic demand, or tissue oxygenation — all of which can vary across cortical regions, even within small spatial scales. We have added a brief note in the Discussion (Subsection: Optical CMRO₂ detection in neonatal care) on page 13 to acknowledge this point:

“Additionally, the spatial heterogeneity in estimated CMRO₂ observed in our data may reflect underlying physiological variability, including differences in vascular structure or metabolic demand across cortical regions. Future studies will aim to further validate and interpret these spatial patterns.”

(7) The justification for using P10 mice in the experiments has not been well presented in the manuscript.

We thank the reviewer for pointing out the need to clarify our choice of developmental stage. We chose P10 mice for our hypoxia-ischemia injury model because this stage is widely recognized as developmentally comparable to human term infants in terms of brain maturation. This approach has been validated by several previous studies (Clancy et al., 2007; Mallard and Vexler, 2015; Sheldon et al., 2018). We have added the following clarification to the Methods section (Subsection: Neonatal Cerebral HI and Hypothermia Treatment) on page 18:

“P10 mice were chosen for our experiments as they are widely used to model near-term infants in humans. At this developmental stage, the brain maturation in mice closely parallels that of near-term infants, making them an appropriate model for studying neonatal brain injury and therapeutic interventions (Clancy et al., 2007; Mallard and Vexler, 2015; Sheldon et al., 2018).”

(8) It was not discussed how the observations made in this manuscript could be affected by the potential discrepancy between the developmental stages of P10 mice and human babies regarding cellular metabolism and neurovascular coupling.

We thank the reviewer for raising this important point regarding developmental differences between P10 mice and human infants. We have discussed this issue by adding the following statement to the Discussion section (newly added subsection: Limitations in this study, the 1^st paragraph) on page 15, where we summarize the overall study design and model selection:

“While P10 mice are widely used to model near-term human infants, developmental differences in cellular metabolism and neurovascular coupling may affect the observed outcomes and limit direct clinical translation (Clancy et al., 2007; Mallard and Vexler, 2015; Sheldon et al., 2018). Nevertheless, the P10 model remains a valuable and widely accepted tool for studying neonatal hypoxia-ischemia mechanisms and evaluating therapeutic interventions.”

(9) Regarding the brain temperature measurements, the authors should use a new cohort of mice, implant the miniature thermocouples 1 mm, 0.5 mm, and immediately below the skull in different mice, and verify the temperature in the brain cortex under conditions applied in the experiments. The same approach could be applied to a few mice undergoing 4-hr-long hypothermia treatment in a chamber, which will provide information about the brain temperature that resulted in observed protection from the injury.

We thank the reviewer for this helpful recommendation. We fully agree that direct intracortical temperature measurement would provide more accurate insight into thermal dynamics during hypothermia treatment. However, the primary aim of this study was not to characterize the precise intracortical temperature response under hypothermic conditions, but rather to examine the effects of hypothermia on CMRO₂ and mitochondrial function. Due to the substantial time and resources required to perform direct intracortical temperature monitoring—and considering the technical focus of the current work—we respectfully suggest reserving such investigations for a future study specifically focused on thermal dynamics in hypoxia-ischemia models.

We have acknowledged this limitation in the subsection Limitations in this study of the Discussion on page 15, noting that skull temperature was used as an approximation of brain temperature and that this approach is consistent with clinical practice, where intracortical temperature is typically not measured. We also note that future studies may benefit from more precise assessments using intracortical probes.

(10) The mean values presented in Fig. 4G are much lower than the peak values in the 2D panels and potentially were calculated as the average values over the entire field of view. Please provide more details on how CMRO₂ was estimated and if the validity of the measurements is expected across the entire field of view. If there are parts of the field of view where the estimation of CMRO₂ is more reliable for technical reasons, maybe one way to compute the mean values is to restrict the usable data to the more centralized part of the field of view.

We thank the reviewer for this thoughtful comment. We confirm that CMRO₂ values shown in Figure 4G were calculated as spatial averages over the entire field of view (FOV; ~5 × 3 mm²) encompassing both hemicortices, as shown in Figure 1C. Regarding the observed CMRO₂ values, The apparent difference likely reflects a comparison between two different post-HI time points. Specifically, the ~0.5 value shown for the 37°C ipsilateral group in Figure 4G reflects the average CMRO₂ measured 24 hours after HI, while the ~1.5 value in Figure 2D (red line) corresponds to CMRO₂ during the early 0–2 hour post-HI period. The temporal difference accounts for the apparent discrepancy in magnitude. We understand the importance of consistency across the field of view and have clarified this point in the subsection Procedures for PAM Imaging in the Methods on page 17 “For the imaging field covering both hemicortices between the Bregma and Lambda of the neonatal mouse (5 × 3 mm² as shown in Figure 1C, with each hemicortex measuring 2.5 × 3 mm²)”, as well as in the Figure 4 legend on page 34 “Correlation of CMRO₂ and post-HI brain infarction in mouse neonates at 24 hours”.

In our model and setup, CMRO₂ estimation is spatially robust across the FOV under standard imaging conditions. We recognize, however, that certain peripheral regions may be more prone to signal attenuation. Future refinement of region selection could further improve spatial averaging strategies. For the current study, full-FOV averaging was used consistently across all groups to maintain comparability.

(11) Minor: Results presented in Supplementary Tables have too many significant digits.

Thank you for the helpful suggestion. We have revised Supplementary Tables S1 and S2 to reduce the number of significant digits and improve clarity.

Reviewer #2 (Public review)

(1) In this study, authors have hypothesized that mitochondrial injury in HIE is caused by OXPHOS-uncoupling, which is the cause of secondary energy failure in HI. In addition, therapeutic hypothermia rescues secondary energy failure. The methodologies used are state-of-the art and include PAM technique in live animal, bioenergetic studies in the isolated mitochondria, and others.

The study is comprehensive and impressive. The article is well written and statistical analyses are appropriate.

We thank the reviewer for the positive feedback.

(2) The manuscript does not discuss the limitation of this animal model study in view of the clinical scenario of neonatal hypoxia-ischemia.

We thank the reviewer for this valuable feedback. In response, we have added a dedicated “Limitations in this study” subsection in the Discussion, where we address the potential limitations of this animal model in the context of the clinical scenario of neonatal hypoxia-ischemia in the first paragraph on page 14, including the developmental differences between P10 mice and human infants.

(3) I see many studies on Pubmed on bioenergetics and HI. Hence, it is unclear what is novel and what is known.

We thank the reviewer for this important comment regarding the novelty of our study in the context of existing research on bioenergetics and hypoxia-ischemia (HI). To better clarify the novel aspects of our work, we have highlighted the relevant content in the Abstract (page 4) and Introduction (page 5). Specifically, while many studies have explored HI-related bioenergetic dysfunction, the mechanisms by which therapeutic hypothermia modulates CMRO₂ and mitochondrial function post-HI remain poorly understood.

Abstract on page 4: “However, it is unclear how post-HI hypothermia helps to restore the balance, as cooling reduces CMRO₂. Also, how transient HI leads to secondary energy failure (SEF) in neonatal brains remains elusive. Using photoacoustic microscopy, we examined the effects of HI on CMRO₂ in awake 10-day-old mice, supplemented by bioenergetic analysis of purified cortical mitochondria.”

Introduction on page 5: “The use of awake mouse neonates avoided the confounding effects of anesthesia on CBF and CMRO₂ (Cao et al., 2017; Gao et al., 2017; Sciortino et al., 2021; Slupe and Kirsch, 2018). In addition, we measured the oxygen consumption rate (OCR), reactive oxygen species (ROS), and the membrane potential of mitochondria that were immediately purified from the same cortical area imaged by PAM. This dual-modal analysis enabled a direct comparison of cerebral oxygen metabolism and cortical mitochondrial respiration in the same animal. Moreover, we compared the effects of therapeutic hypothermia on oxygen metabolism and mitochondrial respiration, and correlated the extent of CMRO₂-reduction with the severity of infarction at 24 hours after HI. Our results suggest that blocking HI-induced OXPHOS-uncoupling is an acute effect of hypothermia and that optical detection of CMRO₂ may have clinical applications in HIE.”

In this study, we propose that uncoupled oxidative phosphorylation (OXPHOS) underlies the secondary energy failure observed after HI, and we demonstrate that hypothermia suppresses this pathological CMRO₂ surge, thereby protecting mitochondrial integrity and preventing injury. Additionally, our use of photoacoustic microscopy (PAM) in awake neonatal mice represents a novel, non-invasive approach to track cerebral oxygen metabolism, with potential clinical relevance for guiding hypothermia therapy.

(4) What are the limitations of ex-vivo mitochondrial studies?

We thank the reviewer for this insightful comment. We acknowledge that ex-vivo mitochondrial assays do not fully replicate in vivo physiological conditions, as they lack systemic factors such as blood flow, cellular interactions, and intact tissue architecture. However, these assays are well-established and widely accepted in the field for evaluating mitochondrial function under controlled conditions (Caspersen et al., 2008; Niatsetskaya et al., 2012). Despite their limitations, they enable direct comparisons of mitochondrial activity across experimental groups and provide valuable mechanistic insights that complement in vivo observations.

(5) PAM technique limits the resolution of the image beyond 500-750 micron depth. Assessing basal ganglia may not be possible with this approach?

We thank the reviewer for this important comment. We agree that the imaging depth of PAM is limited and may not allow assessment of deeper brain structures such as the basal ganglia. However, in our neonatal HI model—as in many clinical cases of HIE—cortical injury is typically more severe and represents a major focus for mechanistic and therapeutic investigations. The cortical regions assessed with PAM are thus highly relevant to the pathophysiology of neonatal HI. We have now acknowledged this depth limitation in the third paragraph of the newly added Limitations in this study subsection of the Discussion on page 15:

“Another limitation of this study is the restricted imaging depth of the PAM technique, which is typically less than 1 mm and therefore does not allow assessment of deeper brain structures such as the basal ganglia. However, in both our neonatal HI model and most clinical cases of neonatal hypoxia-ischemia, cortical injury tends to be more prominent and functionally significant. As such, our cortical measurements remain highly relevant for investigating the mechanisms of injury and evaluating therapeutic interventions.”

(6) Hypothermia in present study reduces the brain temperature from 37 to 29-32 degree centigrade. In clinical set up, head temp is reduced to 33-34.5 in neonatal hypoxia ischemia. Hence a drop in temperature to 29 degrees is much lower relative to the clinical practice. How the present study with greater drop in head temperature can be interpreted for understanding the pathophysiology of therapeutic hypothermia in neonatal HIE. Moreover, in HIE model using higher temperature of 37 and dropping to 29 seems to be much different than the clinical scenario. Please discuss.

We thank the reviewer for raising this important point regarding temperature ranges in our study. In Figure 1, we used a broader temperature range (down to 29°C) to explore the general relationship between temperature and CMRO₂ in uninjured neonatal mice. This experiment was not intended to model therapeutic hypothermia directly, but rather to characterize the baseline physiological responses.

For all experiments involving hypothermia as a therapeutic intervention following HI, we consistently maintained a brain temperature of 32°C, which falls within the clinically accepted mild hypothermia range for neonatal HIE (typically 33–34.5°C). We believe this temperature closely mimics clinical practice and supports the translational relevance of our findings.

(7) NMR was assessed ex-vivo. How does it relate to in vivo assessment. Infants admitted in Neonatal intensive Care Unit, frequently get MRI with spectroscopy. How do the MRS findings in human newborns with HIE correlate with the ex-vivo evaluation of metabolites.

We thank the reviewer for this insightful question. While our study assessed brain metabolites ex vivo, similar metabolic changes have been observed in vivo using proton magnetic resonance spectroscopy (¹H-MRS) in infants with HIE. Specifically, reductions in N-acetylaspartate (NAA) — a marker of neuronal integrity — have been reported in neonates with severe brain injury, aligning with our ex vivo findings. This correlation between in vivo and ex vivo assessments supports the translational relevance of our model for studying metabolic disruption in neonatal HIE. We have added this point to the subsection Using Optically Measured CMRO₂ to Detect Neonatal HI Brain Injury of the Results on page 8, along with a supporting clinical reference (Lally et al., 2019):

“In addition, in vivo proton MRS in infants with HIE has also shown a reduction in NAA, particularly in cases of severe injury (Lally et al., 2019). This reduction in NAA, observed in neonatal intensive care settings, reflects neuronal and axonal loss or dysfunction and serves as a biomarker for injury severity. The alignment between our ex vivo observations and in vivo MRS findings in clinical studies reinforces the translational relevance of our model for investigating metabolic disturbances in neonatal HIE.”

Reviewer #3 (Public review)

(1) In Sun et al. present a comprehensive study using a novel photoacoustic microscopy setup and mitochondrial analysis to investigate the impact of hypoxia-ischemia (HI) on brain metabolism and the protective role of therapeutic hypothermia. The authors elegantly demonstrate three connected findings: (1) HI initially suppresses brain metabolism, (2) subsequently triggers a metabolic surge linked to oxidative phosphorylation uncoupling and brain damage, and (3) therapeutic hypothermia mitigates HI-induced damage by blocking this surge and reducing mitochondrial stress.

The study's design and execution are great, with a clear presentation of results and methods. Data is nicely presented, and methodological details are thorough.

We thank the reviewer for the positive feedback.

(2) However, a minor concern is the extensive use of abbreviations, which can hinder readability. As all the abbreviations are introduced in the text, their overuse may render the text hard to read to non-specialist audiences. Additionally, sharing the custom Matlab and other software scripts online, particularly those used for blood vessel segmentation, would be a valuable resource for the scientific community. In addition, while the study focuses on the short-term effects of HI, exploring the long-term consequences and definitively elucidating HI's impact on mitochondria would further strengthen the manuscript's impact.

We thank the reviewer for these valuable suggestions. Please find our point-by-point responses below:

Abbreviations: To improve readability, we have added a List of Abbreviations on page 3 to help readers, especially non-specialists, navigate the terminology more easily.

MATLAB Code Availability: The methodology for blood vessel segmentation was described in detail in our previous publication (Sun et al., 2020). We have now updated the subsection Quantification of Cerebral Hemodynamics and Oxygen Metabolism by PAM of the Methods on page 18 to provide additional details and have indicated that the MATLAB scripts are available upon request.

“Briefly, this process involves generating a vascular map using signal amplitude from the Hilbert transformation, selecting a region slightly larger than the vessel of interest, and applying Otsu’s thresholding method to remove background pixels. Isolated or spurious boundary fragments are then removed to improve boundary smoothness. The customized MATLAB code used for vessel segmentation is available upon request.”

Long-Term Effects of Hypothermia: We agree that exploring long-term outcomes would enhance the broader impact of this research. While our study focuses on the acute phase following HI, prior studies have shown long-term neuroprotective benefits of therapeutic hypothermia, such as enhanced white matter development (Koo et al., 2017). We have added this point to the fourth paragraph in the subsection Limitations in this study of the Discussion on page 15:

“While our study focuses on the acute effects of hypothermia, previous research has shown long-term neuroprotective benefits, including improved white matter development post-injury (Koo et al., 2017). These findings highlight hypothermia's potential for both immediate and extended recovery, warranting further study of long-term outcomes.”

(3) Extensive use of abbreviations.

Thank you for the helpful suggestion. To improve readability for a broader audience, we have added a List of Abbreviations on page 3 of the manuscript to assist readers in navigating terminology used throughout the text. This has been included as Response #2 to Reviewer #3.

(4) Share code used to conduct the study.

Thank you for the suggestion. The methodology for vessel segmentation was previously published (Sun et al., 2020), and we have noted in the subsection Quantification of Cerebral Hemodynamics and Oxygen Metabolism by PAM of the Methods on page 18 that the MATLAB code is available upon request. This has also been included as Response #2 to Reviewer #3.

Reference:

Cao R, Li J, Kharel Y, Zhang C, Morris E, Santos WL, Lynch KR, Zuo Z, Hu S. 2018. Photoacoustic microscopy reveals the hemodynamic basis of sphingosine 1-phosphate-induced neuroprotection against ischemic stroke. Theranostics 8:6111–6120. doi:10.7150/thno.29435

Caspersen CS, Sosunov A, Utkina-Sosunova I, Ratner VI, Starkov AA, Ten VS. 2008. An Isolation Method for Assessment of Brain Mitochondria Function in Neonatal Mice with Hypoxic-Ischemic Brain Injury. Developmental Neuroscience 30:319–324. doi:10.1159/000121416

Clancy B, Kersh B, Hyde J, Darlington RB, Anand KJS, Finlay BL. 2007. Web-based method for translating neurodevelopment from laboratory species to humans. Neuroinformatics 5:79–94. doi:10.1385/ni:5:1:79

Greenberg RS, Zahurak M, Belden C, Tunkel DE. 1998. Assessment of oropharyngeal distance in children using magnetic resonance imaging. Anesth Analg 87:1048–1051. doi:10.1097/00000539-199811000-00014

Kiyatkin EA. 2007. Brain temperature fluctuations during physiological and pathological conditions. Eur J Appl Physiol 101:3–17. doi:10.1007/s00421-007-0450-7

Koo E, Sheldon RA, Lee BS, Vexler ZS, Ferriero DM. 2017. Effects of therapeutic hypothermia on white matter injury from murine neonatal hypoxia-ischemia. Pediatr Res 82:518–526. doi:10.1038/pr.2017.75

Lally PJ, Montaldo P, Oliveira V, Soe A, Swamy R, Bassett P, Mendoza J, Atreja G, Kariholu U, Pattnayak S, Sashikumar P, Harizaj H, Mitchell M, Ganesh V, Harigopal S, Dixon J, English P, Clarke P, Muthukumar P, Satodia P, Wayte S, Abernethy LJ, Yajamanyam K, Bainbridge A, Price D, Huertas A, Sharp DJ, Kalra V, Chawla S, Shankaran S, Thayyil S, MARBLE consortium. 2019. Magnetic resonance spectroscopy assessment of brain injury after moderate hypothermia in neonatal encephalopathy: a prospective multicentre cohort study. Lancet Neurol 18:35–45. doi:10.1016/S1474-4422(18)30325-9

Lin W, Powers WJ. 2018. Oxygen metabolism in acute ischemic stroke. J Cereb Blood Flow Metab 38:1481–1499. doi:10.1177/0271678X17722095

Mallard C, Vexler Z. 2015. Modeling ischemia in the immature brain: how translational are animal models? Stroke 46:3006–3011. doi:10.1161/STROKEAHA.115.007776

Niatsetskaya ZV, Sosunov SA, Matsiukevich D, Utkina-Sosunova IV, Ratner VI, Starkov AA, Ten VS. 2012. The Oxygen Free Radicals Originating from Mitochondrial Complex I Contribute to Oxidative Brain Injury Following Hypoxia–Ischemia in Neonatal Mice. J Neurosci 32:3235–3244. doi:10.1523/JNEUROSCI.6303-11.2012

Sheldon RA, Windsor C, Ferriero DM. 2018. Strain-Related Differences in Mouse Neonatal Hypoxia-Ischemia. Dev Neurosci 40:490–496. doi:10.1159/000495880

Sun N, Bruce AC, Ning B, Cao R, Wang Y, Zhong F, Peirce SM, Hu S. 2022. Photoacoustic microscopy of vascular adaptation and tissue oxygen metabolism during cutaneous wound healing. Biomed Opt Express, BOE 13:2695–2706. doi:10.1364/BOE.456198

Sun N, Ning B, Bruce AC, Cao R, Seaman SA, Wang T, Fritsche-Danielson R, Carlsson LG, Peirce SM, Hu S. 2020. In vivo imaging of hemodynamic redistribution and arteriogenesis across microvascular network. Microcirculation 27:e12598. doi:10.1111/micc.12598

Sun N, Zheng S, Rosin DL, Poudel N, Yao J, Perry HM, Cao R, Okusa MD, Hu S. 2021. Development of a photoacoustic microscopy technique to assess peritubular capillary function and oxygen metabolism in the mouse kidney. Kidney International 100:613–620. doi:10.1016/j.kint.2021.06.018

eLife Assessment

This valuable study presents a well-designed set of experiments demonstrating how a planthopper salivary carbonic anhydrase can promote rice stripe virus infection by modulating callose deposition in the host plant. The authors provide solid data for the proposed protein-protein interactions, including strengthened evidence for the LssaCA-NP-OsTLP complex and clarified dynamics of LssaCA presence in planta. Overall, the work reveals a mechanistic link whereby a vector salivary protein enhances a plant β-1,3-glucanase to suppress callose-based defense, thereby facilitating early viral establishment.

Reviewer #2 (Public Review):

There is increasing evidence that viruses manipulate vectors and hosts to facilitate transmission. For arthropods, saliva plays an essential role for successful feeding on a host and consequently for arthropod-borne viruses that are transmitted during arthropod feeding on new hosts. This is so because saliva constitutes the interaction interface between arthropod and host and contains many enzymes and effectors that allow feeding on a compatible host by neutralizing host defenses. Therefore, it is not surprising that viruses change saliva composition or use saliva proteins to provoke altered vector-host interactions that are favorable for virus transmission. However, detailed mechanistic analyses are scarce. Here, Zhao and coworkers study transmission of rice stripe virus (RSV) by the planthopper Laodelphax striatellus. RSV infects plants as well as the vector, accumulates in salivary glands and is injected together with saliva into a new host during vector feeding.

The authors present evidence that a saliva-contained enzyme - carbonic anhydrase (CA) - might facilitate virus infection of rice by interfering with callose deposition, a plant defense response. In vitro pull-down experiments, yeast two hybrid assay and binding affinity assays show convincingly interaction between CA and a plant thaumatin-like protein (TLP) that degrades callose. Similar experiments show that CA and TLP interact with the RSV nuclear capsid protein NT to form a complex. Formation of the CA-TLP complex increases TLP activity by roughly 30% and integration of NT increases TLP activity further. This correlates with lower callose content in RSV-infected plants and higher virus titer. Further, silencing CA in vectors decreases virus titers in infected plants. Interestingly, aphid CA was found to play a role in plant infection with two non-persistent non-circulative viruses, turnip mosaic virus and cucumber mosaic virus (Guo et al. 2023 doi.org/10.1073/pnas.2222040120), but the proposed mode of action is entirely different.

Editors' note: this version was assessed by the editors, without further input from the reviewers.

Author response:

The following is the authors’ response to the previous reviews

Reviewer #1 (Public review):

In this study, the authors identified an insect salivary protein LssaCA participating viral initial infection in plant host. LssaCA directly bond to RSV nucleocapsid protein and then interacted with a rice OsTLP that possessed endo-β-1,3-glucanase activity to enhance OsTLP enzymatic activity and degrade callose caused by insects feeding. The manuscript suffers from fundamental logical issues, making its central narrative highly unconvincing.

(1) These results suggested that LssaCA promoted RSV infection through a mechanism occurring not in insects or during early stages of viral entry in plants, but in planta after viral inoculation. As we all know that callose deposition affects the feeding of piercing-sucking insects and viral entry, this is contradictory to the results in Fig. S4 and Fig. 2. It is difficult to understand callose functioned in virus reproduction in 3 days post virus inoculation. And authors also avoided to explain this mechanism.

We appreciate your insightful comment and acknowledge that our initial description may not have been sufficiently clear.

(1) Based on the EPG results, we found that LssaCA deficiency did not significantly affect total feeding time, time to first non-phloem phase, or time to first phloem feeding (Fig. S8A-D in the revised manuscript). However, the continuity of sap ingestion was disturbed—the N4 waveform of dsLssaCA SBPHs was occasionally interrupted for brief periods (newly added Fig. S8E in the revised manuscript), likely due to phloem blockage. In the revised manuscript, we have added this analysis to the Result section (Lines 285-291 and 578-587) and provided the EPG procedure in Material and Methods section (Lines 670-680).

(2) We assessed RSV titers immediately post-feeding to confirm the inoculation viral loads (Fig. 2G) and at 3 dpf (Fig. 2H-I) to assess the in-planta effects following viral inoculation. This did not mean that callose functions in virus reproduction at 3 days post viral inoculation. Rather, callose deposition typically occurs immediately in response to insect feeding and virus inoculation. When measuring callose deposition, we allowed insects to feed for 24 h and quantified the callose levels immediately post feeding. The EPG results showed that sap ingestion continuity was disrupted—the N4 waveform of dsLssaCA-treated SBPHs was occasionally interrupted for brief periods (newly added Fig. S8E in the revised manuscript), likely due to phloem blockage. We have reorganized the description to avoid confusion. Please see Lines 139-144 and Fig. S8E for detail.

(1) Missing significant data. For example, the phenotypes of the transgenic plants, the RSV titers in the transgenic plants (OsTLP OE, ostlp). The staining of callose deposition were also hard to convince. The evidence about RSV NP-LssaCA-OsTLP tripartite interaction to enhance OsTLP enzymatic activity is not enough.

We thank the reviewer for this insightful comment.

(1) We constructed OsTLP overexpression and mutant transgenic plants (OsTLP OE and ostlp) and assessed their phenotypes regarding RSV infection levels. Compared with wild-type plants, OsTLP OE plants exhibited accelerated growth, while ostlp plants showed growth inhibition. Following feeding by viruliferous L. striatellus, OsTLP OE plants had significantly higher RSV titers compared with wild-type plants, whereas ostlp mutant plants exhibited significantly lower RSV titers (Lines 221-228 and new Fig. 3I). These results indicate that OsTLP facilitates RSV infection in planta.

(2) The images showing callose deposition staining are representative of 15 images from 3 independent insect treatments. In addition to the staining images, we quantified fluorescence intensity and measured callose concentration by ELISA.

(2)  Figure 4a, there was the LssaCA signal in the fourth lane of pull-down data. Did MBP also bind LsssCA? The characterization of pull-down methods was rough a little bit. The method of GST pull-down and MBP pull-down should be characterized more in more detail.

We thank the reviewer for this helpful comment. MBP did not bind LssaCA. We have repeated the pull-down experiment and provide clearer figure with improved results. We have also revised and provided more detailed descriptions of the GST pull-down and MBP pull-down methods. Please refer to Lines 744-774 and Figure 4A for details.

eLife Assessment

The medicinal leech preparation is an amenable system in which to understand the neural basis of locomotion. Here a previously identified non-spiking neuron was studied in leech and found to alter the mean firing frequency of a crawl-related motoneuron, which fires during the contraction phase of crawling. The findings are valuable and the experiments were diligently done and considered solid. The results lay a foundation for additional studies in this system.

Reviewer #1 (Public review):

The medicinal leech preparation is an amenable system in which to understand how the underlying cellular networks for locomotion function. A previously identified non-spiking neuron (NS) was studied and found to alter the mean firing frequency of a crawl-related motoneuron (DE-3), which fires during the contraction phase of crawling. The data are solid. Identifying upstream neurons responsible for crawl motor patterning is essential for understanding how rhythmic behavior is controlled.

Reviewer #2 (Public review):

This study by Radice et al., takes advantage of the very well-established leach preparation to investigate questions related to motor control, more precisely the question of how the activity of motoneurons taking part in leach crawling behavior are finely tuned.

The paper is overall well written. The findings are clearly presented, and the data seems solid overall.

Author response:

The following is the authors’ response to the previous reviews

Reviewer #1 (Public review): 

The medicinal leech preparation is an amenable system in which to understand how the underlying cellular networks for locomotion function. A previously identified non-spiking neuron (NS) was studied and found to alter the mean firing frequency of a crawl-related motoneuron (DE-3), which fires during the contraction phase of crawling. The data are mostly solid. Identifying upstream neurons responsible for crawl motor patterning is essential for understanding how rhythmic behavior is controlled.

Review of Revision: 

On a positive note, the rationale for the study is clearer to me now after reading the authors' responses to both reviewers, but that information, as described in the authors' responses, is minimally incorporated into the current revised paper. Incorporating a discussion of previous work on the NS cell has, indeed, improved the paper. 

I suggested earlier that the paper be edited for clarity but not much text has been changed since the first draft. I will provide an example of the types of sentences that are confusing. The title of the paper is: "Phase-specific premotor inhibition modulates leech rhythmic motor output". Are the authors referring to the inhibition created by premotor neurons (e.g., on to the motoneurons) or the inhibition that the premotor neurons receive? 

In this case, this is an interesting ambiguity: NS is inhibited and that inhibition is directly transmitted to the motoneurons because both cells are electrically coupled.  We believe that the title does not disguise the findings conveyed by the manuscript.

I also find the paper still confusing with regard to the suggested "functional homology" with the vertebrate Renshaw cells. When the authors set up this expectation of homology (should be analogy) in the introduction and other sections of the paper, one would assume that the NS cell would be directly receiving excitation from a motoneuron (like DE-3) and, in turn, the motoneuron would then receive some sort of inhibitory input to regulate its firing frequency. Essentially, I have always viewed the Renshaw cells as nature's clever way to monitor the ongoing activity of a motoneuron while also providing recurrent feedback or "recurrent inhibition" to modify that cell's excitatory state. The authors present their initial idea below on line 62. Authors write: "These neurons are present as bilateral pairs in each segmental ganglion and are functional homologs of the mammalian Renshaw cells (Szczupak, 2014). These spinal cord cells receive excitatory inputs from motoneurons and, in turn, transmit inhibitory signals to the motoneurons (Alvarez and Fyffe, 2007)." 

We agree with Reviewer #2: the correct term is "analogous," not "homologous." Thanks for pointing this out. We changed the term throughout the text.

The Reviewer is also right in the appreciation of the role of Renshaw cells. NS plays exactly the role that the Reviewer expresses. The ONLY difference is that NS is inhibited by the motoneurons, and in turn transmits this inhibition to the motoneurons via the rectifying electrical junctions. Attending the confusion that our description caused in the Reviewer, we have modified the cited sentence accordingly now in lines 65-67.

Minor note:

I suggest re-writing this last sentence as "these" is confusing. Change to: 'In the spinal cord, Renshaw interneurons receive excitatory inputs from motoneurons and, in turn, transmit inhibitory signals to them (Alvarez and Fyffe, 2007).'] 

Please, see the changes mentioned above.

Furthermore, the authors note that (line 69 on): "In the context of this circuit the activity of excitatory motoneurons evokes chemically mediated inhibitory synaptic potentials in NS. Additionally, the NS neurons are electrically coupled......In physiological conditions this coupling favors the transmission of inhibitory signals from NS to motoneurons." Based on what is being conveyed here, I see a disconnect with the "functional homology" being presented earlier. I may be missing something, but the Renshaw analogy seems to be quite different compared to what looks like reciprocal inhibition in the leech. If the authors want to make the analogy to Renshaw cells clearer, then they should make a simple ball and stick diagram of the leech system and visually compare it to the Renshaw/motoneuron circuit with regard to functionality. This simple addition would help many readers. 

We have simplified the description regarding the Renshaw cell (lines 65-67) to avoid the “details” of the connectivity between the two circuits.

This report focuses on NS neurons and their role in crawling; we mention the analogy with Renshaw cells to widen the interest of the results. We do not think that making a special diagram to compare how the two neurons play a similar role via different connections among the players is useful in the context of this manuscript.

The Abstract, Authors write (line 19), "Specifically, we analyzed how electrophysiological manipulation of a premotor nonspiking (NS) neuron, that forms a recurrent inhibitory circuit (homologous to vertebrate Renshaw cells)...."

First, a circuit would not be homologous to a cell, and the term homology implies a strict developmental/evolutionary commonality. At best, I would use the term functionally analogous but even then I am still not sure that they are functionally that similar (see comments above). 

Reviewer #2 is right. We changed the sentence in line 20.

Line 22: "The study included a quantitative analysis of motor units active throughout the fictive crawling cycle that shows that the rhythmic motor output in isolated ganglia mirrors the phase relationships observed in vivo." This sentence must be revised to indicate that not all of the extracellular units were demonstrated to be motor units. Revise to: "The study included a quantitative analysis of identified and putative motor units active throughout the fictive crawling cycle that shows.....' 

Line 187 regarding identifying units as motoneurons: Authors write, "While multiple extracellular recordings have been performed previously (Eisenhart et al., 2000), these results (Figure 4) present the first quantitative analysis of motor units activated throughout the crawling cycle in this type of recordings." The authors cannot assume that the units in the recorded nerves belong only to motoneurons. Based on their first rebuttal, the authors seem to be reluctant to accept the idea that the extracellularly recorded units might represent a different class of neurons. They admit that some sensory neurons (with somata located centrally) do, indeed, travel out the same nerves recorded, but go on to explain why they would not be active. 

The leech has a variety of sensory organs that are located in the periphery, and some of these sensory neurons do show rhythmic activity correlated with locomotor activity (see Blackshaw's early work). The numerous stretch receptors, in fact, have very large axons that pass through all the nerves recorded in the current paper. 

In Fig. 4, it is interesting that the waveforms of all the units recorded in the PP nerve exhibit a reversal in waveform as compared to those in the DP nerve, which might indicate (based on bipolar differential recording) that the units in the PP nerve are being propagated in the opposite direction (i.e., are perhaps afferent). Rhythmic presynaptic inhibition and excitation is commonly seen for stretch receptors within the CNS (see the work of Burrows) and many such cells are under modulatory control. 

Most likely, the majority of the units are from motoneurons, but we do not really know at this point. The authors should reframe their statements throughout the paper as: 'While multiple extracellular recordings have been performed previously (Eisenhart et al., 2000), these results (Figure 4) present the first quantitative analysis of multiple extracellular units, using spike sorting methods, which are activated throughout the crawling cycle.' In cases where the identity of the unit is known, then it is fine to state that, but when the identity of the unit is not known, then there should be some qualification and stated as 'putative motor units' 

We understand the concern of Reviewer #2 regarding the type of neurons active during dopamine-induced crawling in isolated ganglia. However, we believe there is sufficient evidence to support that the recorded spikes originate from motoneurons. As readers may share the same concern, we have added a paragraph explaining why spikes from somatic sensory neurons such as P or T cells, or from stretch receptors, are unlikely to contribute (lines 206-214). We included the term putative in the abstract.

The Methods section:

Needs to include the full parameters that were used to assess whether bursting activity was qualified in ways to be considered crawling activity or not. Typically, crawl-like burst periods of no more than 25 seconds have been the limit for their qualification as crawling activity. In Fig 2F, for example, the inter-burst period is over 35 seconds; that coupled with an average 5 second burst duration would bring the burst period to 40 seconds, which is substantially out of range for there to be bursting relevant to crawl activity. Simply put, long DE-3 burst periods are often observed but may not be indicative of a crawl state as the CV motoneurons are no longer out of phase with DE-3. A number of papers have adopted this criterion. 

We now indicate in the methods the range of period values measured in our experiments.  For the reviewer informatio we show here histograms depicting the variability of period and duty cycle values recorded in our experiments (control conditions). The Reviewer can see that the bursting activity of DE-3 fall within what has been published.

Author response image 1.

Crawling in isolated ganglia. A. Histogram of periods end-to-end during crawling in isolated ganglia. The dotted line indicates the mean obtained from the averages of all experiments. The solid black line represents the mean of all cycles across all experiments. B. As in A, for the duty cycle calculated using end-to-end periods.  (n = 210 cycles from 45 ganglia obtained from 32 leeches in all cases).

Reviewer #1 (Recommendations for the authors): 

Minor comments-

Line 100: "In the frame of the recurrent inhibitory circuit, NS is the target of inhibitory signals". Suggestion: 'Within the framework of the recurrent inhibitory circuit, NS is the target of inhibitory signals.' 

Changed as suggested (line 107).

Line 163: "This series of experiments proves that, as predicted based on the known circuit (Figure 164 1C), inhibitory signals onto NS premotor neurons were transmitted to DE-3 motoneurons and counteracted their excitatory drive during crawling, limiting their firing frequency". I think this sentence is too strong plus needs some editing. Suggestion: 'As predicted based on the known circuit (Figure 164 1C), this series of experiments indicates that inhibitory signals onto NS premotor neurons are transmitted to DE-3 motoneurons, thus limiting their firing frequency and counteracting their excitatory drive during crawling."

Changed as suggested.

Lines 86, 292 and 304 and Fig 4 legend: "Different from DE-3, In-Phase units showed a marked decrease in the maximum bFF along time." Suggestion: Replace the word "along" with 'across' time. Also replace those words in the Fig 4 legend and Line 80...."along" (replace with 'across') the different stages of crawling. 

Changed as suggested.

Line 311: "bursts and a concurrent inhibitory input via NS (Figure 7). Coherent with this interpretation, the activity level of the Anti- Phase units was not influenced by these inhibitory signals". Suggestion: Replace the word "coherent" with 'consistent'. 

Changed as suggested.

Line 332: "...offer the particular advantage of allowing electrical manipulation of individual neurons in wildtype adults," I am unsure what the authors are attempting to convey. Not sure what they mean by "wildtype" in this context and why that would matter. 

“wildtype” was eliminated

We thank Reviewer #2 for the suggested edits to the text.

eLife Assessment

This important Research Advance builds on the authors' previous work delineating the roles of the rodent perirhinal cortex and the basolateral amygdala in first- and second-order learning. The convincing results show that serial exposure of non-motivationally relevant stimuli influences how those stimuli are encoded within the perirhinal cortex and basolateral amygdala when paired with a shock. This manuscript will be interesting for researchers in cognitive and behavioral neuroscience.

Reviewer #1 (Public review):

Summary:

This study advances the lab's growing body of evidence exploring higher-order learning and its neural mechanisms. They recently found that NMDA receptor activity in the perirhinal cortex was necessary for integrating stimulus-stimulus associations with stimulus-shock associations (mediated learning) to produce preconditioned fear, but it was not necessary for forming stimulus-shock associations. On the other hand, basolateral amygdala NMDA receptor activity is required for forming stimulus-shock memories. Based on these facts, the authors assessed: 1. why the perirhinal cortex is necessary for mediated learning but not direct fear learning and 2. the determinants of perirhinal cortex versus basolateral amygdala necessity for forming direct versus indirect fear memories. The authors used standard sensory preconditioning and variants designed to manipulate the novelty and temporal relationship between stimuli and shock and, therefore, the attentional state under which associative information might be processed. Under experimental conditions where information would presumably be processed primarily in the periphery of attention (temporal distance between stimulus/shock or stimulus pre-exposure), perirhinal cortex NMDA receptor activation was required for learning indirect associations. On the other hand, when information would likely be processed in focal attention (novel stimulus contiguous with shock), basolateral amygdala NMDA activity was required for learning direct associations. Together, the findings indicate that the perirhinal cortex and basolateral amygdala subserve peripheral and focal attention, respectively. The authors provide support for their conclusions using careful, hypothesis-driven experimental design, rigorous methods, and integrating their findings with the relevant literature on learning theory, information processing, and neurobiology. Therefore, this work will be highly interesting to several fields.

Strengths:

(1) The experiments were carefully constructed and designed to test hypotheses that were rooted in the lab's previous work, in addition to established learning theory and information processing background literature.

(2) There are clear predictions and alternative outcomes. The provided table does an excellent job of condensing and enhancing the readability of a large amount of data.

(3) In a broad sense, attention states are a component of nearly every behavioral experiment. Therefore, identifying their engagement by dissociable brain areas and under different learning conditions is an important area of research.

(4) The authors clearly note where they replicated their own findings, report full statistical measures, effect sizes, and confidence intervals, indicating the level of scientific rigor.

(5) The findings raise questions for future experiments that will further test the authors' hypotheses; this is well discussed.

Reviewer #2 (Public review):

This paper continues the authors' research on the roles of the basolateral amygdala (BLA) and the perirhinal cortex (PRh) in sensory preconditioning (SPC) and second order conditioning (SOC). In this manuscript, the authors explore how prior exposure to stimuli may influence which regions are necessary for conditioning to the second-order cue (S2). The authors perform a series of experiments which first confirm prior results shown by the author - that NMDA receptors in the PRh are necessary in SPC during conditioning of the first-order cue (S1) with shock to allow for freezing to S2 at test; and that NMDA receptors in the BLA are necessary for S1 conditioning during the S1-shock pairings. The authors then set out to test the hypothesis that the PRh encodes associations in a peripheral state of attention whereas the BLA encodes associations in a focal state of attention, similar to the A1 and A2 states in Wagner's theory of SOP. To do this, they show that BLA is necessary for conditioning to S2 when the S2 is first exposed during a serial compound procedure - S2-S1-shock. To determine whether pre-exposure of S2 will shift S2 to a peripheral focal state, the authors run a design in which S2-S1 presentations are given prior to the serial compound phase. The authors show that this restores NMDA receptor activity within the PRh as necessary for fear response to S2 at test. They then test whether the presence of S1 during the serial compound conditioning allows the PRh to support the fear responses to S2 by introducing a delay conditioning paradigm in which S1 is no longer present. The authors find that PRh is no longer required and suggest that this is due to S2 remaining in the primary focal state.

Strengths:

As with their earlier work, the authors have performed a rigorous series of experiments to better understand the roles of the BLA and PRh in the learning of first- and second-order stimuli. The experiments are well-designed and clearly presented, and the results show definitive differences in functionality between the PRh and BLA. The first experiment confirms earlier findings from the lab (and others), and the authors then build on their previous work to more deeply reveal how these regions differ in how they encode associations between stimuli. The authors have done a commendable job on pursuing these questions.

Table 1 is an excellent way to highlight the results and provide the reader with a quick look-up table of the findings.

Reviewer #3 (Public review):

Summary:

This manuscript presents a series of experiments that further investigate the roles of the BLA and PRH in sensory preconditioning, with a particular focus on understanding their differential involvement in the association of S1 and S2 with shock.

Strengths:

The motivation for the study is clearly articulated, and the experimental designs are thoughtfully constructed. I especially appreciate the inclusion of Table 1, which makes the designs easy to follow. The results are clearly presented, and the statistical analyses are rigorous.

During the revision, the authors have adequately addressed my minor suggestions from the original version.

Author response:

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

Reviewer #1 (Public review):

Summary:

This study advances the lab's growing body of evidence exploring higher-order learning and its neural mechanisms. They recently found that NMDA receptor activity in the perirhinal cortex was necessary for integrating stimulus-stimulus associations with stimulus-shock associations (mediated learning) to produce preconditioned fear, but it was not necessary for forming stimulus-shock associations. On the other hand, basolateral amygdala NMDA receptor activity is required for forming stimulus-shock memories. Based on these facts, the authors assessed: (1) why the perirhinal cortex is necessary for mediated learning but not direct fear learning, and (2) the determinants of perirhinal cortex versus basolateral amygdala necessity for forming direct versus indirect fear memories. The authors used standard sensory preconditioning and variants designed to manipulate the novelty and temporal relationship between stimuli and shock and, therefore, the attentional state under which associative information might be processed. Under experimental conditions where information would presumably be processed primarily in the periphery of attention (temporal distance between stimulus/shock or stimulus pre-exposure), perirhinal cortex NMDA receptor activation was required for learning indirect associations. On the other hand, when information would likely be processed in focal attention (novel stimulus contiguous with shock), basolateral amygdala NMDA activity was required for learning direct associations. Together, the findings indicate that the perirhinal cortex and basolateral amygdala subserve peripheral and focal attention, respectively. The authors provide support for their conclusions using careful, hypothesis-driven experimental design, rigorous methods, and integrating their findings with the relevant literature on learning theory, information processing, and neurobiology. Therefore, this work will be highly interesting to several fields.

Strengths:

(1) The experiments were carefully constructed and designed to test hypotheses that were rooted in the lab's previous work, in addition to established learning theory and information processing background literature.

(2) There are clear predictions and alternative outcomes. The provided table does an excellent job of condensing and enhancing the readability of a large amount of data.

(3) In a broad sense, attention states are a component of nearly every behavioral experiment. Therefore, identifying their engagement by dissociable brain areas and under different learning conditions is an important area of research.

(4) The authors clearly note where they replicated their own findings, report full statistical measures, effect sizes, and confidence intervals, indicating the level of scientific rigor.

(5) The findings raise questions for future experiments that will further test the authors' hypotheses; this is well discussed.

Weaknesses:

As a reader, it is difficult to interpret how first-order fear could be impaired while preconditioned fear is intact; it requires a bit of "reading between the lines".

We appreciate the Reviewer’s point and have attempted to address on lines 55-63 of the revised paper: “In a recent pair of studies, we extended these findings in two ways. First, we showed that S1 does not just form an association with shock in stage 2; it also mediates an association between S2 and the shock. Thus, S2 enters testing in stage 3 already conditioned, able to elicit fear responses (Wong et al., 2019). Second, we showed that this mediated S2-shock association requires NMDAR-activation in the PRh, as well as communication between the PRh and BLA (Wong et al., 2025). These findings raise two critical questions: 1) why is the PRh engaged for mediated conditioning of S2 but not for direct conditioning of S1; and 2) more generally, what determines whether the BLA and/or PRh is engaged for conditioning of the S1 and/or S2?”

Reviewer #2 (Public review):

Summary:

This paper continues the authors' research on the roles of the basolateral amygdala (BLA) and the perirhinal cortex (PRh) in sensory preconditioning (SPC) and second-order conditioning (SOC). In this manuscript, the authors explore how prior exposure to stimuli may influence which regions are necessary for conditioning to the second-order cue (S2). The authors perform a series of experiments which first confirm prior results shown by the author - that NMDA receptors in the PRh are necessary in SPC during conditioning of the first-order cue (S1) with shock to allow for freezing to S2 at test; and that NMDA receptors in the BLA are necessary for S1 conditioning during the S1-shock pairings. The authors then set out to test the hypothesis that the PRh encodes associations in a peripheral state of attention, whereas the BLA encodes associations in a focal state of attention, similar to the A1 and A2 states in Wagner's theory of SOP. To do this, they show that BLA is necessary for conditioning to S2 when the S2 is first exposed during a serial compound procedure - S2-S1-shock. To determine whether pre-exposure of S2 will shift S2 to a peripheral focal state, the authors run a design in which S2-S1 presentations are given prior to the serial compound phase. The authors show that this restores NMDA receptor activity within the PRh as necessary for the fear response to S2 at test. They then test whether the presence of S1 during the serial compound conditioning allows the PRh to support the fear responses to S2 by introducing a delay conditioning paradigm in which S1 is no longer present. The authors find that PRh is no longer required and suggest that this is due to S2 remaining in the primary focal state.

Strengths:

As with their earlier work, the authors have performed a rigorous series of experiments to better understand the roles of the BLA and PRh in the learning of first- and second-order stimuli. The experiments are well-designed and clearly presented, and the results show definitive differences in functionality between the PRh and BLA. The first experiment confirms earlier findings from the lab (and others), and the authors then build on their previous work to more deeply reveal how these regions differ in how they encode associations between stimuli. The authors have done a commendable job of pursuing these questions.

Table 1 is an excellent way to highlight the results and provide the reader with a quick look-up table of the findings.

Weaknesses:

The authors have attempted to resolve the question of the roles of the PRh and BLA in SPC and SOC, which the authors have explored in previous papers. Laudably, the authors have produced substantial results indicating how these two regions function in the learning of first- and second-order cues, providing an opportunity to narrow in on possible theories for their functionality. Yet the authors have framed this experiment in terms of an attentional framework and have argued that the results support this particular framework and hypothesis - that the PRh encodes peripheral and the BLA encodes focal states of learning. This certainly seems like a viable and exciting hypothesis, yet I don't see why the results have been completely framed and interpreted this way. It seems to me that there are still some alternative interpretations that are plausible and should be included in the paper.

We appreciate the Reviewer’s point and have attempted to address it on lines 566-594 of the Discussion: “An additional point to consider in relation to Experiments 3A, 3B, 4A and 4B is the level of surprise that rats experienced following presentations of the familiar S2 in stage 2. Specifically, in Experiments 3A and 3B, S2 was followed by the expected S1 (low surprise) and its conditioning required activation of NMDA receptors in the PRh and not the BLA. By contrast, in Experiments 4A and 4B, S2 was followed by omission of the expected S1 (high surprise) and its conditioning required activation of NMDA receptors in the BLA and not the PRh. This raises the possibility that surprise, or prediction error, also influences the way that S2 is processed in focal and peripheral states of attention. When prediction error is low, S2 is processed in the peripheral state of attention: hence, learning under these circumstances requires NMDA receptor activation in the PRh and not the BLA. By contrast, when prediction error is high, S2 is preserved in the focal state of attention: hence, learning under these circumstances requires NMDA receptor activation in the BLA and not the PRh. The impact of prediction error on the processing of S2 could be assessed using two types of designs. In the first design, rats are pre-exposed to S2-S1 pairings in stage 1 and this is followed by S2-S3-shock pairings in stage 2. The important feature of this design is that, in stage 2, the S2 is followed by surprise in omission of S1 and presentation of S3. Thus, if a large prediction error maintains processing of the familiar S2 in the BLA, we might expect that its conditioning in this design would require NMDA receptor activation in the BLA (in contrast to the results of Experiment 3B) and no longer require NMDA receptor activation in the PRh (in contrast to the results of Experiment 3A). In the second design, rats are pre-exposed to S2 alone in stage 1 and this is followed by S2-[trace]-shock pairings in stage 2. The important feature of this design is that, in stage 2, the S2 is not followed by the surprising omission of any stimulus. Thus, if a small prediction error shifts processing of the familiar S2 to the PRh, we might expect that its conditioning in this design would no longer require NMDA receptor activation in the BLA (in contrast to the results of Experiment 4B) but, instead, require NMDA receptor activation in the PRh (in contrast to the results of Experiment 4A). Future studies will use both designs to determine whether prediction error influences the processing of S2 in the focus versus periphery of attention and, thereby, whether learning about this stimulus requires NMDA receptor activation in the BLA or PRh.”

Reviewer #3 (Public review):

Summary:

This manuscript presents a series of experiments that further investigate the roles of the BLA and PRH in sensory preconditioning, with a particular focus on understanding their differential involvement in the association of S1 and S2 with shock.

Strengths:

The motivation for the study is clearly articulated, and the experimental designs are thoughtfully constructed. I especially appreciate the inclusion of Table 1, which makes the designs easy to follow. The results are clearly presented, and the statistical analyses are rigorous. My comments below mainly concern areas where the writing could be improved to help readers more easily grasp the logic behind the experiments.

Weaknesses:

(1) Lines 56-58: The two previous findings should be more clearly summarized. Specifically, it's unclear whether the "mediated S2-shock" association occurred during Stage 2 or Stage 3. I assume the authors mean Stage 2, but Stage 2 alone would not yet involve "fear of S2," making this expression a bit confusing.

We apologise for the confusion and have revised the summary of our previous findings on lines 55-63. The revised text now states: “In a recent pair of studies, we extended these findings in two ways. First, we showed that S1 does not just form an association with shock in stage 2; it also mediates an association between S2 and the shock. Thus, S2 enters testing in stage 3 already conditioned, able to elicit fear responses (Wong et al., 2019). Second, we showed that this mediated S2-shock association requires NMDAR-activation in the PRh, as well as communication between the PRh and BLA (Wong et al., 2025). These findings raise two critical questions: 1) why is the PRh engaged for mediated conditioning of S2 but not for direct conditioning of S1; and 2) more generally, what determines whether the BLA and/or PRh is engaged for conditioning of the S1 and/or S2?”

(2) Line 61: The phrase "Pavlovian fear conditioning" is ambiguous in this context. I assume it refers to S1-shock or S2-shock conditioning. If so, it would be clearer to state this explicitly.

Apologies for the ambiguity - we have omitted the term “Pavlovian” which may have been the source of confusion: The revised text on lines 60-63 now states: “These findings raise two critical questions: 1) why is the PRh engaged for mediated conditioning of S2 but not for direct conditioning of S1; and 2) more generally, what determines whether the BLA and/or PRh is engaged for conditioning of the S1 and/or S2?”

(3) Regarding the distinction between having or not having Stage 1 S2-S1 pairings, is "novel vs. familiar" the most accurate way to frame this? This terminology could be misleading, especially since one might wonder why S2 couldn't just be presented alone on Stage 1 if novelty is the critical factor. Would "outcome relevance" or "predictability" be more appropriate descriptors? If the authors choose to retain the "novel vs. familiar" framing, I suggest providing a clear explanation of this rationale before introducing the predictions around Line 118.

We have incorporated the suggestion regarding “predictability” while also retaining “novelty” as follows. 

L76-85: “For example, different types of arrangements may influence the substrates of conditioning to S2 by influencing its novelty and/or its predictive value at the time of the shock, on the supposition that familiar stimuli are processed in the periphery of attention and, thereby, the PRh (Bogacz & Brown, 2003; Brown & Banks, 2015; Brown & Bashir, 2002; Martin et al., 2013; McClelland et al., 2014; Morillas et al., 2017; Murray & Wise, 2012; Robinson et al., 2010; Suzuki & Naya, 2014; Voss et al., 2009; Yang et al., 2023) whereas novel stimuli are processed in the focus of attention and, thereby, the amygdala (Holmes et al., 2018; Qureshi et al., 2023; Roozendaal et al., 2006; Rutishauser et al., 2006; Schomaker & Meeter, 2015; Wright et al., 2003).”

L116-120: “Subsequent experiments then used variations of this protocol to examine whether the engagement of NMDAR in the PRh or BLA for Pavlovian fear conditioning is influenced by the novelty/predictive value of the stimuli at the time of the shock (second implication of theory) as well as their distance or separation from the shock (third implication of theory; Table 1).”

(4) Line 121: This statement should refer to S1, not S2.

(5) Line 124: This one should refer to S2, not S1.

We have checked the text on these lines for errors and confirmed that the statements are correct. The lines encompassing this text (L121-130) are reproduced here for convenience:

(1) When rats are exposed to novel S2-S1-shock sequences, conditioning of S2 and S1 will be disrupted by a DAP5 infusion into the BLA but not into the PRh (Experiments 2A and 2B);

(2) When rats are exposed to S2-S1 pairings and then to S2-S1-shock sequences, conditioning of S2 will be disrupted by a DAP5 infusion into the PRh but not the BLA whereas conditioning of S1 will be disrupted by a DAP5 infusion into the BLA not the PRh (Experiments 3A and 3B);

(3) When rats are exposed to S2-S1 pairings and then to S2 (trace)-shock pairings, conditioning of S2 will be disrupted by a DAP5 into the BLA not the PRh (Experiments 4A and 4B).

(6) Additionally, the rationale for Experiment 4 is not introduced before the Results section. While it is understandable that Experiment 4 functions as a follow-up to Experiment 3, it would be helpful to briefly explain the reasoning behind its inclusion.

Experiment 4 follows from the results obtained in Experiment 3; and, as noted, the reasoning for its inclusion is provided locally in its introduction. We attempted to flag this experiment earlier in the general introduction to the paper; but this came at the cost of clarity to the overall story. As such, our revised paper retains the local introduction to this experiment. It is reproduced here for convenience:

“In Experiments 3A and 3B, conditioning of the pre-exposed S1 required NMDAR-activation in the BLA and not the PRh; whereas conditioning of the pre-exposed S2 required NMDAR-activation in the PRh and not the BLA. We attributed these findings to the fact that the pre-exposed S2 was separated from the shock by S1 during conditioning of the S2-S1-shock sequences in stage 2: hence, at the time of the shock, S2 was no longer processed in the focal state of attention supported by the BLA; instead, it was processed in the peripheral state of attention supported by the PRh.

“Experiments 4A and 4B employed a modification of the protocol used in Experiments 3A and 3B to examine whether a pre-exposed S1 influences the processing of a pre-exposed S2 across conditioning with S2-S1-shock sequences. The design of these experiments is shown in Figure 4A. Briefly, in each experiment, two groups of rats were exposed to a session of S2-S1 pairings in stage 1 and, 24 hours later, a session of S2-[trace]-shock pairings in stage 2, where the duration of the trace interval was equivalent to that of S1 in the preceding experiments. Immediately prior to the trace conditioning session in stage 2, one group in each experiment received an infusion of DAP5 or vehicle only into either the PRh (Experiment 4A) or BLA (Experiment 4B). Finally, all rats were tested with presentations of the S2 alone in stage 3. If the substrates of conditioning to S2 are determined only by the amount of time between presentations of this stimulus and foot shock in stage 2, the results obtained in Experiments 4A and 4B should be the same as those obtained in Experiments 3A and 3B: acquisition of freezing to S2 will require activation of NMDARs in the PRh and not the BLA. If, however, the presence of S1 in the preceding experiments (Experiments 3A and 3B) accelerated the rate at which processing of S2 transitioned from the focus of attention to its periphery, the results obtained in Experiments 4A and 4B will differ from those obtained in Experiments 3A and 3B. That is, in contrast to the preceding experiments where acquisition of freezing to S2 required NMDAR-activation in the PRh and not the BLA, here acquisition of freezing to S2 should require NMDAR-activation in the BLA but not the PRh.”

Reviewer #1 (Recommendations for the authors):

I greatly enjoyed reading and reviewing this manuscript, and so I only have boilerplate recommendations.

(1) I might add a couple of sentences discussing how/why preconditioned fear could be intact while first-order fear is impaired. Of course, if I am interpreting the provided interpretation correctly, the reason is that peripheral processing is still intact even when BLA NMDA receptors are blocked, and so mediated conditioning still occurs. Does this mean that mediated conditioning does not require learning the first-order relationship, and that they occur in parallel? Perhaps I just missed this, but I cannot help but wonder whether/how the psychological processes at play might change when first-order learning is impaired, so this would be greatly appreciated.

As noted above, we have revised the general introduction (around lines 55-59) to clarify that the direct S1-shock and mediated S2-shock associations form in parallel. Hence, manipulations that disrupt first-order fear to the S1 (such as a BLA infusion of the NMDA receptor antagonist, DAP5) do not automatically disrupt the expression of sensory preconditioned fear to the S2.

(2) Adding to the above - does the SOP or another theory predict serial vs parallel information flow from focal state to peripheral, or perhaps it is both to some extent?

SOP predicts both serial and parallel processing of information in its focal and peripheral states. That is, some proportion of the elements that comprise a stimulus may decay from the focal state of attention to the periphery (serial processing); hence, at any given moment, the elements that comprise a stimulus can be represented in both focal and peripheral states (parallel processing).

Given the nature of the designs and tools used in the present study (between-subject assessment of a DAP5 effect in the BLA or PRh), we selected parameters that would maximize the processing of the S2 and S1 stimuli in one or the other state of activation; hence the results of the present study. We are currently examining the joint processing of stimulus elements across focal and peripheral states using simultaneous recordings of activity in the BLA and PRh. These recordings are collected from rats trained in the different stages of a within-subject sensory preconditioning protocol. The present study created the basis for this work, which will be published separately in due course.

(3) The organization of PRh vs BLA is nice and consistent across each figure, but I would suggest adding any kind of additional demarcation beyond the colors and text, maybe just more space between AB / CD. The figure text indicating PRh/BLA is a bit small.

Thank you for the suggestion – we have added more space between the top and bottom panels of the figure.

(4) Line 496 typo ..."in the BLA but not the BLA".

Apologies for the type - this has been corrected.

Reviewer #2 (Recommendations for the authors):

I found the experiments to be extremely well-designed and the results convincing and exciting. The hypothesis of the focal and peripheral states of attention being encoded by BLA and PRh respectively, is enticing, yet as indicated in the public review, this does not seem to be the only possible interpretation. This is my only serious comment for the authors.

(1) I think it would be worth reframing the article slightly to give credence to alternative hypotheses. Not to say that the authors' intriguing hypothesis shouldn't be an integral part of the introduction, but no alternatives are mentioned. In experiment 2, could the fact that S2 is already being a predictor of S1, not block new learning to S2? In the framework of stimulus-stimulus associations, there would be no surprise in the serial-compound stage of conditioning at the onset of S1. This may prevent direct learning of the S2-shock association within the BLA. This type of association may as well (S2 predicts S1, but it's omitted), which could support learning by S2. fall under the peripheral/focal theory, but I don't think it's necessary to frame this possibility in terms of a peripheral/focal theory. To build on this alternative interpretation, the absence of S1 in experiment 4 may induce a prediction error. The peripheral and focal states appear to correspond to A2 and A1 in SOP extremely well, and I think it would potentially add interest and support. If the authors do intend to make the paper a strong argument for their hypothesis, perhaps a few additional experiments may be introduced. If the novelty of S2 is critical for S2 not to be processed in a focal state during the serial compound stage, could pre-exposure of S2 alone allow for dependence of S2-shock on the PRh? Assuming this is what the authors would predict, this might disentangle the S-S theory mentioned above from the peripheral/focal theory. Or perhaps run an experiment S2-X in stage 1 and S2-S1-shock in stage 2? This said, I think the experiments are more than sufficient for an exciting paper as is, and I don't think running additional experiments is necessary. I would only argue for this if the authors make a hard claim about the peripheral/focal theory, as is the case for the way the paper is currently written.

We appreciate the reviewer’s excellent point and suggestions. We have included an additional paragraph in the Discussion on page 24 (lines 566-594).  “An additional point to consider in relation to Experiments 3A, 3B, 4A and 4B is the level of surprise that rats experienced following presentations of the familiar S2 in stage 2. Specifically, in Experiments 3A and 3B, S2 was followed by the expected S1 (low surprise) and its conditioning required activation of NMDA receptors in the PRh and not the BLA. By contrast, in Experiments 4A and 4B, S2 was followed by omission of the expected S1 (high surprise) and its conditioning required activation of NMDA receptors in the BLA and not the PRh. This raises the possibility that surprise, or prediction error, also influences the way that S2 is processed in focal and peripheral states of attention. When prediction error is low, S2 is processed in the peripheral state of attention: hence, learning under these circumstances requires NMDA receptor activation in the PRh and not the BLA. By contrast, when prediction error is high, S2 is preserved in the focal state of attention: hence, learning under these circumstances requires NMDA receptor activation in the BLA and not the PRh. The impact of prediction error on the processing of S2 could be assessed using two types of designs. In the first design, rats are pre-exposed to S2-S1 pairings in stage 1 and this is followed by S2-S3-shock pairings in stage 2. The important feature of this design is that, in stage 2, the S2 is followed by surprise in omission of S1 and presentation of S3. Thus, if a large prediction error maintains processing of the familiar S2 in the BLA, we might expect that its conditioning in this design would require NMDA receptor activation in the BLA (in contrast to the results of Experiment 3B) and no longer require NMDA receptor activation in the PRh (in contrast to the results of Experiment 3A). In the second design, rats are pre-exposed to S2 alone in stage 1 and this is followed by S2-[trace]-shock pairings in stage 2. The important feature of this design is that, in stage 2, the S2 is not followed by the surprising omission of any stimulus. Thus, if a small prediction error shifts processing of the familiar S2 to the PRh, we might expect that its conditioning in this design would no longer require NMDA receptor activation in the BLA (in contrast to the results of Experiment 4B) but, instead, require NMDA receptor activation in the PRh (in contrast to the results of Experiment 4A). Future studies will use both designs to determine whether prediction error influences the processing of S2 in the focus versus periphery of attention and, thereby, whether learning about this stimulus requires NMDA receptor activation in the BLA or PRh.”

(3) I was surprised the authors didn't frame their hypothesis more in terms of Wagner's SOP model. It was minimally mentioned in the introduction or the authors' theory if it were included more in the introduction. I was wondering whether the authors may have avoided this framing to avoid an expectation for modeling SOP in their design. If this were the case, I think the paper stands on its own without modeling, and at least for myself, a comparison to SOP would not require modeling of SOP. If this was the authors' concern for avoiding it, I would suggest to the authors that they need not be concerned about it.

We appreciate the endorsement of Wagner’s SOP theory as a nice way of framing our results. We are currently working on a paper in which we use simulations to show how Wagner’s theory can accommodate the present findings as well as others in the literature on sensory preconditioning. For this reason, we have not changed the current paper in relation to this point.

eLife Assessment

This study presents an important new approach to quantifying parsimony preferences in human inference. The work provides convincing evidence that humans are sensitive to specific formalizations of parsimony, such as the dimensionality of perceptual shapes. The work is considered timely, well-written, and technically sophisticated, effectively bridging concepts from statistical inference and human decision-making.

Reviewer #1 (Public review):

I have to preface my evaluation with a disclosure that I lack the mathematical expertise to fully assess what seems to be the authors' main theoretical contribution. I am providing this assessment to the best of my ability, but I cannot substitute for a reviewer with more advanced mathematical/physical training.

Summary:

This paper describes a new theoretical framework for measuring parsimony preferences in human judgments. The authors derive four metrics that they associate with parsimony (dimensionality, boundary, volume, and robustness) and measure whether human adults are sensitive to these metrics. In two tasks, adults had to choose one of two flower beds which a statistical sample was generated from, with or without explicit instruction to choose the flower bed perceptually closest to the sample. The authors conduct extensive statistical analyses showing that humans are sensitive to most of the derived quantities, even when the instructions encouraged participants to choose only based on perceptual distance. The authors complement their study with a computational neural network model that learns to make judgments about the same stimuli with feedback. They show that the computational model is sensitive to the tasks communicated by feedback and only uses the parsimony-associated metrics when feedback trains it to do so.

Strengths:

(1) The paper derives and applies new mathematical quantities associated with parsimony. The mathematical rigor is very impressive and is much more extensive than in most other work in the field, where studies often adopt only one metric (such as the number of causes or parameters). These formal metrics can be very useful for the field.

(2) The studies are preregistered, and the statistical analyses are strong.

(3) The computational model complements the behavioral findings, showing that the derived quantities are not simply equivalent to maximum-likelihood inference in the task.

(4) The speculations in the discussion section (e.g., the idea that human sensitivity is driven by the computational demands each metric requires) are intriguing and could usefully guide future work.

Weaknesses:

(1) The paper is very hard to understand. Many of the key details of the derived metrics are in the appendix, with very little accessible explanation in the main text. The figures helped me understand the metrics somewhat, although I am still not sure how some of them (such as boundary or robustness as measured here) are linked to parsimony. I understand that this is addressed by the derivations in the appendix, but as a computational cognitive scientist, I would have benefited from more accessible explanations. Important aspects of the human studies are also missing from the main text, such as the sample size for Experiment 2.

(2) It is not fully clear whether the sensitivity of human participants to some of the quantities convincingly reported here actually means that participants preferred shapes according to the corresponding aspect of parsimony. The title and framing suggest that parsimony "guides" human decision-making, which may lead readers to conclude that humans prefer more parsimonious shapes. I am not sure the sensitivity findings alone support this framing, but it might just be my misunderstanding of the analyses.

(3) The stimulus set included only four combinations of shapes, each designed to diagnostically target one of the theoretical quantities. It is unclear whether the results are robust or specific to these particular 4 stimuli.

(4) The study is framed as measuring "decision-making," but the task resembles statistical inference (e.g., which shape generated the data) or perceptual judgment. This is a minor point since "decision-making" is not well defined in the literature, yet the current framing in the title gave me the initial impression that humans would be making preference choices and learning about them over time with feedback.

Reviewer #2 (Public review):

This manuscript presents a sophisticated investigation into the computational mechanisms underlying human decision-making, and it presents evidence for a preference for simpler explanations (Occam's razor). The authors dissect the simplicity bias into four different components, and they design experiments to target each of them by presenting choices whose underlying models differ only in one of these components. In the learning tasks, participants must infer a "law" (a logical rule) from observed data in a way that operationalizes the process of scientific reasoning in a controlled laboratory setting. The tasks are complex enough to be engaging but simple enough to allow for precise computational modeling.

As a further novel feature, authors derive a further term in the expansion of the log-evidence, which arises from boundary terms. This is combined with a choice model, which is the one that is tested in experiments. Experiments are run, but with humans and with artificial intelligence agents, showing that humans have an enhanced preference for simplicity as compared to artificial neural networks.

Overall, the work is well written, interesting, and timely, bridging concepts in statistical inference and human decision making. Although technical details are rather elaborate, my understanding is that they represent the state of the art.

I have only one main comment that I think deserves more comments. Computing the complexity penalty of models may be hard. It is unlikely that humans can perform such a calculation on the fly. As authors discuss in the final section, while the dimensionality term may be easier to compute, others (e.g., the volume term, which requires an integral) may be considerably harder to compute (it is true that they should be computed once and for all for each task, but still...). I wonder whether the sensitivity of human decision making with reference to the different terms is so different, and in particular whether it aligns with computational simplicity, or with the possibility of approximating each term by simple heuristics. Indeed, the sensitivity to the volume term is significantly and systematically lower than that of other terms. I wonder whether this relation could be made more quantitative using neural networks, using as a proxy of computational hardness the number of samples needed to reach a given error level in learning each of these terms.

Reviewer #3 (Public review):

Summary:

This is a very interesting paper that documents how humans use a variety of factors that penalize model complexity and integrate over a possible set of parameters within each model. By comparison, trained neural networks also use these biases, but only on tasks where model selection was part of the reward structure. In the situation where training emphasizes maximum-likelihood decisions, only neural networks, but not humans, were able to adapt their decision-making. Humans continue to use model integration simplicity biases.

Strengths:

This study used a pre-registered plan for analyzing human data, which exceeds the standards compared to other current studies.

The results are technically correct.

Weaknesses:

The presentation of the results could be improved.

Author response:

Reviewer #1 (Public review)

I have to preface my evaluation with a disclosure that I lack the mathematical expertise to fully assess what seems to be the authors' main theoretical contribution. I am providing this assessment to the best of my ability, but I cannot substitute for a reviewer with more advanced mathematical/physical training.

Summary:

This paper describes a new theoretical framework for measuring parsimony preferences in human judgments. The authors derive four metrics that they associate with parsimony (dimensionality, boundary, volume, and robustness) and measure whether human adults are sensitive to these metrics. In two tasks, adults had to choose one of two flower beds which a statistical sample was generated from, with or without explicit instruction to choose the flower bed perceptually closest to the sample. The authors conduct extensive statistical analyses showing that humans are sensitive to most of the derived quantities, even when the instructions encouraged participants to choose only based on perceptual distance. The authors complement their study with a computational neural network model that learns to make judgments about the same stimuli with feedback. They show that the computational model is sensitive to the tasks communicated by feedback and only uses the parsimony-associated metrics when feedback trains it to do so.

Strengths:

(1)  The paper derives and applies new mathematical quantities associated with parsimony. The mathematical rigor is very impressive and is much more extensive than in most other work in the field, where studies often adopt only one metric (such as the number of causes or parameters). These formal metrics can be very useful for the field.

(2)  The studies are preregistered, and the statistical analyses are strong.

(3)  The computational model complements the behavioral findings, showing that the derived quantities are not simply equivalent to maximum-likelihood inference in the task.

(4)  The speculations in the discussion section (e.g., the idea that human sensitivity is driven by the computational demands each metric requires) are intriguing and could usefully guide future work.

Weaknesses:

(1) The paper is very hard to understand. Many of the key details of the derived metrics are in the appendix, with very little accessible explanation in the main text. The figures helped me understand the metrics somewhat, although I am still not sure how some of them (such as boundary or robustness as measured here) are linked to parsimony. I understand that this is addressed by the derivations in the appendix, but as a computational cognitive scientist, I would have benefited from more accessible explanations. Important aspects of the human studies are also missing from the main text, such as the sample size for Experiment 2.

(2) It is not fully clear whether the sensitivity of human participants to some of the quantities convincingly reported here actually means that participants preferred shapes according to the corresponding aspect of parsimony. The title and framing suggest that parsimony "guides" human decision-making, which may lead readers to conclude that humans prefer more parsimonious shapes. I am not sure the sensitivity findings alone support this framing, but it might just be my misunderstanding of the analyses.

(3) The stimulus set included only four combinations of shapes, each designed to diagnostically target one of the theoretical quantities. It is unclear whether the results are robust or specific to these particular 4 stimuli.

(4) The study is framed as measuring "decision-making," but the task resembles statistical inference (e.g., which shape generated the data) or perceptual judgment. This is a minor point since "decision-making" is not well defined in the literature, yet the current framing in the title gave me the initial impression that humans would be making preference choices and learning about them over time with feedback.

We are grateful for the supportive comments highlighting the rigor of our experimental design and data analysis. The Reviewer lists four points under “weaknesses”, to which we reply below. 

(1)  The paper is very hard to understand

In the revised version of the paper, we will expand the main text to include a more detailed and intuitive description of the terms of the Fisher Information Approximation, in particular clarifying the interpretation of robustness and boundary as parsimony. We also will include more details that are now given only in Methods, such as the sample size for the second experiment. 

(2) Sensitivity of human participants 

We do argue, and believe, that our data show that people tend to prefer simpler shapes. However, giving a well-posed definition of "preference" in this context turns out to be nontrivial.

At the very least, any statement such as "people prefer shape A over B" should be qualified with something like “when the distance of the data from both shapes is the same.” In other words, one should control for goodness-of-fit. Even before making any reference to our behavioral model, this phenomenon (a preference for the simpler model when goodness of fit is matched between models) is visible in Figure 3a, where the effective decision boundary used by human participants is closer to the more complex model than the cyan line representing the locus of points with equal goodness of fit under the two models (or equivalently, with the same Euclidean distance from the two shapes). The goal of our theory and our behavioral model is precisely to systematize this sort of control, extending it beyond just goodness-of-fit and allowing us to control simultaneously for multiple features of model complexity that may affect human behavior in different ways. In other words, it allows us not only to ask whether people prefer shape A over B after controlling for the distance of the data to the shapes, but also to understand to what extent this preference is driven by important geometrical features such as dimensionality, volume, curvature, and boundaries of the shapes. More specifically, and importantly, our theory makes it possible to measure the strength of the preference, rather than merely asserting its existence. In our modeling framework, the existence of a preference for simpler shapes is captured by the fact that the estimated sensitivities to the complexity penalties are positive (and although they differ in magnitude, all are statistically reliable).

(3) Generalization to different shapes  

Thank you for bringing up this important topic. First, note that while dimensionality and volume are global properties of models and only take two possible values in our human tasks, the boundary and robustness penalties depend on the model and on the data and therefore assume a continuum of values through the tasks (note also that the boundary penalty is relevant for all task types, not just the one designed specifically to study it, because all models except the zero-dimensional dot have boundaries). Therefore, our experimental setting is less restrictive of what it may seem, because it explores a range of possible values for two of the four model features. However, we agree that it would be interesting to repeat our experiment with a broader range of models, perhaps allowing their dimensionality and volume to vary more. In the same spirit, it would be interesting to study the dependence of human behavior on the amount of available data. We believe that these are all excellent ideas for further study that exceed the scope of the present paper. We will include these important points in a revised Discussion. 

(4) Usage of “decision making” vs “perceptual judgment”

Thank you. We will clarify better in the text that our usage of “decision making” overlaps with the idea of a perceptual judgment and that our experiments do not tackle sequential aspects of repeated decisions. 

Reviewer #2 (Public review):

This manuscript presents a sophisticated investigation into the computational mechanisms underlying human decision-making, and it presents evidence for a preference for simpler explanations (Occam's razor). The authors dissect the simplicity bias into four different components, and they design experiments to target each of them by presenting choices whose underlying models differ only in one of these components. In the learning tasks, participants must infer a "law" (a logical rule) from observed data in a way that operationalizes the process of scientific reasoning in a controlled laboratory setting. The tasks are complex enough to be engaging but simple enough to allow for precise computational modeling.

As a further novel feature, authors derive a further term in the expansion of the logevidence, which arises from boundary terms. This is combined with a choice model, which is the one that is tested in experiments. Experiments are run, but with humans and with artificial intelligence agents, showing that humans have an enhanced preference for simplicity as compared to artificial neural networks.

Overall, the work is well written, interesting, and timely, bridging concepts in statistical inference and human decision making. Although technical details are rather elaborate, my understanding is that they represent the state of the art.

I have only one main comment that I think deserves more comments. Computing the complexity penalty of models may be hard. It is unlikely that humans can perform such a calculation on the fly. As authors discuss in the final section, while the dimensionality term may be easier to compute, others (e.g., the volume term, which requires an integral) may be considerably harder to compute (it is true that they should be computed once and for all for each task, but still...). I wonder whether the sensitivity of human decision making with reference to the different terms is so different, and in particular whether it aligns with computational simplicity, or with the possibility of approximating each term by simple heuristics. Indeed, the sensitivity to the volume term is significantly and systematically lower than that of other terms. I wonder whether this relation could be made more quantitative using neural networks, using as a proxy of computational hardness the number of samples needed to reach a given error level in learning each of these terms.

Thank you. The computational complexity associated with calculating the different terms and its potential connection to human sensitivity to the terms is an intriguing topic. As we hinted at in the discussion, we agree with the reviewer that this is a natural candidate for further research, which likely deserves its own study and exceeds the scope of the present paper. 

As a minor aside, at least for the present task the volume term may not be that hard to compute, because it can be expressed with the number of distinguishable probability distributions in the model (Balasubramanian 1996). Given the nature of our task, where noise is Gaussian, isotropic and with known variance, the geometry of the model is actually the Euclidean geometry of the plane, and the volume is simply the (log of the) length of the line that represents the one-dimensional models, measured in units of the standard deviation of the noise.

Reviewer #3 (Public review):

Summary:

This is a very interesting paper that documents how humans use a variety of factors that penalize model complexity and integrate over a possible set of parameters within each model. By comparison, trained neural networks also use these biases, but only on tasks where model selection was part of the reward structure. In the situation where training emphasizes maximum-likelihood decisions, only neural networks, but not humans, were able to adapt their decision-making. Humans continue to use model integration simplicity biases.

Strengths:

This study used a pre-registered plan for analyzing human data, which exceeds the standards compared to other current studies.

The results are technically correct.

Weaknesses:

The presentation of the results could be improved.

We thank the reviewer for their appreciation of our experimental design and methodology, and for pointing out (in the separate "recommendations to authors") a few passages of the paper where the presentation could be improved. We will clarify these passages in the revision.

eLife Assessment

This valuable study successfully decoded visual representations of facial expressions and stereoscopic depth information from electroencephalogram (EEG) signals recorded in an immersive virtual reality (VR) environment. The evidence is solid in demonstrating the technical feasibility of integrating state-of-the-art EEG decoding and VR with eye tracking. This work will interest neuroscience researchers, as well as engineers developing brain-machine interfaces and/or virtual reality displays.

Reviewer #1 (Public review):

Summary:

The study by Klotzsche et al. examines whether emotional facial expressions can be decoded from EEG while participants view 3D faces in immersive VR and whether stereoscopic depth cues affect these neural representations. Participants viewed computer-generated faces (three identities, four emotions) rendered either stereoscopically or monoscopically, while performing an emotion recognition task. Time-resolved multivariate decoding revealed above-chance decodability of facial expressions from EEG. Importantly, decoding accuracy did not differ between monoscopic and stereoscopic viewing. This indicates that the neural representation of expressions is robust against stereoscopic disparity for the relevant features. However, a separate classifier could distinguish the depth condition (mono vs. stereo) from EEG, i.e., the pattern of neuronal activity differs between conditions, but not in ways relevant for the decoding of emotions. It had an early peak and a temporal profile similar to identity decoding, suggesting that early, task-irrelevant visual differences are captured neurally. Cross-decoding further demonstrated that expression decoders trained in one depth condition could generalize to the other, supporting the idea of representational invariance. Eye-tracking analyses showed that expressions and identities could be decoded from gaze patterns, but not the depth condition, and EEG- and gaze-based decoding performances were not correlated across participants. Overall, this work shows that EEG decoding in VR is feasible and sensitive, and suggests that stereoscopic cues are represented in the brain but do not influence the neural processing of facial expressions. This study addresses a relevant question with state-of-the-art experimental and data analysis techniques.

Strengths:

(1) It combines EEG, virtual reality stereoscoptic and monoscopic presentation of visual stimuli, and advanced data analysis methods to address a timely question.

(2) The figures are of very high quality.

(3) The reference list is appropriate and up to date.

Weaknesses:

(1) The introduction-results-discussion-methods order makes it hard to follow the Results without repeatedly consulting the Methods. Please introduce minimal, critical methodological context at the start of each Results subsection; reserve technical details for Methods/Supplement.

(2) Many Results subsections begin with a crisp question and present rich analyses, but end without a short synthesis. Please add 1-2 sentences that explicitly answer the opening question and state what the analyses demonstrate.

(3) The Results compellingly show that (a) expressions are decodable from EEG and (b) mono vs stereo trials are decodable from EEG; yet expression decoding is comparable across mono and stereo. It would help if you articulate why depth is neurally distinguishable while leaving expression representations unchanged. Maybe improve the discussion of the results of source localization and give a more detailed connection to what we already know about the processing of disparity.

Reviewer #2 (Public review):

Summary:

The authors' main aim was to determine the extent to which the emotional expression of face images could be inferred from electrophysiological data under the viewing conditions imposed by immersive virtual reality displays. Further, given that stereoscopic depth cues can be easily manipulated in such displays, the authors wished to investigate whether successful emotion decoding was affected by the presence or absence of these depth cues, and also if the presence/absence of depth cues was itself a property of the viewing experience that could be decoded from neural data.

Overall, the authors use fairly standard approaches to decoding neural data to demonstrate that above-chance results (slightly above the 0.5 chance threshold for their measure of choice) are in general achievable for emotion decoding, decoding the identity of faces from neural data, and decoding the presence/absence of depth cues in an immersive virtual reality display. They further examine the contribution of specific components of the response to visual stimuli with similar outcomes.

Strengths:

The main contribution of the manuscript is methodological. Rather than shedding particular light on the neural mechanisms supporting depth processing or face perception, what is on offer is primarily a straightforward examination of an applied question. With regard to the goal of answering that applied question, I think the paper succeeds. The overall experimental design is not novel, but in this case, that is a good thing. The authors have used relatively unadorned tasks and previous approaches to applying decoding tools to EEG data to see what they can get out of the neural data collected under these viewing conditions. While I would say that there is not a great deal that is especially surprising about these results, the authors do meet the goal they set for themselves.

Weaknesses:

Some of the key weaknesses I see are points that the authors raise themselves in their discussion, particularly with regard to the generalizability of their results. In particular, the 3D faces they have employed here perhaps exhibit a somewhat limited repertoire of emotional expression and do not necessarily cover a representative gamut of emotional face appearances, such as one would encounter in naturalistic settings. Then again, part of the goal of the paper was to examine the decodability of emotional expression in a specific, non-natural viewing environment - a viewing environment in which one could reasonably expect to encounter artificial faces like these. Still, the limitations of the stimuli potentially limit the scope of the conclusions one should draw from the data. I also think that there is a great deal of room for low-level image properties to drive the decoding results for faces, which could have been addressed in a number of ways (matching power spectra, for example, or using an inverted-image control condition). The absence of such control comparisons means that it is difficult to know if this is really a result that reflects face processing or much lower-level image differences that are diagnostic of emotion or identity in this subset of images. Again, to some extent, this is potentially acceptable - if one is mostly interested in whether this result is achievable at all (by hook or by crook), then it is not so important how the goal is met. Then again, one would perhaps like to know if what has been measured here is more a reflection of spatial vision vs. face processing mechanisms.

Reviewer #3 (Public review):

Summary:

This study investigates two main questions:

(1) whether brain activity recorded during immersive virtual reality can differentiate facial expressions and stereoscopic depth, and

(2) whether depth cues modulate facial information processing.

The results show that both expression and depth information can be decoded from multivariate EEG recorded in a head-mounted VR setup. However, the results show that the decoding performance of facial expressions does not benefit from depth information.

Strengths:

The study is technically strong and well executed. EEG data are of high quality despite the challenges of recording inside a head-mounted VR system. The work effectively combines stereoscopic stimulus presentation, eye-tracking to monitor gaze behavior, and time-resolved multivariate decoding techniques. Together, these elements provide an exemplary demonstration of how to collect and analyze high-quality EEG data in immersive VR environments.

Weaknesses:

The major limitation concerns the theoretical question about how stereoscopic depth modulates facial expression processing. While previous work has suggested that stereoscopic depth cues can shape natural face perception and emphasize the importance of binocular information in recognizing facial expressions (lines 95-97), the present study reports a null effect of depth. However, the stimulus configuration they used likely constrained the ability to detect any depth-related effects. All facial stimuli were static, frontal, and presented at a fixed distance. This design leads to near-ceiling behavioral performance and no behavioral effect of depth on expression recognition. It makes the null modulation of depth on expression processing unsurprising and limits the theoretical reach of the study. Adding more subtle or naturalistic features (such as various viewing angles and dynamic expressions) to the stimulus set if the authors aim to advance a strong theoretical claim about the role of binocular disparity. Or reframing the work as a technical validation of EEG decoding in this context.

Another issue relates to the claim that eye movements cannot explain the EEG decoding results. It is a real challenge to remove eye-movement-related artifacts and confounds, as the VR setup tends to encourage viewers to explore the environment freely. However, nearly half of the eye-tracking datasets were lost (usable in only 17 of 33 participants), which substantially weakens the evidence for EEG-gaze dissociation. Moreover, it would be almost impossible to decode facial information from only two-dimensional gaze direction, given that with 60 EEG channels, the decoding accuracy was modest (AUC ≈ 0.60). These two factors together limited the strength of the reported null correlation between neural and eye-data decoding.

The decoding analysis appears to use all 60 EEG channels as input features. I wonder why the authors did not examine using more spatially specific channel subsets. Facial expression and depth cues are known to preferentially engage occipito-temporal regions (e.g., N170-related sites), yet the current approach treats all sensors equally. Including all the channels may add noise and irrelevant signals to facial information decoding. Besides, using a subset of spatial-specific channels would align more directly with the subsequent source reconstruction.

Author response:

We thank the reviewers for their thoughtful and constructive comments. We are pleased that they found the study technically strong and the integration of EEG decoding, immersive VR, and eye tracking valuable.

Across all three reviews, several points of clarification emerged. In our revision, we will focus on:

(1) Improving clarity and structure of the manuscript (Reviewer #1).

We will strengthen the flow between the Methods and Results subsections and include explicit concluding statements for the single results.

(2) Emphasize methodological scope and limitations in terms of stimulus set and generalizability (Reviewers #2 and #3).

We will further emphasize that a key objective was to establish, for the first time, the methodological feasibility of decoding facial features (especially emotional expressions) under VR conditions, and that our stimulus set (consisting of facial expressions that were easy to distinguish) limits (a) the task-relevance (and thus possibly the neural integration) of depth information and (b) the generalizability to less easily distinguishable settings. We appreciate the suggestion of an inverted-face control to further investigate the extent to which the decoding results were based on low-level features; however, we do not plan a follow-up experiment at this stage; instead, we will discuss this limitation more explicitly.

We believe these revisions will substantially strengthen the manuscript and further highlight its methodological focus.

eLife Assessment

This important study reveals that mitotic release of an ER-microtubule tether is critical for normal mitotic progression. Manipulating CLIMP63 phosphorylation, the authors provide convincing evidence that persistent microtubule-ER contacts activate the spindle assembly checkpoint and, if mitosis is forced to proceed, drive severe micronucleation. While the study provides new mechanistic insights, some evidence is indirect, and additional experiments would further refine the model.

Reviewer #1 (Public review):

Summary:

In the present manuscript, de Bos and Kutay investigate the functional implications of persistent microtubule-ER contacts as cells go through mitosis. To do so, they resorted to investigating phosphorylation mutants of the ER-Microtubule crosslinker Climp63. They found that phosphodeficient Climp63 mutants induce a severe SAC-dependent mitotic delay after normal chromosome alignment, with an impressive mitotic index of approximately 75%. Strikingly, this was often associated with massive nuclear fragmentation into up to 30 micronuclei that are able to recruit both core and non-core nuclear envelope components. One particular residue (S17) that is phosphorylated by Cdk1 seems to account for most, if not all, these phenotypes. Furthermore, the authors use the impact on mitosis as an indirect way to map the microtubule binding domain of Climp63, which has remained controversial, and found that it is mostly restricted to the N-terminal 28 residues of Climp63. Of note, despite the strong impact on mitosis, persistent microtubule-ER contacts did not affect the distribution of other organelles during mitosis, such as mitochondria or lysosomes.

Strengths:

Overall, this work provides important mechanistic insight into the functional implications of ER-microtubule network remodelling during mitosis and should be of great interest to a vast readership of cell biologists.

Weaknesses:

Some of the key findings appear somewhat preliminary and would be worth exploring further to substantiate some of the claims and clarify the respective impact on mitosis and nuclear envelope reassembly on the resulting micronuclei.

The following suggestions would significantly clarify some key points:

(1) The striking increase in mitotic index in cells expressing the Climp63 phosphodefective mutant, together with their live cell imaging data indicating extensive mitotic delays that can be relieved by SAC inhibition, suggests that SAC silencing is significantly delayed or even impossible to achieve. The fact that most chromosomes align in 12 min, irrespective of the expression of the Climp63 phosphodefective mutant, suggests that initial microtubule-kinetochore interactions are not compromised, but maybe cannot be stably maintained. Alternatively, the stripping of SAC proteins from kinetochores by dynein along attached microtubules might be compromised, despite normal microtubule-kinetochore attachments. The authors allude to both these possibilities, but unfortunately, they never really test them. This could easily be done by immunofluorescence with a Mad1 or c-Mad2 antibody to inspect which fraction of kinetochores (co-stained with a constitutive kinetochore marker, such as CENP-A or CENP-C) are positive for these SAC proteins. If just a small fraction, then the stability of some attachments is likely the cause. If most/all kinetochores retain Mad1/c-Mad2, then it is probably an issue of silencing the SAC.

(2) The authors use the increase in mitotic index (H3 S10 phosphorylation levels) as a readout for the MT binding efficiency of Climp63 and respective mutants. Although suggestive, this is fairly indirect and requires additional confirmation. For example, the authors could perform basic immunofluorescence in fixed cells to inspect co-localization of Climp63 (and its mutants) with microtubules.

(3) The authors refer in the discussion that the striking nuclear fragmentation seen upon mitotic exit of cells expressing Climp63 phosphodefective mutant has not been reported before, and yet it is strikingly similar to what has been previously observed in cells treated with taxol (they cite Samwer et al. 2017, but they might elect to cite also Mitchison et al., Open Biol, 2017 and most relevantly Jordan et al., Cancer Res, 1996). This striking similarity and given the extensive mitotic delay observed in the Climp63 phosphodefective mutant, it is tempting to speculate that these cells are undergoing mitotic slippage (i.e., cells exit mitosis without ever satisfying the SAC) because they are unable to silence/satisfy the SAC. Indeed, the scattered micronuclei morphology has also been observed in cells undergoing mitotic slippage (e.g., Brito and Rieder, Curr Biol., 2006). The experiment suggested in point #1 should also shed light on this problem. The authors might want to consider discussing this possible explanation to interpret the observed phenotypes.

(4) One of the most significant implications of the findings reported in this paper is that microtubule proximity does not seem to impact the assembly of either core or non-core nuclear envelope proteins on micronuclei (that possibly form due to mitotic slippage, rather than normal anaphase). These results challenge some models explaining nuclear envelope defects in micronuclei derived from lagging chromosomes due to the proximity of microtubules, and, as the authors point out at the very end, other reasons might underlie these defects. Along this line, the authors might elect to cite Afonso et al. Science, 2014, and Orr et al., Cell Reports, 2022, who provide evidence that a spindle midzone-based Aurora B gradient, rather than microtubules per se, underlie the nuclear envelope defects commonly seen in micronuclei derived from lagging chromosomes during anaphase.

Reviewer #2 (Public review):

Mitotic phosphorylation of the ER-microtubule linker CLIMP63 was discovered decades ago and was shown to release CLIMP63 from microtubules. Here, the authors describe for the first time the significance of CLIMP63 phosphorylation for mitotic division in cells. Expression of non-phosphorylatable CLIMP63 led to a massive re-localization of ER into the area of the mitotic spindle. This was not unexpected, as another ER-microtubule linker, STIM1, is phosphorylated during mitosis to release it from microtubules, and unphosphorylatable STIM1 also leads to an invasion of the ER into the spindle. The authors map CLIMP63's microtubule-binding domain and define S17 as the critical residue that needs to be phosphorylated for release from microtubules and as a target of Cdk1, albeit with an indirect assay that is based on the ability of overexpressed mutants to disrupt mitosis. The authors further demonstrate that aberrant, microtubule-tethered membranes in the spindle disrupt spindle function. This is in line with the group's prior findings that chromosome-tethered membranes lead to severe chromosome segregation defects. Cells overexpressing phospho-deficient CLIMP63 arrested in prometaphase with an active checkpoint. When these cells were forced to exit mitosis, a large number of micronuclei formed. Interestingly, these micronuclei had different compositions and properties from previously described ones, suggesting that there are diverse paths for a cell to become multinucleated. Lastly, the authors asked whether mitochondria and lysosomes depend on ER for their distribution in mitotic cells. However, the position of these other organelles was unchanged in cells in which ER was re-localized due to the overexpression of phospho-deficient CLIMP63. This is an interesting observation in the context of how the interior organisation of mitotic cells is achieved.

Suggestions:

(1) The authors should confirm the mapping of the microtubule-binding domain by more direct assays, such as microtubule co-pelleting or proximity ligation assays.

(2) The authors should clarify why they performed phenotypic studies and live microscopy experiments (Figures 4 and 5) using the CLIMP63(3A) mutant, despite knowing that the relevant phosphorylation site was S17. Were the phenotypes different for S17A versus the triple mutant?

eLife Assessment

This study provides useful insights into addressing the question of whether the prevalence of autoimmune disease could be driven by sex differences in the T cell receptor (TCR) repertoire, correlating with higher rates of autoimmune disease in females. The authors compare male and female TCR repertoires using bulk RNA sequencing, from sorted thymocyte subpopulations in pediatric and adult human thymuses; however, the results do not provide sufficient analytical rigor and incompletely support the central claims.

Reviewer #1 (Public review):

Summary:

The goal of this paper was to determine whether the T cell receptor (TCR) repertoire differs between a male and a female human. To address this, this group sequenced TCRs from double-positive and single-positive thymocytes in male and female humans of various ages. Such an analysis on sorted thymocyte subsets has not been performed in the past. The only comparable dataset is a pediatric thymocyte dataset where total thymocytes were sorted.

They report on participant ages and sexes, but not on ethnicity, race, nor provide information about HLA typing of individuals. Though the experiments themselves are heroic, they do represent a relatively small sampling of diverse humans. They observed no differences in TCRbeta or TCRalpha usage, combinational diversity, or differences in the length of the CDR3 region, or amino acid usage in the CD3aa region between males or females. Though they observed some TCRbeta CD3aa sequence motifs that differed between males and females, these findings could not be replicated using an external dataset and therefore were not generalizable to the human population.

They also compared TCRbeta sequences against those identified in the past using computational approaches to recognize cancer-, bacterial-, viral-, or autoimmune-antigens. They found very little overlap of their sequences with these annotated sequences (depending on the individual, ranging from 0.82-3.58% of sequences). Within the sequences that were in overlap, they found that certain sequences against autoimmune or bacterial antigens were significantly over-represented in female versus male CD8 SP cells. Since no other comparable dataset is available, they could not conclude whether this is a finding that is generalizable to the human population.

Strengths:

This is a novel dataset. Overall, the methodologies appear to be sound. There was an attempt to replicate their findings in cases where an appropriate dataset was available. I agree that there are no gross differences in TCR diversity between males and females.

Weaknesses:

Overall, the sample size is small given that it is an outbred population. The cleaner experiment would have been to study the impact of sex in a number of inbred MHC I/II identical mouse strains or in humans with HLA-identical backgrounds.

It is unclear whether there was consensus between the three databases they used regarding the antigens recognized by the TCR sequences. Given the very low overlap between the TCR sequences identified in these databases and their dataset, and the lack of replication, they should tone down their excitement about the CD8 T cell sequences recognizing autoimmune and bacterial antigens being over-represented in females.

The dataset could be valuable to the community.

Reviewer #2 (Public review):

Summary:

This study addresses the hypothesis that the strikingly higher prevalence of autoimmune diseases in women could be the result of biased thymic generation or selection of TCR repertoires. The biological question is important, and the hypothesis is valuable. Although the topic is conceptually interesting and the dataset is rich, the study has a number of major issues that require substantial improvement. In several instances, the authors conclude that there are no sex-associated differences for specific parameters, yet inspection of the data suggests visible trends that are not properly quantified. The authors should either apply more appropriate statistical approaches to test these trends or provide stronger evidence that the observed differences are not significant. In other analyses, the authors report the differences between sexes based on a pulled analysis of TCR sequences from all the donors, which could result in differences driven by one or two single donors (e.g., having particular HLA variants) rather than reflect sex-related differences.

Strengths:

The key strength of this work is the newly generated dataset of TCR repertoires from sorted thymocyte subsets (DP and SP populations). This approach enables the authors to distinguish between biases in TCR generation (DP) and thymic selection (SP). Bulk TCR sequencing allows deeper repertoire coverage than single-cell approaches, which is valuable here, although the absence of TRA-TRB pairing and HLA context limits the interpretability of antigen specificity analyses. Importantly, this dataset represents a valuable community resource and should be openly deposited rather than being "available upon request."

Weaknesses:

Major:

(1) The authors state that there is "no clear separation in PCA for both TRA and TRB across all subsets." However, Figure 2 shows a visible separation for DP thymocytes (especially TRA, and to a lesser degree TRB) and also for TRA of Tregs. This apparent structure should be acknowledged and discussed rather than dismissed.

(2) Supplementary Figures 2-5 involve many comparisons, yet no correction for multiple testing appears to be applied. After appropriate correction, all the reported differences would likely lose significance. These analyses must be re-evaluated with proper multiple-testing correction, and apparent differences should be tested for reproducibility in an external dataset (for example, the pediatric thymus and peripheral blood repertoires later used for motif validation).

(3) Supplementary Figure 6 suggests that women consistently show higher Rényi entropies across all subsets. Although individual p-values are borderline, the consistent direction of change is notable. The authors should apply an integrated statistical test across subsets (for example, a mixed-effects model) to determine whether there is an overall significant trend toward higher diversity in females.

(4) Figures 4B and S8 clearly indicate enrichment of hydrophobic residues in female CDR3s for both TRA and TRB (excluding alanine, which is not strongly hydrophobic). Because CDR3 hydrophobicity has been linked to increased cross-reactivity and self-reactivity (see, e.g., Stadinski et al., Nat Immunol 2016), this observation is biologically meaningful and consistent with higher autoimmune susceptibility in females.

(5) The majority of "hundreds of sex-specific motifs" are probably donor-specific motifs confounded by HLA restriction. This interpretation is supported by the failure to validate motifs in external datasets (pediatric thymus, peripheral blood). The authors should restrict analysis to public motifs (shared across multiple donors) and report the number of donors contributing to each motif.

(6) When comparing TCRs to VDJdb or other databases, it is critical to consider HLA restriction. Only database matches corresponding to epitopes that can be presented by the donor's HLA should be counted. The authors must either perform HLA typing or explicitly discuss this limitation and how it affects their conclusions.

(7) Although the age distributions of male and female donors are similar, the key question is whether HLA alleles are similarly distributed. If women in the cohort happen to carry autoimmune-associated alleles more often, this alone could explain observed repertoire differences. HLA typing and HLA comparison between sexes are therefore essential.

(8) In some analyses (e.g., Figures 8C-D) data are shown per donor, while others (e.g., Fig. 8A-B) pool all sequences. This inconsistency is concerning. The apparent enrichment of autoimmune or bacterial specificities in females could be driven by one or two donors with particular HLAs. All analyses should display donor-level values, not pooled data.

(9) The reported enrichment of matches to certain specificities relative to the database composition is conceptually problematic. Because the reference database has an arbitrary distribution of epitopes, enrichment relative to it lacks biological meaning. HLA distribution in the studied patients and HLA restrictions of antigens in the database could be completely different, which could alone explain enrichment and depletions for particular specificities. Moreover, differences in Pgen distributions across epitopes can produce apparent enrichment artifacts. Exact matches typically correspond to high-Pgen "public" sequences; thus, the enrichment analysis may simply reflect variation in Pgen of specific TCRs (i.e., fraction of high-Pgen TCRs) across epitopes rather than true selection. Consequently, statements such as "We observed a significant enrichment of unique TRB CDR3aa sequences specific to self-antigens" should be removed.

(10) The overrepresentation of self-specific TCRs in females is the manuscript's most interesting finding, yet it is not described in detail. The authors should list the corresponding self-antigens, indicate which autoimmune diseases they relate to, and show per-donor distributions of these matches.

(11) The concept of polyspecificity is controversial. The authors should clearly explain how polyspecific TCRs were defined in this study and highlight that the experimental evidence supporting true polyspecificity is very limited (e.g., just a single TCR from Figure 5 from Quiniou et al.).

Minor:

(1) Clarify why the Pgen model was used only for DP and CD8 subsets and not for others.

(2) The Methods section should define what a "high sequence reliability score" is and describe precisely how the "harmonized" database was constructed.

(3) The statement "we generated 20,000 permuted mixed-sex groups" is unclear. It is not evident how this permutation corrects for individual variation or sex bias. A more appropriate approach would be to train the Pgen model separately for each individual's nonproductive sequences (if the number of sequences is large enough).

eLife Assessment

The authors ask whether a simple whole-head spectral power analysis of human magnetoencephalography data recorded at rest in a large cohort of adults shows robust effects of age, and their results provide compelling evidence that it does. The relative simplicity of the analysis is a major strength of the paper, and the authors are careful to control for many different confounds - although perhaps highly correlated factors like brain anatomy still pose a slight issue. The paper provides a valuable power analysis framework that should inform researchers across the broader neuroimaging community

Reviewer #1 (Public review):

Summary:

This is a careful, well-powered treatment of age effects in resting-state MEG. Rather than extracting (say) complex connectivity measures, the authors look at the 'simplest possible thing': changes in the overall power spectrum across age.

Strengths:

They find significant age-related changes at different frequency bands: broadly, attenuation at low-frequency (alpha) and increased beta. These patterns are identified in a large dataset (CamCAN) and then verified in other public data.

Weaknesses:

Some secondary interpretations (what is "unique" to age vs global anatomy) may go beyond what the statistics strictly warrant in the current form, but these can be tightened with (I think, fairly quick) additions already foreshadowed by the authors' own analyses.

Aims:

The authors set out to replace piecemeal, band-by-band ageing claims with t-maps, and Cohen's f2 over sensors×frequency ("GLM-Spectrum").

On CamCAN, six spatio-spectral peaks survive relatively strict statistical controls. The larger effects are in low-frequency and upper-alpha/beta ranges (f2 approx 0.2-0.3), while lower-alpha and gamma reach significance but with small practical impact (f2 < 0.075). A nice finding is that the same qualitative profile appears in three additional independent datasets.

Two analyses are especially interesting. First, the authors show a difference between absolute and relative spectral magnitude (basically, within-subject normalization). Relative scaling sharpens the spectral specificity of the spatial maps, while absolute magnitude is dominated by a broad spatial mode that correlates positively across frequencies, likely reflecting head-position/field-spread factors. The replication of the main age profile is robust to preprocessing decisions (e.g., SSS movement compensation choices) - the bigger determinant of the effect is whether they apply sensor normalization (relative vs absolute).

Second, lots of brain-related things might be related to age, and the authors spend some time trying to back out confounds/covariates. This section is handled transparently (in general, I found the writing style very clear throughout) - they examine single covariates (sex, BP, GGMV, etc.) and compare simple vs partial age effects. For example, aging is correlated with reductions in global grey-matter volume (GGMV), but it would be nice to find a measure that is independent of this: controlling for GGMV (via a linear model) reduces age-related effect sizes heterogeneously across space/frequency but does not eliminate them, a nuance the authors treat carefully.

This is a nice paper, and I have only a few concrete suggestions:

(1) High-gamma:

There can be a lot of EMG / eye movement contamination (I know these were RS eyes closed data, but still..) above 30-40 Hz, and these effects are the weakest anyway. Could you add an analysis (e.g., ICA/label-based muscle component removal) and show the gamma band's sensitivity to that step? Or just note this point more clearly?

(2) GGMV confound control:

Controlling for GGMV reduces, but does not eliminate, age effects. I have a few questions about this: a) Could we see the residuals as a function of age? I wonder if there are non-linear effects or something else that the regression is not accounting for. Also, b) GGMV and age are highly colinear - is this an issue? Can regression really split them apart robustly? I think by some cunning orthogonalisation, you can compute the effect of age independent of GGVM. I don't think this is the same as the effect 'adjusted' for GGMV (which is what is shown here if I'm reading it correctly). Finally, of course, GGMV might actually be the thing you want to look at (because it might more accurately reflect clinical issues) - so strong correlations are not really a problem: I think really the focus might even be on using MEG to predict GGMV and controlling for age.

Reviewer #2 (Public review):

This paper describes the application of the "GLM-Spectrum" mass univariate approach to examine the effects of age on M/EEG power spectra. Its strengths include promotion of the unbiased approach, suitable for future meta/mega-analyses, and the provision of effect sizes for powering future studies. These are useful contributions to the literature. What is perhaps lacking is a discussion of the limitations of this approach, in comparison to other methods.

An analogy is the mass univariate approach to spatial localisation of effects in fMRI/PET images. This approach is unbiased by prior assumptions about the organisation of the brain, but potentially also less sensitive, by ignoring that prior knowledge. For example, a voxelwise univariate approach is less sensitive to detecting effects in functionally homogeneous brain regions, where SNR can be increased by averaging over voxels. In the context of power spectra, the authors' approach deliberately ignores knowledge about the dominant frequency bands/oscillations in human power spectra. This is in contrast to approaches like FOOOF and IRASA, which explicitly parametrise frequency components. I am not saying these methods are better; I just think that the authors should acknowledge that these approaches have advantages over their mass univariate approach (in sensitivity and interpretation; see below). I guess it is a type of bias-sensitivity trade-off: the authors want to avoid bias, but they should acknowledge the corresponding loss of sensitivity, as well as loss of interpretation compared to model-based approaches (i.e, models that parameterise frequency; I don't mean the statistical models for each frequency separately).

An example of the interpretational loss can be seen in the authors' observation of opposite-signed effects of age around the alpha peak. While the authors acknowledge that this pattern can arise from a reduction in alpha frequency with age, this is an indirect inference, and a direct (and likely much more sensitive) approach would be to parametrise and estimate the peak alpha frequency directly for each participant, as done with FOOOF for example (possibly with group priors, as in Medrano et al, 2025, EJN). The authors emphasise the nonlinear effects of age in Figure 2A, but their approach cannot test this directly (e.g., in terms of plotting effects of age on frequency, magnitude, and width for each participant), so for me, this figure illustrates a weakness of their approach, not a strength.

Then I think the section "Two dissociable and opposite effects in the alpha range" in the Discussion section is confusing, because if there is a single reduction in alpha peak frequency and magnitude with age, then there is only one "effect", not "two dissociable" ones. If the authors do want to claim that there are two dissociable age effects within the alpha range, then they need to do a statistical test, e.g., that the topographies of low and high alpha are significantly different. This then reveals another limitation of the mass univariate approach - that space (channel) is not parametrised either - so one cannot test for significant channel x effect interactions within this framework, as necessary to really claim a dissociation (e.g., in underlying neural generators).

While the authors show that normalisation of each person's power spectra by the sum across frequencies helps improve some statistics, they might want to say more about disadvantages of this approach, e.g., loss of sensitivity to any effects (eg of age) that are broadly distributed across majority of frequencies, loss of real SI units (absolute effect sizes) (as well as problems if normalisation were used for techniques like FOOOF, where the 1/f exponent would be affected).

The authors should give more information on how artifactual ICs were defined. This may be important for cardiac artefacts, since Schmidt et al (2004, eLife) have pointed out how "standard" ICA thresholds can fail to remove all cardiac effects. This is very important for the effects of age, given that age affects cardiac dynamics (even though the focus of Schmidt et al is the 1/f exponent, could residual cardiac effects cause artifactual age effects in current results, even above ~1Hz?).

The authors should clarify the precise maxfilter arguments, and explain what "reference" was used for the "trans" option - e.g., did the authors consider transforming the data to match a sphere at the centre of the helmet, which might not only remove some of the global power differences due to different head positions, but also be best for generalisation of the effect sizes they report to future studies (assuming the centre of the helmet is the most likely location on average)? And on that matter, did head positions actually differ by age at all?

eLife Assessment

This study explores how exogenous attention operates at the finest spatial scale of vision, within the foveola - a topic that has not been previously explored. The question is important for understanding how attention shapes perception, and how it differs between the periphery and the central regions of highest visual acuity. The evidence is compelling, as shown by carefully designed experiments with state-of-the-art eye tracking to monitor attended locations just a few tens of minutes of arc away from the fixation target, but additional clarification regarding analyses and implications for vision and oculomotor control would broaden the impact of the study.

Reviewer #1 (Public review):

Summary:

The manuscript investigates how exogenous attention modulates spatial frequency sensitivity within the foveola. Using high-precision eye-tracking and gaze-contingent stimulus control, the authors show that exogenous attention selectively improves contrast sensitivity for low- to mid-range spatial frequencies (4-8 cycles/degree), but not for higher frequencies (12-20 CPD). In contrast, improvements in asymptotic performance at the highest contrast levels occur across all spatial frequencies. These results suggest that, even within the foveola, exogenous attention operates through a mechanism similar to that observed in peripheral vision, preferentially enhancing lower spatial frequencies.

Strengths:

The study shows strong methodological rigor. Eye position was carefully controlled, and the stimulus generation and calibration were highly precise. The authors also situate their work well within the existing literature, providing a clear rationale for examining the fine-grained effects of exogenous attention within the foveola. The combination of high spatial precision, gaze-contingent presentation, and detailed modeling makes this a valuable technical contribution.

Weaknesses:

The manipulation of attention raises some interpretive concerns. Clarifying this issue, together with additional detail about statistics, participant profiles, other methodological elements, and further discussion in relation to oculomotor control in general, could broaden the impact of the findings.

Reviewer #2 (Public review):

Summary:

This study aims to test whether foveal and non-foveal vision share the same mechanisms for endogenous attention. Specifically, they aim to test whether they can replicate at the foveola previous results regarding the effects of exogenous attention for different spatial frequencies.

Strengths:

Monitoring the exact place where the gaze is located at this scale requires very precise eye-tracking methods and accurate and stable calibration. This study uses state-of-the-art methods to achieve this goal. The study builds on many other studies that show similarities between foveal vision and non-foveal vision, adding more data supporting this parallel.

Weaknesses:

The study lacks a discussion of the strength of the effect and how it relates to previous studies done away from the fovea. It would be valuable to know if not just the range of frequencies, but the size of the effect is also comparable.

Reviewer #3 (Public review):

Summary:

This paper explores how spatial attention affects foveal information processing across different spatial frequencies. The results indicate that exogenously directed attention enhances contrast sensitivity for low- to mid-range spatial frequencies (4-8 CPD), with no significant benefits for higher spatial frequencies (12-20 CPD). However, asymptotic performance increased as a result of spatial attention independently of spatial frequency.

Strengths:

The strengths of this article lie in its methodological approach, which combines a psychophysical experiment with precise control over the information presented in the foveola.

Weaknesses:

The authors acknowledge that they used the standard approach of analyzing observer-averaged data, but recognize that this method has limitations: it ignores the uncertainty associated with parameter estimates and the relationships between different parameters of the psychometric model. This may affect the interpretation of attentional effects. In the future, mixed-effects models at the trial level could overcome these limitations.

eLife Assessment

This valuable study provides solid evidence for deficits in aversive taste learning and taste coding in a mouse model of autism spectrum disorders. Specifically, the authors found that Shank3 knockout mice exhibit behavioral deficits in learning and extinction of conditioned taste aversion, and calcium imaging of the gustatory cortex identified impaired neuronal responses to taste stimuli. This paper will likely be of interest to researchers studying how learning and sensory processes are affected by genetic causes of autism spectrum disorders.

Reviewer #1 (Public review):

Summary:

The study from Wu and Turrigiano investigates how disruption of taste coding in a mouse model of autism spectrum disorders (ASDs) affects aversive learning in the context of a conditioned taste aversion (CTA) paradigm. The experiments combine 2-photon calcium imaging of neurons in the gustatory portion of the anterior insular cortex (i.e., gustatory cortex) with behavioral training and testing. The authors rely on Shank3 knockout mice as a model for ASDs. The authors found that Shank3 mice learn CTA more slowly and extinguish the memory more rapidly than control subjects. Calcium imaging identified impairments in taste-evoked activity associated with memory encoding and extinction. During memory encoding, the authors found less suppressed neuronal activity and increased correlated variability in Shank3 mice compared to controls. During extinction, they observed a faster loss of taste selectivity and degradation of taste discriminability in mutants compared to controls.

Strengths:

This is a well-written manuscript that presents interesting findings. The results on the learning and extinction deficits in Shank3 mice are of particular interest. Analyses of neural activity are well conducted and provide important information on the type of impaired cortical activity that may correlate with behavioral deficits.

Weaknesses:

(1) The experiments rely on three groups: CS-only WT, CTA WT, and CTA KO. Can the authors provide a rationale for not having a CS-only KO group?

(2) The authors design an effective behavioral paradigm comparing consumption of water and saccharin and tracking extinction (Figure 3). This paradigm shows differences in licking across distinct behavioral conditions. For instance, during T1, licking to water strongly differs from licking to saccharin for both WT and KO. During T2, licking to water strongly differs from licking to saccharin only for WT (much less for KO), and licking to saccharin in WT differs from that in KO. These differences in taste sampling across conditions could contribute to some of the effects on neural activity and discriminability reported in Figures 5 and 6. That is sucrose and water trials may be highly discriminable because in one case the mouse licks and in the other it does not (or licks much less). The author may want to address this issue.

(3) Are there any omission trials following CTA? If so, they should be quantified and reported. How are the omission trials treated with regard to the analyses?

(4) The authors describe the extinction paradigm as "alternative choice". In decision-making, alternative choice paradigms typically require 2 lateral spouts to report decisions following the sampling from a central spout. To avoid confusion, the authors may want to define their paradigm as alternative sampling.

(5) Figure 4 reports that CTA increases the proportion of neurons that consistently respond to saccharin and water across days. While the saccharin result could be an effect of aversive learning, it is less clear why the phenomenon would generalize to water as well. Can the authors provide an explanation?

(6) The recordings are performed in the part of the anterior insular cortex that is typically defined as "gustatory cortex" (GC). Given the functional heterogeneity of the anterior insular cortex (AIC) and given that the authors do not sample all of the anteroposterior extent of AIC, I would suggest being more explicit about their positioning in GC. Also, some citations (e.g., Gogolla et al, 2014) refer to the posterior insular cortex, which is considered more inherently multimodal than GC. GC multimodality is typically associative in nature, as only a few neurons respond to sound and light in naïve animals.

(7) It would be useful to add summary figures showing the extent of viral spread as well as GRIN lens placement.

(8) I encourage the authors to add Ns every time percentages are reported. How many neurons have been recorded in each condition? Can the authors provide the average number of neurons recorded per session and per animal?

(9) It looks like some animals learned more than others (Figure 1E or Figure 3C). Is it possible to compare neural activity across animals that showed different degrees of learning?

Reviewer #2 (Public review):

Wu and Turrigiano investigated how cortical taste coding during conditioned taste aversion (CTA) learning is affected in Shank3 knockout (KO) mice, a model of monogenic ASD. Using longitudinal two-photon calcium imaging of AIC neurons, the authors show that Shank3 KO mice exhibit reduced suppression of activity in a subset of neurons and a higher correlated variability in neural activity. This is accompanied by slower learning and faster extinction of aversive taste memories. These results suggest that Shank3 loss compromises the flexibility and stability of cortical representations underlying adaptive behaviour.

Major strengths:

(1) Conceptual significance: The study connects a molecular ASD risk gene (Shank3) to flexible sensory encoding, bridging genetics, systems neuroscience, and behaviour.

(2) Technical rigour: Longitudinal calcium imaging with cell-registration across learning and extinction sessions is technically demanding and well-executed.

(3) Behavioural paradigm: The use of both acquisition and extinction paradigms provides a more nuanced picture of learning dynamics.

(4) Analyses: Correlated variability, discriminability indices, and population decoding analyses are robust and appropriate for addressing behavioural and network-level coding changes.

Major weaknesses:

(1) Causality: The paper infers that increased correlated variability causes learning deficits, but no causal tests (e.g., optogenetic modulation of inhibition or interneuron rescue) are presented to confirm this.

(2) Behavioural scope: The study focuses exclusively on taste aversion; generalisation to other flexible learning paradigms (e.g., reversal or probabilistic tasks) is not addressed.

(3) Mechanistic insights: While providing interesting findings of altered sensory perception and extinction of learning-related signals in AIC, it offered nearly no mechanistic insights. This makes the interpretation, especially on how generalisable these findings are, difficult. Also, different reported findings are "potentially" connected, but the exact relation between increased correlated variability and faster loss of taste selectivity cannot be assessed.

Reviewer #3 (Public review):

In this study, Wu & Turrigiano investigate an ethologically relevant form of associative learning (conditioned taste aversion - CTA) and its extinction in the Shank3 KO mouse model of ASD. They also examine the underlying circuits in the anterior insular cortex (AIC) simultaneously, using two-photon calcium imaging through a GRIN lens. They report that Shank3 KO mice learn CTA slower and suggest that this is mediated by a reduction in tastant-stimulus activity suppression of AIC neurons and a reduced signal-to-noise ratio due to increased noise correlations in AIC neurons. Interestingly, once Shank3 KO mice acquire CTA, they extinguish the aversive memory more rapidly than wild-type mice. This accelerated extinction is accompanied by a faster loss of neuronal and population-level taste selectivity and coding in the AIC compared to WT mice.

This is an important study that uses in vivo methods to assess circuit dysfunction in a mouse model of ASD, related to sensory perception valence (in this case, taste). The study is well executed, the data are of high quality, and the analytical procedures are detailed. Furthermore, the behavioural paradigm is well thought out, particularly the approach for assessing extinction through repeated retrieval sessions (T1-T5), which effectively tests discrimination between saccharin and water rather than relying solely on lick counts or total consumption as a measure of extinction. Finally, the statistical tests used are appropriate and justified.

There is, however, a missing link between the behavioural findings and the underlying mechanisms. More specifically:

(1) The authors don't make a causal link between the behaviour and AIC neurophysiology, both the percentage of suppressed cells and the coactivity measurements. For the % of suppressed cells, it seems that both WT and KO cells are suppressed in the transition between CST1 and CST2 (Figure 1L), yet only the WT mice exhibit CTA (at least by CST2). For the taste-elicited coactivity measure, it seems that there is an increase in coactivity from CST1 to CST2 in WT (Figure 2C - blue, although not statistically tested?), but persistently higher coactivity in KO. Is this change of coactivity in WT important for the expression of CTA? Plotting behavioral performance (from Figure 1G) against coactivity (from Figure 2C) for each animal would be informative.

(2) Shank3 KO cells already show an increase in baseline coactivity (Figure 2- figure supplement 1), and the authors never examine CS-only responses in the KO group, therefore making it difficult to determine whether elevated coactivity and noise correlations reflect a generalized AIC abnormality in Shank3 KOs (perhaps through impaired PV-mediated inhibition in insular cortex - Gogolla et al, 2014) that is not directly responsible/related to CTA?

(3) How do the authors interpret the large range of lick ratios (Figure 1G) for WT (almost bi-modal distribution)? Is there a within-subject correlation with any of the neurophysiological measurements to suggest a relationship between AIC neurophysiology and behavioural expression of CTA?

(4) Indeed, CTA appears to be successfully achieved for Shank3 KO mice delayed by 1 day, as the level of saccharin aversion during the first retrieval session (T1) is comparable between Shank3 KO and WTs. In this context, not extending the first part of the paradigm to include CST3 seems to be a missed opportunity. Doing so would have allowed for within-cell and within-subject comparison of taste-elicited pairwise correlation across the learning and to investigate the neural mechanism of delayed extinction in KOs more effectively.

(5) How to interpret Figure 5F: Absolute discriminability is lower for T5 for CTA WT and CTA KO compared to CS-only? Why would AIC neurons have less information on taste identity by the end of extinction than during the unconditioned (CS-only) condition? And if that is the case, how is decoding accuracy in Figure 6C higher in T5 for CTA WT vs CS-only?

eLife Assessment

This important study shows that different forms and mixtures of cardenolide toxins in tropical milkweed, especially nitrogen- and sulfur-containing types, change how monarch caterpillars eat, grow, and store these chemicals under laboratory conditions. It provides solid evidence to demonstrate that chemical diversity within a single group of plant toxins (cardenolides) can have combined effects on even highly specialized herbivores that are different from what one would expect from each toxin alone. However, as all experiments used leaf-disc assays with fixed "natural" toxin ratios and only one adapted herbivore species, tests on living plants, other mixture designs, and non-adapted herbivores would make the broader conclusions stronger.

Reviewer #1 (Public review):

Summary:

In the ecological interactions between wild plants and specialized herbivorous insects, structural innovation-based diversification of secondary metabolites often occurs. In this study, Agrawal et al. utilized two milkweed species (Asclepias curassavica and Asclepias incarnata) and the specialist Monarch butterfly (Danaus plexippus) as a model system to investigate the effects of two N,S-cardenolides - formed through structural diversification and innovation in A. curassavica-on the growth, feeding, and chemical sequestration of D. plexippus, compared to other conventional cardenolides. Additionally, the study examined how cardenolide diversification resulting from the formation of N,S-cardenolides influences the growth and sequestration of D. plexippus. On this basis, the research elucidates the ecophysiological impact of toxin diversity in wild plants on the detoxification and transport mechanisms of highly adapted herbivores.

Strengths:

The study is characterized by the use of milkweed plants and the specialist Monarch butterfly, which represent a well-established model in chemical ecology research. On one hand, these two organisms have undergone extensive co-evolutionary interactions; on the other hand, the butterfly has developed a remarkable capacity for toxin sequestration. The authors, building upon their substantial prior research in this field and earlier observations of structural evolutionary innovation in cardenolides in A. curassavica, proposed two novel ecological hypotheses. While experimentally validating these hypotheses, they introduced the intriguing concept of a "non-additive diversity effect" of trace plant secondary metabolites when mixed, contrasting with traditional synergistic perspectives, in their impact on herbivores.

Weaknesses:

The manuscript has two main weaknesses. First, as a study reliant on the control of compound concentrations, the authors did not provide sufficient or persuasive justification for their selection of the natural proportions (and concentrations) of cardenolides. The ratios of these compounds likely vary significantly across different environmental conditions, developmental stages, pre- and post-herbivory, and different plant tissues. The ecological relevance of the "natural proportions" emphasized by the authors remains questionable. Furthermore, the same compound may even exert different effects on herbivorous insects at different concentrations. The authors should address this issue in detail within the Introduction, Methods, or Discussion sections.

Second, the study was conducted using leaf discs in an in vitro setting, which may not accurately reflect the responses of Monarch butterflies on living plants. This limitation undermines the foundation for the novel ecological theory proposed by the authors. If the observed phenomena could be validated using specifically engineered plant lines-such as those created through gene editing, knockdown, or overexpression of key enzymes involved in the synthesis of specific N,S-cardenolides - the findings would be substantially more compelling.

Reviewer #2 (Public review):

This study examined the effects of several cardenolides, including N,S-ring containing variants, on sequestration and performance metrics in monarch larvae. The authors confirm that some cardenolides, which are toxic to non-adapted herbivores, are sequestered by monarchs and enhance performance. Interestingly, N,S-ring-containing cardenolides did not have the same effects and were poorly sequestered, with minimal recovery in frass, suggesting an alternate detoxification or metabolic strategy. These N,S-containing compounds are also known to be less potent defences against non-adapted herbivores. The authors further report that mixtures of cardenolides reduce herbivore performance and sequestration compared to single compounds, highlighting the important role of phytochemical diversity in shaping plant-herbivore interactions.

Overall, this study is clearly written, well-conducted and has the potential to make a valuable contribution to the field. However, I have one major concern regarding the interpretations of the mixture results. From what I understand of the methods, all tested mixtures contain all five compounds. As such, it is not possible to determine whether reduced performance and sequestration result from the complete mixture or from the presence of a single compound, such as voruscharin for performance and uscharin for sequestration. For instance, if all compounds except voruscharin (or uscharin) were combined, would the same pattern emerge? I suspect not, since the effects of the individual N,S-containing compounds alone are generally similar to those of the full mixture (Figure S3). By taking the average of all single compounds, the individual effects of the N,S-containing ones are being inflated by the non-N,S-containing ones (in the main text, Figure 4). In the mix, of course, they are not being 'diluted', as they are always present. This interpretation is further supported by the fact that in the equimolar mix, the relative proportion of voruscharin decreases (from 50% in the 'real mix'), and the target measurements of performance and sequestration tend to increase in the equimolar mix compared to the real mix.

Despite this issue, the discussion of mixtures in the context of plant defence against both adapted and non-adapted herbivores is fascinating and convincing. The rationale that mixtures may serve as a chemical tool-kit that targets different sets of herbivores is compelling. The non-N,S cardenolides are effective against non-adapted herbivores and the N,S-containing cardenolides are effective against adapted herbivores. However, the current experiments focus exclusively on an adapted species. It would be especially interesting to test whether such mixtures reduce overall herbivory when both adapted and non-adapted species are present.

It remains possible that mixtures, even in the absence of voruscharin or uscharin, genuinely reduce sequestration or performance; however, this would need to be tested directly to address the abovementioned concern.

Author response:

Thanks for these insightful reviews and your summary assessment. We certainly agree that ours was a laboratory study with a single specialized insect, and both mixtures types had all five compounds (controlling for total toxin concentration). Thus, our conclusion that combined effects of naturally occurring toxins (within the cardenolide class) have non-additive effects for the specialized sequestering monarch are constrained by our experimental conditions. In our assay we used two mixture types, equimolar and “natural” proportions. We acknowledge that the natural proportions will vary with plant age, damage history, etc. of the host plant, Asclepias curassavica. Our proportions were based on growing the plants a few different times under variable conditions. Although we did not conduct these experiments on non-adapted insects, we discuss a related experiment that was conducted with wild-type and genetically engineered Drosophila (Lopez-Goldar et al. 2024, PNAS). In sum, we appreciate the reviewers’ comments.

eLife Assessment

This study provides a useful advance in generating mouse oligodendrocytes by direct lineage conversion from cortical astrocytes. The authors demonstrate that Sox10 converts astrocytes to MBP+ oligodendrocytes, whereas Olig2 expression converts astrocytes to PDFRalpha+ oligodendrocyte progenitor cells. The data supporting the conclusions are solid, but there are concerns regarding select figures and the absence of functional validation.

Reviewer #1 (Public review):

Bajohr and colleagues propose a transcription factor-driven approach to generating bonafide oligodendrocyte lineage cells (OLCs) from primary mouse astrocytes. Ectopic expression of Olig2, Sox10, or Nkx6.2 in isolated astrocytes produced a range of OLC-like cell states, with Sox10 emerging from lineage tracing and single cell RNA sequencing experiments as the most successful transcription factor in driving direct lineage reprogramming. The authors strengthened their claims with an unbiased, deep learning perturbation model to predict genetic drivers of the astrocyte cluster to OLC cluster transition observed in their scRNA seq dataset. Here, Sox10 surfaced in the top ten correlated genes, and the top transcription factor, mediating this fate shift. Altogether, this paper presents an interesting approach to generate OLCs, a cell type historically difficult to procure, from primary mouse astrocytes to study this lineage in development and disease and perhaps repopulate it in dysmyelinating conditions. While this certainly addresses a technical gap in the field, authors defined iOLCs as ones with lineage-specific gene expression and morphological characteristics, lacking any functional analysis to assess the reprogrammed cells' capacity to myelinate. This comment and other critiques are discussed below.

While Sox10 and Mbp expression in iOLCs, as confirmed by IHC, is a promising result suggesting that ectopic Sox10 instructs transduced cells to develop into cells of myelinating potential, functional confirmation is essential. As mentioned in the discussion, the absence of a substrate for myelination may have also contributed to the low DLR efficiency. Co-culturing Sox10 iOLCs with primary neurons and examining the cells' potential to engage and enwrap axons would greatly strengthen the authors' claim that this could be an effective therapeutic approach to myelin regeneration in vivo, or even a technical approach to studying myelin dynamics in vitro.

In Figure 1B, it appears that Mbp expression in tdTomato+ cells decreases in Sox10 transduced iOLs during the observed time period. Can the authors elaborate on this result, given that MBP expression is crucial for myelination and should, if anything, increase with time?

The authors acknowledge that there is a conversion of tdTomato- zsGreen+ cells with an astrocyte-like morphology to OLC cells expressing Mbp following Sox10 induction (Supplementary figure 5C,D). While they note the diversity of the astrocyte lineage in the discussion, further analysis should be applied to this subset of cells to confirm the subset of astrocyte or progenitor-like cell type that gives rise to their cell endpoint of interest (Sox10-driven Mbp+ iOLs).

Finally, ectopic expression of Olig2 and Sox10 in primary astrocytes resulted in very different OLC subtypes, as evidenced by OLC marker expression seen in IHC and the subclustering of these cell types in scRNA seq. Although this diversity in OLC type and generation efficiency follows with previous reports showing that these two transcription factors vary in effect, might the authors further discuss this discrepancy given that the two transcription factors regulate one another (as mentioned in the introduction) and should theoretically give rise to more similar cells? Perhaps due to the lower specificity of Olig2 in marking a pure OLC population relative to Sox10?

Author response:

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

Reviewer #1 (Public Review):

Faiz et al. investigate small molecule-driven direct lineage reprogramming of mouse postnatal mouse astrocytes to oligodendrocyte lineage cells (OLCs). They use a combination of in vitro, in vivo, and computational approaches to confirm lineage conversion and to examine the key underlying transcription factors and signaling pathways. Lentiviral delivery of transcription factors previously reported to be essential in OLC fate determination-Sox10, Olig2, and Nkx2.2-to astrocytes allows for lineage tracing. They found that these transcription factors are sufficient in reprogramming astrocytes to iOLCs, but that the OLCs range in maturity level depending on which factor they are transfected with. They followed up with scRNA-seq analysis of transfected and control cultures 14DPT, confirming that TF-induced astrocytes take on canonical OLC gene signatures. By performing astrocyte lineage fate mapping, they further confirmed that TF-induced astrocytes give rise to iOLCs. Finally, they examined the distinct genetic drivers of this fate conversion using scRNA-seq and deep learning models of Sox10- astrocytes at multiple time points throughout the reprogramming. These findings are certainly relevant to diseases characterized by the perturbation of OLC maturation and/or myelination, such as Multiple Sclerosis and Alzheimer's Disease. Their application of such a wide array of experimental approaches gives more weight to their findings and allows for the identification of additional genetic drivers of astrocyte to iOLC conversion that could be explored in future studies. Overall, I find this manuscript thoughtfully constructed and only have a few questions to be addressed. 

(1) The authors suggest that Sox10- and Olig2- transduced astrocytes result in distinct subpopulations iOLCs. Considering it was discussed in the introduction that these TFs cyclically regulate one another throughout differentiation, could they speculate as to why such varying iOLCs resulted from the induction of these two TFs? 

We thank the Reviewer for the opportunity to speculate. We hypothesize that Sox10 and Olig2 may induce different OLCs as a result of differential activation of downstream genes within the gene regulatory network, which are important for OPC, committed OLC and mature OL identity [1]. In support of this, we found different expression levels of genes involved in downstream OLC specification networks [1], including Sox6, Tcfl2 and Myrf, at D14 (Author response image 1), following further analysis of our RNA-seq data.

Author response image 1.

Expression of OLC regulatory network genes in Sox10- and Olig2- cultures. Violin plots show gene expression levels (log-normalized) of downstream OLC regulatory genes (Sox6, Zeb2, Tcf7l2, Myrf, Zfp488, Nfatc2, Hes5, Id2) between Sox10 and Olig2 treated OLCs at 14 days post transduction. Analysis was performed on oligodendrocyte progenitor and mature oligodendrocyte clusters (from Manuscript Figure 1D, clusters 3 and 8).

(2) In Figure 1B it appears that the Sox10- MBP+ tdTomato+ cells decreases from D12 to D14. Does this make sense considering MBP is a marker of more mature OLCs? 

Thank you for this comment. To address this, we compared the number of MBP+tdTomato+ Sox10 cells across reprogramming timepoints. We saw no difference between the number of MBP+tdTomato+ OLs at D12 and D14 (Author response image 2, p = 0.2314). However,  we do see a [nonsignificant] decrease in MBP+tdTomato+ Sox10 cells from D12 to D22 (Manuscript Supplementary Figure 3B, Author response image 2, p= 0.0543), which suggests that culture conditions are not optimal for longer-term cell survival [2], [3], [4].  

Author response image 2.

Comparison of Sox10- induced MBP+tdTomato+ iOLCs over time. Quantification of MBP⁺tdTomato⁺ iOLs in Sox10 cultures at D8 (n=5), D10 (n=5), D12 (n=5), D14 (n=7) and D22 (n=3) post transduction. Data are presented as mean ± SEM, each data point represents one individual cell culture experiment, Brown-Forsythe and Welch ANOVA on transformed percentages with Dunnett’s T3 multiple comparisons test (*= p<0.05).  

(3) Previous studies have shown that MBP expression and myelination in vitro occurs at the earliest around 4-6 weeks of culturing. When assessing whether further maturation would increase MBP positivity, authors only cultured cells up to 22 DPT and saw no significant increase. Has a lengthier culture timeline been attempted? 

We agree with the Reviewer that previous studies of pluripotent stem cell derived (hESCs or iPSCs) have shown MBP+ OLCs in vitro around 4-6 weeks [5], [6], [7]. However,  studies of neural stem cells [8] or fibroblasts [9] conversion show OLC appearance after 7 and 24 days, respectively, demonstrating that OLCs can be generated in vitro within 1-3 weeks of plating. Moreover, as noted above in response to #2, we see fewer MBP+ cells at  22DPT, suggesting that extended time in culture may require additional factors for support. Therefore, we did not attempt longer timepoints. 

(4) Figure S4D is described as "examples of tdTomatonegzsGreen+OLCmarker+ cells that arose from a tdTomatoneg cell with an astrocyte morphology." The zsGreen+ tdTomato- cell is not convincingly of "astrocyte morphology"; it could be a bipolar OLC. To strengthen the conclusions and remove this subjectivity, more extensive characterizations of astrocyte versus OLC morphology in the introduction or results are warranted. This would make this observation more convincing since there is clearly an overlap in the characteristics of these cell types.  

We thank the reviewer for this excellent suggestion. To assess astrocyte morphology, we measured the cell size, nucleus size, number of branches and branch thickness of 70 Aldh1l1+tdTomato+ astrocytes in tamoxifen-labelled Aldh1l1-CreERT2;Ai14 cultures (new Supplemental Table 1). To assess OPC morphology, we  performed IHC for PDGFRa in iOLC cultures and measured the same parameters in 70 PDGFRa+ OPCs (new Supplemental Table 1).  We found that astrocytes were characterized by larger branch thickness, cell length and nucleus size, while OPCs showed a larger number of branches (new Supplemental Figure 1, and Author response image 3 below). Based on this framework, the AAV9-GFAP::zsGreen^posAldh1l1-tdTomato^neg and AAV9-GFAP::zsGreen^posAldh1l1-tdTomato^posstarting cells tracked fall within the bounds of ‘astrocytes’. We have revised the manuscript to include this more rigorous characterization (Line 119-124, Page 4; Line 307-312, Page 9; Line 323-326, Page 9). We also demonstrate (below) that the GFAP::zsGreen^pos Aldh1l1-tdTomato^pos and GFAP::zsGreen^posAldh1l1-tdTomato^neg starting cell depicted in Figure 2G and Supplemental Figure 5D is consistent with astrocyte morphology (Author response image 3). 

Author response image 3.

Morphological characterization of astrocytes, oligodendrocyte lineage cells, and starting cells. Quantification of the (A) cell length, (B) nucleus size, (C) number of branches, and (D) branch thickness iAldh1l1+tdTomato+ and PDGFRα+ OPCs (n= 70 per cell type, data are presented as mean ± SEM). Orange line indicates parameter value for GFAP::zsGreen^posAldh1l1-tdTomato^pos starting cell in Figure 2G. Green line indicates parameter value for GFAP::zsGreen^pos Aldh1l1-tdTomato^neg starting cell in Supplemental Figure 5D.

Reviewer #2 (Public Review):             

The study by Bajohr investigates the important question of whether astrocytes can generate oligodendrocytes by direct lineage conversion (DLR). The authors ectopically express three transcription factors - Sox10, Olig2 and Nkx6.2 - in cultured postnatal mouse astrocytes and use a combination of Aldh1|1-astrocyte fate mapping and live cell imaging to demonstrate that Sox10 converts astrocytes to MBP+ oligodendrocytes, whereas Olig2 expression converts astrocytes to PDFRalpha+ oligodendrocyte progenitor cells. Nkx6.2 does not induce lineage conversion. The authors use single-cell RNAseq over 14 days post-transduction to uncover molecular signatures of newly generated iOLs.  

The potential to convert astrocytes to oligodendrocytes has been previously analyzed and demonstrated. Despite the extensive molecular characterization of the direct astrocyteoligodendrocyte lineage conversion, the paper by Bajohr et al. does not represent significant progress. The entire study is performed in cultured cells, and it is not demonstrated whether this lineage conversion can be induced in astrocytes in vivo, particularly at which developmental stage (postnatal, adult?) and in which brain region. The authors also state that generating oligodendrocytes from astrocytes could be relevant for oligodendrocyte regeneration and myelin repair, but they don't demonstrate that lineage conversion can be induced under pathological conditions, particularly after white matter demyelination. Specific issues are outlined below. 

We thank the reviewer for this summary. We agree that there are a handful of reports of astrocytelike cells to OLC conversion [10], [11]. However, our study is the first study to confirm bonafide astrocyte to OLC conversion, which is important given the recent controversy in the field of in vivo astrocyte to neuron reprogramming [12]. In addition, the extensive characterization of the molecular timeline of reprogramming, highlights that although conversion of astrocytes is possible by ectopic expression of any of the three factors, the subtypes of astrocytes converted and maturity of OLCs produced may vary depending on the choice of TF delivered. Our findings will inform future in vivo studies of iOLC generation that aim to understand the impact of brain region, age, pathology, and sex, which are especially important given the diversity of astrocyte responses to disease [13], [14], [15].

(1) The authors perform an extensive characterization of Sox10-mediated DLR by scRNAseq and demonstrate a clear trajectory of lineage conversion from astrocytes to terminally differentiated MBP+ iOLCs. A similar type of analysis should be performed after Olig2 transduction, to determine whether transcriptomics of olig2 conversion overlaps with any phase of sox10 conversion.

We thank the Reviewer for this excellent comment. We chose to include an in-depth analysis of Sox10 in the manuscript, as Sox10-transduced cultures showed a higher percentage of mature iOLCs compared to Olig2 in our studies. We have added this specific rationale to the manuscript (Line 329-330-Page 9). 

Nonetheless, we also agree that understanding the underpinnings of Olig2-mediated conversion is important. Therefore, we used Cell Oracle [16] to understand the regulation of cell identity by Olig2.  in silico overexpression of Olig2 in our control time course dataset (D0, D3, D8 and D14) showed cell movement from cluster 1, characterized by astrocyte genes [Mmd2[17], Entpd2[18], H2-D1[19]], towards cluster 5, characterized by OPC genes [Pdgfra[20], Myt1[21]] validating astrocyte to OLC conversion by Olig2 (Author response image 4).

We hypothesize that reprogramming via Sox10 and Olig2 take different conversion paths to oligodendrocytes for the following reasons. 

(1) Differential astrocyte gene expression at D14 when cells are exposed to Sox10 and Olig2 (Manuscript Figure 1D-E [Sox10 characterized by Lcn2[19], C3[19]; Olig2 characterized by Slc6a11[22], Slc1a2[23]].

(2) Differential expression of key OLC gene regulatory network genes at D14 between cells treated with Sox10 and Olig2 (Author response image 1). 

Author response image 4.

in silico modeling of Olig2 reprogramming (A) UMAP clustering of Cre control treated cells from 0, 3, 8, and 14 days post transduction (DPT). (B) UMAP clustering from (A) overlayed with timepoint and treatment group. (C) Cell Oracle modeling of predicted cell trajectories following Olig2 knock in (KI), overlaid onto UMAP plot. Arrows indicate cell movement prediction with Olig2 KI perturbation.  

(2) A complete immunohistochemical characterization of the cultures should be performed at different time points after Sox10 and Olig2 transduction to confirm OL lineage cell phenotypes. 

We performed a complete immunohistochemical characterization of Ai14 cultures transduced with GFAP::Sox10-Cre and GFAP::Olig2-Cre. This system allows permanent labelling and therefore, enabled the tracking of transduced cells through the process or DLR, which we believe is the most appropriate way to characterize iOLC conversion efficiencies. We then confirmed the conversion of Aldh1l1+ astrocytes in Aldh1l1-CreERT2;Ai14 cultures transduced with GFAP::Sox10-zsGreen and GFAP::Olig2-zsGreen. In this system, GFAP drives the expression of zsGreen, and therefore, may not faithfully track all cells and lead to an underestimate of the numbers of converted cells. For example, iOLCs from Aldh1l1^neg astrocytes or iOLCs that have lost zsGreen expression following conversion. Therefore we use this system only to confirm astrocyte origin.

Nonetheless, we appreciate this comment and recognize that there may be differences in conversion efficiencies when analyzing Aldh1l1+ astrocytes versus all transduced cells. Therefore, we have softened the language in the manuscript (see below) regarding Olig2 and Sox10 generating different OLC phenotypes and now claim iOLC generation from both Sox10 and Olig2. We thank the Reviewer for this comment, and believe it has strengthened the discussion. 

Line 240, Page 7

Line 261-263, Page 8

Line 304-307, Page 8/9

Line 413-414, Page 11

References

(1) E. Sock and M. Wegner, “Using the lineage determinants Olig2 and Sox10 to explore transcriptional regulation of oligodendrocyte development,” Dev Neurobiol, vol. 81, no. 7, pp. 892–901, Oct. 2021, doi: 10.1002/dneu.22849.

(2) B. A. Barres, M. D. Jacobson, R. Schmid, M. Sendtner, and M. C. Raff, “Does oligodendrocyte survival depend on axons?,” Current Biology, vol. 3, no. 8, pp. 489–497, Aug. 1993, doi: 10.1016/0960-9822(93)90039-Q.

(3) A.-N. Cho et al., “Aligned Brain Extracellular Matrix Promotes Differentiation and Myelination of Human-Induced Pluripotent Stem Cell-Derived Oligodendrocytes,” ACS Appl. Mater. Interfaces, vol. 11, no. 17, pp. 15344–15353, May 2019, doi: 10.1021/acsami.9b03242.

(4) E. G. Hughes and M. E. Stockton, “Premyelinating Oligodendrocytes: Mechanisms Underlying Cell Survival and Integration,” Front. Cell Dev. Biol., vol. 9, Jul. 2021, doi: 10.3389/fcell.2021.714169.

(5) M. Ehrlich et al., “Rapid and efficient generation of oligodendrocytes from human induced pluripotent stem cells using transcription factors,” Proc Natl Acad Sci U S A, vol. 114, no. 11, pp. E2243–E2252, Mar. 2017, doi: 10.1073/pnas.1614412114.

(6) Y. Liu, P. Jiang, and W. Deng, “OLIG gene targeting in human pluripotent stem cells for motor neuron and oligodendrocyte differentiation,” Nat Protoc, vol. 6, no. 5, pp. 640–655, May 2011, doi: 10.1038/nprot.2011.310.

(7) S. A. Goldman and N. J. Kuypers, “How to make an oligodendrocyte,” Development, vol. 142, no. 23, pp. 3983–3995, Dec. 2015, doi: 10.1242/dev.126409.

(8) M. Faiz, N. Sachewsky, S. Gascón, K. W. A. Bang, C. M. Morshead, and A. Nagy, “Adult Neural Stem Cells from the Subventricular Zone Give Rise to Reactive Astrocytes in the Cortex after Stroke,” Cell Stem Cell, vol. 17, no. 5, pp. 624–634, Nov. 2015, doi:10.1016/j.stem.2015.08.002.

(9) F. J. Najm et al., “Transcription factor–mediated reprogramming of fibroblasts to expandable, myelinogenic oligodendrocyte progenitor cells,” Nat Biotechnol, vol. 31, no. 5, pp. 426–433, May 2013, doi: 10.1038/nbt.2561.

(10) A. Mokhtarzadeh Khanghahi, L. Satarian, W. Deng, H. Baharvand, and M. Javan, “In vivo conversion of astrocytes into oligodendrocyte lineage cells with transcription factor Sox10; Promise for myelin repair in multiple sclerosis,” PLoS One, vol. 13, no. 9, p. e0203785, Sep. 2018, doi: 10.1371/journal.pone.0203785.

(11) S. Farhangi, S. Dehghan, M. Totonchi, and M. Javan, “In vivo conversion of astrocytes to oligodendrocyte lineage cells in adult mice demyelinated brains by Sox2,” Mult Scler Relat Disord, vol. 28, pp. 263–272, Feb. 2019, doi: 10.1016/j.msard.2018.12.041.

(12) L.-L. Wang, C. Serrano, X. Zhong, S. Ma, Y. Zou, and C.-L. Zhang, “Revisiting astrocyte to neuron conversion with lineage tracing in vivo,” Cell, vol. 184, no. 21, pp. 5465-5481.e16, Oct. 2021, doi: 10.1016/j.cell.2021.09.005.

(13) I  Matias, J. Morgado, and F. C. A. Gomes, “Astrocyte Heterogeneity: Impact to Brain Aging and Disease,” Front. Aging Neurosci., vol. 11, Mar. 2019, doi: 10.3389/fnagi.2019.00059.

(14) N. Habib et al., “Disease-associated astrocytes in Alzheimer’s disease and aging,” Nat Neurosci, vol. 23, no. 6, pp. 701–706, Jun. 2020, doi: 10.1038/s41593-020-0624-8.

(15)  M. A. Wheeler et al., “MAFG-driven astrocytes promote CNS inflammation,” Nature, vol. 578, no. 7796, pp. 593–599, Feb. 2020, doi: 10.1038/s41586-020-1999-0.

(16) K. Kamimoto, B. Stringa, C. M. Hoffmann, K. Jindal, L. Solnica-Krezel, and S. A. Morris, “Dissecting cell identity via network inference and in silico gene perturbation,” Nature, vol. 614, no. 7949, pp. 742–751, Feb. 2023, doi: 10.1038/s41586-022-05688-9.

(17) P. Kang et al., “Sox9 and NFIA coordinate a transcriptional regulatory cascade during the initiation of gliogenesis,” Neuron, vol. 74, no. 1, pp. 79–94, Apr. 2012, doi:10.1016/j.neuron.2012.01.024.

(18) K. Saito et al., “Microglia sense astrocyte dysfunction and prevent disease progression in an Alexander disease model,” Brain, vol. 147, no. 2, pp. 698–716, Nov. 2023, doi:10.1093/brain/awad358.

(19) S. A. Liddelow et al., “Neurotoxic reactive astrocytes are induced by activated microglia,” Nature, vol. 541, no. 7638, pp. 481–487, Jan. 2017, doi: 10.1038/nature21029.

(20) Q. Zhu et al., “Genetic evidence that Nkx2.2 and Pdgfra are major determinants of the timing of oligodendrocyte differentiation in the developing CNS,” Development, vol. 141, no. 3, pp. 548–555, Feb. 2014, doi: 10.1242/dev.095323.

(21) J. A. Nielsen, J. A. Berndt, L. D. Hudson, and R. C. Armstrong, “Myelin transcription factor 1 (Myt1) modulates the proliferation and differentiation of oligodendrocyte lineage cells,” Mol Cell Neurosci, vol. 25, no. 1, pp. 111–123, Jan. 2004, doi:10.1016/j.mcn.2003.10.001.

(22) J. Liu, X. Feng, Y. Wang, X. Xia, and J. C. Zheng, “Astrocytes: GABAceptive and GABAergic Cells in the Brain,” Front. Cell. Neurosci., vol. 16, Jun. 2022, doi:10.3389/fncel.2022.892497.

(23) A. Sharma et al., “Divergent roles of astrocytic versus neuronal EAAT2 deficiency on cognition and overlap with aging and Alzheimer’s molecular signatures,” Proceedings of the National Academy of Sciences, vol. 116, no. 43, pp. 21800–21811, Oct. 2019, doi:10.1073/pnas.1903566116

eLife Assessment

In this valuable study, Wandler et al. provide convincing theoretical evidence for alternate mechanisms of rhythm generation by CPGs. Their model shows that cell-type-specific connectivity and an inhibitory drive could underlie rhythm generation. Excitatory input could act to enhance the frequency range of these rhythms. This modeling study could motivate further experimental investigation of these mechanisms to understand CPG rhythmogenesis.

Reviewer #1 (Public review):

This study explores the connectivity patterns that could lead to fast and slow undulating swim patterns in larval zebrafish using a simplified theoretical framework. The authors show that a pattern of connectivity based only on inhibition is sufficient to produce realistic patterns with a single frequency. Two such networks couple with inhibition but with distinct time constants can produce a range of frequencies. Adding excitatory connections further increases the range of obtainable frequencies, albeit at the expense of sudden transitions in mid-frequency range.

Strengths:

(1) This is an eloquent approach to answering the question of how spinal locomotor circuits generate coordinated activity using a theoretical approach based on moving bump models of brain activity.

(2) The models make specific predictions on patterns of connectivity while discounting the role of connectivity strength or neuronal intrinsic properties in shaping the pattern.

(3) The models also propose that there is an important association between cell-type-specific intersegmental patterns and the recruitment of speed-selective subpopulations of interneurons.

(4) Having a hierarchy of models creates a compelling argument for explaining rhythmicity at the network level. Each model builds on the last and reveals a new perspective on how network dynamics can control rhythmicity. I liked that each model can be used to probe questions in the next/previous model.

Comments on revisions:

I am very happy to see the simplified biophysical model supporting the original findings. The authors have done an excellent job addressing my comments.

Just a small note, please change C. Elegans to C. elegans.

Reviewer #2 (Public review):

Summary:

The authors aimed to show that connectivity patterns within spinal circuits composed of specific excitatory and inhibitory connectivity and with varying degrees of modularity could achieve tail beats at various frequencies as well as proper left-right coordination and rostrocaudal propagation speeds.

Strengths:

The model is simple and the connectivity patterns explored are well supported by the literature

The conclusions are intuitive and support many experimental studies on zebrafish spinal circuits for swimming. The simulations provide strong support for the sufficiency of connectivity patterns to produce and control many hallmark features of swimming in zebrafish

Weaknesses:

The authors have addressed my previous concerns well. I have no further concerns.

Reviewer #3 (Public review):

Summary:

Central pattern generator (CPG) circuits underly rhythmic motor behaviors. Till date, it is thought that these CPG networks are rather local and multiple CPG circuits are serially connected to allow locomotion across the entire body. Distributed CPG networks that incorporate long-range connections have not been proposed although such connectivity has been experimentally shown for several different spinal populations. In this manuscript, the authors use this existing literature on long-range spinal interneuron connectivity to build a new computational model that reproduces basic features of locomotion like left-right alternation, rostrocaudal propagation and independent control of frequency and amplitude. Interestingly, the authors show that a model solely based on inhibitory neurons can recapitulate these basic locomotor features. Excitatory sources were then added that increased the dynamic range of frequencies generated. Finally, the authors were also able to reproduce experimentally observed consequences of cell-type-specific ablations showing that local and long range, cell-type-specific connectivity could be sufficient for generating locomotion.

Strengths:

This work is novel, providing an interesting alternative of distributed CPGs to the local networks traditionally predicted. It shows cell type-specific network connectivity is as important if not more than intrinsic cell properties for rhythmogenesis and that inhibition plays a crucial role in shaping locomotor features. Given the importance of local CPGs in understanding motor control, this alternative concept will be of broad interest to the larger motor control field including invertebrate and vertebrate species.

Weaknesses:

The main weaknesses were addressed in the revision.

Author response:

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

Reviewer #1 (Public review):

(1)How is this simplified model representative of what is observed biologically? A bump model does not naturally produce oscillations. How would the dynamics of a rhythm generator interact with this simplistic model?

Bump models naturally produce sequential activity, and can be engineered to repeat this sequential activity periodically (Zhang, 1996; Samsonovich and McNaughton, 1997; Murray and Escola, 2017). This is the basis for the oscillatory behavior in the model presented here. As we describe in our paper, such a model is consistent with numerous neurobiological observations about cell-type-specific connectivity patterns. The reviewer is, however, correct to point out that our model does not incorporate other key neurobiological features--in particular, intracellular dynamical properties--that have been shown to play important roles in rhythm generation. Our aim in this work is to establish a circuit-level mechanism for rhythm generation, complementary to classical models that rely on intracellular dynamics for rhythm generation. Whether and how these mechanisms work together is something that we plan to explore in future work, and we have added a sentence to the Discussion to this effect.

(2) Would this theoretical construct survive being expressed in a biophysical model? It seems that it should, but even a simple biological model with the basic patterns of connectivity shown here would greatly increase confidence in the biological plausibility of the theory.

We thank the reviewer for pointing out this way to strengthen our paper. We implemented the connectivity developed in the rate models in a spiking neuron model which used EI-balanced Poisson noise as input drive. We found that we could reproduce all the main results of our analysis. In particular, with a realistic number of neurons, we observed swimming activity characterized by (i) left-right alternation, (ii) rostal-caudal propagation, and (iii) variable speed control with constant phase lag. The spiking model demonstrates that the connectivity-motif based mechanisms for rhythmogenesis that we propose are robust in a biophysical setting.

We included these results in the updated manuscript in a new Results subsection titled “Robustness in a biophysical model.”

(3) How stable is this model in its output patterns? Is it robust to noise? Does noise, in fact, smooth out the abrupt transitions in frequency in the middle range?

The newly added spiking model implementation of the network demonstrates that the core mechanisms of our models are robust to noise,  since the connectivity is randomly chosen and the input drive is Poisson noise.

To test the effect of noise as it is parametrically varied, we also added noise directly to the rate models in the form of white noise input to each unit. Namely, the rate model was adapted to obey the stochastic differential equation

\[

\tau_i \frac{dr_i(t)}{dt} = -r_i(t) + \left[ \sum_j W_{ij} r_j(t - \Delta_{ij}) + D_i + \sigma\xi_t \right]_+

\]

Here $\xi_t$ is a standard Gaussian white noise and $\sigma$ sets the strength of the noise. We found that the swimming patterns were robust at all frequencies up to $\sigma =  0.05$. Above this level, coherent oscillations started to break down for some swim frequencies. To investigate whether the noise smoothed out abrupt transitions, we swept through different values of noise and modularity of excitatory connections. The results showed very minor improvement in controllability (see figure below), but this was not significant enough to include in the manuscript.

Author response image 1.

(4) All figure captions are inadequate. They should have enough information for the reader to understand the figure and the point that was meant to be conveyed. For example, Figure 1 does not explain what the red dot is, what is black, what is white, or what the gradations of gray are. Or even if this is a representative connectivity of one node, or if this shows all the connections? The authors should not leave the reader guessing.

All figure captions have been updated to enhance clarity and address these concerns.

Reviewer #2 (Public review):

(1) Figure 1A, if I interpret Figure 1B correctly, should there not be long descending projections as well that don't seem to be illustrated?

Thank you for highlighting this potential point of confusion. The diagram in question was only intended to be a rough schematic of the types of connections present in the model. We have added additional descending connections as requested

(2)Page 5, It would be good to define what is meant by slow and fast here, as this definition changes with age in zebrafish (what developmental age)?

We have updated the manuscript to include the sentence: “These values were chosen to coincide with observed ranges from larval zebrafish.” with appropriate citation.

Reviewer #3 (Public review):

(1) The authors describe a single unit as a neuron, be it excitatory or inhibitory, and the output of the simulation is the firing rate of these neurons. Experimentally and in other modeling studies, motor neurons are incorporated in the model, and the output of the network is based on motor neuron firing rate, not the interneurons themselves. Why did the authors choose to build the model this way?

We chose to leave out the motor neurons from our models for a few reasons. While motor neurons read out the rhythmic activity generated by the interneurons and may provide some feedback, they are not required for rhythmogenesis. In fact, interneuron activity (especially in the excitatory V2a neurons (Agha et al., 2024)) is highly correlated with the ventral root bursts within the same segment. This suggests that motor neurons are primarily a local readout of the rhythmic activity of interneurons; therefore, the rhythmic swimming activity can be deduced directly from the interneurons themselves.

Moreover, there is a lack of experimental observation of the connectivity between all the cell types considered in our model and motor neurons. Hence, it was unclear how we should include them in the model. To address this, we are currently developing a data-driven approach that will determine the proper connectivity between the motor neurons and the interneurons, including intrasegmental connections.

(2) In the single population model (Figure 1), the authors use ipsilateral inhibitory connections that are long-range in an ascending direction. Experimentally, these connections have been shown to be local, while long-range ipsilateral connections have been shown to be descending. What were the reasons the authors chose this connectivity? Do the authors think local ascending inhibitions contribute to rostrocaudal propagation, and how?

The long-range ascending ipsilateral inhibitory connections arises from a limitation of our modeling framework. The V1 neurons that provide these connections have been shown experimentally to fire later than other neurons (especially descending V2a  neurons) within the same hemisegment (Jay et al., J Neurosci, 2023); however, our model can only produce synchronized local activity. Hence, we replace local phase offsets with spatial offsets to produce correctly structured recurrent phasic inputs. We are currently investigating a data-driven method for determining intrasegmental connectivity which should be able to produce the local phase offset and address this concern; however, this is beyond the scope of the current paper.

(3) In the two-population model, the authors show independent control of frequency and rhythm, as has been reported experimentally. However, in these previous experimental studies, frequency and amplitude are regulated by different neurons, suggesting different networks dedicated to frequency and amplitude control. However, in the current model, the same population with the same connections can contribute to frequency or amplitude depending on relative tonic drive. Can the authors please address these differences either by changes in the model or by adding to the Discussion?

Our prior  experimental results that suggested a separation of frequency and amplitude control circuits focus on motor neuron recruitment, instead of interneuron activity (Jay et al., J Neurosci 2023; Menelaou and McLean, Nat Commun 2019). To avoid potential confusion about amplitudes of interneurons vs. of motor neurons, we have removed the results from Figure 3 about control of amplitude in the 2-population model, instead focusing this figure on the control of frequency via speed-module recruitment. For the same reason, we have removed the panel showing the effects of targeted ablations on interneuron amplitudes in Figure 7. We have kept the result about amplitude control in our Supplemental Figure S2 for the 8-population model, but we try to make it clear in the text that any relationship between interneuron amplitude and motor neuron amplitude would depend on how motor neurons are modeled, which we do not pursue in this work.

(4) It would be helpful to add a paragraph in the Discussion on how these results could be applicable to other model systems beyond zebrafish. Cell intrinsic rhythmogenesis is a popular concept in the field, and these results show an interesting and novel alternative. It would help to know if there is any experimental evidence suggesting such network-based propagation in other systems, invertebrates, or vertebrates.

We have expanded a paragraph in the Discussion to address these questions. In particular, we highlight how a recent study of mouse locomotor circuits produced a model with similar key features (Komi et al., 2024). These authors made direct use of experimentally determined connectivity structure and cell-type distributions, which informed a model that produced purely network-based rhythmogenesis. We also point out that inhibition-dominated connectivity has been used for understanding oscillatory behavior in neural circuits outside the context of motor control (Zhang, 1996; Samsonovich and McNaughton, 1997; Murray and Escola, 2017). Finally, we address a study that used the cell-type specific connectivity within the C. Elegans locomotor circuit as the architecture for an artificial motor control system and found that the resulting system could more efficiently learn motor control tasks than general machine learning architectures (Bhattasali et al. 2022). Like our model, the Komi et al. and Bhattasali et al. models generate rhythm via structured connectivity motifs rather than via intracellular dynamical properties, suggesting that these may be a key mechanism underlying locomotion across species.

Reviewer #1 (Recommendations for the authors):

(1) Express this modeling construct in a simple biophysical model.

See the new Results subsection titled “Robustness in a biophysical model.”

(2) Please cite the classic models of Kopell, Ermentrout, Williams, Sigvardt etc., especially where you say "classic models".

We have added relevant citations including the mentioned authors.

(3) "Rhythmogenesis remain incompletely understood" changed to "Rhythmogenesis remains incompletely understood".

We chose not to make this change since the ‘remain’ refers to the plural ‘core mechanisms’ not the singular ‘rhythmogenesis’.

Reviewer #3 (Recommendations for the authors):

(1) The figures are well made; however, it would help to add more details to the figure legends. For example, what neuron's firing rate is shown in Figure 1C? What is the red dot in 1B? Figures 3E,F,G: what is being plotted? Mean and SD? Blue dot in Figure 5C?

All figure captions have been updated to enhance clarity and address these concerns.

(2) A, B text missing in Figure 7.

We have revised this figure and its caption; please see our response to Comment 3 above.

(3) It would be nice to see the tonic drive pattern that is fed to the model for each case, along with the different firing rates in the figures. It would help understand how the tonic drive is changed to rhythmic activity.

The tonic drive in the rate models is implemented as a constant excitatory input that is uniform across all units within the same speed-population. There is no patterning in time or location to this drive.

References

(1) Moneeza A Agha, Sandeep Kishore, and David L McLean. Cell-type-specific origins of locomotor rhythmicity at different speeds in larval zebrafish. eLife, July 2024

(2) Nikhil Bhattasali, Anthony M Zador, and Tatiana Engel. Neural circuit architectural priors for embodied control. In S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh, editors, Advances in Neural Information Processing Systems, volume 35, pages 12744–12759. Curran Associates, Inc., 2022.

(3) Salif Komi, August Winther, Grace A. Houser, Roar Jakob Sørensen, Silas Dalum Larsen, Madelaine C. Adamssom Bonfils, Guanghui Li, and Rune W. Berg. Spatial and network principles behind neural generation of locomotion. bioRxiv, 2024

(4) James M Murray and G Sean Escola. Learning multiple variable-speed sequences in striatum via cortical tutoring. eLife, 6:e26084, May 2017.

(5) Alexei Samsonovich and Bruce L McNaughton. Path integration and cognitive mapping in a continuous attractor neural network model. Journal of Neuroscience, 17(15):5900–5920, 1997.

(6) K Zhang. Representation of spatial orientation by the intrinsic dynamics of the head-direction cell ensemble: a theory. Journal of Neuroscience, 16(6):2112–2126, 1996.

eLife Assessment

This global study compares environmental niche model outputs of avian influenza pathogen niche constructed for two distinct periods, and uses differences between those outputs to suggest that the changed case numbers and distribution relate to intensification of chicken and duck farming, and extensive cultivation. While a useful update to existing niche models of highly pathogenic avian influenza, the justification for the use of environmental niche models to explore land cover change as a driver of changed case epidemiology is incomplete.

Reviewer #1 (Public review):

The authors aim to predict ecological suitability for transmission of highly pathogenic avian influenza (HPAI) using ecological niche models. This class of models identify correlations between the locations of species or disease detections and the environment. These correlations are then used to predict habitat suitability (in this work, ecological suitability for disease transmission) in locations where surveillance of the species or disease has not been conducted. The authors fit separate models for HPAI detections in wild birds and farmed birds, for two strains of HPAI (H5N1 and H5Nx) and for two time periods, pre- and post-2020. The authors also validate models fitted to disease occurrence data from pre-2020 using post-2020 occurrence data.

Reviewer #2 (Public review):

Summary:

The geographic range of highly pathogenic avian influenza cases changed substantially around the period 2020, and there is much interest in understanding why. Since 2020 the pathogen irrupted in the Americas and the distribution in Asia changed dramatically. This study aimed to determine which spatial factors (environmental, agronomic and socio-economic) explain the change in numbers and locations of cases reported since 2020 (2020--2023). That's a causal question which they address by applying correlative environmental niche modelling (ENM) approach to the avian influenza case data before (2015--2020) and after 2020 (2020--2023) and separately for confirmed cases in wild and domestic birds. To address their questions they compare the outputs of the respective models, and those of the first global model of the HPAI niche published by Dhingra et al 2016.

ENM is a correlative approach useful for extrapolating understandings based on sparse geographically referenced observational data over un- or under-sampled areas with similar environmental characteristics in the form of a continuous map. In this case, because the selected covariates about land cover, use, population and environment are broadly available over the entire world, modelled associations between the response and those covariates can be projected (predicted) back to space in the form of a continuous map of the HPAI niche for the entire world.

Strengths:

The authors are clear about expected bias in the detection of cases, such geographic variation in surveillance effort (testing of symptomatic or dead wildlife, testing domestic flocks) and in general more detections near areas of higher human population density (because if a tree falls in a forest and there is no-one there, etc), and take steps to ameliorate those. The authors use boosted regression trees to implement the ENM, which typically feature among the best performing models for this application (also known as habitat suitability models). They ran replicate sets of the analysis for each of their model targets (wild/domestic x pathogen variant), which can help produce stable predictions. Their code and data is provided, though I did not verify that the work was reproducible.

The paper can be read as a partial update to the first global model of H5Nx transmission by Dhingra and others published in 2016 and explicitly follows many methodological elements. Because they use the same covariate sets as used by Dhingra et al 2016 (including the comparisons of the performance of the sets in spatial cross-validation) and for both time periods of interest in the current work, comparison of model outputs is possible. The authors further facilitate those comparisons with clear graphics and supplementary analyses and presentation. The models can also be explored interactively at a weblink provided in text, though it would be good to see the model training data there too.

The authors' comparison of ENM model outputs generated from the distinct HPAI case datasets is interesting and worthwhile, though for me, only as a response to differently framed research questions.

Weaknesses:

This well-presented and technically well-executed paper has one major weakness to my mind. I don't believe that ENM models were an appropriate tool to address their stated goal, which was to identify the factors that "explain" changing HPAI epidemiology.

Comments on the revised version from the editors:

We are extremely grateful to the authors for presenting a thoughtful and respectful point by point rebuttal to the prior reviewers' comments. After reading these comments carefully, we conclude that there is a straightforward strongly held disagreement between the authors and the reviewers as to the validity of the methods (Ecological Niche Modeling) for this particular dataset. Please note that the two reviewers have substantial expertise in the area of Ecologic Niche Modeling. We elected not to reach out to the reviewers for a third set of comments as we do not think their overall opinions will change, and wish to be respectful of their time.

To allow readers a balanced assessment of the paper, we intend to publish your rebuttal comments in full. It is our hope that interested readers can weigh both sides of this respectful and interesting debate in order to reach their own conclusions about the strength of evidence presented in your manuscript.

Author response:

The following is the authors’ response to the previous reviews

Public Reviews:

We thank the Reviewers for their thorough attention to our paper and the interesting discussion about the findings. Before responding to more specific comments, here some general points we would like to clarify:

(1) Ecological niche models are indeed correlative models, and we used them to highlight environmental factors associated with HPAI outbreaks within two host groups. We will further revise the terminology that could still unintentionally suggest causal inference. The few remaining ambiguities were mainly in the Discussion section, where our intent was to interpret the results in light of the broader scientific literature. Particularly, we will change the following expressions:

-  “Which factors can explain…” to  “Which factors are associated with…” (line 75);

-  “the environmental and anthropogenic factors influencing” to “the environmental and anthropogenic factors that are correlated with” (line 273);

-  “underscoring the influence” to “underscoring the strong association” (line 282).

(2) We respectfully disagree with the suggestion that an ecological niche modelling (ENM) approach is not appropriate for this work and the research question addressed therein. Ecological niche models are specifically designed to estimate the spatial distribution of the environmental suitability of species and pathogens, making them well suited to our research questions. In our study, we have also explicitly detailed the known limitations of ecological niche models in the Discussion section, in line with prior literature, to ensure their appropriate interpretation in the context of HPAI.

(3) The environmental layers used in our models were restricted to those available at a global scale, as listed in Supplementary Information Resources S1 (https://github.com/sdellicour/h5nx\_risk\_mapping/blob/master/Scripts\_%26\_data/SI\_Resource\_S1.xlsx). Naturally, not all potentially relevant environmental factors could be included, but the selected layers are explicitly documented and only these were assessed for their importance. Despite this limitation, the performance metrics indicate that the models performed well, suggesting that the chosen covariates capture meaningful associations with HPAI occurrence at a global scale.

Reviewer #1 (Public review):

The authors aim to predict ecological suitability for transmission of highly pathogenic avian influenza (HPAI) using ecological niche models. This class of models identify correlations between the locations of species or disease detections and the environment. These correlations are then used to predict habitat suitability (in this work, ecological suitability for disease transmission) in locations where surveillance of the species or disease has not been conducted. The authors fit separate models for HPAI detections in wild birds and farmed birds, for two strains of HPAI (H5N1 and H5Nx) and for two time periods, pre- and post-2020. The authors also validate models fitted to disease occurrence data from pre-2020 using post-2020 occurrence data. I thank the authors for taking the time to respond to my initial review and I provide some follow-up below.

Detailed comments:

In my review, I asked the authors to clarify the meaning of "spillover" within the HPAI transmission cycle. This term is still not entirely clear: at lines 409-410, the authors use the term with reference to transmission between wild birds and farmed birds, as distinct to transmission between farmed birds. It is implied but not explicitly stated that "spillover" is relevant to the transmission cycle in farmed birds only. The sentence, "we developed separate ecological niche models for wild and domestic bird HPAI occurrences ..." could have been supported by a clear sentence describing the transmission cycle, to prime the reader for why two separate models were necessary.

We respectfully disagree that the term “spillover” is unclear in the manuscript. In both the Methods and Discussion sections (lines 387-391 and 409-414), we explicitly define “spillover” as the introduction of HPAI viruses from wild birds into domestic poultry, and we distinguish this from secondary farm-to-farm transmission. Our use of separate ecological niche models for wild and domestic outbreaks reflects not only the distinction between primary spillover and secondary transmission, but also the fundamentally different ecological processes, surveillance systems, and management implications that shape outbreaks in these two groups. We will clarify this choice in the revised manuscript when introducing the separate models. Furthermore, on line 83, we will add “as these two groups are influenced by different ecological processes, surveillance biases, and management contexts”.

I also queried the importance of (dead-end) mammalian infections to a model of the HPAI transmission risk, to which the authors responded: "While spillover events of HPAI into mammals have been documented, these detections are generally considered dead-end infections and do not currently represent sustained transmission chains. As such, they fall outside the scope of our study, which focuses on avian hosts and models ecological suitability for outbreaks in wild and domestic birds." I would argue that any infections, whether they are in dead-end or competent hosts, represent the presence of environmental conditions to support transmission so are certainly relevant to a niche model and therefore within scope. It is certainly understandable if the authors have not been able to access data of mammalian infections, but it is an oversight to dismiss these infections as irrelevant.

We understand the Reviewer’s point, but our study was designed to model HPAI occurrence in avian hosts only. We therefore restricted our analysis to wild birds and domestic poultry, which represent the primary hosts for HPAI circulation and the focus of surveillance and control measures. While mammalian detections have been reported, they are outside the scope of this work.

Correlative ecological niche models, including BRTs, learn relationships between occurrence data and covariate data to make predictions, irrespective of correlations between covariates. I am not convinced that the authors can make any "interpretation" (line 298) that the covariates that are most informative to their models have any "influence" (line 282) on their response variable. Indeed, the observation that "land-use and climatic predictors do not play an important role in the niche ecological models" (line 286), while "intensive chicken population density emerges as a significant predictor" (line 282) begs the question: from an operational perspective, is the best (e.g., most interpretable and quickest to generate) model of HPAI risk a map of poultry farming intensity?

We agree that poultry density may partly reflect reporting bias, but we also assumed it a meaningful predictor of HPAI risk. Its importance in our models is therefore expected. Importantly, our BRT framework does more than reproduce poultry distribution: it captures non-linear relationships and interactions with other covariates, allowing a more nuanced characterisation of risk than a simple poultry density map. Note also that we distinguished in our models intensive and extensive chicken poultry density and duck density. Therefore, it is not a “map of poultry farming intensity”. 

At line 282, we used the word “influence” while fully recognising that correlative models cannot establish causality. Indeed, in our analyses, “relative influence” refers to the importance metric produced by the BRT algorithm (Ridgeway, 2020), which measures correlative associations between environmental factors and outbreak occurrences. These scores are interpreted in light of the broader scientific literature, therefore our interpretations build on both our results and existing evidence, rather than on our models alone. However, in the next version of the paper, we will revise the sentence as: “underscoring the strong association of poultry farming practices with HPAI spread (Dhingra et al., 2016)”. 

I have more significant concerns about the authors' treatment of sampling bias: "We agree with the Reviewer's comment that poultry density could have potentially been considered to guide the sampling effort of the pseudo-absences to consider when training domestic bird models. We however prefer to keep using a human population density layer as a proxy for surveillance bias to define the relative probability to sample pseudo-absence points in the different pixels of the background area considered when training our ecological niche models. Indeed, given that poultry density is precisely one of the predictors that we aim to test, considering this environmental layer for defining the relative probability to sample pseudo-absences would introduce a certain level of circularity in our analytical procedure, e.g. by artificially increasing to influence of that particular variable in our models." The authors have elected to ignore a fundamental feature of distribution modelling with occurrence-only data: if we include a source of sampling bias as a covariate and do not include it when we sample background data, then that covariate would appear to be correlated with presence. They acknowledge this later in their response to my review: "...assuming a sampling bias correlated with poultry density would result in reducing its effect as a risk factor." In other words, the apparent predictive capacity of poultry density is a function of how the authors have constructed the sampling bias for their models. A reader of the manuscript can reasonably ask the question: to what degree are is the model a model of HPAI transmission risk, and to what degree is the model a model of the observation process? The sentence at lines 474-477 is a helpful addition, however the preceding sentence, "Another approach to sampling pseudo-absences would have been to distribute them according to the density of domestic poultry," (line 474) is included without acknowledgement of the flow-on consequence to one of the key findings of the manuscript, that "...intensive chicken population density emerges as a significant predictor..." (line 282). The additional context on the EMPRES-i dataset at line 475-476 ("the locations of outbreaks ... are often georeferenced using place name nomenclatures") is in conflict with the description of the dataset at line 407 ("precise location coordinates"). Ultimately, the choices that the authors have made are entirely defensible through a clear, concise description of model features and assumptions, and precise language to guide the reader through interpretation of results. I am not satisfied that this is provided in the revised manuscript.

We thank the Reviewer for this important point. To address it, we compared model predictive performance and covariate relative influences obtained when pseudo-absences were weighted by poultry density versus human population density (Author response table 1). The results show that differences between the two approaches are marginal, both in predictive performance (ΔAUC ranging from -0.013 to +0.002) and in the ranking of key predictors (see below Author response images 1 and 2). For instance, intensive chicken density consistently emerged as an important predictor regardless of the bias layer used.

Note: the comparison was conducted using a simplified BRT configuration for computational efficiency (fewer trees, fixed 5-fold random cross-validation, and standardised parameters). Therefore, absolute values of AUC and variable importance may differ slightly from those in the manuscript, but the relative ranking of predictors and the overall conclusions remain consistent.

Given these small differences, we retained the approach using human population density. We agree that poultry density partly reflects surveillance bias as well as true epidemiological risk, and we will clarify this in the revised manuscript by noting that the predictive role of poultry density reflects both biological processes and surveillance systems. Furthermore, on line 289, we will add “We note, however, that intensive poultry density may reflect both surveillance intensity and epidemiological risk, and its predictive role in our models should be interpreted in light of both processes”.

Author response table 1.

Comparison of model predictive performances (AUC) between pseudo-absence sampling were weighted by poultry density and by human population density across host groups, virus types, and time periods. Differences in AUC values are shown as the value for poultry-weighted minus human-weighted pseudo-absences.

Author response image 1.

Comparison of variable relative influence (%) between models trained with pseudo-absences weighted by poultry density (red) and human population density (blue) for domestic bird outbreaks. Results are shown for four datasets: H5N1 (<2020), H5N1 (>2020), H5Nx (<2020), and H5Nx (>2020).

Author response image 2.

Comparison of variable relative influence (%) between models trained with pseudo-absences weighted by poultry density (red) and human population density (blue) for wild bird outbreaks. Results are shown for three datasets: H5N1 (>2020), H5Nx (<2020), and H5Nx (>2020).

The authors have slightly misunderstood my comment on "extrapolation": I referred to "environmental extrapolation" in my review without being particularly explicit about my meaning. By "environmental extrapolation", I meant to ask whether the models were predicting to environments that are outside the extent of environments included in the occurrence data used in the manuscript. The authors appear to have understood this to be a comment on geographic extrapolation, or predicting to areas outside the geographic extent included in occurrence data, e.g.: "For H5Nx post-2020, areas of high predicted ecological suitability, such as Brazil, Bolivia, the Caribbean islands, and Jilin province in China, likely result from extrapolations, as these regions reported few or no outbreaks in the training data" (lines 195-197). Is the model extrapolating in environmental space in these regions? This is unclear. I do not suggest that the authors should carry out further analysis, but the multivariate environmental similarly surface (MESS; see Elith et al., 2010) is a useful tool to visualise environmental extrapolation and aid model interpretation.

On the subject of "extrapolation", I am also concerned by the additions at lines 362-370: "...our models extrapolate environmental suitability for H5Nx in wild birds in areas where few or no outbreaks have been reported. This discrepancy may be explained by limited surveillance or underreporting in those regions." The "discrepancy" cited here is a feature of the input dataset, a function of the observation distribution that should be captured in pseudo-absence data. The authors state that Kazakhstan and Central Asia are areas of interest, and that the environments in this region are outside the extent of environments captured in the occurrence dataset, although it is unclear whether "extrapolation" is informed by a quantitative tool like a MESS or judged by some other qualitative test. The authors then cite Australia as an example of a region with some predicted suitability but no HPAI outbreaks to date, however this discussion point is not linked to the idea that the presence of environmental conditions to support transmission need not imply the occurrence of transmission (as in the addition, "...spatial isolation may imply a lower risk of actual occurrences..." at line 214). Ultimately, the authors have not added any clear comment on model uncertainty (e.g., variation between replicated BRTs) as I suggested might be helpful to support their description of model predictions.

Many thanks for the clarification. Indeed, we interpreted your previous comments in terms of geographic extrapolations. We thank the Reviewer for these observations. We will adjust the wording to further clarify that predictions of ecological suitability in areas with few or no reported outbreaks (e.g., Central Asia, Australia) are not model errors but expected extrapolations, since ecological suitability does not imply confirmed transmission (for instance, on Line 362: “our models extrapolate environmental suitability” will be changed to “Interestingly, our models extrapolate geographical”). These predictions indicate potential environments favorable to circulation if the virus were introduced.

In our study, model uncertainty is formally assessed when comparing the predictive performances of our models (Fig. S3, Table S1), the relative influence (Table S3) and response curves (Fig. 2) associated with each environmental factor (Table S2). All the results confirming a good converge between these replicates. Finally, we indeed did not use a quantitative tool such as a MESS to assess extrapolation but did rely on qualitative interpretation of model outputs.

All of my criticisms are, of course, applied with the understanding that niche modelling is imperfect for a disease like HPAI, and that data may be biased/incomplete, etc.: these caveats are common across the niche modelling literature. However, if language around the transmission cycle, the niche, and the interpretation of any of the models is imprecise, which I find it to be in the revised manuscript, it undermines all of the science that is presented in this work.

We respectfully disagree with this comment. The scope of our study and the methods employed are clearly defined in the manuscript, and the limitations of ecological niche modelling in this context are explicitly acknowledged in the Discussion section. While we appreciate the Reviewer’s concern, the comment does not provide specific examples of unclear or imprecise language regarding the transmission cycle, niche, or interpretation of the models. Without such examples, it is difficult to identify further revisions that would improve clarity.

Reviewer #2 (Public review):

The geographic range of highly pathogenic avian influenza cases changed substantially around the period 2020, and there is much interest in understanding why. Since 2020 the pathogen irrupted in the Americas and the distribution in Asia changed dramatically. This study aimed to determine which spatial factors (environmental, agronomic and socio-economic) explain the change in numbers and locations of cases reported since 2020 (2020--2023). That's a causal question which they address by applying correlative environmental niche modelling (ENM) approach to the avian influenza case data before (2015--2020) and after 2020 (2020--2023) and separately for confirmed cases in wild and domestic birds. To address their questions they compare the outputs of the respective models, and those of the first global model of the HPAI niche published by Dhingra et al 2016.

We do not agree with this comment. In the manuscript, it is well established that we are quantitatively assessing factors that are associated with occurrences data before and after 2020. We do not claim to determine the causality. One sentence of the Introduction section (lines 75-76) could be confusing, so we intend to modify it in the final revision of our manuscript. 

ENM is a correlative approach useful for extrapolating understandings based on sparse geographically referenced observational data over un- or under-sampled areas with similar environmental characteristics in the form of a continuous map. In this case, because the selected covariates about land cover, use, population and environment are broadly available over the entire world, modelled associations between the response and those covariates can be projected (predicted) back to space in the form of a continuous map of the HPAI niche for the entire world.

We fully agree with this assessment of ENM approaches.

Strengths:

The authors are clear about expected bias in the detection of cases, such geographic variation in surveillance effort (testing of symptomatic or dead wildlife, testing domestic flocks) and in general more detections near areas of higher human population density (because if a tree falls in a forest and there is no-one there, etc), and take steps to ameliorate those. The authors use boosted regression trees to implement the ENM, which typically feature among the best performing models for this application (also known as habitat suitability models). They ran replicate sets of the analysis for each of their model targets (wild/domestic x pathogen variant), which can help produce stable predictions. Their code and data is provided, though I did not verify that the work was reproducible.

The paper can be read as a partial update to the first global model of H5Nx transmission by Dhingra and others published in 2016 and explicitly follows many methodological elements. Because they use the same covariate sets as used by Dhingra et al 2016 (including the comparisons of the performance of the sets in spatial cross-validation) and for both time periods of interest in the current work, comparison of model outputs is possible. The authors further facilitate those comparisons with clear graphics and supplementary analyses and presentation. The models can also be explored interactively at a weblink provided in text, though it would be good to see the model training data there too.

The authors' comparison of ENM model outputs generated from the distinct HPAI case datasets is interesting and worthwhile, though for me, only as a response to differently framed research questions.

Weaknesses:

This well-presented and technically well-executed paper has one major weakness to my mind. I don't believe that ENM models were an appropriate tool to address their stated goal, which was to identify the factors that "explain" changing HPAI epidemiology.

Here is how I understand and unpack that weakness:

(1) Because of their fundamentally correlative nature, ENMs are not a strong candidate for exploring or inferring causal relationships.

(2) Generating ENMs for a species whose distribution is undergoing broad scale range change is complicated and requires particular caution and nuance in interpretation (e.g., Elith et al, 2010, an important general assumption of environmental niche models is that the target species is at some kind of distributional equilibrium (at time scales relevant to the model application). In practice that means the species has had an opportunity to reach all suitable habitats and therefore its absence from some can be interpreted as either unfavourable environment or interactions with other species). Here data sets for the response (N5H1 or N5Hx case data in domestic or wild birds ) were divided into two periods; 2015--2020, and 2020--2023 based on the rationale that the geographic locations and host-species profile of cases detected in the latter period was suggestive of changed epidemiology. In comparing outputs from multiple ENMs for the same target from distinct time periods the authors are expertly working in, or even dancing around, what is a known grey area, and they need to make the necessary assumptions and caveats obvious to readers.

We thank the Reviewer for this observation. First, we constrained pseudo-absence sampling to countries and regions where outbreaks had been reported, reducing the risk of interpreting non-affected areas as environmentally unsuitable. Second, we deliberately split the outbreak data into two periods (2015-2020 and 2020-2023) because we do not assume a single stable equilibrium across the full study timeframe. This division reflects known epidemiological changes around 2020 and allows each period to be modeled independently. Within each period, ENM outputs are interpreted as associations between outbreaks and covariates, not as equilibrium distributions. Finally, by testing prediction across periods, we assessed both niche stability and potential niche shifts. These clarifications will be added to the manuscript to make our assumptions and limitations explicit.

Line 66, we will add: “Ecological niche model outputs for range-shifting pathogens must therefore be interpreted with caution (Elith et al., 2010). Despite this limitation, correlative ecological niche models  remain useful for identifying broad-scale associations and potential shifts in distribution. To account for this, we analysed two distinct time periods (2015-2020 and 2020-2023).”

Line 123, we will revise “These findings underscore the ability of pre-2020 models in forecasting the recent geographic distribution of ecological suitability for H5Nx and H5N1 occurrences” to “These results suggest that pre-2020 models captured broad patterns of suitability for H5Nx and H5N1 outbreaks, while post-2020 models provided a closer fit to the more recent epidemiological situation”.

(3) To generate global prediction maps via ENM, only variables that exist at appropriate resolution over the desired area can be supplied as covariates. What processes could influence changing epidemiology of a pathogen and are their covariates that represent them? Introduction to a new geographic area (continent) with naive population, immunity in previously exposed populations, control measures to limit spread such as vaccination or destruction of vulnerable populations or flocks? Might those control measures be more or less likely depending on the country as a function of its resources and governance? There aren't globally available datasets that speak to those factors, so the question is not why were they omitted but rather was the authors decision to choose ENMs given their question justified? How valuable are insights based on patterns of correlation change when considering different temporal sets of HPAI cases in relation to a common and somewhat anachronistic set of covariates?

We agree that the ecological niche models trained in our study are limited to environmental and host factors, as described in the Methods section with the selection of predictors. While such models cannot capture causality or represent processes such as immunity, control measures, or governance, they remain a useful tool for identifying broad associations between outbreak occurrence and environmental context. Our study cannot infer the full mechanisms driving changes in HPAI epidemiology, but it does provide a globally consistent framework to examine how associations with available covariates vary across time periods.

(4) In general the study is somewhat incoherent with respect to time. Though the case data come from different time periods, each response dataset was modelled separately using exactly the same covariate dataset that predated both sets. That decision should be understood as a strong assumption on the part of the authors that conditions the interpretation: the world (as represented by the covariate set) is immutable, so the model has to return different correlative associations between the case data and the covariates to explain the new data. While the world represented by the selected covariates \*may\* be relatively stable (could be statistically confirmed), what about the world not represented by the covariates (see point 3)?

We used the same covariate layers for both periods, which indeed assumes that these environmental and host factors are relatively stable at the global scale over the short timeframe considered. We believe this assumption is reasonable, as poultry density, land cover, and climate baselines do not change drastically between 2015 and 2023 at the resolution of our analysis. We agree, however, that unmeasured processes such as control measures, immunity, or governance may have changed during this time and are not captured by our covariates.

Recommendations for the Authors:

Reviewer #1 (Recommendations for the authors):

- Line 400-401: "over the 2003-2016 periods" has an extra "s"; "two host species" (with reference to wild and domestic birds) would be more precise as "two host groups".

- Remove comma line 404

Many thanks for these comments, we have modified the text accordingly.

Reviewer #2 (Recommendations for the authors):

Most of my work this round is encapsulated in the public part of the review.

The authors responded positively to the review efforts from the previous round, but I was underwhelmed with the changes to the text that resulted. Particularly in regard to limiting assumptions - the way that they augmented the text to refer to limitations raised in review downplayed the importance of the assumptions they've made. So they acknowledge the significance of the limitation in their rejoinder, but in the amended text merely note the limitation without giving any sense of what it means for their interpretation of the findings of this study.

The abstract and findings are essentially unchanged from the previous draft.

I still feel the near causal statements of interpretation about the covariates are concerning. These models really are not a good candidate for supporting the inference that they are making and there seem to be very strong arguments in favour of adding covariates that are not globally available.

We never claimed causal interpretation, and we have consistently framed our analyses in terms of associations rather than mechanisms. We acknowledge that one phrasing in the research questions (“Which factors can explain…”) could be misinterpreted, and we are correcting this in the revised version to read “Which factors are associated with…”. Our approach follows standard ecological niche modelling practice, which identifies statistical associations between occurrence data and covariates. As noted in the Discussion section, these associations should not be interpreted as direct causal mechanisms. Finally, all interpretive points in the manuscript are supported by published literature, and we consider this framing both appropriate and consistent with best practice in ecological niche modelling (ENM) studies.

We assessed predictor contributions using the “relative influence” metric, the terminology reported by the R package “gbm” (Ridgeway, 2020). This metric quantifies the contribution of each variable to model fit across all trees, rescaled to sum to 100%, and should be interpreted as an association rather than a causal effect.

L65-66 The general difficulty of interpreting ENM output with range-shifting species should be cited here to alert readers that they should not blithely attempt what follows at home.

I believe that their analysis is interesting and technically very well executed, so it has been a disappointment and hard work to write this assessment. My rough-cut last paragraph of a reframed intro would go something like - there are many reasons in the literature not to do what we are about to do, but here's why we think it can be instructive and informative, within certain guardrails.

To acknowledge this comment and the previous one, we revised lines 65-66 to: “However, recent outbreaks raise questions about whether earlier ecological niche models still accurately predict the current distribution of areas ecologically suitable for the local circulation of HPAI H5 viruses. Ecological niche model outputs for range-shifting pathogens must therefore be interpreted with caution (Elith et al., 2010). Despite this limitation, correlative ecological niche models  remain useful for identifying broad-scale associations and potential shifts in distribution.”

We respectfully disagree with the Reviewer’s statement that “there are many reasons in the literature not to do what we are about to do”. All modeling approaches, including mechanistic ones, have limitations, and the literature is clear on both the strengths and constraints of ecological niche models. Our manuscript openly acknowledges these limits and frames our findings accordingly. We therefore believe that our use of an ENM approach is justified and contributes valuable insights within these well-defined boundaries.

Reference: Ridgeway, G. (2007). Generalized Boosted Models: A guide to the gbm package. Update, 1(1), 2007.

eLife Assessment

Davis and colleagues describe findings that are fundamental to the understanding of pressure mechanosensation in lymphatic vessels and are of significant importance to other areas of mechanosensory physiology. Based on many different knockout mouse models and rigorous state-of-the-art pressure myography recordings, they present compelling evidence that mechano-activation of GNAQ/GNA11-coupled GPCRs generates IP3, which induces Ca2+ release from internal stores through IP3R1 and drives depolarization through the activation of ANO1 Cl- channels to induce lymphatic vessel contractility. Nevertheless, some aspects of the manuscript are incomplete. The specific identity of the GPCR(s) involved remains to be uncovered, as evidence of frequency-pressure impairment is only demonstrated with abolition of GNAQ/GNA11action, not the receptors per se.

Reviewer #1 (Public review):

Summary:

Davis and co-authors used many mouse models to investigate mechanisms that regulate the contractility of mouse popliteal collecting vessels, primarily chronotropy. Many of the mechanisms studied were previously shown to regulate pressure-induced constriction in small arteries. The authors use prior literature from the vasculature as a framework to test similar concepts in lymphatic vessels. The mouse models used provide evidence for and against the involvement of multiple proteins in regulating chronotropy and other contractile properties in lymphatic vessels. They propose that mechano-activation of GNAQ/GNA11-coupled GPCRs generates IP3, which induces Ca2+ release through IP3R1 and drives depolarization through the activation of ANO1 Cl- channels. Major concerns include the author's major conclusion that GNAQ/GNA11-coupled GPCRs contribute to chronotropy. This conclusion is not supported by the data presented.

Strengths:

One major strength of the study lies in the vast number of mouse knockout models that were used to test the importance of ion channels and G protein signaling pathways in the regulation of lymphatic vessel contractility. In this regard, the study is a valiant effort. The authors achieved several objectives to find that ANO1 and IP3R1 regulate chronotropy, and many other potential proteins do not regulate chronotropy. This study will have a major impact on the field if additional support for G proteins is provided.

Weaknesses:

Major conclusions concerning the involvement of G proteins are drawn from the global Gna11 knockout mouse models. This conclusion is weak. Global Gna11 knockout mice are highly likely to have a multifactorial phenotype that could create significant differences in the data. Control experiments need to be performed on vessels from the global knockout mice if these major conclusions are to be made. Similarly, pharmacological tools or alternative approaches to manipulate G proteins should be used to support the data from these mouse models to draw these major conclusions.

The Gnaq smKO mice are the most specific G protein model studied here. However, there is no phenotype. Do not discuss trends in the data. If the data are not significant, conclude so. If more experiments are required to reach significance, provide more data in the manuscript.

The conclusions repeatedly refer to a signaling pathway wherein the upstream component is GPCRs, which activate G proteins. While this may be the case, no GPCRs were identified here, and the involvement of G proteins is questionable, as the authors outline in lines 693-695 and noted above. The conclusions should be tempered, including in the abstract, unless additional experiments are performed to support the involvement of G proteins. Perhaps then the authors may be able to infer that GPCRs are involved.

Line 318. The point regarding the choice to use popliteal vessels versus IALVs will be unclear to the uninitiated, particularly as the authors previously used IALVs. Including additional justification in the text and/or data from IALVs in Figure 1, which compares IALVs to popliteal vessels, would better explain the logic.

The conclusions drawn for TRPC6 and TRPC3 are less convincing. Germline global knockout mice, which are known to undergo compensation, were used, and high data variability is apparent. Using TRPC3 and TRPC6 blockers in the mouse models studied in Figure 4 would strengthen the arguments made regarding these proteins.

Did you perform power analysis to ensure that experimental numbers were sufficient to conclude that no statistical difference exists between datasets? If not, this needs to be done. For example, data shown in Figure 5C for tone and 6C for frequency and tone appear to be significantly different, but are concluded not to be so.

At the end of each result section, a concluding statement is made regarding the effects on pressure-induced chronotrophy. In many cases, there are additional effects of manipulating protein expression on other contractile properties. One example is for TRPC3 and TRPC6 (lines 414-416), but others are TRPV4, TRPV3, ENaC, Kir, Cav3.1/3.2, etc. Some interpretation is in the Discussion, but the concluding statements at the end of each result section should be expanded to summarize what the authors think the other significant differences in the data represent.

Kv7.4 channels. You state you have data (not shown) with linopiridine and XE991. Why not show those results here to support the experiments with the Kcnq4 smKO mice? Otherwise, I suggest you remove the statement from the unpublished data.

Figure 13A. Kcnj2 is modestly expressed in LECs, but very little is present in LMCs. This likely underlies the effect of barium. If you remove the endothelium, does the effect of barium disappear? While this is not the major focus of the study, the effects of barium are dramatic, and it should be made clear whether this is due to inhibition of Kir channels in smooth muscle or endothelial cells.

Figure 18C tone. Several values for losartan look different but are not labelled as such. Please clarify and discuss if different.

The manuscript should include raw data traces in figures that show the major pathways that you conclude regulate chronotropy.

Reviewer #2 (Public review):

Summary:

In this study, Davis et al. embarked on the quest for the molecular elements responsible for the regulation of lymphatic phasic contractile activity in response to variation of transmural pressure, a mechanism (termed pressure-induced lymphatic chronotropy by the authors) critical for drainage of interstitial fluid from the tissue and transport of lymph back to the blood circulation. Their aim was to investigate the mechanism(s) involved in the pressure-induced regulation of lymphatic pumping, and test whether activation of cation channels, shown in other systems to play mechanosensitive roles are directly at play, and/or whether mechano-activation of GNAQ/GNA11-coupled GPCRs is necessary to generate second messengers to activate those channels, as it has been suggested for the regulation of myogenic tone in arteries. To achieve their goal, the authors used their well-described, highly reliable protocols of mouse lymphatic vessel isolation, pressure myography, and data acquisition to obtain frequency-pressure relationships and other contractile function parameters from transgenic mice where specific channels or molecular elements of interest have been ablated. They combined these data with scRNAseq analysis of these gene targets to determine their respective role and levels of expression in lymphatic muscle cells. Their conclusion is that none of the exhaustive list of tested ion channels was critical, except ANO1 Cl channels, part of the contractile pacemaker mechanism, but that transmural pressure activates GNAQ/GNA11-coupled GPCRs, which generate IP3 to induce SR Ca2+ release through IP3R1 and activate ANO1-mediated depolarization.

Strengths:

The manuscript's strengths reside primarily in very robust, clean, and unequivocal pressure myography data and analysis. The research team is mastering these techniques they developed more than a decade ago and have implemented in mouse lymphatics to study their contractile properties, with consistent and convincing outcomes. They also provide data from an impressive list of transgenic mice in order to determine the role of the targeted gene in pressure-induced lymphatic chronotropy, relying on pharmacological small molecule inhibitors only when necessary. Finally, the use of scRNAseq analysis they gathered from previously published datasets brings novelty with respect to the expression of the genes of interest in all populations of cells comprising the lymphatic vessels, but more critically, to validate or contrast the potential impact of genetic alteration of the given gene on the ability of lymphatic muscles to respond to a change in pressure.

Weaknesses:

The main weakness may reside in the fact that while the authors provide a convincing demonstration that GNAQ/GNA11 are involved in the regulation of the F-P relationship, they give little evidence of the involvement of "upstream" receptors. Indeed, inhibition of AT1R, shown to be involved in myogenic regulation of arteries (a phenomenon the authors rightfully compare to pressure-induced lymphatic chronotropy), didn't lead toa similar effect (decrease in F-P) in lymphatic vessels. Arguably, other GPCRs might be involved in lymphatic vessels, but as such information is not provided in the manuscript, the author's conclusions should be dampened. More in-depth discussion would be required. In fact, it can be argued that the discussion is very restricted with respect to the amount of data and information the manuscript provides.

Overall, the authors convincingly achieved their aim by performing an impressive number of technically challenging experiments, leading to solid datasets. While these support their main conclusions, a more elaborate discussion might be required to refine them.

This study is likely to have an important impact on the field as it provides some answers to the lingering question of how lymphatic vessels regulate their contractile activity to variation in transmural pressure and certainly proposes an experimental means to further explore and address that question.

Reviewer #3 (Public review):

In this manuscript, Davis and colleagues aimed to identify the molecular sensors and signaling cascade that enable collecting lymphatic vessels to increase their spontaneous contraction frequency in response to intraluminal pressure (pressure-induced chronotropy). They tested whether the process is similar to blood vessel myogenic constriction by relying on cation channels (TRPC6, TRPM4, PKD2, PIEZO1, etc.) or instead require the activation of G-protein-coupled receptors (presumably mechanosensitive GNAQ/GNA11-coupled receptors), using ex vivo pressure myography of mouse popliteal lymphatics, smooth muscle-specific conditional knockouts, quantitative PCR validation, and single-cell RNA sequencing for target prioritization. The authors convincingly demonstrate that pressure-induced chronotropy does not require the cation channels implicated in arterial myogenic tone but is blunted by deletion of GNAQ/GNA11 or IP3 receptor 1, supporting a model of GPCR > IP3 > Ca2+ release > Cl⁻ channel activation > depolarization. The core conclusion is robust. The work redefines lymphatic pacemaking as G-protein-coupled receptor-dependent mechanotransduction, distinct from arterial mechanisms, and provides a genetically validated toolkit that is useful for studying lymphatic function and dysfunction.

Strengths:

(1) The data are of high quality and highly sensitive functional readouts

(2) The systematic genetic targeting is a major strength that overcomes pharmacological artifacts

(3) Careful quantitative analyses of frequency-pressure slopes

Weaknesses:

(1) The use of inguinal-axillary vessels for single-cell RNA sequencing rather than the popliteal segment studied functionally.

(2) No direct testing of the specific G-protein-coupled receptor involved.

Author response:

We thank the reviewers and editors for their insightful comments on our manuscript. We intend to submit a revised manuscript that addresses all concerns raised by the reviewers. A major limitation identified by the reviewers was our inability to identify one or more specific mechanosensitive GPCRs in lymphatic muscle cells (LMCs). To address this concern, we plan to include several additional figures in the revised manuscript. One figure will list the 136 GPCRs identified in LMCs by our scRNAseq analysis, based on the list of validated GPCRs in https://esbl.nhlbi.nih.gov/Databases/GPCRs/index.html and olfactory GPCRs listed in https://esbl.nhlbi.nih.gov/Databases/GPCRs/MouseHumanRatORs.html. We plan to arrange the data in a hierarchical manner according to their expression level and denote their heterotrimeric GTP-binding protein alpha subunit(s), if known. To reinforce our finding that pressure-induced chronotropy in LMCs is mediated through Gq/11, we will present additional data testing the effects of acute Gq/11  inhibition with YM-254890 (a selective Gq/11 inhibitor) on the frequency-pressure relationship of popliteal vessels, as suggested by one reviewer. We will address concerns regarding the potential regional differences in lymphatic contractile regulation arising from our use of popliteal lymphatic vessels for contraction assays and expression analysis of LMCs obtained from Inguinal-Axillary lymphatic vessels (IALVs). To account for possible differences between the two, we will test pressure responses of IALVs from double Gq/11 knockout mice and test responses of wild-type IALVs to acute administration of YM-25489.

Our preliminary analysis of the 136 GPCRs in LMCs revealed a shorter list of 10 GPCRs that are expressed in at least 50% of LMCs (based on the IALV scRNAseq dataset). Since existing evidence from our studies, and those of other investigators, suggests that any LMC is capable of initiating pacemaking, we consider it reasonable to impose this requirement.

Author response table 1.

We plan to use pharmacologic inhibitors to test as many of these candidates as possible. Unfortunately, inhibitors are not available for many of the GPCRs listed above, but we will test Npr3, Npy1R, and Ednra; a negative result for Tbxa2r has already been documented in a previous study (Schulz et al. ATVB 2025). Even if this strategy does not lead to identification of one or more specific GPCRs involved in LMC pressure transduction, it will narrow the list of possible candidates that need to be tested in future experiments.

eLife Assessment

This study offers important insights into how outer membrane vesicles (OMVs) secreted by Serratia marcescens, which carry various virulence factors, contribute to pathogenicity. The experiments provide solid preliminary support for OMV-mediated pathogenic effects, with a critical role for the metalloprotease virulence factor PrtA. However, the evidence remains incomplete, and the current level of validation limits confidence in the strength of the conclusions.

Reviewer #1 (Public review):

Summary:

The work of Bechara Rahme and colleagues provides an explanation as to how bacterially infected flies eventually die. While widespread tissue and multiorgan damage are to be expected in the latest stages of a systemic infection, the mechanisms leading to the host's death remain unresolved. To this end, this work illustrates the role of PrtA, a metalloproteinase found within Outer Membrane Vesicles (OMVs) secreted by Serratia marcescens, in inducing neuronal apoptosis and paralysis before death. Another interesting aspect of the work is the compromise of blood blood-brain barrier (BBB) by OMVs. BBB is different between mammals and flies; however, it merits scientific attention.

Strengths:

The strength of evidence lies in a wealth of experiments involving disparate innate immune mechanisms that either contribute (Imd, PPO1/2, Nox, Duox, SOD2) or oppose (hemocytes and Hayan protease) host defense. Moreover, the role of neuronal JNK and apoptic signaling is shown to contribute to host death.

Genetics is supported by experiments using chemical treatments (Vitamin C and mito-TEMPO) as host-protecting antioxidants, and the biochemical purification and quantification of OMVs and the PrtA protease.

Weaknesses:

However, the reliance on non-isogenised flies to provide quantitative data is unsafe, and at this point, the strength of the evidenceis apparently incomplete. The mutant flies used for the genes Key, Myd88, Hayan, and Nos are doubtfully comparable to the control fly strains used in terms of the general genetic background. The latter is of utmost importance in assessing quantitative traits.

The general background difference between control and test flies is also an issue when using tissue-specific expression via GAL4/UAS, because the UAS lines used are only apparently but not truly isogenic to the w flies used as controls.

Reviewer #2 (Public review):

Summary:

In this manuscript, the authors investigate the mechanisms underlying the virulence of OMVs using a Drosophila model. They reveal a complex interplay between host defenses and OMV pathogenicity. Although the study enhances our understanding of Drosophila innate immunity, additional evidence is needed to strengthen the conclusions.

Strengths:

(1) In Figure 1, Toll pathway mutants infected with OMVs displayed three distinct phenotypic outcomes: mildly enhanced resistance to OMV infection, a response similar to that of the control, or increased susceptibility. Therefore, in addition to Imd and Kenny mutants from the Imd pathway, further mutants, such as Relish and PGRP-LC, should be examined to assess whether the Imd pathway is involved in host defense against OMVs.

(2) Plasmatocytes clear particles via phagocytosis or endocytosis. However, flies lacking all hemocytes showed increased resistance to OMV challenge, raising the question of whether hemocytes actually aid the pathogen. To explore this hypothesis, the uptake of fluorescently tagged OMVs should be examined.

(3) Hayan cleaves PPO into active PO. However, Hayan and PPO mutants exhibit opposite phenotypes upon OMV injection, raising the question of whether OMV-induced pathogenesis is linked to melanization.

(4) Puckered mRNA levels were used as a read-out for JNK pathway activity. A transient induction of the JNK pathway was observed in head and thorax tissues. It would be beneficial if the authors could directly examine JNK activation in neuronal cells using immunostaining for pJNK.

(5) In Figure 4B, the kayak was knocked down using the pan-neuronal driver elav-Gal4. To confirm the specificity and validity of this observation, the experiment should be repeated using another neural-specific driver.

Weaknesses:

It is unclear how many Serratia marcescens cells a 69 nL injection of 0.1 ng/nL OMVs corresponds to.

Reviewer #3 (Public review):

Summary:

The authors investigate deficiencies in various immune responses, and also the prtA toxin's role in OMV toxicity. Some key interpretations are that the Imd pathway contributes to preventing OMV toxicity, but not Toll, and that Hayan and Eater somehow mediate OMV or PrtA toxicity. This descriptive effort is a solid set of experiments, although some experimental results may require further validation.

Strengths:

The breadth of experiments tests multiple immune parameters, providing a systematic effort that ensures a number of potentially relevant interactions can be recovered. Certain findings, such as the PrtA toxicity to flies, appear solid, and some interesting findings regarding Hayan and eater will be of interest to the fly immunity field.

Weaknesses:

It appears almost all results rely on the use of a single mutant representing the deletion of the gene. It's not clear if the mutations are always in the same genetic background, but this can be clarified. There are a couple of results that are confusing and may be internally contradicting, and should be additionally validated and clarified.

Author response:

We thank the reviewers and editors for the careful evaluation of our manuscript. Below, we provide a first refutation of some of the concerns expressed by reviewers.

Both reviewer 1 &3 underscore the importance of controlling for genetic backgrounds. This is actually an issue only for a limited part of the study and this criticism should not apply to major findings of this study, with some exceptions, as detailed below.

It is important to note that we have identified ourselves several of the mutant lines we have been using. For instance, key and MyD88 mutant alleles have been identified in the Exelixis transposon insertion collection that we have screened in collaboration with this firm (e.g., [3, 4, 5]). This resource has been generated in a isogenized w [A5001] strain[6], which we are using as matched control for these mutants (Figs 1B,D). Of note, while they share a common genetic background, the phenotypes of key and MyD88 are opposite in terms of sensitivity to OMV challenge. The imd^shadok null allele had been identified during our chemical mutagenesis screen with EMS in a yw cn bw background [5, 7, 8, 9], which was used as a control (FigS1A).

With respect to Hayan (Fig. 2C, Fig. S2C) and eater (Fig. S2A-B) mutants[10, 11, 12], we find a similarly strong phenotype with two independent mutants in distinct genetic backgrounds (actually three for Hayan, as we have not included in our original manuscript the Hayan^SK3allele generated in the Lemaitre laboratory in which OMVs displayed also impaired virulence). We have shown that the Hayan mutants do display the expected phenotype in terms of PPO cleavage (Fig. S2D). Please, also note that in Fig. S2C the two mutant alleles are tested in the same experiment: even though there is some variation between the w¹¹¹⁸ and the w[A5001] strains, the two mutants behave in a remarkably similar manner. As regards the role of the cellular response, we note that we obtained results similar to those obtained with eater mutants using genetic ablation of hemocytes (Fig. 2A) or by saturating the phagocytosis apparatus (Fig. 2B), a confirmation by two totally-independent approaches.

Of note, the observed eater and Hayan phenotypes are strong and not relatively small and thus unlikely to be due to the genetic background.

The PPO mutants have been isogenized in the w¹¹¹⁸ by the lab of Bruno Lemaitre[13, 14] and are also validated biochemically in Fig. S2D. These mutants have been extensively tested in the Lemaitre laboratory[13, 14, 15].

With respect to RNAi silencing driven ubiquitously or in specific tissues using the UAS-Gal4 system, we have mostly used transgenes from the Trip collection and have used as a control the mCherry RNAi provided by this resource[16]. As the RNAi transgenes have been generated in the same genetic background, it follows that independently of the driver used, the genetic background used in mCherry and genes-of-interest (Duox, Nox, Jafrac2) silenced flies is controlled for (Fig. 3D,E).

For UAS-Gal4-mediated overexpression of fly superoxide dismutase genes, we have used SOD1 and SOD2 transgenes that have both been generated by the same laboratory (Phillips laboratory, University of Guelph) presumably in the same genetic background. Using two distinct drivers we find a strongly enhanced susceptibility phenotype when using UAS-SOD2 but not UAS-SOD1 transgenes (Fig. 3F, Fig. 4E). Importantly, the former is associated with mitochondria whereas the other is expressed in the endoplasmic reticulum: we independently confirm this phenotype using the mitoTempo mitochondrial ROS inhibitor.

We shall thus address the criticism with NOS mutants, where genetic background control is indeed critical and for the UAS-kay RNAi line using a Trip line and its associated mCherry RNAi control transgene.

With respect to the Toll pathway mutants, we agree that some of the variability of the phenotypes may be due to the genetic background, especially as regards tube and pelle. The SPE and grass mutants have been retrieved in a screen performed by the group of Jean-Marc Reichhart in our Research Unit. They thus have been generated in the same genetic background, yet grass displays a mildly decreased virulence of injected OMVs whereas SPE mutants display an opposite phenotype (compare Fig. S1E to S1I; the survival experiment shave been performed in the same set of experiments and have been separated for clarity). We do not intend to analyze further the mutants of the Toll pathway as our data suggest that the canonical Toll pathway, likely activated through psh (Fig. S1F) appears to be activated to detectable levels too late by comparison with the time course of OMV pathogenicity. In our opinion, the contribution of the Toll pathway in the host defense against OMV pathogenicity is minor, albeit we acknowledge that some of the findings, especially with SPE are puzzling.

With respect to the IMD pathway, we shall test also PGRP-LC and Relish mutants, as suggested by reviewers 2&3.

Reviewer 2 query: “It is unclear how many Serratia marcescens cells a 69 nL injection of 0.1 ng/nL OMVs corresponds to.”

OMVs were purified from 600 mL of SmDb11 cultures grown to an average OD₆₀₀ of 2.0. Based on a cell density of 0.8 × 10⁸ cells/mL per OD unit, this corresponds to approximately 9.6 × 10¹⁰ total bacterial cells.

Each OMV preparation was concentrated into a final volume of 400 µL, resulting in a concentration factor of ~1500× relative to the original culture. Therefore, an injection dose of 69 nL of OMVs is equivalent to 0.1 mL of the starting bacterial culture, which corresponds to:

0.2 OD units

Approximately 1.6 × 10⁷ bacterial cells

It is likely that such high concentrations occur only toward the end of the infection, if OMVs are produced at the same rate in the host and in vitro.

With respect to other Reviewer 2 queries, we shall give a try at labeling OMVs with the FM4-64 lipophilic dye and examining whether they are taken up by hemocytes. However, an issue may arise with potentially high background, which has been encountered in cell culture. Of note, OMVs are known to attack cultured human THP1 cells, a monocyte cell line [17].Of note, determining whether OMVs are taken up by hemocytes may only be a starting point to understand how they promote the pathogenicity of OMVs. This question constitutes the topic of a full study that we are currently unable to undertake.

We shall also test whether we can document phospho-JNK expression in neural tissues.

Finally, we shall also confirm the data obtained with two elav-Gal4 drivers (including an inducible one) with the nsyb-Gal4 driver line.

References

(1) Xu R, et al. The Toll pathway mediates Drosophila resilience to Aspergillus mycotoxins through specific Bomanins. EMBO Rep 24, e56036 (2023).

(2) Huang J, et al. A Toll pathway effector protects Drosophila specifically from distinct toxins secreted by a fungus or a bacterium. Proc Natl Acad Sci U S A 120, e2205140120 (2023).

(3) Gobert V, et al. Dual Activation of the Drosophila Toll Pathway by Two Pattern Recognition Receptors. Science 302, 2126-2130 (2003).

(4) Gottar M, et al. Dual Detection of Fungal Infections in Drosophila via Recognition of Glucans and Sensing of Virulence Factors. Cell 127, 1425-1437 (2006).

(5) Gottar M, et al. The Drosophila immune response against Gram-negative bacteria is mediated by a peptidoglycan recognition protein. Nature 416, 640-644 (2002).

(6) Thibault ST, et al. A complementary transposon tool kit for Drosophila melanogaster using P and piggyBac. Nat Genet 36, 283-287 (2004).

(7) Rutschmann S, Jung AC, Hetru C, Reichhart J-M, Hoffmann  JA, Ferrandon D. The Rel protein DIF mediates the antifungal, but not the antibacterial,  response in Drosophila. Immunity 12, 569-580 (2000).

(8) Rutschmann S, Jung AC, Rui Z, Silverman N, Hoffmann JA, Ferrandon D. Role of Drosophila IKKg in a Toll-independent antibacterial immune response. Nat Immunology 1, 342-347 (2000).

(9) Jung A, Criqui M-C, Rutschmann S, Hoffmann J-A, Ferrandon D. A microfluorometer assay to measure the expression of ß-galactosidase and GFP reporter genes in single Drosophila flies. Biotechniques 30, 594- 601 (2001).

(10) Nam HJ, Jang IH, You H, Lee KA, Lee WJ. Genetic evidence of a redox-dependent systemic wound response via Hayan protease-phenoloxidase system in Drosophila. Embo J 31, 1253-1265 (2012).

(11) Kocks C, et al. Eater, a transmembrane protein mediating phagocytosis of bacterial pathogens in Drosophila. Cell 123, 335-346 (2005).

(12) Bretscher AJ, et al. The Nimrod transmembrane receptor Eater is required for hemocyte attachment to the sessile compartment in Drosophila melanogaster. Biology open 4, 355-363 (2015).

(13) Binggeli O, Neyen C, Poidevin M, Lemaitre B. Prophenoloxidase activation is required for survival to microbial infections in Drosophila. PLoS Pathog 10, e1004067 (2014).

(14) Dudzic JP, Kondo S, Ueda R, Bergman CM, Lemaitre B. Drosophila innate immunity: regional and functional specialization of prophenoloxidases. BMC Biol 13, 81 (2015).

(15) Dudzic JP, Hanson MA, Iatsenko I, Kondo S, Lemaitre B. More Than Black or White: Melanization and Toll Share Regulatory Serine Proteases in Drosophila. Cell reports 27, 1050-1061 e1053 (2019).

(16) Perkins LA, et al. The Transgenic RNAi Project at Harvard Medical School: Resources and Validation. Genetics 201, 843-852 (2015).

(17) Goman A, et al. Uncovering a new family of conserved virulence factors that promote the production of host-damaging outer membrane vesicles in gram-negative bacteria. J Extracell Vesicles 14, e270032 (2025).

eLife Assessment

This valuable study presents an analysis of the gene regulatory networks that contribute to tumour heterogeneity and tumor plasticity in Ewing sarcoma, with key implications for other fusion-driven sarcomas. The authors convincingly employed orthogonal approaches, including single-cell sequencing and xenografts, to reveal the existence and plasticity of specific gene regulatory networks (e.g., TGF-beta signaling) within Ewing sarcoma, as well as significant differences that exist between cell lines and patient tumors.

Reviewer #1 (Public review):

The investigators elegantly utilized a single-cell co-assay of RNA and ATAC seq to unveil the heterogeneous gene regulatory networks in Ewing sarcoma. The authors should be commended on their ability to identify multiple unique modules of gene regulation of Ewing sarcoma utilizing complex computational methods between numerous Ewing sarcoma cell lines. Additionally, they complemented their single-cell findings with xenografts as well as primary Ewing sarcoma patient tumors - validating the intratumoral heterogeneous gene regulatory networks of Ewing sarcoma. More importantly, they have revealed that exogenous TGF-β may modify these distinct epigenetic and transcriptional signatures within Ewing sarcoma tumors. Overall, the manuscript highlights an important discovery of the heterogenous gene regulatory programming of Ewing sarcoma and further highlights the role that TGFB plays within the tumor microenvironment of Ewing sarcoma. There are some areas of ambiguity that require clarification to increase the impact of the manuscript.

Reviewer #2 (Public review):

Summary:

This work by Waltner et. al. provides a comprehensive single-cell multiomics analysis of plasticity in gene regulatory networks present in Ewing sarcoma using single-cell RNA-sequencing (scRNA-seq) and single-cell assay for transposase accessible chromatin with sequencing (scATAC-seq). They find that Ewing sarcoma cell line models have distinct patterns of chromatin accessibility compared to non-Ewing sarcoma models, and that there is significant variability across Ewing sarcoma cell lines, and sometimes within a single cell line. These differences across models are linked to 3 distinct gene regulatory modules, 2 of which are present across the range of model systems studied here. The first modules present across models are activated when the fusion is expressed and include genes enriched for the known EWSR1::FLI1 response element, GGAA microsatellites, along with other neural crest transcription factors. The other module primarily consists of genes repressed by EWSR1::FLI1, which are activated in EWSR1::FLI1-low states. Interestingly, EWSR1::FLI1-low cells have already been tied to more migratory and metastatic phenotypes, and the data here suggest these cells are more responsive to external signals from TGF-β, and this may be mediated through FOSL2-mediated gene regulation. While there are some minor additional validation studies that can be performed to strengthen a few individual analyses, this is a technically rigorous study, with a variety of different analytical techniques used to address similar questions, and this approach elevates confidence in the answers provided. This is further strengthened by the diverse set of model systems used, including patient-derived cell lines, cell line xenograft models, patient-derived xenografts, mining available single-cell data from patient samples, and validation of the gene modules identified in a larger set of patient microarray samples. In whole, this study provides a valuable resource for understanding heterogeneity, plasticity, and gene expression networks in Ewing sarcoma. This may be useful for future studies of metastatic disease and may also provide a framework for similar questions in other fusion-driven sarcomas.

Strengths:

There are a few core strengths in this study. First is the number and diversity of Ewing sarcoma models studied, spanning commonly used cell lines, patient-derived xenografts, and patient samples. The second is the large array of rigorous and orthogonal approaches used to uncover the identity and function of various gene modules. This includes an array of informatics techniques, as well as specific modulation of cell line models in culture. A third is confirmation that different gene expression programs are present in the same tumor using spatial transcriptomic analysis. Lastly, the authors have made all of their data and code accessible, enabling continued use of this dataset as a resource for others.

Weaknesses:

As highlighted by the authors, this study is somewhat limited by the small number of single-cell data from patient samples that are publicly available. Much of the analysis comes from cell lines. Additionally, they focus only on one type of signal that may modulate cell plasticity, and there are likely to be many others. Lastly, there are a few weak spots in the data. Some of this likely arises from the underlying complexity of the data, the generally sparse nature of scATAC data, and the biological heterogeneity present in the cell lines studied. The most pronounced weakness was in the analysis of transcription factors that dictate gene expression in the distinct modules, as well as the response to TGF-β. While some specific transcription factors showed module-specific expression consistent with the computational prediction in Figure 2, others did not likely due to additional factors not tested here. Likewise, the same transcription factors did not always show consistent enrichment in the gene modules that responded to TGF-β treatment when analyzed across cell lines. On the whole, these are relatively minor weaknesses and do not diminish the value of this study.

eLife Assessment

This study tested the specific hypothesis that age-related changes to hearing involve a partial loss of synapse connections between sensory cells in the ear and the nerve fibers that carry information about sounds to the brain, and that this interferes with the ability to discriminate rapid temporal fluctuations in sounds. Physiological, behavioral, and histological analyses provide a powerful combination to test this hypothesis in gerbils. Contrary to previous suggestions, it was found that chemically-induced isolated synaptopathy (at similar levels as observed in aged gerbils) did not result in worse performance on a behavioral task measuring sensitivity to temporal fine-structure, nor did it produce degradations in auditory-nerve fiber encoding of fine structure. Aged gerbils showed degraded behavior and stronger than normal envelope responses, but temporal fine-structure coding was not affected; interpreted by the authors as suggesting central processing contributions to aging effects on discrimination. These findings are important for advancing our knowledge of the mechanistic bases for age-related changes to hearing, and the evidence provided is solid with the results largely supporting the claims made and minor limitations related to possible confounds discussed in reasonable depth.

Reviewer #1 (Public review):

Summary:

The authors investigate the effects of aging on auditory system performance in understanding temporal fine structure (TFS), using both behavioral assessments and physiological recordings from the auditory periphery, specifically at the level of the auditory nerve. This dual approach aims to enhance understanding of the mechanisms underlying observed behavioral outcomes. The results indicate that aged animals exhibit deficits in behavioral tasks for distinguishing between harmonic and inharmonic sounds, which is a standard test for TFS coding. However, neural responses at the auditory nerve level do not show significant differences when compared to those in young, normal-hearing animals. The authors suggest that these behavioral deficits in aged animals are likely attributable to dysfunctions in the central auditory system, potentially as a consequence of aging.To further investigate this hypothesis, the study includes an animal group with selective synaptic loss between inner hair cells and auditory nerve fibers, a condition known as cochlear synaptopathy (CS). CS is a pathology associated with aging and is thought to be an early indicator of hearing impairment. Interestingly, animals with selective CS showed physiological and behavioral TFS coding similar to that of the young normal-hearing group, contrasting with the aged group's deficits. Despite histological evidence of significant synaptic loss in the CS group, the study concludes that CS does not appear to affect TFS coding, either behaviorally or physiologically.

Strengths:

This study addresses a critical health concern, enhancing our understanding of mechanisms underlying age-related difficulties in speech intelligibility, even when audiometric thresholds are within normal limits. A major strength of this work is the comprehensive approach, integrating behavioral assessments, auditory nerve (AN) physiology, and histology within the same animal subjects. This approach enhances understanding of the mechanisms underlying the behavioral outcomes and provides confidence in the actual occurrence of synapse loss and its effects.The study carefully manages controlled conditions by including five distinct groups: young normal-hearing animals, aged animals, animals with CS induced through low and high doses, and a sham surgery group. This careful setup strengthens the study's reliability and allows for meaningful comparisons across conditions. Overall, the manuscript is well-structured, with clear and accessible writing that facilitates comprehension of complex concepts.

Weakness:

The stimulus and task employed in this study are very helpful for behavioral research, and using the same stimulus setup for physiology is advantageous for mechanistic comparisons. However, I have some concerns about the limitations in auditory nerve (AN) physiology. Due to practical constraints, it is not feasible to record from a large enough population of fibers that covers a full range of best frequencies (BFs) and spontaneous rates (SRs) within each animal. This raises questions about how representative the physiological data are for understanding the mechanism in behavioral data. I am curious about the authors' interpretation of how this stimulus setup might influence results compared to methods used by Kale and Heinz (2010), who adjusted harmonic frequencies based on the characteristic frequency (CF) of recorded units. While, the harmonic frequencies in this study are fixed across all CFs, meaning that many AN fibers may not be tuned closely to the stimulus frequencies. If units are not responsive to the stimulus further clarification on detecting mistuning and phase locking to TFS effects within this setup would be valuable. Given the limited number of units per condition-sometimes as few as three for certain conditions-I wonder if CF-dependent variability might impact the results of the AN data in this study and discussing this factor can help with better understanding the results. While the use of the same stimuli for both behavioral and physiological recordings is understandable, a discussion on how this choice affects interpretation would be beneficial. In addition a 60 dB stimulus could saturate high spontaneous rate (HSR) AN fibers, influencing neural coding and phase-locking to TFS. Potentially separating SR groups, could help address these issues and improve interpretive clarity.

A deeper discussion on the role of fiber spontaneous rate could also enhance the study. How might considering SR groups affect AN results related to TFS coding? While some statistical measures are included in the supplement, a more detailed discussion in the main text could help in interpretation.

Although Figure S2 indicates no change in median SR, the high-dose treatment group lacks LSR fibers, suggesting a different distribution based on SR for different animal groups, as seen in similar studies on other species. A histogram of these results would be informative, as LSR fiber loss with CS-whether induced by ouabain in gerbils or noise in other animals-is well documented (e.g., Furman et al., 2013).

Although ouabain effects on gerbils have been explored in previous studies, since these data is already seems to be recorded for the animal in this study, a brief description of changes in auditory brainstem response (ABR) thresholds, wave 1 amplitudes, and tuning curves for animals with cochlear synaptopathy (CS) in this study would be beneficial. This would confirm that ouabain selectively affects synapses without impacting outer hair cells (OHCs). For aged animals, since ABR measurements were taken, comparing hearing differences between normal and aged groups could provide insights into the pathologies besides CS in aged animals. Additionally, examining subject variability in treatment effects on hearing and how this correlates with behavior and physiology would yield valuable insights. If limited space maybe a brief clarification or inclusion in supplementary could be good enough.

Another suggestion is to discuss the potential role of MOC efferent system and effect of anesthesia in reducing efferent effects in AN recordings. This is particularly relevant for aged animals, as CS might affect LSR fibers, potentially disrupting the medial olivocochlear (MOC) efferent pathway. Anesthesia could lessen MOC activity in both young and aged animals, potentially masking efferent effects that might be present in behavioral tasks. Young gerbils with functional efferent systems might perform better behaviorally, while aged gerbils with impaired MOC function due to CS might lack this advantage. A brief discussion on this aspect could potentially enhance mechanistic insights.

Lastly, although synapse counts did not differ between the low-dose treatment and NH I sham groups, separating these groups rather than combining them with the sham might reveal differences in behavior or AN results, particularly regarding the significance of differences between aged/treatment groups and the young normal-hearing group.

Reviewer #2 (Public review):

Summary:

Using a gerbil model, the authors tested the hypothesis that loss of synapses between sensory hair cells and auditory nerve fibers (which may occur due to noise exposure or aging) affects behavioral discrimination of the rapid temporal fluctuations of sounds. In contrast to previous suggestions in the literature, their results do not support this hypothesis; young animals treated with a compound that reduces the number of synapses did not show impaired discrimination compared to controls. Additionally, their results from older animals showing impaired discrimination suggest that age-related changes aside from synaptopathy are responsible for the age-related decline in discrimination.

Strengths:

(1) The rationale and hypothesis are well-motivated and clearly presented.

(2) The study was well conducted with strong methodology for the most part, and good experimental control. The combination of physiological and behavioral techniques is powerful and informative. Reducing synapse counts fairly directly using ouabain is a cleaner design than using noise exposure or age (as in other studies), since these latter modifiers have additional effects on auditory function.

(3) The study may have a considerable impact on the field. The findings could have important implications for our understanding of cochlear synaptopathy, one of the most highly researched and potentially impactful developments in hearing science in the past fifteen years.

Weaknesses:

(1) I have concerns that the gerbils may not have been performing the behavioral task using temporal fine structure information.

Human studies using the same task employed a filter center frequency that was (at least) 11 times the fundamental frequency (Marmel et al., 2015; Moore and Sek, 2009). Moore and Sek wrote: "the default (recommended) value of the centre frequency is 11F0." Here, the center frequency was only 4 or 8 times the fundamental frequency (4F0 or 8F0). Hence, relative to harmonic frequency, the harmonic spacing was considerably greater in the present study. However, gerbil auditory filters are thought to be broader than those in human. In the revised version of the manuscript, the authors provide modelling results suggesting that the excitation patterns were discriminable for the 4F0 conditions, but may not have been for the 8F0 conditions. These results provide some reassurance that the 8F0 discriminations were dependent on temporal cues, but the description of the model lacks detail. Also, the authors state that "thus, for these two conditions with harmonic number N of 8 the gerbils cannot rely on differences in the excitation patterns but must solve the task by comparing the temporal fine structure." This is too strong. Pulsed tone intensity difference limens (the reference used for establishing whether or not the excitation pattern cues were usable) may not be directly comparable to profile-analysis-like conditions, and it has been argued that frequency discrimination may be more sensitive to excitation pattern cues than predicted from a simple comparison to intensity difference limens (Micheyl et al. 2013, https://doi.org/10.1371/journal.pcbi.1003336).

I'm also somewhat concerned that the masking noise used in the present study was too low in level to mask cochlear distortion products. Based on their excitation pattern modelling, the authors state (without citation) that "since the level of excitation produced by the pink noise is less than 30 dB below that produced by the complex tones, distortion products will be masked." The basis for this claim is not clear. In human, distortion products may be only ~20 dB below the levels of the primaries (referenced to an external sound masker / canceller, which is appropriate, assuming that the modelling reported in the present paper did not include middle-ear effects; see Norman-Haignere and McDermott, 2016, doi: 10.1016/j.neuroimage.2016.01.050). Oxenham et al. (2009, doi: 10.1121/1.3089220) provide further cautionary evidence on the potential use of distortion product cues when the background noise level is too low (in their case the relative level of the noise in the compromised condition was only a little below that used in the present study). The masking level used in the present study may have been sufficient, but it would be useful to have some further reassurance on this point.

(2) The synapse reductions in the high ouabain and old groups were relatively small (mean of 19 synapses per hair cell compared to 23 in the young untreated group). In contrast, in some mouse models of the effects of noise exposure or age, a 50% reduction in synapses is observed, and in the human temporal bone study of Wu et al. (2021, https://doi.org/10.1523/JNEUROSCI.3238-20.2021) the age-related reduction in auditory nerve fibres was ~50% or greater for the highest age group across cochlear location. It could be simply that the synapse loss in the present study was too small to produce significant behavioral effects. Hence, although the authors provide evidence that in the gerbil model the age-related behavioral effects are not due to synaptopathy, this may not translate to other species (including human).

(3) The study was not pre-registered, and there was no a priori power calculation, so there is less confidence in replicability than could have been the case. Only three old animals were used in the behavioral study, which raises concerns about the reliability of comparisons involving this group. Statistical analyses on very small samples can be unreliable due to problems of power, generalisability, and susceptibility to outliers.

Reviewer #3 (Public review):

This study is a part of the ongoing series of rigorous work from this group exploring neural coding deficits in the auditory nerve, and dissociating the effects of cochlear synaptopathy from other age-related deficits. They have previously shown no evidence of phase-locking deficits in the remaining auditory nerve fibers in quiet-aged gerbils. Here, they study the effects of aging on the perception and neural coding of temporal fine structure cues in the same Mongolian gerbil model.

They measure TFS coding in the auditory nerve using the TFS1 task which uses a combination of harmonic and tone-shifted inharmonic tones which differ primarily in their TFS cues (and not the envelope). They then follow this up with a behavioral paradigm using the TFS1 task in these gerbils. They test young normal hearing gerbils, aged gerbils, and young gerbils with cochlear synaptopathy induced using the neurotoxin ouabain to mimic synapse losses seen with age.

In the behavioral paradigm, they find that aging is associated with decreased performance compared to the young gerbils, whereas young gerbils with similar levels of synapse loss do not show these deficits. When looking at the auditory nerve responses, they find no differences in neural coding of TFS cues across any of the groups. However, aged gerbils show an increase in the representation of periodicity envelope cues (around f0) compared to young gerbils or those with induced synapse loss. The authors hence conclude that synapse loss by itself doesn't seem to be important for distinguishing TFS cues, and rather the behavioral deficits with age are likely having to do with the misrepresented envelope cues instead.

The manuscript is well written, and the data presented are robust. Some of the points below will need to be considered while interpreting the results of the study, in its current form. These considerations are addressable if deemed necessary, with some additional analysis in future versions of the manuscript.

Spontaneous rates - Figure S2 shows no differences in median spontaneous rates across groups. But taking the median glosses over some of the nuances there. Ouabain (in the Bourien study) famously affects low spont rates first, and at a higher degree than median or high spont rates. It seems to be the case (qualitatively) in figure S2 as well, with almost no units in the low spont region in the ouabain group, compared to the other groups. Looking at distributions within each spont rate category and comparing differences across the groups might reveal some of the underlying causes for these changes. Given that overall, the study reports that low-SR fibers had a higher ENV/TFS log-z-ratio, the distribution of these fibers across groups may reveal specific effects of TFS coding by group.

[Update: The revised manuscript has addressed these issues]

Threshold shifts - It is unclear from the current version if the older gerbils have changes in hearing thresholds, and whether those changes may be affecting behavioral thresholds. The behavioral stimuli appear to have been presented at a fixed sound level for both young and aged gerbils, similar to the single unit recordings. Hence, age-related differences in behavior may have been due to changes in relative sensation level. Approaches such as using hearing thresholds as covariates in the analysis will help explore if older gerbils still show behavioral deficits.

[Update: The issue of threshold shifts with aging gerbils is still unresolved in my opinion. From the revised manuscript, it appears that aged gerbils have a 36dB shift in thresholds. While the revised manuscript provides convincing evidence that these threshold shifts do not affect the auditory nerve tuning properties, the behavioral paradigm was still presented at the same sound level for young and aged animals. But a potential 36 dB change in sensation level may affect behavioral results. The authors may consider adding thresholds as covariates in analyses or present any evidence that behavioral thresholds are plateaued along that 30dB range].

Task learning in aged gerbils - It is unclear if the aged gerbils really learn the task well in two of the three TFS1 test conditions. The d' of 1 which is usually used as the criterion for learning was not reached in even the easiest condition for aged gerbils in all but one condition for the aged gerbils (Fig. 5H) and in that condition, there doesn't seem to be any age-related deficits in behavioral performance (Fig. 6B). Hence dissociating the inability to learn the task from the inability to perceive TFS 1 cues in those animals becomes challenging.

[Update: The revised manuscript sufficiently addresses these issues, with the caveat of hearing threshold changes affecting behavioral thresholds mentioned above].

Increased representation of periodicity envelope in the AN - the mechanisms for increased representation of periodicity envelope cues is unclear. The authors point to some potential central mechanisms but given that these are recordings from the auditory nerve what central mechanisms these may be is unclear. If the authors are suggesting some form of efferent modulation only at the f0 frequency, no evidence for this is presented. It appears more likely that the enhancement may be due to outer hair cell dysfunction (widened tuning, distorted tonotopy). Given this increased envelope coding, the potential change in sensation level for the behavior (from the comment above), and no change in neural coding of TFS cues across any of the groups, a simpler interpretation may be -TFS coding is not affected in remaining auditory nerve fibers after age-related or ouabain induced synapse loss, but behavioral performance is affected by altered outer hair cell dysfunction with age.

[Update: The revised manuscript has addressed these issues]

Emerging evidence seems to suggest that cochlear synaptopathy and/or TFS encoding abilities might be reflected in listening effort rather than behavioral performance. Measuring some proxy of listening effort in these gerbils (like reaction time) to see if that has changed with synapse loss, especially in the young animals with induced synaptopathy, would make an interesting addition to explore perceptual deficits of TFS coding with synapse loss.

[Update: The revised manuscript has addressed these issues]

Author response:

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

Reviewer #2 (Public review):

Summary:

Using a gerbil model, the authors tested the hypothesis that loss of synapses between sensory hair cells and auditory nerve fibers (which may occur due to noise exposure or aging) affects behavioral discrimination of the rapid temporal fluctuations of sounds. In contrast to previous suggestions in the literature, their results do not support this hypothesis; young animals treated with a compound that reduces the number of synapses did not show impaired discrimination compared to controls. Additionally, their results from older animals showing impaired discrimination suggest that age-related changes aside from synaptopathy are responsible for the age-related decline in discrimination.

Strengths:

(1) The rationale and hypothesis are well-motivated and clearly presented.

(2) The study was well conducted with strong methodology for the most part, and good experimental control. The combination of physiological and behavioral techniques is powerful and informative. Reducing synapse counts fairly directly using ouabain is a cleaner design than using noise exposure or age (as in other studies), since these latter modifiers have additional effects on auditory function.

(3) The study may have a considerable impact on the field. The findings could have important implications for our understanding of cochlear synaptopathy, one of the most highly researched and potentially impactful developments in hearing science in the past fifteen years.

Weaknesses:

(1) I have concerns that the gerbils may not have been performing the behavioral task using temporal fine structure information.

Human studies using the same task employed a filter center frequency that was (at least) 11 times the fundamental frequency (Marmel et al., 2015; Moore and Sek, 2009). Moore and Sek wrote: "the default (recommended) value of the centre frequency is 11F0." Here, the center frequency was only 4 or 8 times the fundamental frequency (4F0 or 8F0). Hence, relative to harmonic frequency, the harmonic spacing was considerably greater in the present study. However, gerbil auditory filters are thought to be broader than those in human. In the revised version of the manuscript, the authors provide modelling results suggesting that the excitation patterns were discriminable for the 4F0 conditions, but may not have been for the 8F0 conditions. These results provide some reassurance that the 8F0 discriminations were dependent on temporal cues, but the description of the model lacks detail. Also, the authors state that "thus, for these two conditions with harmonic number N of 8 the gerbils cannot rely on differences in the excitation patterns but must solve the task by comparing the temporal fine structure." This is too strong. Pulsed tone intensity difference limens (the reference used for establishing whether or not the excitation pattern cues were usable) may not be directly comparable to profile-analysis-like conditions, and it has been argued that frequency discrimination may be more sensitive to excitation pattern cues than predicted from a simple comparison to intensity difference limens (Micheyl et al. 2013, https://doi.org/10.1371/journal.pcbi.1003336

We can assume that our conclusions based on the excitation patterns are adequate when putting gerbil auditory filter data, frequency difference limens and intensity difference limens together into perspective. Kittel et al. (2002) observed an about factor 2 larger auditory-filter bandwidth in the gerbil than in humans reducing the number of independent frequency channels in the analysis of excitation patterns. The gerbil frequency-difference limen for pure tones being an indicator for the sensitivity to make use of excitation patterns is more than an order of magnitude larger than the corresponding human frequency difference limen (Klinge and Klump 2009, https://doi.org/10.1121/1.3021315). Finally, the gerbil intensity-difference limen of 2.8 dB observed for 1-kHz pure tones is considerably larger than the 0.75 dB observed for humans in the same study (Sinnott et al. 1992). Thus, taken together these lines of evidence indicate that our conclusions regarding the potential use of excitation patterns are not too strong.

I'm also somewhat concerned that the masking noise used in the present study was too low in level to mask cochlear distortion products. Based on their excitation pattern modelling, the authors state (without citation) that "since the level of excitation produced by the pink noise is less than 30 dB below that produced by the complex tones, distortion products will be masked." The basis for this claim is not clear. In human, distortion products may be only ~20 dB below the levels of the primaries (referenced to an external sound masker / canceller, which is appropriate, assuming that the modelling reported in the present paper did not include middle-ear effects; see Norman-Haignere and McDermott, 2016, doi: 10.1016/j.neuroimage.2016.01.050). Oxenham et al. (2009, doi: 10.1121/1.3089220) provide further cautionary evidence on the potential use of distortion product cues when the background noise level is too low (in their case the relative level of the noise in the compromised condition was only a little below that used in the present study). The masking level used in the present study may have been sufficient, but it would be useful to have some further reassurance on this point.

In the method section, we provide the citation for estimating the size of the distortion products and the estimated signal-to-noise ratio making the basis for our estimates clear.

We consulted Oxenham et al. (2009, doi: 10.1121/1.3089220) who suggested that distortion products may have been used in human subjects. However, in Fig. 1 of their paper, they convincingly demonstrate that even for humans that have more narrow auditory filters than gerbils, spectral cues cannot be used to evaluate the frequency shift in harmonic complex tones. We are confident that the same limitation applies to gerbils that have wider auditory filters than humans and a lower ability to use spectral cues as indicated by their higher frequency-difference limens and intensity-difference limens compared to humans.

(2) The synapse reductions in the high ouabain and old groups were relatively small (mean of 19 synapses per hair cell compared to 23 in the young untreated group). In contrast, in some mouse models of the effects of noise exposure or age, a 50% reduction in synapses is observed, and in the human temporal bone study of Wu et al. (2021, https://doi.org/10.1523/JNEUROSCI.3238-20.2021) the age-related reduction in auditory nerve fibres was ~50% or greater for the highest age group across cochlear location. It could be simply that the synapse loss in the present study was too small to produce significant behavioral effects. Hence, although the authors provide evidence that in the gerbil model the age-related behavioral effects are not due to synaptopathy, this may not translate to other species (including human).

(3) The study was not pre-registered, and there was no a priori power calculation, so there is less confidence in replicability than could have been the case. Only three old animals were used in the behavioral study, which raises concerns about the reliability of comparisons involving this group.

Reviewer #3 (Public review):

This study is a part of the ongoing series of rigorous work from this group exploring neural coding deficits in the auditory nerve, and dissociating the effects of cochlear synaptopathy from other age-related deficits. They have previously shown no evidence of phase-locking deficits in the remaining auditory nerve fibers in quiet-aged gerbils. Here, they study the effects of aging on the perception and neural coding of temporal fine structure cues in the same Mongolian gerbil model.

They measure TFS coding in the auditory nerve using the TFS1 task which uses a combination of harmonic and tone-shifted inharmonic tones which differ primarily in their TFS cues (and not the envelope). They then follow this up with a behavioral paradigm using the TFS1 task in these gerbils. They test young normal hearing gerbils, aged gerbils, and young gerbils with cochlear synaptopathy induced using the neurotoxin ouabain to mimic synapse losses seen with age.

In the behavioral paradigm, they find that aging is associated with decreased performance compared to the young gerbils, whereas young gerbils with similar levels of synapse loss do not show these deficits. When looking at the auditory nerve responses, they find no differences in neural coding of TFS cues across any of the groups. However, aged gerbils show an increase in the representation of periodicity envelope cues (around f0) compared to young gerbils or those with induced synapse loss. The authors hence conclude that synapse loss by itself doesn't seem to be important for distinguishing TFS cues, and rather the behavioral deficits with age are likely having to do with the misrepresented envelope cues instead.

The manuscript is well written, and the data presented are robust. Some of the points below will need to be considered while interpreting the results of the study, in its current form. These considerations are addressable if deemed necessary, with some additional analysis in future versions of the manuscript.

Spontaneous rates - Figure S2 shows no differences in median spontaneous rates across groups. But taking the median glosses over some of the nuances there. Ouabain (in the Bourien study) famously affects low spont rates first, and at a higher degree than median or high spont rates. It seems to be the case (qualitatively) in figure S2 as well, with almost no units in the low spont region in the ouabain group, compared to the other groups. Looking at distributions within each spont rate category and comparing differences across the groups might reveal some of the underlying causes for these changes. Given that overall, the study reports that low-SR fibers had a higher ENV/TFS log-z-ratio, the distribution of these fibers across groups may reveal specific effects of TFS coding by group.

[Update: The revised manuscript has addressed these issues]

Threshold shifts - It is unclear from the current version if the older gerbils have changes in hearing thresholds, and whether those changes may be affecting behavioral thresholds. The behavioral stimuli appear to have been presented at a fixed sound level for both young and aged gerbils, similar to the single unit recordings. Hence, age-related differences in behavior may have been due to changes in relative sensation level. Approaches such as using hearing thresholds as covariates in the analysis will help explore if older gerbils still show behavioral deficits.

[Update: The issue of threshold shifts with aging gerbils is still unresolved in my opinion. From the revised manuscript, it appears that aged gerbils have a 36dB shift in thresholds. While the revised manuscript provides convincing evidence that these threshold shifts do not affect the auditory nerve tuning properties, the behavioral paradigm was still presented at the same sound level for young and aged animals. But a potential 36 dB change in sensation level may affect behavioral results. The authors may consider adding thresholds as covariates in analyses or present any evidence that behavioral thresholds are plateaued along that 30dB range].

Since we do not have behavioural detection thresholds from our individual animals, only CAP thresholds that represent the auditory-nerve data and cannot be translated to behavioural thresholds directly, we want to refrain from using these indirect measures as covariates in the present analysis. In addition, the study by Hamann et al. (2002, https://doi.org/10.1016/S0378-5955(02)00454-9) indicates that age-related behavioural threshold increases are smaller than threshold increases obtained from auditory brainstem response measurements. Finally, statistical analyses on very small samples can be unreliable due to problems of power, generalisability, and susceptibility to outliers.

Moore and Sek (2009) in their paper on the TFS1 test pointed out that the effect of signal level on the TFS1 threshold in normal hearing human subjects was small when the signal-to-noise ratio between the broadband masking noise and the complex tone was kept constant. Furthermore, the masking noise will raise the thresholds of normal hearing gerbils and old gerbils with an audibility threshold increase to about the same signal-to-noise ratio. Thus, as long as the signal remains audible to the behaviourally tested gerbil which can be expected at an overall signal level of 68 dB SPL, we expect little effect of raised audibility thresholds on the TFS1 threshold. The lack of temporal processing deficits in the auditory-nerve fibers of old, mildly hearing impaired gerbils compared to those in normal hearing young adult gerbils further strengthens this argument.

Task learning in aged gerbils - It is unclear if the aged gerbils really learn the task well in two of the three TFS1 test conditions. The d' of 1 which is usually used as the criterion for learning was not reached in even the easiest condition for aged gerbils in all but one condition for the aged gerbils (Fig. 5H) and in that condition, there doesn't seem to be any age-related deficits in behavioral performance (Fig. 6B). Hence dissociating the inability to learn the task from the inability to perceive TFS 1 cues in those animals becomes challenging.

[Update: The revised manuscript sufficiently addresses these issues, with the caveat of hearing threshold changes affecting behavioral thresholds mentioned above].

As we argued above, an audibility threshold increase in the old gerbils is unlikely to explain the raised TFS1 thresholds in the old gerbils.

Increased representation of periodicity envelope in the AN - the mechanisms for increased representation of periodicity envelope cues is unclear. The authors point to some potential central mechanisms but given that these are recordings from the auditory nerve what central mechanisms these may be is unclear. If the authors are suggesting some form of efferent modulation only at the f0 frequency, no evidence for this is presented. It appears more likely that the enhancement may be due to outer hair cell dysfunction (widened tuning, distorted tonotopy). Given this increased envelope coding, the potential change in sensation level for the behavior (from the comment above), and no change in neural coding of TFS cues across any of the groups, a simpler interpretation may be -TFS coding is not affected in remaining auditory nerve fibers after age-related or ouabain induced synapse loss, but behavioral performance is affected by altered outer hair cell dysfunction with age.

[Update: The revised manuscript has addressed these issues]

Emerging evidence seems to suggest that cochlear synaptopathy and/or TFS encoding abilities might be reflected in listening effort rather than behavioral performance. Measuring some proxy of listening effort in these gerbils (like reaction time) to see if that has changed with synapse loss, especially in the young animals with induced synaptopathy, would make an interesting addition to explore perceptual deficits of TFS coding with synapse loss.

[Update: The revised manuscript has addressed these issues]

Reviewer #3 (Recommendations for the authors):

Thank you for your revisions. They largely address most of my initial concerns. The issue of threshold shifts potentially affecting behavioral thresholds still remains unresolved in my opinion. The new data about unaltered tuning curves is convincing that the auditory nerve fiber recordings are unaffected by threshold shifts. But am I correct in my understanding that the threshold shift with age was 36 dB relative to the young (L168)? If so, wouldn't the fact that behavior was performed at 68 dB SPL regardless of group affect the behavioral thresholds with age? Is there any additional evidence that suggests that behavioral performance plateaus along that ~30dB range that the authors could include to strengthen this claim?

In our response above to reviewer #3 and to reviewer #2 we provided additional arguments why we think that an audibility threshold increase in old gerbils cannot explain their compromised TFS1 thresholds.

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

Reviewer #1(Public review)  

Summary:  

The authors investigate the effects of aging on auditory system performance in understanding temporal fine structure (TFS), using both behavioral assessments and physiological recordings from the auditory periphery, specifically at the level of the auditory nerve. This dual approach aims to enhance understanding of the mechanisms underlying observed behavioral outcomes. The results indicate that aged animals exhibit deficits in behavioral tasks for distinguishing between harmonic and inharmonic sounds, which is a standard test for TFS coding. However, neural responses at the auditory nerve level do not show significant differences when compared to those in young, normalhearing animals. The authors suggest that these behavioral deficits in aged animals are likely attributable to dysfunctions in the central auditory system, potentially as a consequence of aging. To further investigate this hypothesis, the study includes an animal group with selective synaptic loss between inner hair cells and auditory nerve fibers, a condition known as cochlear synaptopathy (CS).CS is a pathology associated with aging and is thought to be an early indicator of hearing impairment. Interestingly, animals with selective CS showed physiological and behavioral TFS coding similar to that of the young normal-hearing group, contrasting with the aged group's deficits. Despite histological evidence of significant synaptic loss in the CS group, the study concludes that CS does not appear to affect TFS coding, either behaviorally or physiologically.  

We agree with the reviewer’s summary.

Strengths:  

This study addresses a critical health concern, enhancing our understanding of mechanisms underlying age-related difficulties in speech intelligibility, even when audiometric thresholds are within normal limits. A major strength of this work is the comprehensive approach, integrating behavioral assessments, auditory nerve (AN) physiology, and histology within the same animal subjects. This approach enhances understanding of the mechanisms underlying the behavioral outcomes and provides confidence in the actual occurrence of synapse loss and its effects. The study carefully manages controlled conditions by including five distinct groups: young normal-hearing animals, aged animals, animals with CS induced through low and high doses, and a sham surgery group. This careful setup strengthens the study's reliability and allows for meaningful comparisons across conditions. Overall, the manuscript is well-structured, with clear and accessible writing that facilitates comprehension of complex concepts.

Weaknesses:

The stimulus and task employed in this study are very helpful for behavioral research, and using the same stimulus setup for physiology is advantageous for mechanistic comparisons. However, I have some concerns about the limitations in auditory nerve (AN) physiology. Due to practical constraints, it is not feasible to record from a large enough population of fibers that covers a full range of best frequencies (BFs) and spontaneous rates (SRs) within each animal. This raises questions about how representative the physiological data are for understanding the mechanism in behavioral data. I am curious about the authors' interpretation of how this stimulus setup might influence results compared to methods used by Kale and Heinz (2010), who adjusted harmonic frequencies based on the characteristic frequency (CF) of recorded units. While, the harmonic frequencies in this study are fixed across all CFs, meaning that many AN fibers may not be tuned closely to the stimulus frequencies. If units are not responsive to the stimulus further clarification on detecting mistuning and phase locking to TFS effects within this setup would be valuable. Since the harmonic frequencies in this study are fixed across all CFs, this means that many AN fibers may not be tuned closely to the stimulus frequencies, adding sampling variability to the results.

We chose the stimuli for the AN recordings to be identical to the stimuli used in the behavioral evaluation of the perceptual sensitivity. Only with this approach can we directly compare the response of the population of AN fibers with perception measured in behavior.

The stimuli are complex, i.e., comprise of many frequency components AND were presented at 68 dB SPL. Thus, the stimuli excite a given fiber within a large portion of the fiber’s receptive field. Furthermore, during recordings, we assured ourselves that fibers responded to the stimuli by audiovisual control. Otherwise it would have cost valuable recording time to record from a nonresponsive AN fiber.

Given the limited number of units per condition-sometimes as few as three for certain conditions - I wonder if CF-dependent variability might impact the results of the AN data in this study and discussing this factor can help with better understanding the results. While the use of the same stimuli for both behavioral and physiological recordings is understandable, a discussion on how this choice affects interpretation would be beneficial. In addition a 60 dB stimulus could saturate high spontaneous rate (HSR) AN fibers, influencing neural coding and phase-locking to TFS. Potentially separating SR groups, could help address these issues and improve interpretive clarity.  

A deeper discussion on the role of fiber spontaneous rate could also enhance the study. How might considering SR groups affect AN results related to TFS coding? While some statistical measures are included in the supplement, a more detailed discussion in the main text could help in interpretation.  We do not think that it will be necessary to conduct any statistical analysis in addition to that already reported in the supplement.  

We considered moving some supplementary information back into the main manuscript but decided against it. Our single-unit sample was not sufficient, i.e. not all subpopulations of auditory-nerve fibers were sufficiently sampled for all animal treatment groups, to conclusively resolve every aspect that may be interesting to explore. The power of our approach lies in the direct linkage of several levels of investigation – cochlear synaptic morphology, single-unit representation and behavioral performance – and, in the main manuscript, we focus on the core question of synaptopathy and its relation to temporal fine structure perception. This is now spelled out clearly in lines 197 - 203 of the main manuscript.  

Although Figure S2 indicates no change in median SR, the high-dose treatment group lacks LSR fibers, suggesting a different distribution based on SR for different animal groups, as seen in similar studies on other species. A histogram of these results would be informative, as LSR fiber loss with CS-whether induced by ouabain in gerbils or noise in other animals-is well documented (e.g., Furman et al., 2013).  

Figure S2 was revised to avoid overlap of data points and show the distributions more clearly. Furthermore, the sample sizes for LSR and HSR fibers are now provided separately.

Although ouabain effects on gerbils have been explored in previous studies, since these data already seems to be recorded for the animal in this study, a brief description of changes in auditory brainstem response (ABR) thresholds, wave 1 amplitudes, and tuning curves for animals with cochlear synaptopathy (CS) in this study would be beneficial. This would confirm that ouabain selectively affects synapses without impacting outer hair cells (OHCs). For aged animals, since ABR measurements were taken, comparing hearing differences between normal and aged groups could provide insights into the pathologies besides CS in aged animals. Additionally, examining subject variability in treatment effects on hearing and how this correlates with behavior and physiology would yield valuable insights. If limited space maybe a brief clarification or inclusion in supplementary could be good enough.  

We thank the reviewer for this constructive suggestion. The requested data were added in a new section of the Results, entitled “Threshold sensitivity and frequency tuning were not affected by the synapse loss.” (lines 150 – 174). Our young-adult, ouabain-treated gerbils showed no significant elevations of CAP thresholds and their neural tuning was normal. Old gerbils showed the typical threshold losses for individuals of comparable age, and normal neural tuning, confirming previous reports. Thus, there was no evidence for relevant OHC impairments in any of our animal groups.   

Another suggestion is to discuss the potential role of MOC efferent system and effect of anesthesia in reducing efferent effects in AN recordings. This is particularly relevant for aged animals, as CS might affect LSR fibers, potentially disrupting the medial olivocochlear (MOC) efferent pathway. Anesthesia could lessen MOC activity in both young and aged animals, potentially masking efferent effects that might be present in behavioral tasks. Young gerbils with functional efferent systems might perform better behaviorally, while aged gerbils with impaired MOC function due to CS might lack this advantage. A brief discussion on this aspect could potentially enhance mechanistic insights.  

Thank you for this suggestion. The potential role of olivocochlear efferents is now discussed in lines 597 - 613.

Lastly, although synapse counts did not differ between the low-dose treatment and NH I sham groups, separating these groups rather than combining them with the sham might reveal differences in behavior or AN results, particularly regarding the significance of differences between aged/treatment groups and the young normal-hearing group.  

For maximizing statistical power, we combined those groups in the statistical analysis. These two groups did not differ in synapse number, threshold sensitivity or neural tuning bandwidths.

Reviewer #2 (Public review):

Summary:  

Using a gerbil model, the authors tested the hypothesis that loss of synapses between sensory hair cells and auditory nerve fibers (which may occur due to noise exposure or aging) affects behavioral discrimination of the rapid temporal fluctuations of sounds. In contrast to previous suggestions in the literature, their results do not support this hypothesis; young animals treated with a compound that reduces the number of synapses did not show impaired discrimination compared to controls. Additionally, their results from older animals showing impaired discrimination suggest that agerelated changes aside from synaptopathy are responsible for the age-related decline in discrimination. 

We agree with the reviewer’s summary.

Strengths: 

(1) The rationale and hypothesis are well-motivated and clearly presented. 

(2) The study was well conducted with strong methodology for the most part, and good experimental control. The combination of physiological and behavioral techniques is powerful and informative. Reducing synapse counts fairly directly using ouabain is a cleaner design than using noise exposure or age (as in other studies), since these latter modifiers have additional effects on auditory function. 

(3) The study may have a considerable impact on the field. The findings could have important implications for our understanding of cochlear synaptopathy, one of the most highly researched and potentially impactful developments in hearing science in the past fifteen years.  

Weaknesses: 

(1) My main concern is that the stimuli may not have been appropriate for assessing neural temporal coding behaviorally. Human studies using the same task employed a filter center frequency that was (at least) 11 times the fundamental frequency (Marmel et al., 2015; Moore and Sek, 2009). Moore and Sek wrote: "the default (recommended) value of the centre frequency is 11F0." Here, the center frequency was only 4 or 8 times the fundamental frequency (4F0 or 8F0). Hence, relative to harmonic frequency, the harmonic spacing was considerably greater in the present study. By my calculations, the masking noise used in the present study was also considerably lower in level relative to the harmonic complex than that used in the human studies. These factors may have allowed the animals to perform the task using cues based on the pattern of activity across the neural array (excitation pattern cues), rather than cues related to temporal neural coding. The authors show that mean neural driven rate did not change with frequency shift, but I don't understand the relevance of this. It is the change in response of individual fibers with characteristic frequencies near the lowest audible harmonic that is important here.  

The auditory filter bandwidth of the gerbil is about double that of human subjects. Because of this, the masking noise has a larger overall level than in the human studies in the filter, prohibiting the use of distortion products. The larger auditory filter bandwidth precludes that the gerbils can use excitation patterns, especially in the condition with a center frequency of 1600 Hz and a fundamental of 200 Hz and in the condition with a center frequency of 3200 Hz and a fundamental of 400 Hz. In the condition with a center frequency of 1600 Hz and a fundamental of 400 Hz, it is possible that excitation patterns are exploited. We have now added  modeling of the excitation patterns, and a new figure showing their change at the gerbils’ perception threshold, in the discussion of the revised version (lines 440 - 446 and Fig. 8).

The case against excitation pattern cues needs to be better made in the Discussion. It could be that gerbil frequency selectivity is broad enough for this not to be an issue, but more detail needs to be provided to make this argument. The authors should consider what is the lowest audible harmonic in each case for their stimuli, given the level of each harmonic and the level of the pink noise. Even for the 8F0 center frequency, the lowest audible harmonic may be as low as the 4th (possibly even the 3rd). In human, harmonics are thought to be resolvable by the cochlea up to at least the 8th.  

This issue is now covered in the discussion, see response to the previous point.

(2) The synapse reductions in the high ouabain and old groups were relatively small (mean of 19 synapses per hair cell compared to 23 in the young untreated group). In contrast, in some mouse models of the effects of noise exposure or age, a 50% reduction in synapses is observed, and in the human temporal bone study of Wu et al. (2021, https://doi.org/10.1523/JNEUROSCI.3238-20.2021) the age-related reduction in auditory nerve fibres was ~50% or greater for the highest age group across cochlear location. It could be simply that the synapse loss in the present study was too small to produce significant behavioral effects. Hence, although the authors provide evidence that in the gerbil model the age-related behavioral effects are not due to synaptopathy, this may not translate to other species (including human). This should be discussed in the manuscript. 

We agree that our results apply to moderate synaptopathy, which predominantly characterizes early stages of hearing loss or aged individuals without confounding noise-induced cochlear damage. This is now discussed in lines 486 – 498.

It would be informative to provide synapse counts separately for the animals who were tested behaviorally, to confirm that the pattern of loss across the group was the same as for the larger sample.  

Yes, the pattern was the same for the subgroup of behaviorally tested animals. We have added this information to the revised version of the manuscript (lines 137 – 141).

(3) The study was not pre-registered, and there was no a priori power calculation, so there is less confidence in replicability than could have been the case. Only three old animals were used in the behavioral study, which raises concerns about the reliability of comparisons involving this group.  

The results for the three old subjects differed significantly from those of young subjects and young ouabain-treated subjects. This indicates a sufficient statistical power, since otherwise no significant differences would be observed.

Reviewer #3 (Public review):

This study is a part of the ongoing series of rigorous work from this group exploring neural coding deficits in the auditory nerve, and dissociating the effects of cochlear synaptopathy from other agerelated deficits. They have previously shown no evidence of phase-locking deficits in the remaining auditory nerve fibers in quiet-aged gerbils. Here, they study the effects of aging on the perception and neural coding of temporal fine structure cues in the same Mongolian gerbil model. 

They measure TFS coding in the auditory nerve using the TFS1 task which uses a combination of harmonic and tone-shifted inharmonic tones which differ primarily in their TFS cues (and not the envelope). They then follow this up with a behavioral paradigm using the TFS1 task in these gerbils. They test young normal hearing gerbils, aged gerbils, and young gerbils with cochlear synaptopathy induced using the neurotoxin ouabain to mimic synapse losses seen with age. 

In the behavioral paradigm, they find that aging is associated with decreased performance compared to the young gerbils, whereas young gerbils with similar levels of synapse loss do not show these deficits. When looking at the auditory nerve responses, they find no differences in neural coding of TFS cues across any of the groups. However, aged gerbils show an increase in the representation of periodicity envelope cues (around f0) compared to young gerbils or those with induced synapse loss. The authors hence conclude that synapse loss by itself doesn't seem to be important for distinguishing TFS cues, and rather the behavioral deficits with age are likely having to do with the misrepresented envelope cues instead.  

We agree with the reviewer’s summary.

The manuscript is well written, and the data presented are robust. Some of the points below will need to be considered while interpreting the results of the study, in its current form. These considerations are addressable if deemed necessary, with some additional analysis in future versions of the manuscript. 

Spontaneous rates - Figure S2 shows no differences in median spontaneous rates across groups. But taking the median glosses over some of the nuances there. Ouabain (in the Bourien study) famously affects low spont rates first, and at a higher degree than median or high spont rates. It seems to be the case (qualitatively) in Figure S2 as well, with almost no units in the low spont region in the ouabain group, compared to the other groups. Looking at distributions within each spont rate category and comparing differences across the groups might reveal some of the underlying causes for these changes. Given that overall, the study reports that low-SR fibers had a higher ENV/TFS log-zratio, the distribution of these fibers across groups may reveal specific effects of TFS coding by group.  

As the reviewer points out, our sample from the group treated with a high concentration of ouabain showed very few low-spontaneous-rate auditory-nerve fibers, as expected from previous work. However, this was also true, e.g., for our sample from sham-operated animals, and may thus well reflect a sampling bias. We are therefore reluctant to attach much significance to these data distributions. We now point out more clearly the limitations of our auditory-nerve sample for the exploration of  interesting questions beyond our core research aim (see also response to Reviewer 1 above).  

Threshold shifts - It is unclear from the current version if the older gerbils have changes in hearing thresholds, and whether those changes may be affecting behavioral thresholds. The behavioral stimuli appear to have been presented at a fixed sound level for both young and aged gerbils, similar to the single unit recordings. Hence, age-related differences in behavior may have been due to changes in relative sensation level. Approaches such as using hearing thresholds as covariates in the analysis will help explore if older gerbils still show behavioral deficits.  

Unfortunately, we did not obtain behavioral thresholds that could be used here. We want to point out that the TFS 1 stimuli had an overall level of 68 dB SPL, and the pink noise masker would have increased the threshold more than expected from the moderate, age-related hearing loss in quiet. Thus, the masked thresholds for all gerbil groups are likely similar and should have no effect on the behavioral results.

Task learning in aged gerbils - It is unclear if the aged gerbils really learn the task well in two of the three TFS1 test conditions. The d' of 1 which is usually used as the criterion for learning was not reached in even the easiest condition for aged gerbils in all but one condition for the aged gerbils (Fig. 5H) and in that condition, there doesn't seem to be any age-related deficits in behavioral performance (Fig. 6B). Hence dissociating the inability to learn the task from the inability to perceive TFS 1 cues in those animals becomes challenging.  

Even in the group of gerbils with the lowest sensitivity, for the condition 400/1600 the animals achieved a d’ of on average above 1. Furthermore, stimuli were well above threshold and audible, even when no discrimination could be observed. Finally, as explained in the methods, different stimulus conditions were interleaved in each session, providing stimuli that were easy to discriminate together with those being difficult to discriminate. This approach ensures that the gerbils were under stimulus control, meaning properly trained to perform the task. Thus, an inability to discriminate does not indicate a lack of proper training.  

Increased representation of periodicity envelope in the AN - the mechanisms for increased representation of periodicity envelope cues is unclear. The authors point to some potential central mechanisms but given that these are recordings from the auditory nerve what central mechanisms these may be is unclear. If the authors are suggesting some form of efferent modulation only at the f0 frequency, no evidence for this is presented. It appears more likely that the enhancement may be due to outer hair cell dysfunction (widened tuning, distorted tonotopy). Given this increased envelope coding, the potential change in sensation level for the behavior (from the comment above), and no change in neural coding of TFS cues across any of the groups, a simpler interpretation may be -TFS coding is not affected in remaining auditory nerve fibers after age-related or ouabain induced synapse loss, but behavioral performance is affected by altered outer hair cell dysfunction with age. 

A similar point was made by Reviewer #1. As indicated above, new data on threshold sensitivity and neural tuning were added in a new section of the Results which indirectly suggest that significant OHC pathologies were not a concern, neither in our young-adult, synaptopathic gerbils nor in the old gerbils.  

Emerging evidence seems to suggest that cochlear synaptopathy and/or TFS encoding abilities might be reflected in listening effort rather than behavioral performance. Measuring some proxy of listening effort in these gerbils (like reaction time) to see if that has changed with synapse loss, especially in the young animals with induced synaptopathy, would make an interesting addition to explore perceptual deficits of TFS coding with synapse loss.  

This is an interesting suggestion that we now explore in the revision of the manuscript. Reaction times can be used as a proxy for listening effort and were recorded for all responses. The the new analysis now reported in lines 378 - 396 compared young-adult control gerbils with young-adult gerbils that had been treated with the high concentration of ouabain. No differences in response latencies was found, indicating that listening effort did not change with synapse loss.  

Reviewer #1 (Recommendations for the authors): 

Figure 2: The y-axis labeled as "Frequency" is potentially misleading since there are additional frequency values on the right side of the panels. It would be helpful to clarify more in the caption what these right-side frequency values represent. Additionally, the legend could be positioned more effectively for clarity.

Thank you for your suggestion. The axis label was rephrased.

Figure 7: This figure is a bit unclear, as it appears to show two sets of gerbil data at 1500 Hz, yet the difference between them is not explained.  

We added the following text to the figure legend: „The higher and lower thresholds shown for the gerbil data reflect thresholds at  fc of 1600 Hz for fundamentals f0 of 200 Hz and 400 Hz, respectively.“

Maybe a short description of fmax that is used in Figure 4 could help or at least point to supplementary for finding the definition.  

We thank the reviewer for pointing out this typo/inaccuracy. The correct terminology in line with the remainder of the manuscript is “fmaxpeak”. We corrected the caption of figure 5 (previously figure 4) and added the reference pointing to figure 11 (previously figure 9), which explains the terms.

I couldn't find information about the possible availability of data. 

The auditory-nerve recordings reported in this paper are part of a larger study of single-unit auditorynerve responses in gerbils, formally described and published by Heeringa (2024) Single-unit data for sensory neuroscience: Responses from the auditory nerve of young-adult and aging gerbils. Scientific Data 11:411, https://doi.org/10.1038/s41597-024-03259-3. As soon as the Version of Record will be submitted, the raw single-unit data can be accessed directly through the following link:  https://doi.org/10.5061/dryad.qv9s4mwn4. The data that are presented in the figures of the present manuscript and were statistically analyzed are uploaded to the Zenodo repository (https://doi.org/10.5281/zenodo.15546625).  

Reviewer #2 (Recommendations for the authors): 

L22. The term "hidden hearing loss" is used in many different ways in the literature, from being synonymous with cochlear synaptopathy, to being a description of any listening difficulties that are not accounted for by the audiogram (for which there are many other / older terms). The original usage was much more narrow than your definition here. It is not correct that Schaette and McAlpine defined HHL in the broad sense, as you imply. I suggest you avoid the term to prevent further confusion.  

We eliminated the term hidden hearing loss.

L43. SNHL is undefined.

Thank you for catching that. The term is now spelled out.

L64. "whether" -> "that"  

We corrected this issue.

L102. It would be informative to see the synapse counts (across groups) for the animals tested in the behavioral part of the study. Did these vary between groups in the same way?  

Yes, the pattern was the same for the subgroup of behaviorally tested animals. We have added this information to the revised version of the manuscript (lines 137 – 141).

L108. How many tests were considered in the Bonferroni correction? Did this cover all reported tests in the paper?  

The comparisons of synapse numbers between treatment groups were done with full Bonferroni correction, as in the other tests involving posthoc pair-wise comparisons after an ANOVA.

Figure 1 and 6 captions. Explain meaning of * and ** (criteria values).  

The information was added to the figure legends of now Figs. 1 and 7. 

L139. I don't follow the argument - the mean driven rate is not important. It is the rate at individual CFs and how that changes with frequency shift that provides the cue.

L142. I don't follow - individual driven rates might have been a cue (some going up, some down, as frequency was shifted).  

Yes, theoretically it is possible that the spectral pattern of driven rates (i.e., excitation pattern) can be specifically used for profile analysis and subsequently as a strong cue for discriminating the TFS1 stimuli. In order to shed some light on this question with regard to the actual stimuli used in this study, we added a comprehensive figure showing simulated excitation patterns (figure 8). The excitation patterns were generated with a gammatone filter bank and auditory filter bandwidths appropriate for gerbils (Kittel et al. 2002). The simulated excitation patterns allow to draw some at least semi-quantitative conclusions about the possibility of profile analysis: 1. In the 200/1600 Hz and 400/3200 Hz conditions (i.e., harmonic number of fc is 8), the difference between all inharmonic excitation patterns and the harmonic reference excitation pattern is far below the threshold for intensity discrimination (Sinnott et al. 1992). 2. In the same conditions, the statistics of the pink noise make excitation patterns differences at or beyond the filter slopes (on both high and low frequency limits) useless for frequency shift discrimination. 3. In the 400/1600 Hz condition (i.e., harmonic number of fc is 4), there is a non-negligible possibility that excitation pattern differences were a main cue for discrimination. All of these conclusions are compatible with the results of our study.

L193. Is this p-value Bonferroni corrected across the whole study? If not, the finding could well be spurious given the number of tests reported.  

Yes, it is Bonferroni corrected

L330. TFS is already defined.  

L346. AN is already defined.  

L408. "temporal fine structure" -> "TFS"  

It was a deliberate decision to define these terms again in the Discussion, for readers who prefer to skip most of the detailed Results. 

L364-366. This argument is somewhat misleading. Cochlear resolvability largely depends on the harmonic spacing (i.e., F0) relative to harmonic frequency (in other words, on harmonic rank). Marmel et al. (2015) and Moore and Sek (2009) used a center frequency (at least) 11 times F0. Here, the center frequency was only 4 or 8 times F0. In human, this would not be sufficient to eliminate excitation pattern cues.  

We have now included results from modeling the excitation patterns in the discussion with a new figure demonstrating that at a center frequency of 8 times F0, excitation patterns provide no useful cue while this is a possibility at  a center frequency of 4 times F0 (Fig. 8, lines 440 - 446).

L541. Was that a spectrum level of 20 dB SPL (level per 1-Hz wide band) at 1 kHz? Need to clarify.  

The power spectral density of the pink noise at 1 kHz (i.e., the level in a 1 Hz wide band centered at 1 kHz) was 13.3 dB SPL. The total level of the pink noise (including edge filters at 100 Hz and 11 kHz) was 50 dB SPL.

L919. So was the correction applied across only the tests within each ANOVA? Don't you need to control the study-wise error rate (across all primary tests) to avoid spurious findings?  

We added information about the family-wise error rate (line 1077 - 1078). Since the ANOVAs tested different specific research questions, we do not think that we need to control the study-wise error rate.

Reviewer #3 (Recommendations for the authors): 

There was no difference in TFS sensitivity in the AN fiber activity across all the groups. Potential deficits with age were only sound in the behavioral paradigm. Given that, it might make it clearer to specify that the deficits or lack thereof are in behavior, in multiple instances in the manuscript where it says synaptopathy showed no decline in TFS sensitivity (For example Line 342-344).  

We carefully went through the entire text and clarified a couple more instances.

L353 - this statement is a bit too strong. It implies causality when there is only a co-occurrence of increased f0 representation and age-related behavioral deficits in TFS1 task.  

The statement was rephrased as “Thus, cue representation may be associated with the perceptual deficits, but not reduced synapse numbers, as originally proposed.”

L465-467 - while this may be true, I think it is hard to say this with the current dataset where only AN fibers are being recorded from. I don't think we can say anything about afferent central mechanisms with this data set.  

We agree. However, we refer here to published data on central inhibition to provide a possible explanation. 

Hearing thresholds with ABRs are mentioned in the methods, but that data is not presented anywhere. Would be nice to see hearing thresholds across the various groups to account or discount outer hair cell dysfunction. 

This important point was made repeatedly and we thank the Reviewers for it. As indicated above, new data on threshold sensitivity and neural tuning were added in a new section of the Results which indirectly suggest that significant OHC pathologies were not a concern, neither in our young-adult, synaptopathic gerbils nor in the old gerbils.

eLife Assessment

This valuable study presents a theoretical model of how punctuated mutations influence multistep adaptation, supported by empirical evidence from some TCGA cancer cohorts. This solid model points to the case of possible punctuated evolution rather than gradual genomic change. There was some disagreement amongst the reviewers in terms of how closely the theoretical results apply to the phenomena examined empirically, and alternative explanations should be considered in the future.

Reviewer #1 (Public review):

Summary:

Grasper et al. present a combined analysis of the role of temporal mutagenesis in cancer, which includes both theoretical investigation and empirical analysis of point mutations in TCGA cancer patient cohorts. They find that temporal elevated mutation rates contribute to cancer fitness by allowing fast adaptation when the fitness drops (due to previous deleterious mutations). This may be relevant in the case of tumor suppressor genes (TSG), which follow the 2-hit hypothesis (i.e., biallelic 2 mutations are necessary to deactivate TS), and in cases where temporal mutagenesis occurs (e.g. high APOBEC, ROS). They provide evidence that this scenario is likely to occur in patients, in some cancer types. This is an interesting and potentially important result that merits the attention of the target audience. Nonetheless, I have some questions (detailed below) regarding the design of the study, the tools and parametrization of the theoretical analysis and the empirical analysis - that I think if addressed would make the paper more solid and the conclusion more substantiated.

Strengths:

Combined theoretical investigation with empirical analysis of cancer patients

Weaknesses:

Parametrization and systematic investigation of theoretical tools and their relevant to tumor evolution

Comments on revisions:

The authors have adequately addressed my suggestions. I think some of the details provided in some of the replies to my comments (specifically with regard to my points 1, 4, 6ii; minor point 6) could be integrated into relevant text in the introduction , discussion and methods, to help the readers follow better the model and its interpretation - but this is up to the authors to decide what to emphasize.

Reviewer #2 (Public review):

This work presents theoretical results concerning the effect of punctuated mutation on multistep adaptation along with empirical analysis of multistep adaptation in cancer. The empirical results are claimed to demonstrate the acceleration of multistep adaptation predicted theoretically. However, there is an important disconnect between the theoretical results and the empirical observations, such that it is not clear that punctuated mutation can produce the phenomena observed empirically. Furthermore, there are other plausible explanations for the empirical observations.

The theoretical work emphasizes the positive effect of punctuated mutation on the rate of crossing a "fitness valley", i.e., multistep adaptation where the first mutation is deleterious. The empirical work, however, focuses on inactivation of both alleles of a tumor suppressor gene (TSG), for which the first mutation--inactivation of one gene copy--is expected to be neutral or slightly advantageous, not maladaptive as suggested by the authors. Pairs of genes with putative synergystic effects were also analyzed, but there is no indication that these generally involve fitness valleys either.

This disconnect is most glaring in Figure 4, in which the simulations are supposed to confirm that punctuated mutation can produce the empirical phenomena reported for TSG inactivation. If this is the case, it should be possible to produce such results in simulations in which inactivation of just one allele is neutral. Instead, simulations assuming a substantial fitness penalty (0.05) for the first mutation are presented. Contrary to what is claimed in the text (line 212), this is not a "biologically realistic" parameter value for TSG inactivation. The insensitivity of results to the size the fitness penalty is irrelevant: a substantial fitness penalty is qualitatively different from no penalty at all.

The paper does report a small (15%) effect of punctuation on the rate of multistep adaptation in the absence of a fitness valley. This effect is much smaller than the fourfold increase in the presence of a fitness valley. The results presented--a single stochastic run for each condition--are insufficient to establish that there is any effect at all: if we assume that the number of pairs of fixations (about 150-180 in each simulation) is Poisson distributed, the 15% difference is not statistically significant.

Assuming that this effect is genuine, it is likely due to a mutation rate that is unrealisitcally high (considering that "rescue" requires inactivation of a particular gene). Theoretical considerations suggest that punctuated mutation has little or no effect in the absence of a fitness valley in the limit of low mutation rate:

(A1) The authors' theoretical results for a Galton-Watson process (SI2) imply that there is no effect without a fitness valley in that limit. This is so because there is no effect in the "supercritical" regime. Cancer cells must be supercritical (otherwise there would be no net growth), and a neutral or advantangeous mutant would remain in the supercritical regime.

(A2) Fig. S2D indicates, as far as I can tell from the colors, that punctuation makes little or no difference to the rate of adaptation in the absence of a fitness valley, i.e., for vertical axis values of 1 or more. I am not sure why the authors (line 129) point to this figure as evidence that punctuation speeds two-step adaptation when the first mutation is not maladaptive; the figure appears to say that it does not. The fraction of events due to "stochastic tunneling" of course increases with punctuation, but that does not change the fact that adaptation is no faster.

(A3) The authors' verbal argument to the contrary (line 124ff) is flawed. Despite the fact that even a mildly advantageous mutant is likely to go extinct, its expected frequency only increases with time, and that of a neutral allele remains constant over time. Thus, the average number of opportunities for a second mutation does not decrease with time since the first mutation, as it does when the first muation is deleterious.

(A4) I ran some simulations for a Wright-Fisher population, and they seem to confirm the lack of an effect in the low mutation rate limit.

Thus, it is unclear whether punctuated mutation can explain the reported phenomena or should be expected to have major effects on the rate or nature of cancer cell adaptation.

I would also note that routes to inactivation of both copies of a TSG that are not accelerated by punctuation will dilute any effects of punctuation. An example is a single somatic mutation followed by loss of heterozygosity. Such mechanisms are not included in the theoretical analysis nor assessed empirically. If, for example, 90% of double inactivations were the result of such mechanisms with a constant mutation rate, a factor of two effect of punctuated mutagenesis would increase the overall rate by only 10%. Consideration of the rate of apparent inactivation of just one TSG copy and of deletion of both copies would shed some light on the importance of this consideration.

Several factors besides the effects of punctuated mutation might explain or contribute to the empirical observations. Though these are now mentioned in the paper, I will list them here for clarity:

(B1) High APOBEC3 activity can select for inactivation of TSGs (references in Butler and Banday 2023, PMID 36978147). This could explain the empirical correlations.

(B2) Without punctuation, the rate of multistep adaptation is expected to rise more than linearly with mutation rate. Thus, if APOBEC signatures are correlated with a high mutation rate due to the action of APOBEC, this alone could explain the correlation with TSG inactivation.

(B3) The nature of mutations caused by APOBEC might explain the results. Notably, one of the two APOBEC mutation signatures, SBS13, is particularly likely to produce nonsense mutations. The authors count both nonsense and missense mutations, but nonsense mutations are more likely to inactivate the gene, and hence to be selected.

Author response:

The following is the authors’ response to the original reviews

eLife Assessment

his valuable study presents a theoretical model of how punctuated mutations influence multistep adaptation, supported by empirical evidence from some TCGA cancer cohorts. This solid model is noteworthy for cancer researchers as it points to the case for possible punctuated evolution rather than gradual genomic change. However, the parametrization and systematic evaluation of the theoretical framework in the context of tumor evolution remain incomplete, and alternative explanations for the empirical observations are still plausible.

We thank the editor and the reviewers for their thorough engagement with our work. The reviewers’ comments have drawn our attention to several important points that we have addressed in the updated version. We believe that these modifications have substantially improved our paper.

There were two major themes in the reviewers’ suggestions for improvement. The first was that we should demonstrate more concretely how the results in the theoretical/stylized modelling parts of our paper quantitatively relate to dynamics in cancer.

To this end, we have now included a comprehensive quantification of the effect sizes of our results across large and biologically-relevant parameter ranges. Specifically, following reviewer 1’s suggestion to give more prominence to the branching process, we have added two figures (Fig S3-S4) quantifying the likelihood of multi-step adaptation in a branching process for a large range of mutation rates and birth-death ratios. Formulating our results in terms of birth-death ratios also allowed us to provide better intuition regarding how our results manifest in models with constant population size vs models of growing populations. In particular, the added figure (Fig S3) highlights that the effect size of temporal clustering on the probability of successful 2-step adaptation is very sensitive to the probability that the lineage of the first mutant would go extinct if it did not acquire a second mutation. As a result, the phenomenon we describe is biologically likely to be most effective in those phases during tumor evolution in which tumor growth is constrained. This important pattern had not been described sufficiently clearly in the initial version of our manuscript, and we thank both reviewers for their suggestions to make these improvements.

The second major theme in the reviewers’ suggestions was focused on how we relate our theoretical findings to readouts in genomic data, with both reviewers pointing to potential alternative explanations for the empirical patterns we describe.

We have now extended our empirical analyses following some of the reviewers’ suggestions. Specifically, we have included analyses investigating how the contribution of reactive oxygen species (ROS)-related mutation signatures correlates with our proxies for multi-step adaptation; and we have included robustness checks in which we use Spearman instead of Pearson correlations. Moreover, we have included more discussion on potential confounds and the assumptions going into our empirical analyses as well as the challenges in empirically identifying the phenomena we describe.

Below, we respond in detail to the individual comments made by each reviewer.

Public Reviews:

Reviewer #1 (Public review):

Summary:

Grasper et al. present a combined analysis of the role of temporal mutagenesis in cancer, which includes both theoretical investigation and empirical analysis of point mutations in TCGA cancer patient cohorts. They find that temporally elevated mutation rates contribute to cancer fitness by allowing fast adaptation when the fitness drops (due to previous deleterious mutations). This may be relevant in the case of tumor suppressor genes (TSG), which follow the 2-hit hypothesis (i.e., biallelic 2 mutations are necessary to deactivate TS), and in cases where temporal mutagenesis occurs (e.g., high APOBEC, ROS). They provide evidence that this scenario is likely to occur in patients with some cancer types. This is an interesting and potentially important result that merits the attention of the target audience. Nonetheless, I have some questions (detailed below) regarding the design of the study, the tools and parametrization of the theoretical analysis, and the empirical analysis, which I think, if addressed, would make the paper more solid and the conclusion more substantiated.

Strengths:

Combined theoretical investigation with empirical analysis of cancer patients.

Weaknesses:

Parametrization and systematic investigation of theoretical tools and their relevance to tumor evolution.

We sincerely thank Reviewer 1 for their comments. As communicated in more detail in the point-by-point replies to the “Recommendations for the authors”, we have revised the paper to address these comments in various ways. To summarize, Reviewer 1 asked for (1) more comprehensive analyses of the parameter space, especially in ranges of small fitness effects and low mutation rates; (2) additional clarifications on details of mechanisms described in the manuscript; and (3) suggested further robustness checks to our empirical analyses. We have addressed these points as follows: we have added detailed analyses of dynamics and effect sizes for branching processes (see Sections SI2 and SI3 in the Supplementary Information, as well as Figures S3 and S4). As suggested, these additions provide characterizations of effect sizes in biologically relevant parameter ranges (low mutation rates and smaller fitness effect sizes), and extend our descriptions to processes with dynamically changing population sizes. Moreover, we have added further clarifications at suggested points in the manuscript, e.g. to elaborate on the non-monotonicities in Fig 3. Lastly, we have undertaken robustness checks using Spearman rather than Pearson correlation coefficients to quantify relations between TSG deactivation and APOBEC signature contribution, and have performed analyses investigating dynamics of reactive oxygen species-associated mutagenesis instead of APOBEC.

Reviewer #2 (Public review):

This work presents theoretical results concerning the effect of punctuated mutation on multistep adaptation and empirical evidence for that effect in cancer. The empirical results seem to agree with the theoretical predictions. However, it is not clear how strong the effect should be on theoretical grounds, and there are other plausible explanations for the empirical observations.

Thank you very much for these comments. We have now substantially expanded our investigations of the parameter space as outlined in the response to the “eLife Assessment” above and in the detailed comments below (A(1)-A(3)) to convey more quantitative intuition for the magnitude of the effects we describe for different phases of tumor evolution. We agree that there could be potential additional confounders to our empirical investigations besides the challenges regarding quantification that we already described in our initial version of the manuscript. We have thus included further discussion of these in our manuscript (see replies to B(1)-B(3)), and we have expanded our empirical analyses as outlined in the response to the “eLife Assessment”.

For various reasons, the effect of punctuated mutation may be weaker than suggested by the theoretical and empirical analyses:

(A1) The effect of punctuated mutation is much stronger when the first mutation of a two-step adaptation is deleterious (Figure 2). For double inactivation of a TSG, the first mutation--inactivation of one copy--would be expected to be neutral or slightly advantageous. The simulations depicted in Figure 4, which are supposed to demonstrate the expected effect for TSGs, assume that the first mutation is quite deleterious. This assumption seems inappropriate for TSGs, and perhaps the other synergistic pairs considered, and exaggerates the expected effects.

Thank you for highlighting this discrepancy between Figure 2 and Figure 4. For computational efficiency and for illustration purposes, we had opted for high mutation rates and large fitness effects in Figure 2; however, our results are valid even in the setting of lower mutation rates and fitness effects. To improve the connection to Figure 4, and to address other related comments regarding parameter dependencies, we have now added more detailed quantification of the effects we describe (Figures SF3 and SF4) to the revised manuscript. These additions show that the effects illustrated in Figure 2 retain large effect sizes when going to much lower mutation rates and much smaller fitness effects. Indeed, while under high mutation rates we only see the large relative effects if the first mutation is highly deleterious, these large effects become more universal when going to low mutation rates.

In general, it is correct that the selective disadvantage (or advantage) conveyed by the first mutation affects the likelihood of successful 2-step adaptations. It is also correct that the magnitude of the ‘relative effect’ of temporal clustering on valley-crossing is highest if the lineage with only the first of the two mutations is vanishingly unlikely to produce a second mutant before going extinct. If the first mutation is strongly deleterious, the lineage of such a first mutant is likely to quickly go extinct – and therefore also more likely to do so before producing a second mutant.

However, this likelihood of producing the second mutant is also low if the mutation rate is low. As our added figure (Figure SF3) illustrates, at low mutation rates appropriate for cancer cells, is insensitive to the magnitude of the fitness disadvantage for large parts of the parameter space. Especially in populations of constant size (approximated by a birth/death ratio of 1), the relative effects for first mutations that reduce the birth rate by 0.5 or by 0.05 are indistinguishable (Figure SF3f).

Moreover, the absolute effect , as we discuss in the paper (Figures SF2 and SF3) is largest in regions of the parameter space in which the first mutant is not infinitesimally unlikely to produce a second mutant (and 𝑓_𝑘 and 𝑓₁ would be infinitesimally small), but rather in parameter regions in which this first mutant has a non-negligible chance to produce a second mutant. The absolute effect therefore peaks around fitness-neutral first mutations. While the next comment (below) says that our empirical investigations more closely resemble comparisons of relative effects and not absolute effects, we would expect that the observations in our data come preferentially from multi-step adaptations with large absolute effect since the absolute effect is maximal when both 𝑓_𝑘 and 𝑓₁are relatively high.

In summary, we believe Figure 2, while having exaggerated parameters for very defendable reasons, is not a misleading illustration of the general phenomenon or of its applicability in biological settings, as effect sizes remain large when moving to biologically realistic parameter ranges. To clarify this issue, we have largely rewritten the relevant paragraphs in the results section and have added two additional figures (Figures SF3 and SF4) as well as a section in the SI with detailed discussion (SI2).

(A2) More generally, parameter values affect the magnitude of the effect. The authors note, for example, that the relative effect decreases with mutation rate. They suggest that the absolute effect, which increases, is more important, but the relative effect seems more relevant and is what is assessed empirically.

Thank you for this comment. As noted in the replies to the above comments, we have now included extensive investigations of how sensitive effect sizes are to different parameter choices. We also apologize for insufficiently clearly communicating how the quantities in Figure 4 relate to the findings of our theoretical models.

The challenge in relating our results to single-timepoint sequencing data is that we only observe the mutations that a tumor has acquired, but we do not directly observe the mutation rate histories that brought about these mutations. As an alternative readout, we therefore consider (through rough proxies: TSGs and APOBEC signatures) the amount of 2-step adaptations per acquired/retained mutation. While we unfortunately cannot control for the average mutation rate in a sample, we motivate using this “TSG-deactivation score” by the hypothesis that for any given mutation rate, we expect a positive relationship between the amount of temporal clustering and the amount of 2-step adaptations per acquired/retained mutation. This hypothesis follows directly from our theoretical model where it formally translates to the statement that for a fixed , is increasing in .

However, while both quantities 𝑓_𝑘/𝑓₁  or from our theoretical model relate to this hypothesis – both are increasing in 𝑘–, neither of them maps directly onto the formulation of our empirical hypothesis.

We have now rewritten the relevant passages of the manuscript to more clearly convey our motivation for constructing our TSG deactivation score in this form (P. 4-6).

(A3) Routes to inactivation of both copies of a TSG that are not accelerated by punctuation will dilute any effects of punctuation. An example is a single somatic mutation followed by loss of heterozygosity. Such mechanisms are not included in the theoretical analysis nor assessed empirically. If, for example, 90% of double inactivations were the result of such mechanisms with a constant mutation rate, a factor of two effect of punctuated mutagenesis would increase the overall rate by only 10%. Consideration of the rate of apparent inactivation of just one TSG copy and of deletion of both copies would shed some light on the importance of this consideration.

This is a very good point, thank you. In our empirical analyses, the main motivation was to investigate whether we would observe patterns that are qualitatively consistent with our theoretical predictions, i.e. whether we would find positive associations between valley-crossing and temporal clustering. Our aim in the empirical analyses was not to provide a quantitative estimate of how strongly temporally clustered mutation processes affect mutation accumulation in human cancers. We hence restricted attention to only one mutation process which is well characterized to be temporally clustered (APOBEC mutagenesis) and to only one category of (epi)genomic changes (SNPs, in which APOBEC signatures are well characterized). Of course, such an analysis ignores that other mutation processes (e.g. LOH, copy number changes, methylation in promoter regions, etc.) may interact with the mechanisms that we consider in deactivating Tumor suppressor genes.

We have now updated the text to include further discussion of this limitation and further elaboration to convey that our empirical analyses are not intended as a complete quantification of the effect of temporal clustering on mutagenesis in-vivo (P. 10,11).

Several factors besides the effects of punctuated mutation might explain or contribute to the empirical observations:

(B1) High APOBEC3 activity can select for inactivation of TSGs (references in Butler and Banday 2023, PMID 36978147). This selective force is another plausible explanation for the empirical observations.

Thank you for making this point. We agree that increased APOBEC3 activity, or any other similar perturbation, can change the fitness effect that any further changes/perturbations to the cell would bring about. Our empirical analyses therefore rely on the assumption that there are no major confounding structural differences in selection pressures between tumors with different levels of APOBEC signature contributions. We have expanded our discussion section to elaborate on this potential limitation (P. 10-11).

While the hypothesis that APOBEC3 activity selects for inactivation of TSGSs has been suggested, there remain other explanations. Either way, the ways in which selective pressures have been suggested to change would not interfere relevantly with the effects we describe. The paper cited in the comment argues that “high APOBEC3 activity may generate a selective pressure favoring” TSG mutations as “APOBEC creates a high [mutation] burden, so cells with impaired DNA damage response (DDR) due to tumor suppressor mutations are more likely to avert apoptosis and continue proliferating”. To motivate this reasoning, in the same passage, the authors cite a high prevalence of TP53 mutations across several cancer types with “high burden of APOBEC3-induced mutations”, but also note that “this trend could arise from higher APOBEC3 expression in p53-mutated tumors since p53 may suppress APOBEC3B transcription via p21 and DREAM proteins”.

Translated to our theoretical framework, this reasoning builds on the idea that APOBEC3 activity increases the selective advantage of mutants with inactivation of both copies of a TSG. In contrast, the mechanism we describe acts by altering the chances of mutants with only one TSG allele inactivated to inactivate the second allele before going extinct. If homozygous inactivation of TSGs generally conveys relatively strong fitness advantages, lineages with homozygous inactivation would already be unlikely to go extinct. Further increasing the fitness advantage of such lineages would thus manifest mostly in a quicker spread of these lineages, rather than in changes in the chance that these lineages survive. In turn, such a change would have limited effect on the “rate” at which such 2-step adaptations occur, but would mostly affect the speed at which they fixate. It would be interesting to investigate these effects empirically by quantifying the speed of proliferation and chance of going extinct for lineages that newly acquired inactivating mutations in TSGs.

Beyond this explicit mention of selection pressures, the cited paper also discusses high occurrences of mutations in TSGs in relation to APOBEC. These enrichments, however, are not uniquely explained by an APOBEC-driven change in selection pressures. Indeed, our analyses would also predict such enrichments.

(B2) Without punctuation, the rate of multistep adaptation is expected to rise more than linearly with mutation rate. Thus, if APOBEC signatures are correlated with a high mutation rate due to the action of APOBEC, this alone could explain the correlation with TSG inactivation.

Thank you for making this point. Indeed, an identifying assumption that we make is that average mutation rates are balanced between samples with a higher vs lower APOBEC signature contribution. We cannot cleanly test this assumption, as we only observe aggregate mutation counts but not mutation rates. However, the fact that we observe an enrichment for APOBEC-associated mutations among the set of TSG-inactivating mutations (see Figure 4F) would be consistent with APOBEC-mutations driving the correlations in Fig 4D, rather than just average mutation rates. We have now added a paragraph to our manuscript to discuss these points (P. 10-11).

(B3) The nature of mutations caused by APOBEC might explain the results. Notably, one of the two APOBEC mutation signatures, SBS13, is particularly likely to produce nonsense mutations. The authors count both nonsense and missense mutations, but nonsense mutations are more likely to inactivate the gene, and hence to be selected.

Thank you for making this point.  We have included it in our discussion of potential confounders/limitations in the revised manuscript (P. 10-11).  

Recommendations for the authors:

Reviewer #1 (Recommendations for the authors):

Specific questions/comments/suggestions:

(1) For the theoretical investigation, the authors use the Wright-Fisher model with specific parameters for the decrease/increase in the fitness (0.5,1.5). This model is not so relevant to cancer, because it assumes a constant population size, while in cancer, the population is dynamic (increasing, if the tumor grows). Although I see they mention relevance to the branching process (in SI), I think the branching process should be bold in the main text and the Wright-Fisher in SI (or even dropped).

Thank you for this comment. We agree that too little attention had been given to the branching process in the original version of our manuscript. While the Wright-Fisher process is computationally efficient to simulate and thus lends itself to clean simulations for illustrative examples, it did lead us to put undue emphasis on populations of constant size.

The added Figures SF2 and SF3 now focus on branching processes, and we have substantially expanded our discussion of how dynamics differ as a function of the population-size trajectory (constant vs growing; SI2, P. 4,9,10). Generally, we do believe that it is appropriate to consider both regimes. If tumors evolve from being confined within their site of origin to progressively invading adjacent tissues and organ compartments, they traverse different regions of the birth-death ratio parameter space. Moreover, the timing of transitions between phases of more or less constrained growth is likely closely tied to adaptation dynamics, since breaching barriers to expansion requires adapting to novel environments and selection pressures.

We hope that the revised version of the manuscript conveys these points more clearly, and thank you for alerting us to this imbalance in the original version of our manuscript.

(2) The parameters 0.5 (decrease in fitness) and 1.5 (increase in fitness) seem exaggerated (the typical values for the selective advantage are usually much lower (by an order of magnitude). The same goes for the mutation rate. The authors chose values of the order 0.001, while in cancer (and generally) it is much lower than that (10-5 - 10-6). I think that generally, the authors should present a more systematic analysis of the sensitivity of the results to these parameters.

Thank you very much for this very important comment. We have made this a major focus in our revisions (see our reply to the editor’s comments). As suggested, we have now added further analyses to explore more biologically relevant parameter regimes. Reviewer 2 has made a similar remark, and to avoid redundancies, we point for a more detailed response to our response to that comment (A1).

(3) In Figure 3, the authors explore the sensitivity to mu (mutation rate) and k (temporal clustering) and find a non-monotonic behavior (Figure 3C). However, this behavior is not well explained. I think some more explanations are required here.

Thank you for pointing this out. We had initially relegated the more detailed explanations to the SI2 (which in the revised manuscript became SI4), but are happy to provide more elaboration in the main text, and have done so now (P. 5).

For , the non-monotonicity reflects the exploration-exploitation tradeoff that this section is dedicated to very small  values (little exploration) prevent the population from finding fitness peaks. In contrast, once a fitness peak is reached, excessively large  values (little exploitation) scatter the population away from this peak to points of lower fitness.

For , the most relevant dynamic is that at high , the population becomes unable to find close-by fitness improvements (1-step adaptations) if it is not in a burst. As 𝑘 increases, this delay in adaptation (until a burst occurs) eventually comes to outweigh the benefits of high 𝑘 (better ability to undergo multi-step adaptations). Additionally, if 𝑘 ∙ μ becomes very large, clonal interference eventually leads to diminishing exploration-returns when 𝑘 is increased further (Fig 5C), as the per-cell likelihood of finding a specific fitness peak eventually saturates and increasing  only causes multiple cells to find the same peak, rather than one cell finding this peak and its lineage fixating in the population.

(4) In Figure 5, where the authors show the accumulation of the first (red; deleterious mutation) and second (blue; advantageous mutation), it seems that the fraction of deleterious mutations is much lower than that of advantageous mutations. This is opposite to the case of cancer, where most of the mutations are 'passengers', (slightly) deleterious or neutral mutations. Can the author explain this discrepancy and generally the relation of their parametrization to deleterious vs. advantageous mutations?

Thank you for this comment. In general, we have focused attention in our paper on sequences of mutations that bring about a fitness increase. We call those sequences ‘adaptations’ and categorize these as one-step or multi-step, depending on whether or not they contain intermediates states with a fitness disadvantage.

In our modelling, we do not consider mutations that are simply deleterious and are not a necessary part of a multi-step adaptation sequence. The motivation for this abstraction is, firstly, to focus on adaptation dynamics, and secondly, that in certain limits (small mu and large constant population sizes), lineages with only deleterious mutations have a probability close to one of going extinct, so that any emerging deleterious mutant would likely be 'washed out’ of the population before a new mutation emerges.

However, whether the dynamics of how neutral or deleterious passenger mutations are acquired also vary relevantly with the extent of temporal clustering is a valid and interesting question that would warrant its own study. The types of theoretical arguments for such an investigation would be very similar to the ones we use in our paper.

(5) The theoretical investigation assumes a multi/2-step adaptation scenario where the first mutation is deleterious and the second is advantageous. I think this should be generalized and further explored. For example, what happens when there are multiple mutations that are slightly deleterious (as probably is the case in cancer) and only much later mutations confer a selective advantage? How stable is the "valley crossing" if more deleterious mutations occur after the 2 steps?

This is also an important point and relates in part to the previous comment (4).  For discussion of interactions with deleterious mutations, please see the reply to comment (4).  

Regarding generalizations of this valley-crossing scenario, note that any sequence of mutations that increases fitness can be decomposed into sequences of either one-step or multi-step adaptations, as defined  in the paper. Therefore, if all intermediate states before the final selectively advantageous state have a selective disadvantage making the lineages of such cells likely to go extinct, then our derivations in S1 apply, and the relative effect of temporal clustering becomes where n is the number of intermediate states. If, conversely, any of the intermediate states already had a selective advantage, then our model would consider the subsequence until this first mutation with a selective advantage as its individual (one-step or multi-step) “adaptation”.

The second question, “How stable is the "valley crossing" if more deleterious mutations occur after the 2 steps?”, touches on a different property of the population dynamics, namely on how the fate of a mutant lineage depends on how this lineage emerged. In our paper, we compare different levels of temporal clustering for a fixed average mutation rate. This choice implies that, if we assume that the mutant that emerges from a valley-crossing does not go extinct, then the number of deleterious mutations expected to occur in this lineage, once emerged, will not depend on the extent of temporal clustering. However, if in-burst mutation rates increased the expected burden of early acquired deleterious mutations sufficiently much to affect the probability that the lineage with a multi-step adaptation goes extinct before the burst ends, then there may indeed be an interaction between effects of deleterious passengers and temporal clustering. We would, however, expect effects on this probability of early extinction to be relatively minor, since such a lineage with a selective advantage would quickly grow to large cell-numbers implying that it would require a large number of co-occurring and sufficiently deleterious mutations across these cells for the lineage to go extinct.

(6) For the empirical analysis of TCGA cohorts, the authors focus on the contribution of APOBEC mutations (via signature analysis) to temporal mutagenesis. They find only a few cancer types (Figure 4D) that follow their prediction (in Figure 4C) of a correlation between TSG deactivation and temporal mutations in bursts. I think two main points should be addressed:

Thank you for this comment. We will respond in detail to the corresponding points below, but would like to note here that while we find this correlation “in only a few cancer types”, we also show that only few cancer types have relevant proportions of mutations caused by APOBEC, and it is precisely in these cancer types that we find a correlation.  We have clarified this aspect in the revised version of the manuscript (P.7).

(i) APOBEC is not the only cause for temporal mutagenesis. For example, elevated ROS and hypoxia are also potential contributors - it might therefore be important to extend the signature analysis (to include more possible sources for temporal mutagenesis). Potentially, such an extension may show that more cancer types follow the author's prediction.

Thank you for this interesting suggestion. We have now included analogous analyses for contributions of signature SBS18 which is associated with ROS mutagenesis, and for the joint contribution of signatures SBS17a, SBS17b, SBS18 and SBS36, which all have been shown (some in a more context-dependent manner) to be associated with ROS mutagenesis. When doing so, we do not find a clear trend. However, we also do not find these signatures to account for substantial proportions of the acquired mutations, meaning that ROS mutagenesis likely also does not account for much of the variation in how temporally clustered the mutation rate trajectories of different tumors are. We have incorporated these results and their discussion in the manuscript (SI5 and Fig S8).

(ii) The TSG deactivation score used by the authors only counts the number of mutations and does not consider if the 2 mutations are biallelic, which is highly important in this case. There are ways to investigate the specific allele of mutations in TCGA data (for example, see Ciani et al. Cell Sys 2022 PMID: 34731645). Given the focus on TSG of this study, I think it is important to account for this in the analysis.

Thank you for making this point. We did initially consider inferring allele-specific mutation status, but decided against it as this would have shrunk our dataset substantially, thus potentially introducing unwanted biases. Determining whether two mutations lie on the same or on different alleles requires either (1) observing sequencing reads that either cover the loci of both mutations, or (2) tracing whether (sets of) other SNPs on the same gene co-occur exclusively with one of the two considered mutations. These requirements lead to a substantial filtering of the observed mutations. Moreover, this filtering would be especially strong for tumors with a small overall mutation burden, as these would have fewer co-occurring SNPs to leverage in this inference. We would have hence preferentially filtered out TSG-deactivating mutations in tumors with low mutation burden. We have modified the text to address this point (P.14).

(7) To continue point 4. I wonder why some known cancer types with high APOBEC signatures (e.g., lung, mentioned in the introduction) do not appear in the results of Figure 4. Can the author explain why it is missed?

We do provide complete results for all categories in Supplementary Figure 3. To not overwhelm the figure in the main text, we only show the four categories with the highest average APOBEC signature contribution, beyond those four, average APOBEC signature contributions quickly drop. Lung-related categories do not feature in these top four (Lung squamous cell carcinoma are fifth and Lung adenocarcinoma are eighth in this ordering).

Minors:

(1) It is worth mentioning the relevance to resistance to treatment (see https://www.nature.com/articles/s41588-025-02187-1).

Thank you for this suggestion. We have included a mention of the relation to this paper in the discussion section (P. 11).

(2) Some of the figures' resolution should be improved - specifically, Figures 4, S1, and S5, which are not clear/readable.

Thank you for pointing this out. This was the result of conversion to a word document. We will provide tif files in the revisions to have better resolution.

(3) Regarding Figure 3e,f. How come that moving from K=1 to K=I doesn't show any changes in fitness - it looks as if in both cases the value fluctuates around comparable mean fitness? Is that the case?

While fitness differences between simulations with different k manifest robustly over long time-horizons (see Fig 3C with results over  generations), there are various sources of substantial stochasticity that make the fitness values in these short-term plots (Fig3D-F) imperfect illustrations of how long-term average fitness behaves. For instance, fitness landscapes are drawn randomly which introduces variability in how high and how close-by different fitness peaks are. Similarly, there is substantial randomness since both the type (direction on the 2-D fitness landscape) and the timing of mutation are stochastic.

The short-term plots in Fig3D-F are intended to showcase representative dynamics of transitions between points on the genotype space with different fitness values following a redrawing of the landscape – but not necessarily to provide a comparison between the height of the attained (local) fitness-maxima.  

(4) Figures 4c,d - correlation should be Spearman, not Pearson (it's not a linear relationship).

Thank you for this comment. As a robustness check, we have generated the same figures using Spearman and not Pearson correlations and find results that are qualitatively consistent with the initially shown results. Indeed, using Spearman correlations, all four cancer types from Fig 4D have significant correlations.

(5) Typo for E) "...in samples of the cancer types in (C) were caused by APOBEC" - it should be D (not C) I guess.

Thank you for catching this. We fixed the typo.

(6) Figure 5 - the mutation rate is too high (0.001), sensitivity to that? Also the fitness change is exaggerated (0.5, 1.5), and the division of mutations to 100 and 100 (200 in total) loci is not clear.

Thank you for making this point. In this simulation setting it is unfortunately computationally prohibitively expensive to perform simulations at biologically realistic mutation rates. Therefore, we have scaled up the mutation rate while scaling down the population size. Moreover, the choice of model here is not meant to resemble a biologically realistic dynamic, but rather to create a stylized setting to be able to consider the interplay between clonal interference and facilitated valley-crossing in isolation. The key result from this figure is the separation of time scales at which low or high temporal clustering maximizes adaptability.

However, known parameter dependencies in these models allow us to reason about how tuning individual parameters of this stylized model would affect the relative importance of effects of clonal interference. This relative importance is largest when mutants are likely to co-occur on different competing clones in a population. The likelihood of such co-occurrences decreases substantially if decreasing the mutation rate to biologically realistic values. However, this likelihood also sensitively depends on the time that it takes a clone with a one-step adaptation to spread through the population. Smaller fitness advantages, as well as larger population sizes, slow down this process of taking over the population, which increases the likelihood of clonal interference. We now discuss these points in our revised manuscript (P. 8).

7) In the results text (last section) "Performing simulations for 2-step adaptations, we found that fixation rates are non-monotone in k. While at low k increasing k leads to a steep increase in the fixation rate, this trend eventually levels off and becomes negative, with further increases in k leading to a decrease in the fixation rate". Where are the results of this? It should be bold and apparent.

Thank you for alerting us that this is unclear. The relevant figure reference is indeed Fig 5C as in the preceding passage in the manuscript. However, we noticed that due to the presence of the steadily decreasing black line for 1-step adaptations, it is not easy to see that also the blue line is downward sloping. We have added a further reference to Fig 5C, and have adapted the grid spacing in the background of that figure-panel to make this trend more easily visible.

(8) Although not inconceivable, conclusions regarding resistance in the discussion are overstated. If you want to make this statement, you need to show that in resistant tumors, the temporal mutagenesis is responsible for progression vs. non-resistant/sensitive cases (is that the case), otherwise this should be toned down.

Thank you for pointing this out. We have tempered these conclusions in the revised version of the manuscript (P. 11).

Reviewer #2 (Recommendations for the authors):

(1) It might be useful to look specifically at X-linked TSGs. On the authors' interpretation, their relative inactivation rates should not be correlated with APOBEC signatures in males (but should be in females), though the size of the dataset may preclude any definite conclusions.

Thank you for this suggestion. Indeed, the size of the dataset unfortunately makes such analyses infeasible. Moreover, it is not clear whether X-linked TSGs might have structurally different fitness dynamics than TSGs on other chromosomes. However, this is an interesting suggestion worth following up on as more synergistic pairs confined to the X-chromosome are getting identified.

(2) Might there be value in distinguishing tumors that carry mutations expected to increase APOBEC expression from those that do not? Among several reasons, an APOBEC signature due to such a mutation and an APOBEC signature due to abortive viral infection may differ with respect to the degree of punctuation.

This is also an interesting suggestion for future investigations, but for which we unfortunately do not have sufficient information to build a meaningful analysis. In particular, it is unclear to what extent the degree and manifestation of episodicity/punctuation varies between these different mechanisms. Burst duration and intensity, as well as out-of-burst baseline rates of APOBEC mutagenesis likely differ in ways that are yet insufficiently characterized, which would make any result of analyses like these in Fig 4 hard to interpret.

(3) Also, in that paragraph, is "proportional to" used loosely to mean "an increasing function of"?

Thank you for this comment. We are not quite sure which paragraph is meant, but we use the term “proportional” in a literal sense at every point it is mentioned in the paper.

For the occurrences of the term on pages 3, 10 and 11, the word is used in reference to probabilities of reproduction (division in the branching process, or ‘being drawn to populate a spot in the next generation’ in the WF process) being “proportional” to fitness. These probabilities are constructed by dividing each individual cell’s fitness by the total fitness summed across all cells in the population. As the population acquires fitness-enhancing mutations, the resulting proportionality constant (1/total_fitness) changes, so that the mapping from ‘fitness’ to probability of reproduction in the next reproduction event changes over time. Nevertheless, this mapping always remains fitness-proportional.

On page 4, the term is used as follows: “the absolute rates 𝑓_𝑘 and 𝑓₁ are proportional to µ^n+1”. Here, proportionality in the literal sense follows from the equations on page 20, when setting , so that the second factor becomes µⁿ⁺¹.  We have included a clarifying sentence to address this in the derivations (SI1).

(4) It could be mentioned in the main text that the time between bursts (d) must not be too short in order for the effect to be substantial. I would think that the relevant timescale depends on how deleterious the initial mutation is.

Thank you for making this interesting and very relevant point. We have included a section (SI3) and Figure (Fig S4) in the supplement to investigate the dependence on d. In short, we find that effects are weaker for small inter-burst intervals. The sensitivity to the burst size is highest for inter-burst intervals that are sufficiently small so that the lineage of the first mutant has relevant probability of surviving long enough to experience multiple burst phases.

(5) Why not report that relative rate for Figure 2E as for 2D, as the former would seem to be more relevant to TSGs? And why was it assumed that the first inactivation is deleterious in the simulations in Figure 4 if the goal is to model TSGs?

Thank you for noting this. For how we revised the paper to better connect Figures 2 and 4, please see our comment (A1) above. In general, neither 2E nor 2D should serve as quantitative predictions for what effect size we should expect in real world data, but are rather curated illustrations of the general phenomenon that we describe: we chose high mutation rates and exaggerated fitness effects so that dynamics become visually tractable in small simulation examples.

For figure 4, assuming that the first inactivation is deleterious achieves that the branching process for the mutant lineage becomes subcritical, which keeps the simulation example simple and illustrative. For more comprehensive motivation of the approach in 4D, and especially the discussion of how fitness effects of different magnitudes may or may not be subject to the effects we describe depending on whether the population is in a phase of constant or growing population size, we refer the reader to our added section SI2, and the added discussion on pages 6 and 10.

(6) Figure 2, D and E. I'm not sure why heatmaps with height one were provided rather than simple plots over time. It is difficult, for example, to determine from a heatmap whether the increase is linear or the relative rates with and without punctuation.

Thank you for this comment. These are not heatmaps with height one, but rather for every column of pixels, different segments of that column correspond to different clones within that population. This approach is intended to convey the difference in dynamics between the results in Fig 2 and the analogous results for a branching process in Fig S1. In Fig 2, valley-crossings happen sequentially, with subsequent fixations of adapted mutants. In Fig S1, with a growing population size, multiple clones with different numbers of adaptations coexist. We have now adapted the caption of Fig 2 to clarify this point.

(7) Page 3: "High mutation rates are known to limit the rate of 1-step adaptations due to clonal interference." This is a bit misleading, as it makes it sound like increasing the mutation rate decreases the rate of one-step adaptations.

Thank you for alerting us to this poor phrasing. We have changed it in the revised version of the manuscript (P. 3).

(8) Page 4: "proportional to \mu^{n+1}" Is "proportional" being used loosely for "an increasing function of"?

It is meant in the literal mathematical sense (see response to comment (3))

(9) Page 5, near bottom: "at least two mutations across the population". In the same genome?

We counted mutations irrespective of whether they emerged in the same genome, to remain analogous to the TCGA analyses for which we also do not have single cell-resolved information.

(10) Page 6: "missense or nonsense mutation". What about indels? If these are not affected by APOBEC, omitting them will exaggerate the effect of punctuation.

Thank you for pointing out that this focus on single nucleotide substitutions conveys an exaggerated image of the importance of this effect of APOBEC-driven mutagenesis. There are of course several other classes of (epi)genomic alterations (e.g. chromatin modifications, methylation changes, copy number changes) that we do not consider in this part of our analysis. APOBEC mutagenesis serves as an example of a temporally clustered mutation process, which we investigate in its domain of action.

We have added further discussion (P. 10-11) to convey that our empirical results merely constitute an investigation of whether empirical patterns are consistent with our hypothesis, but that the narrow focus on only SNVs, only TSGs, and only APOBEC mutagenesis does not allow for a general quantitative statement about the in-vivo relevance of the phenomena we describe.

(11) Page 6: "normalized by the total number of single nucleotide substitutions." It is difficult to know how to normalize correctly, but I might think that the square of the number of substitutions would be more appropriate. Perhaps the total numbers are close enough that it matters little.

Thank you for noting this. In the revised manuscript we have now expanded this passage in the text to more clearly convey our motivations for why we normalize by the total number of single nucleotide substitutions. While the likelihood for crossing a fitness valley with 2 mutations is indeed proportional to the square of the mutation rate, we do not directly observe mutation rates from our data.  Rather, we observe the number of acquired single nucleotide substitutions for every tumor sample, but since tumors in our data differ in the time since initiation and therefore differ in the numbers of divisions their cells have undergone before being sequenced, we cannot directly infer mutation rates. One way to phrase our main result about valley-crossing is that temporally clustered mutation processes have an increased rate of successful valley-crossings per attempted valley crossing. Our TSG deactivation score is constructed to reflect this idea. The number of TSGs serves as a proxy for successful valley-crossings and the total mutation burden serves as a proxy for attempted valley-crossings.

To convey these points more clearly, we have rewritten the first paragraph in the Section “Proxies for valley crossing and for temporal clustering found in patient data” (P.6)

(12) Perhaps embed links to the COSMIC web pages for SBS2 and SBS13 in the text.

Thank you for this suggestion. We have embedded the links at the first mention of SBS2 and SBS13 in the text.

eLife Assessment

This important study by Jeong and Choi studied neural activity in the medial prefrontal cortex (mPFC) while rats performed a foraging paradigm in which rats forage for rewards in the absence or presence of a threatening object (Lobsterbot). The authors present interesting observations suggesting that the mPFC population activity switches between distinct functional modes conveying distinct task variables- such as the distance to the reward location and types of threat-avoidance behaviors-depending on the location of the animal. The reviewers thought that the results are overall convincing, appreciated the value of studying neural coding in naturalistic settings, and felt that this work offers significant insights into how the mPFC operates during foraging behavior involving reward-threat conflict.

Reviewer #1 (Public review):

Summary:

In this study, Jeong and Choi examine neural correlates of behavior during a naturalistic foraging task in which rats must dynamically balance resource acquisition (foraging) with the risk of threat. Rats first learn to forage for sucrose reward from a spout, and when a threat is introduced (an attack-like movement from a "LobsterBot"), they adjust their behavior to continue foraging while balancing exposure to the threat, adopting anticipatory withdraw behaviors to avoid encounter with the LobsterBot. Using electrode recordings targeting the medial prefrontal cortex (mPFC), they identify heterogenous encoding of task variables across prelimbic and infralimbic cortex neurons, including correlates of distance to the reward/threat zone and correlates of both anticipatory and reactionary avoidance behavior. Based on analysis of population responses, they show that prefrontal cortex switches between different regimes of population activity to process spatial information or behavioral responses to threat in a context-dependent manner. Characterization of the heterogenous coding scheme by which frontal cortex represents information in different goal states is an important contribution to our understanding of brain mechanisms underlying flexible behavior in ecological settings.

Strengths:

As many behavioral neuroscience studies employ highly controlled task designs, relatively less is generally known about how the brain organizes navigation and behavioral selection in naturalistic settings, where environment states and goals are more fluid. Here, the authors take advantage of a natural challenge faced by many animals - how to forage for resources in an unpredictable environment - to investigate neural correlates of behavior when goal states are dynamic. They investigate how prefrontal cortex (mPFC) activity is structured to support different functional "modes" (here, between a navigational mode and a threat-sensitive foraging mode) for flexible behavior. Overall, an important strength and real value of this study is the design of the behavioral experiment, which is trial-structured, permitting strong statistical methods for neural data analysis, yet still rich enough for unconstrained, natural behavior structured by the animal's volitional goals. The experiment is also phased to measure behavioral changes as animals first encounter a threat, and then learn to adapt their foraging strategy to its presence. Characterization of this adaptation process is itself quite interesting and sets a foundation for further study of threat learning and risk management in the foraging context. Finally, the characterization of single-neuron and population dynamics in mPFC in this naturalistic setting with fluid goal states is an important contribution to the field. Previous studies have identified neural correlates of spatial and behavioral variables in frontal cortex, but how these representations are structured, or how they are dynamically adjusted when animals shift their goals, has been less clear. The authors synthesize their main conclusions into a conceptual model for how mPFC could encode task variables in a context-dependent manner, and provide a useful framework for thinking about circuit-level mechanisms that may support mode switching.

Weaknesses:

The task design in this study is intentionally stimulus-rich and places minimal constraint on the animal to preserve naturalistic behavior, and this introduces some confounds that place some limits on the interpretability of neural responses. For example, some variables which are the target of neural correlation analysis, such as spatial/proximity coding and coding of threat and threat-related behaviors, are naturally entwined. In their revisions, the authors have included extensive analyses and control conditions to disambiguate these confounds. Within the limits of their task design, this provides compelling evidence that mPFC neurons encode threat, decision, and spatial information in a context-dependent manner. Future experiment designs, which intentionally separate task contexts (e.g. navigation vs. foraging), could serve to further clarify the structure of coding across contexts and/or goal states.

While the study provides an important advance in our understanding of mPFC coding structure under naturalistic conditions, the study still lacks functional manipulations to establish any form of causality. This limitation is acknowledged in the text, and the report is careful not to over interpret suggestions of causal contribution, instead setting a foundation for future investigations.

Reviewer #2 (Public review):

Summary:

Jeong & Choi (2023) use a semi-naturalistic paradigm to tackle the question of how the activity of neurons in the mPFC might continuously encode different functions. They offer two possibilities: either there are separate dedicated populations encoding each function, or cells alter their activity dependent on the current goal of the animal. In a threat-avoidance task rats procurred sucrose in an area of a chamber where, after remaining there for some amount of time, a 'Lobsterbot' robot attacked. In order to initiate the next trial rats had to move through the arena to another area before returning to the robot encounter zone. Therefore the task has two key components: threat avoidance and navigating through space. Recordings in the IL and PL of the mPFC revealed encoding that depended on what stage of the task the animal was currently engaged in. When animals were navigating, neuronal ensembles in these regions encoded distance from the threat. However, whilst animals were directly engaged with the threat and simultaneously consuming reward, it was possible to decode from a subset of the population whether animals would evade the threat. Therefore the authors claim that neurons in the mPFC switched between two functional modes: representing allocentric spatial information, and representing egocentric information pertaining to the reward and threat. Finally, the authors propose a conceptual model based on these data whereby this switching of population encoding is driven by either bottom-up sensory information or top-down arbitration.

Strengths:

Whilst these multiple functions of activity in the mPFC have generally been observed in tasks dedicated to the study of a singular function, less work has been done in contexts where animals continuously switch between different modes of behaviour in a more natural way. Being able to assess whether previous findings of mPFC function apply in natural contexts is very valuable to the field, even outside of those interested in the mPFC directly. This also speaks to the novelty of the work; although mixed selectivity encoding of threat assessment and action selection has been demonstrated in some contexts (e.g. Grunfeld & Likhtik, 2018) understanding the way in which encoding changes on-the-fly in a self-paced task is valuable both for verifying whether current understanding holds true and for extending our models of functional coding in the mPFC.

The authors are also generally thoughtful in their analyses and use a variety of approaches to probe the information encoded in the recorded activity. In particular, they use relatively close analysis of behaviour as well as manipulating the task itself by removing the threat to verify their own results. The use of such a rich task also allows them to draw comparisons, e.g. in different zones of the arena or different types of responses to threat, that a more reduced task would not otherwise allow. Additional in-depth analyses in the updated version of the manuscript, particularly the feature importance analysis, as well as complimentary null findings (a lack of cohesive place cell encoding, and no difference in location coding dependent on direction of trajectory) further support the authors' conclusion that populations of cells in the mPFC are switching their functional coding based on task context rather than behaviour per se. Finally, the authors' updated model schematic proposes an intriguing and testable implementation of how this encoding switch may be manifested by looking at differentiable inputs to these populations.

Weaknesses:

The main existing weakness of this study is that its findings are correlational (as the authors highlight in the discussion). Future work might aim to verify and expand the authors' findings - for example, whether the elevated response of Type 2 neurons directly contributes to the decision-making process or just represents fear/anxiety motivation/threat level - through direct physiological manipulation. However, I appreciate the challenges of interpreting data even in the presence of such manipulations and some of the additional analyses of behaviour, for example the stability of animals' inter-lick intervals in the E-zone, go some way towards ruling out alternative behavioural explanations. Yet the most ideal version of this analysis is to use a pose estimation method such as DeepLabCut to more fully measure behavioural changes. This, in combination with direct physiological manipulation, would allow the authors to fully validate that the switching of encoding by this population of neurons in the mPFC has the functional attributes as claimed here.

Reviewer #3 (Public review):

Summary:

This study investigates how various behavioral features are represented in the medial prefrontal cortex (mPFC) of rats engaged in a naturalistic foraging task. The authors recorded electrophysiological responses of individual neurons as animals transitioned between navigation, reward consumption, avoidance, and escape behaviors. Employing a range of computational and statistical methods, including artificial neural networks, dimensionality reduction, hierarchical clustering, and Bayesian classifiers, the authors sought to predict from neural activity distinct task variables (such as distance from the reward zone and the success or failure of avoidance behavior). The findings suggest that mPFC neurons alternate between at least two distinct functional modes, namely spatial encoding and threat evaluation, contingent on the specific location.

Strengths:

This study attempt to address an important question: understanding the role of mPFC across multiple dynamic behaviors. The authors highlight the diverse roles attributed to mPFC in previous literature and seek to explain this apparent heterogeneity. They designed an ethologically relevant foraging task that facilitated the examination of complex dynamic behavior, collecting comprehensive behavioral and neural data. The analyses conducted are both sound and rigorous.

Weaknesses:

Because the study still lacks experimental manipulation, the findings remain correlational. The authors have appropriately tempered their claims regarding the functional role of the mPFC in the task. The nature of the switch between functional modes encoding distinct task variables (i.e., distance to reward, and threat-avoidance behavior type) is not established. Moreover, the evidence presented to dissociate movement from these task variables is not fully convincing, particularly without single-session video analysis of movement. Specifically, while the new analyses in Figure 7 are informative, they may not fully account for all potential confounding variables arising from changes in context or behavior.

Comments on revisions:

The authors have addressed my previous recommendations.

Author response:

The following is the authors’ response to the original reviews

Public Reviews:

Reviewer #1 (Public review):

Summary:

In this study, Jeong and Choi examine neural correlates of behavior during a naturalistic foraging task in which rats must dynamically balance resource acquisition (foraging) with the risk of threat. Rats first learn to forage for sucrose reward from a spout, and when a threat is introduced (an attack-like movement from a "LobsterBot"), they adjust their behavior to continue foraging while balancing exposure to the threat, adopting anticipatory withdraw behaviors to avoid encounter with the LobsterBot. Using electrode recordings targeting the medial prefrontal cortex (PFC), they identify heterogenous encoding of task variables across prelimbic and infralimbic cortex neurons, including correlates of distance to the reward/threat zone and correlates of both anticipatory and reactionary avoidance behavior. Based on analysis of population responses, they show that prefrontal cortex switches between different regimes of population activity to process spatial information or behavioral responses to threat in a context-dependent manner. Characterization of the heterogenous coding scheme by which frontal cortex represents information in different goal states is an important contribution to our understanding of brain mechanisms underlying flexible behavior in ecological settings.

Strengths:

As many behavioral neuroscience studies employ highly controlled task designs, relatively less is generally known about how the brain organizes navigation and behavioral selection in naturalistic settings, where environment states and goals are more fluid. Here, the authors take advantage of a natural challenge faced by many animals - how to forage for resources in an unpredictable environment - to investigate neural correlates of behavior when goal states are dynamic. Related to his, they also investigate prefrontal cortex (PFC) activity is structured to support different functional "modes" (here, between a navigational mode and a threat-sensitive foraging mode) for flexible behavior. Overall, an important strength and real value of this study is the design of the behavioral experiment, which is trial-structured, permitting strong statistical methods for neural data analysis, yet still rich enough to encourage natural behavior structured by the animal's volitional goals. The experiment is also phased to measure behavioral changes as animals first encounter a threat, and then learn to adapt their foraging strategy to its presence. Characterization of this adaptation process is itself quite interesting and sets a foundation for further study of threat learning and risk management in the foraging context. Finally, the characterization of single-neuron and population dynamics in PFC in this naturalistic setting with fluid goal states is an important contribution to the field. Previous studies have identified neural correlates of spatial and behavioral variables in frontal cortex, but how these representations are structured, or how they are dynamically adjusted when animals shift their goals, has been less clear. The authors synthesize their main conclusions into a conceptual model for how PFC activity can support mode switching, which can be tested in future studies with other task designed and functional manipulations.

Weaknesses:

While the task design in this study is intentionally stimulus-rich and places minimal constraint on the animal to preserve naturalistic behavior, this also introduces confounds that limit interpretability of the neural analysis. For example, some variables which are the target of neural correlation analysis, such as spatial/proximity coding and coding of threat and threat-related behaviors, are naturally entwined. To their credit, the authors have included careful analyses and control conditions to disambiguate these variables and significantly improve clarity.

The authors also claim that the heterogenous coding of spatial and behavioral variables in PFC is structured in a particular way that depends on the animal's goals or context. As the authors themselves discuss, the different "zones" contain distinct behaviors and stimuli, and since some neurons are modulated by these events (e.g., licking sucrose water, withdrawing from the LobsterBot, etc.), differences in population activity may to some extent reflect behavior/event coding. The authors have included a control analysis, removing timepoints corresponding to salient events, to substantiate the claim that PFC neurons switch between different coding "modes." While this significantly strengthens evidence for their conclusion, this analysis still depends on relatively coarse labeling of only very salient events. Future experiment designs, which intentionally separate task contexts (e.g. navigation vs. foraging), could serve to further clarify the structure of coding across contexts and/or goal states.

Finally, while the study includes many careful, in-depth neural and behavioral analyses to support the notion that modal coding of task variables in PFC may play a role in organizing flexible, dynamic behavior, the study still lacks functional manipulations to establish any form of causality. This limitation is acknowledged in the text, and the report is careful not to over interpret suggestions of causal contribution, instead setting a foundation for future investigations.

Thank you for the positive comment. We also acknowledge the inherent drawbacks of studying naturalistic behavior. As you also mentioned in the second round of review, separating navigation and foraging tasks in a larger apparatus, such as the one illustrated below, could better distinguish neural activity patterns associated with these different task types. To address the limitations of the current study, we have revised the report to avoid overinterpretation or unwarranted assumptions, and we appreciate that you have recognized this effort.

Author response image 1.

Reviewer #2 (Public review):

Summary:

Jeong & Choi (2023) use a semi-naturalistic paradigm to tackle the question of how the activity of neurons in the mPFC might continuously encode different functions. They offer two possibilities: either there are separate dedicated populations encoding each function, or cells alter their activity dependent on the current goal of the animal. In a threat-avoidance task rats procurred sucrose in an area of a chamber where, after remaining there for some amount of time, a 'Lobsterbot' robot attacked. In order to initiate the next trial rats had to move through the arena to another area before returning to the robot encounter zone. Therefore the task has two key components: threat avoidance and navigating through space. Recordings in the IL and PL of the mPFC revealed encoding that depended on what stage of the task the animal was currently engaged in. When animals were navigating, neuronal ensembles in these regions encoded distance from the threat. However, whilst animals were directly engaged with the threat and simultaneously consuming reward, it was possible to decode from a subset of the population whether animals would evade the threat. Therefore the authors claim that neurons in the mPFC switched between two functional modes: representing allocentric spatial information, and representing egocentric information pertaining to the reward and threat. Finally, the authors propose a conceptual model based on these data whereby this switching of population encoding is driven by either bottom-up sensory information or top-down arbitration.

Strengths:

Whilst these multiple functions of activity in the mPFC have generally been observed in tasks dedicated to the study of a singular function, less work has been done in contexts where animals continuously switch between different modes of behaviour in a more natural way. Being able to assess whether previous findings of mPFC function apply in natural contexts is very valuable to the field, even outside of those interested in the mPFC directly. This also speaks to the novelty of the work; although mixed selectivity encoding of threat assessment and action selection has been demonstrated in some contexts (e.g. Grunfeld & Likhtik, 2018) understanding the way in which encoding changes on-the-fly in a self-paced task is valuable both for verifying whether current understanding holds true and for extending our models of functional coding in the mPFC.

The authors are also generally thoughtful in their analyses and use a variety of approaches to probe the information encoded in the recorded activity. In particular, they use relatively close analysis of behaviour as well as manipulating the task itself by removing the threat to verify their own results. The use of such a rich task also allows them to draw comparisons, e.g. in different zones of the arena or different types of responses to threat, that a more reduced task would not otherwise allow. Additional in-depth analyses in the updated version of the manuscript, particularly the feature importance analysis, as well as complimentary null findings (a lack of cohesive place cell encoding, and no difference in location coding dependent on direction of trajectory) further support the authors' conclusion that populations of cells in the mPFC are switching their functional coding based on task context rather than behaviour per se. Finally, the authors' updated model schematic proposes an intriguing and testable implementation of how this encoding switch may be manifested by looking at differentiable inputs to these populations.

Weaknesses:

The main existing weakness of this study is that its findings are correlational (as the authors highlight in the discussion). Future work might aim to verify and expand the authors' findings - for example, whether the elevated response of Type 2 neurons directly contributes to the decision-making process or just represents fear/anxiety motivation/threat level - through direct physiological manipulation. However, I appreciate the challenges of interpreting data even in the presence of such manipulations and some of the additional analyses of behaviour, for example the stability of animals' inter-lick intervals in the E-zone, go some way towards ruling out alternative behavioural explanations. Yet the most ideal version of this analysis is to use a pose estimation method such as DeepLabCut to more fully measure behavioural changes. This, in combination with direct physiological manipulation, would allow the authors to fully validate that the switching of encoding by this population of neurons in the mPFC has the functional attributes as claimed here.

I wanted to add a minor comment about interpreting the two possible accounts presented in fig. 8 to suggest a third possibility: that both bottom-up sensory and top-down arbitration mechanisms can occur simultaneously to influence whether the activity of the population switches. Indeed, a model where these inputs are balanced or pitted against each other, so to speak, to continuously modulate encoding in the mPFC seems both adaptive and likely. Further, some speculation on the source of the 'arbitrator' in the top-down account would make this model more tractable for future testing of its validity.

We thank the reviewer for highlighting this important perspective. We fully agree that an intricate and recurrent interaction between bottom-up and top-down modulations is a highly plausible account of how the mPFC changes its encoding mode. In line with this suggestion, we have incorporated this idea as a third possibility in the revised Discussion, alongside an updated version of Figure 8 that explicitly illustrates this competitive model.

Although we were unable to identify a definitive study directly measuring how the mPFC switches encoding modes across tasks, we did find relevant human EEG and fMRI studies addressing this issue. Based on these findings, we now propose the anterior cingulate cortex (ACC) as a potential hub for top-down arbitration. We have added a paragraph in the Discussion describing this possibility and its implications for future testing.

“Which brain region might act as this arbitrator? Evidence from human neuroimaging studies implicates the anterior cingulate cortex (ACC) as a central hub for switching cognitive modes. During task switching, the ACC shows increased activation (Hyafil et al., 2009), enhances connectivity with task-specific regions (Aben et al., 2020), correlates with multitask performance (Kondo et al., 2004), and monitors the reliability of competing decision systems (Lee et al., 2014). Collectively, these findings point to a pivotal role for the ACC in coordinating task assignment. Rodent studies also link the ACC to strategic mode switching (Tervo et al., 2014), suggesting that the rodent ACC could similarly arbitrate between strategies, determining which task-relevant variables are represented in the ventral mPFC, including the PL and IL. Future studies combining multi-context tasks with causal manipulations will be essential to determine whether these functional shifts are driven primarily by top-down arbitration or by bottom-up sensory inputs.”

Reviewer #3 (Public review):

Summary:

This study investigates how various behavioral features are represented in the medial prefrontal cortex (mPFC) of rats engaged in a naturalistic foraging task. The authors recorded electrophysiological responses of individual neurons as animals transitioned between navigation, reward consumption, avoidance, and escape behaviors. Employing a range of computational and statistical methods, including artificial neural networks, dimensionality reduction, hierarchical clustering, and Bayesian classifiers, the authors sought to predict from neural activity distinct task variables (such as distance from the reward zone and the success or failure of avoidance behavior). The findings suggest that mPFC neurons alternate between at least two distinct functional modes, namely spatial encoding and threat evaluation, contingent on the specific location.

Strengths:

This study attempt to address an important question: understanding the role of mPFC across multiple dynamic behaviors. The authors highlight the diverse roles attributed to mPFC in previous literature and seek to explain this apparent heterogeneity. They designed an ethologically relevant foraging task that facilitated the examination of complex dynamic behavior, collecting comprehensive behavioral and neural data. The analyses conducted are both sound and rigorous.

Weaknesses:

Because the study still lacks experimental manipulation, the findings remain correlational. The authors have appropriately tempered their claims regarding the functional role of the mPFC in the task. The nature of the switch between functional modes encoding distinct task variables (i.e., distance to reward, and threat-avoidance behavior type) is not established. Moreover, the evidence presented to dissociate movement from these task variables is not fully convincing, particularly without single-session video analysis of movement. Specifically, while the new analyses in Figure 7 are informative, they may not fully account for all potential confounding variables arising from changes in context or behavior.

Regarding the claim of highly stereotyped behavior, there are some inconsistencies. While the authors assert this, Figure 1F shows inter-animal variability, and the PETHs, representing averaged activity, may not fully capture the variability of the behavior across sessions and animals. To strengthen this aspect, a more detailed analysis that examines the relationship between behavior and neural activity on a trial-by-trial basis, or at minimum, per session, could help.

We thank the reviewer for this thoughtful recommendation and the opportunity to clarify our use of the term “stereotyped behavior.” By this, we were specifically referring to the animals’ consistent licking behavior in the E-zone, rather than to the latency of head withdrawal, which indeed varied across trials and animals. Because licking tempo and body posture during sucrose consumption were highly consistent, the decision to avoid or stay (AW vs. EW) could not be predicted from overt behavior alone. This consistency strengthens our conclusion that the significant predictive power of the Bayesian decoding analysis reflects intrinsic firing patterns of the mPFC neural network, rather than simple behavioral correlates of avoidance.

We also note that the Bayesian model was conducted on a trial-by-trial basis, and the reported prediction accuracy of 73% represents the average across all individual trials (Figure 6B, C). Thus, the analysis inherently captures variability across trials and animals, directly addressing the reviewer’s concern.

The reviewer is correct that the PETHs shown in Figure 5 are based on session-averaged activity aligned to head-entry and head-withdrawal events. The purpose of this analysis was to illustrate that certain modulation patterns could be grouped into 2–3 distinct categories. While averaged activity can provide insight into collective responses to external events, we agree that trial-based analyses provide a more rigorous demonstration of the link between neural ensemble activity and behavioral decisions. This is precisely why we complemented the PETH analysis with Bayesian decoding, which provides stronger evidence that mPFC ensemble activity is predictive of the animal’s choice to avoid or stay.

Similarly, the claim regarding the limited scope of extraneous behavior (beyond licking) requires further substantiation. It would be more convincing to quantify potential variations in licking vigor and to provide evidence for the absence of significant postural changes.

To address this concern, we quantified licking vigor using the inter-lick interval (ILI) as an indirect index. A lick was defined as the period from tongue contact with the IR beam (Lick-On) to withdrawal (Lick-Off), and the ILI was calculated as the time between a Lick-Off and the subsequent Lick-On. Across all animals, ILIs were clustered within a narrow range with a median of 0.155 s (see Author response image 4, left panel).

We analyzed licking vigor at two levels: within trials and within sessions. Because reduced vigor or satiation would lengthen ILIs, comparing the first half and the last half of ILIs within a trial or within a session provides a sensitive proxy for licking consistency.

Within trials: For each of 2,820 trials, we compared the mean ILI of the first half of licks to that of the second half. The average difference was only ~ 17 ms (middle panel). Across sessions: Trial-averaged ILIs were compared between the first and last halves of each session, yielding a mean difference ~ 1.7 ms per session (right panel).

These analyses demonstrate that rats maintained stable licking vigor whenever they entered the E-zone, regardless of avoidance outcome.

Author response image 2.

Concerning the ANN model, while I understand the choice of a 4-layer network for its performance, the study could have benefited from exploring simpler models. A model where weight corresponds directly to individual neurons could improve interpretability and facilitate the investigation of dynamic changes in neuronal 'modes' (i.e., weight adjustments) over time.

We fully agree with the reviewer on the importance of biologically interpretable models. While artificial neural networks (ANNs) share certain similarities with neural computation, they are not intended to capture biological realism. For example, the error correction mechanism used in ANNs, such as backpropagation has no direct counterpart in mammalian neural circuits. Although we considered approaches that would link each computational node more directly to the activity of individual neurons, building such a model would require temporally sensitive, mechanistic frameworks (e.g., leaky integrate-and-fire networks) and an extensive behavioral alignment effort, which is beyond the scope of the current study.

Our use of an ANN was intended solely as an analytical tool to uncover hidden patterns in multi-unit activity that may not be detectable with traditional methods. Among various machine-learning algorithms, we selected a four-layer ANN regressor because it achieved significantly lower decoding errors (Supplementary Figure S3) and showed robustness to hyperparameter variation (Glaser et al., 2020). To acknowledge the limitations of this approach and suggest future directions, we have revised the Results section to explicitly discuss these points.

“Among various machine learning algorithms, we selected a robust tool for decoding underlying patterns in the data, rather than to model the architecture of the mPFC. We implemented a four-layer artificial neural network regressor (ANN; see Materials and Methods for a detailed structure), as the ANN achieves significantly lower decoding errors (Supplementary Figure S3) and has robustness to hyperparameter changes (Glaser et al., 2020).”

eLife Assessment

This important study investigates nerve-injury-induced allodynia by studying the role of a subpopulation of excitatory dorsal horn CCK+ neurons that express the estrogen receptor GPR30 and potentially modulate nociceptive sensitivity via direct inputs from primary somatosensory cortex. In this revised version, the authors addressed many of the critiques raised through added analyses that convincingly support the notion that spinal GPR30 neurons are indeed an excitatory subpopulation of CCK+ neurons that contribute to neuropathic pain. While evidence of a direct functional corticospinal projection to CCK+/GPR30+neurons is not fully demonstrated, this work will be of broad interest to researchers interested in the neural circuitry of pain.

Reviewer #1 (Public review):

In this manuscript, Chen et al. investigate the role of the membrane estrogen receptor GPR30 in spinal mechanisms of neuropathic pain. Using a wide variety of techniques, they first provide convincing evidence that GPR30 expression is restricted to neurons within the spinal cord, and that GPR30 neurons are well-positioned to receive descending input from the primary sensory cortex (S1). In addition, the authors put their findings in the context the previous knowledge in the field, presenting evidence demonstrating that GRP30 is expressed in the majority of CCK-expressing spinal neurons. Overall, this manuscript furthers our understanding of neural circuity that underlies neuropathic pain and will be of broad interest to neuroscientists, especially those interested in somatosensation. Nevertheless, the manuscript would be strengthened by additional analyses and clarification of data that is currently presented.

Strengths:

The authors present convincing evidence for expression of GPR30 in the spinal cord that is specific to spinal neurons. Similarly, complementary approaches including pharmacological inhibition and knockdown of GPR30 are used to demonstrate a role for the receptor in driving nerve injury-induced pain in rodent models.

Weaknesses:

Although steps were taken to put their data into the broader context of what is already known about the spinal circuitry of pain, more considerations and analyses would help the authors better achieve their goal. For instance, to determine whether GPR30 is expressed in excitatory or inhibitory neurons, more selective markers for these subtypes should be used over CamK2. Moreover, quantitative analysis of the extent of overlap between GPR30+ and CCK+ spinal neurons is needed to understand the potential heterogeneity of the GPR30 spinal neuron population, and to interpret experiments characterizing descending SI inputs onto GPR30 and CCK spinal neurons. Filling these gaps in knowledge would make their findings more solid.

Revised Manuscript Update:

In their revised manuscript, Chen et al. have added additional data that establishes GPR30 spinal neurons as a population of excitatory neurons, half of which express CCK. These data help to position GPR30 neurons in the existing framework of spinal neuron populations that contribute to neuropathic pain, strengthening the author's findings.

I have no new recommendations to the author's following this round of revisions.

Reviewer #3 (Public review):

Summary:

The authors convincingly demonstrate that a population of CCK+ spinal neurons in the deep dorsal horn express the G protein coupled estrogen receptor GPR30 to modulate pain sensitivity in the chronic constriction injury (CCI) model of neuropathic pain in mice. Using complementary pharmacological and genetic knockdown experiments they convincingly show that GPR30 inhibition or knockdown reverses mechanical, tactile and thermal hypersensitivity, conditioned place aversion, and c-fos staining in the spinal dorsal horn after CCI. They propose that GPR30 mediates an increase in postsynaptic AMPA receptors after CCI using slice electrophysiology which may underlie the increased behavioral sensitivity. They then use anterograde tracing approaches to show that CCK and GPR30 positive neurons in the deep dorsal horn may receive direct connections from primary somatosensory cortex. Chemogenetic activation of these dorsal horn neurons proposed to be connected to S1 increased nociceptive sensitivity in a GPR30 dependent manner. Overall, the data are very convincing and the experiments are well conducted and adequately controlled. The potential role of direct connections from S1 for descending modulation of pain and the endogenous mechanism(s) activating GPR30 will be interesting to test in future studies.

Strengths:

The experiments are very well executed and adequately controlled throughout the manuscript. The data are nicely presented and supportive of a role for GPR30 signaling in the spinal dorsal horn influencing nociceptive sensitivity following CCI. The authors also did an excellent job of using complementary approaches to rigorously test their hypothesis.

Weaknesses:

While the viral tracing demonstrates a potential connection between S1 and CCK+ or GPR30+ spinal neurons, no direct evidence is provided for S1 in facilitating any activity of these neurons in the dorsal horn.

Comments on the latest version:

The authors have done a good job addressing previous critiques and have appropriately revised the manuscript and conclusions.

Author response:

The following is the authors’ response to the previous reviews

Reviewer #1 (Public review): 

In their revised manuscript, Chen et al. have added additional data that establishes GPR30 spinal neurons as a population of excitatory neurons, half of which express CCK. These data help to position GPR30 neurons in the existing framework of spinal neuron populations that contribute to neuropathic pain, strengthening the author's findings.

Thank you very much for your positive feedback and for recognizing the value of our additional data.

Reviewer #3 (Public review):

The authors did an excellent job addressing many of the critiques raised. Despite acknowledging that a direct functional corticospinal projection to CCK/GPR30+neurons is not supported by the data and revising the title, these claims still persist throughout the manuscript. Manipulating gene expression or the activity of postsynaptic neurons through a trans-synaptic labeling strategy does not directly support any claim that those upstream neurons are directly modulating spinal neurons through the proposed pathway. Indeed they might, but that is not demonstrated here.

We sincerely thank the reviewer for this critical insight. We fully agree that our trans-synaptic approach does not provide a direct functional connection. In response, we have revised the manuscript to remove any overstated claims of "direct" modulation and instead emphasize the critical role of spinal GPR30+ neurons. Moreover, we have added a statement in the Discussion to acknowledge this limitation and to highlight that the precise function role of this connection requires further investigation in further studies.

Reviewer #1 (Recommendations for the authors): 

I recommend 2 minor corrections to the text and figures

(1)  Line 131 : "What's more, near-universal CCK+ neurons were co-localized with GPR30 (Fig 2F and G)."

The additional quantification of the overlap between GPR30 and tdTomato provided by the authors is useful, but there are inconsistencies with how the data are reported in the figures and text, making them difficult to interpret. 2F supports the author's conclusion that approximately 90% of CCK⁺ neurons express GPR30, and about 50% of GPR30⁺ neurons co-express CCK. However, the x-axis labels in 2G appear to have been switched, and suggest that the opposite is true (i.e., most GRPR neurons are CCK+, while only 50% of CCK neurons are GPR30+). Please clarify which is correct throughout the results and discussion sections.

Thank you for identifying this important error. We apologized for the confusion caused by the mislabeled x-axis in Fig. 2G. The x-axis labels were indeed inadvertently switched. The correct data is that approximately 90% of CCK⁺ neurons express GPR30. We have corrected the figure and have carefully reviewed the entire manuscript to ensure all related descriptions and discussions are consistent with the accurate quantification.

(2) The following sentence describing Figure 5 was hard to follow: Lines 190-192, "Consistent with prior observations, we found that these SDH downstream neurons exhibited colocalization with CCK+ neurons, with 28.1% of mCherry+ neurons expressing CCK (Fig 5I and J)." Since the authors are describing a common population of neurons, a statement describing the coexpression (rather than the colocalization" would more simply summarize their data.

We thank the reviewer for this helpful suggestion. We fully agree that "coexpression" is a more precise term for the description. We have revised the sentence on Lines 189-190 to read: "Consistent with prior observations, we found that 28.1% of mCherry+ S1-SDH downstream neurons coexpressed CCK (Fig 5I and J)."

Reviewer #3 (Recommendations for the authors): 

Additional Recommendations

The authors did a commendable job revising the manuscript text to improve readability; however, several informal phrases from the original version still persist, or were added (e.g. "by the way").

We thank the reviewer for this valuable feedback regarding the language. We have conducted a line-by-line review of the entire manuscript to identify all remaining informal phrases, and replaced them with more appropriate phrasing.

It should be clearly mentioned that spontaneous E/IPSCs were recorded in Figure 4 and Fig S5.

We thank the reviewer for this helpful suggestion. We have now clearly indicated the spontaneous E/IPSCs in Fig. 4 and Fig. S5 and manuscript.

The rationale for recording EPSCs from GFP-labeled CCK+ neurons because "a significant proportion of spinal CCK+ neurons form excitatory synapses with upstream neurons" does not make any sense. Do the authors instead mean that CCK neurons receive excitatory inputs from other spinal neurons and intend to test if those synaptic connections are modulated by GPR30?

We thank the reviewer for this critical correction. Our intended meaning was indeed that CCK⁺ neurons receive excitatory inputs from other neurons, and we aimed to test whether those synaptic connections are modulated by GPR30. To avoid confusion, we have revised the manuscript to remove the erroneous statement “Since CCK+ neurons mainly receive excitatory synaptic inputs from upstream neurons, we then intended to test whether GPR30 modulated these synaptic connections.”

I am confused by the statement on Page 8 "to examine whether GPCR30-mediated EPSCs depend on AMPA mediated currents." Given that sEPSCs were recorded at -70 mV in low Cl internal I'm not sure what other glutamate receptor would be involved. Perhaps the intention was to more directly test whether GPR30 activation acutely modulates AMPAR-mediated EPSCs? However, as the authors acknowledged, this experiment does not necessarily support a solely post-synaptic AMPAR-dependent mechanism.

We thank the reviewer for this insightful comment and apologize for the lack of clarity. Our intention was indeed to test whether GPR30 activation modulates AMPAR-mediated currents. We have revised the text. In addition, we also emphasize in the Discussion that our data did not rule out the potential pre-synaptic contributions to this effect.

An elevation in EPSCs within a cell does not necessarily mean that the cell is more excitable, only that it is receiving more excitatory inputs or has an increase in synaptic receptors. The cell may scale down its activity to compensate for this increase. I recommend only drawing conclusions from what the experiments actually tested.

We thank the reviewer for this crucial clarification. We have revised the manuscript to remove any claims that the cells were "more excitable". Our conclusions now strictly focus on the specific findings that GPR30 activation enhanced the excitatory transmission onto CCK⁺ neurons.

eLife Assessment

Cryptovaranoides, a Late Triassic animal (some 230 Ma old), was originally described as a possibly anguimorph squamate, i.e., more closely related to snakes and some extant lizards than to other extant lizards, making Squamata much older than previously thought and providing a new calibration date inside it. Following a rebuttal and a defense, this fourth important contribution to the debate makes a convincing argument that Cryptovaranoides is not a squamate. Further comparisons to potentially closely related animals such as early lepidosauromorphs would greatly benefit this study, and parts of the text require clarification.

Reviewer #1 (Public review):

In the Late Triassic (around 230 Ma ago), southern Wales and adjacent parts of England were a karst landscape. The caves and crevices accumulated remains of small vertebrates. These fossil-rich fissure fills are being exposed in limestone quarrying. In 2022 (reference 13 of the article), a partial articulated skeleton and numerous isolated bones from one fissure fill were named Cryptovaranoides microlanius and described as the oldest known squamate - the oldest known animal, by some 50 Ma, that is more closely related to snakes and some extant lizards than to other extant lizards. This would have considerable consequences for our understanding of the evolution of squamates and their closest relatives, especially for its speed and absolute timing, and was supported in the same paper by phylogenetic analyses based on different datasets.

In 2023, the present authors published a rebuttal (ref. 18) to the 2022 paper, challenging anatomical interpretations and the irreproducible referral of some of the isolated bones to Cryptovaranoides. Modifying the datasets accordingly, they found Cryptovaranoides outside Squamata and presented evidence that it is far outside. In 2024 (ref. 19), the original authors defended most of their original interpretation and presented some new data, some of it from newly referred isolated bones. The present article discusses anatomical features and the referral of isolated bones in more detail, documents some clear misinterpretations, argues against the widespread but not justifiable practice of referring isolated bones to the same species as long as there is merely no known evidence to the contrary, further argues against comparing newly recognized fossils to lists of diagnostic characters from the literature as opposed to performing phylogenetic analyses and interpreting the results, and finds Cryptovaranoides outside Squamata again.

Although a few of the character discussions can probably still be improved, I see no sign that the discussion is going in circles or otherwise becoming unproductive. I can even imagine that the present contribution will end it.

Reviewer #2 (Public review):

Congratulations on this revised manuscript on the phylogenetic affinities of Cryptovaranoides, and thank you for your modifications to this manuscript following review.

This manuscript offers a careful review of the features used to hypothesize the placement of Cryptovaranoides within crown Squamata and instead suggests that this taxon represents an earlier-diverging reptile. This work therefore reconciles morphological and molecular data regarding lizard origins, which is an important contribution to the field of vertebrate paleontology.

The authors have improved their manuscript following reviewer comments and now provide more thorough comparisons with other early reptiles and archosauromorphs, an improvement over early versions of this paper. Changes to these comparative descriptions provide important rationale concerning the absence of superficially squamate-like features in Cryptovaranoides.

The evolutionary relationships of Cryptovaranoides among reptiles will certainly be a matter of debate until detailed anatomical descriptions of this taxon and other putative lepidosauromorphs are published. However, it can now be said with confidence that the presence of any crown squamate in the Permian or Triassic is unlikely and should be met with skepticism, the same sort of skepticism provided in this manuscript.

Reviewer #3 (Public review):

Summary:

The study provides an interesting contribution to our understanding of Cryptovaranoides relationships, which is a matter of intensive debate among researchers. The authors have modified the manuscript according to most of my suggestions. My main concerns are about the wording of some statements but the authors have the right to put it as they want in the end. Overall the discussion and data are well prepared. I would recommend to publish the manuscript after very minor revisions.

Strengths:

Detailed analysis of the discussed characters. Illustrations of some comparative materials.

Weaknesses:

Abstract: "Our team challenged this identification and instead suggested †Cryptovaranoides had unclear affinities to living reptiles"

Unfortunately I have to disagree again. "unclear affinities to living reptiles" can mean anything including a crown lizard. First, the 2023 paper clearly rejected the squamate hypothesis and presented some evidence that potentially places Cryptovaranoides among Archosauromorpha. In this context "unclear where it would belong within the latter" does not really matter. Second, we are not discussing here if Cryptovaranoides is a squamate or a stem-squamate. We have many more options on the table, so "unclear affinities" is too imprecise. Please change it to "could be an archosauromorph or an indeterminate neodiapsid" in the abstract to show the scale of conflicting evidence.

Author response:

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

Reviewer #1 (Public review):

In the Late Triassic and Early Jurassic (around 230 to 180 Ma ago), southern Wales and adjacent parts of England were a karst landscape. The caves and crevices accumulated remains of small vertebrates. These fossil-rich fissure fills are being exposed in limestone quarrying. In 2022 (reference 13 of the article), a partial articulated skeleton and numerous isolated bones from one fissure fill of end-Triassic age (just over 200 Ma) were named Cryptovaranoides microlanius and described as the oldest known squamate - the oldest known animal, by some 20 to 30 Ma, that is more closely related to snakes and some extant lizards than to other extant lizards. This would have considerable consequences for our understanding of the evolution of squamates and their closest relatives, especially for their speed and absolute timing, and was supported in the same paper by phylogenetic analyses based on different datasets.

In 2023, the present authors published a rebuttal (reference 18) to the 2022 paper, challenging anatomical interpretations and the irreproducible referral of some of the isolated bones to Cryptovaranoides. Modifying the datasets accordingly, they found Cryptovaranoides outside Squamata and presented evidence that it is far outside. In 2024 (reference 19), the original authors defended most of their original interpretation and presented some new data, some of it from newly referred isolated bones. The present article discusses anatomical features and the referral of isolated bones in more detail, documents some clear misinterpretations, argues against the widespread but not justifiable practice of referring isolated bones to the same species as long as there is merely no known evidence to the contrary, further argues against comparing newly recognized fossils to lists of diagnostic characters from the literature as opposed to performing phylogenetic analyses and interpreting the results, and finds Cryptovaranoides outside Squamata again.

Although a few of the character discussions and the discussion of at least one of the isolated bones can probably still be improved (and two characters are addressed twice), I see no sign that the discussion is going in circles or otherwise becoming unproductive. I can even imagine that the present contribution will end it.

We appreciate the positive response from reviewer 1!

Reviewer #2 (Public review):

Congratulations on this thorough manuscript on the phylogenetic affinities of Cryptovaranoides.

Thank you.

Recent interpretations of this taxon, and perhaps some others, have greatly changed the field's understanding of reptile origins- for better and (likely) for worse.

We agree, and note that while it is possible for challenges to be worse than the original interpretations, both the original and subsequent challenges are essential aspects of what make science, science.

This manuscript offers a careful review of the features used to place Cryptovaranoides within Squamata and adequately demonstrates that this interpretation is misguided, and therefore reconciles morphological and molecular data, which is an important contribution to the field of paleontology. The presence of any crown squamate in the Permian or Triassic should be met with skepticism, the same sort of skepticism provided in this manuscript.

We agree and add that every testable hypothesis requires skepticism and testing.

I have outlined some comments addressing some weaknesses that I believe will further elevate the scientific quality of the work. A brief, fresh read‑through to refine a few phrases, particularly where the discussion references Whiteside et al. could also give the paper an even more collegial tone.

We have followed Reviewer 2’s recommendations closely (see below) and have justified in our responses if we do not fully follow a particular recommendation.

This manuscript can be largely improved by additional discussion and figures, where applicable. When I first read this manuscript, I was a bit surprised at how little discussion there was concerning both non-lepidosauromorph lepidosaurs as well as stem-reptiles more broadly. This paper makes it extremely clear that Cryptovaranoides is not a squamate, but would greatly benefit in explaining why many of the characters either suggested by former studies to be squamate in nature or were optimized as such in phylogenetic analyses are rather widespread plesiomorphies present in crownward sauropsids such as millerettids, younginids, or tangasaurids. I suggest citing this work where applicable and building some of the discussion for a greatly improved manuscript. In sum:

(1) The discussion of stem-reptiles should be improved. Nearly all of the supposed squamate features in Cryptovaranoides are present in various stem-reptile groups. I've noted a few, but this would be a fairly quick addition to this work. If this manuscript incorporates this advice, I believe arguments regarding the affinities of Cryptovaranoides (at least within Squamata) will be finished, and this manuscript will be better off for it.

(2) I was also surprised at how little discussion there was here of putative stem-squamates or lepidosauromorphs more broadly. A few targeted comparisons could really benefit the manuscript. It is currently unclear as to why Cryptovaranoides could not be a stem-lepidosaur, although I know that the lepidosaur total-group in these manuscripts lacks character sampling due to their scarcity.

We are responding to (1) and (2) together. We agree with the Reviewer that a thorough comparison of Cryptovaranoides to non-lepidosaurian reptiles is critical. This is precisely what we did in our previous study: Brownstein et al. (2023)— see main text and supplementary information therein. As addressed therein, there is a substantial convergence between early lepidosaurs and some groups of archosauromorphs (our inferred position for Cryptovaranoides). Many of those points are not addressed in detail here in order to avoid redundancy and are simply referenced back to Brownstein et al. (2023). Secondly, stem reptiles (i.e., non-lepidosauromorphs and non-archosauromorphs), such as suggested above (millerettids, younginids, or tangasaurids), are substantially more distantly related to Cryptovaranoides (following any of the published hypotheses). As such, they share fewer traits (either symplesiomorphies or homoplasies), and so, in our opinion, we would risk directing losing the squamate-focus of our study.

We thus respectfully decline to engage the full scope of the problem in this contribution, but do note that this level of detailed work would make for an excellent student dissertation research program.

(3) This manuscript can be improved by additional figures, such as the slice data of the humerus. The poor quality of the scan data for Cryptovaranoides is stated during this paper several times, yet the scan data is often used as evidence for the presence or absence of often minute features without discussion, leaving doubts as to what condition is true. Otherwise, several sections can be rephrased to acknowledge uncertainty, and probably change some character scorings to '?' in other studies.

We strongly agree with the reviewer. Unfortunately, the original publication (Whiteside et al., 2021) did not make available the raw CT scan data to make this possible. As noted below in the Responses to Recommendations Section, we only have access to the mesh files for each segmented element. While one of us has observed the specimens personally, we have not had the opportunity to CT scan the specimens ourselves.

Reviewer #3 (Public review):

Summary:

The study provides an interesting contribution to our understanding of Cryptovaranoides relationships, which is a matter of intensive debate among researchers. My main concerns are in regard to the wording of some statements, but generally, the discussion and data are well prepared. I would recommend moderate revisions.

Strengths:

(1) Detailed analysis of the discussed characters.

(2) Illustrations of some comparative materials.

Thank you for noting the strengths inherent to our study.

Weaknesses:

Some parts of the manuscript require clarification and rewording.

One of the main points of criticism of Whiteside et al. is using characters for phylogenetic considerations that are not included in the phylogenetic analyses therein. The authors call it a "non-trivial substantive methodological flaw" (page 19, line 531). I would step down from such a statement for the reasons listed below:

(1) Comparative anatomy is not about making phylogenetic analyses. Comparative anatomy is about comparing different taxa in search of characters that are unique and characters that are shared between taxa. This creates an opportunity to assess the level of similarity between the taxa and create preliminary hypotheses about homology. Therefore, comparative anatomy can provide some phylogenetic inferences.

That does not mean that tests of congruence are not needed. Such comparisons are the first step that allows creating phylogenetic matrices for analysis, which is the next step of phylogenetic inference. That does not mean that all the papers with new morphological comparisons should end with a new or expanded phylogenetic matrix. Instead, such papers serve as a rationale for future papers that focus on building phylogenetic matrices.

We agree completely. We would also add that not every study presenting comparative anatomical work need be concluded with a phylogenetic analysis.

Our criticism of Whiteside et al. (2022) and (2024) is that these studies provided many unsubstantiated claims of having recovered synapomorphies between Cryptovaranoides and crown squamates without actually having done so through the standard empirical means (i.e., phylogenetic analysis and ancestral state reconstruction). Both Whiteside et al. (2022) and (2024) indicate characters presented as ‘shared with squamates’ along with 10 characters presented as synapomorphies (10). However, their actual phylogenetically recovered synapomorphies were few in number (only 3) and these were not discussed.

Furthermore, Whiteside et al. (2022) and (2024) comparative anatomy was restricted to comparing †Cryptovaranoides to crown squamates., based on the assumption that †Cryptovaranoides was a crown squamate and thus only needed to be compared to crown squamates.

In conclusion, we respectfully, we maintain such efforts are “non-trivial substantive methodological flaw(s)”.

(2) Phylogenetic matrices are never complete, both in terms of morphological disparity and taxonomic diversity. I don't know if it is even possible to have a complete one, but at least we can say that we are far from that. Criticising a work that did not include all the possibly relevant characters in the phylogenetic analysis is simply unfair. The authors should know that creating/expanding a phylogenetic matrix is a never-ending work, beyond the scope of any paper presenting a new fossil.

Respectfully, we did not criticize previous studies for including an incomplete phylogeny. Instead, we criticized the methodology behind the homology statements made in Whiteside et al. (2022) and Whiteside et al. (2024).

(3) Each additional taxon has the possibility of inducing a rethinking of characters. That includes new characters, new character states, character state reordering, etc. As I said above, it is usually beyond the scope of a paper with a new fossil to accommodate that into the phylogenetic matrix, as it requires not only scoring the newly described taxon but also many that are already scored. Since the digitalization of fossils is still rare, it requires a lot of collection visits that are costly in terms of time.

We agree on all points, but we are unsure of what the Reviewer is asking us to do relative to this study.

(4) If I were to search for a true flaw in the Whiteside et al. paper, I would check if there is a confirmation bias. The mentioned paper should not only search for characters that support Cryptovaranoides affinities with Anguimorpha but also characters that deny that. I am not sure if Whiteside et al. did such an exercise. Anyway, the test of congruence would not solve this issue because by adding only characters that support one hypothesis, we are biasing the results of such a test.

We would refer the Reviewer to their section (1) on comparative anatomy. As we and the Reviewer have pointed out, Whiteside et al. did not perform comparative anatomical statements outside of crown Squamata in their original study. More specifically, Whiteside et al. (2022, Fig. 8) presented a phylogeny where Cryptovaranoides formed a clade with Xenosaurus within the crown of Anguimorpha or what they termed “Anguiformes”, and made comparisons to the anatomies of the legless anguids, Pseudopus and Ophisaurus. Whiteside et al. (2024), abandoned “Anguiformes”, maintained comparisons to Pseudopus and emphasized affinities with Anguimorpha (but almost all of their phylogenies as published, they do not recover a monophyletic Angumimorpha unless amphisbaenians and snakes are considered to be anguimorphans. Thus, we agree that confirmation bias was inherent in their studies.

To sum up, there is nothing wrong with proposing some hypotheses about character homology between different taxa that can be tested in future papers that will include a test of congruence. Lack of such a test makes the whole argumentation weaker in Whiteside et al., but not unacceptable, as the manuscript might suggest. My advice is to step down from such strong statements like "methodological flaw" and "empirical problems" and replace them with "limitations", which I think better describes the situation.

We agree with the first sentence in this paragraph – there is nothing wrong with proposing character homologies between different taxa based on comparative anatomical studies. However, that is not what Whiteside et al. (2022) and (2024) did. Instead, they claimed that an ad hoc comparison of Cryptovaranoides to crown Squamata confirmed that Cryptovaranoides is in fact a crown squamate and likely a member of Anguimorpha. Their study did not recognize limitations, but rather, concluded that their new taxon pushed the age of crown Squamata into the Triassic.

As noted by Reviewer 2, such a claim, and the ‘data’ upon which it is based, should be treated with skepticism. We have elected to apply strong skepticism and stringent tests of falsification to our critique.

Reviewer #1 (Recommendations for the authors):

(1) Lines 596-598 promise the following: "we provide a long[-]form review of these and other features in Cryptovaranoides that compare favorably with non-squamate reptiles in Supplementary Material." You have kindly informed me that all this material has been moved into the main text; please amend this passage.

This has been deleted.

(2) Comments on science

41: I would rather say "an additional role".

This has been edited accordingly.

43: Reconstructing the tree entirely from extant organisms and adding fossils later is how Hennig imagined it, because he was an entomologist, and fossil insects are, on average,e extremely rare and usually very incomplete (showing a body outline and/or wing venation and little or nothing else). He was wrong, indeed wrong-headed. As a historical matter, phylogenetic hypotheses were routinely built on fossils by the mid-1860s, pretty much as soon as the paleontologists had finished reading On the Origin of Species, and this practice has never declined, let alone been interrupted. As a theoretical matter, including as many extinct taxa as possible in a phylogenetic analysis is desirable because it breaks up long branches (as most recently and dramatically shown by Mongiardino Koch & Parry 2020), and while some methods and some kinds of data are less susceptible to long-branch attraction and long-branch repulsion than others, none are immune; and while missing data (on average more common in fossils) can actively mislead parametric methods, this is not the case with parsimony, and even in Bayesian inference the problem is characters with missing data, not taxa with missing data. Some of you have, moreover, published tip-dated phylogenetic analyses. As a practical matter, molecular data are almost never available from fossils, so it is, of course, true that analyses which only use molecular data can almost never include fossils; but in the very rare exceptions, there is no reason to treat fossil evidence as an afterthought.

We agree and have changed “have become” to “is.”

49-50, 59: The ages of individual fissure fills can be determined by biostratigraphy; as far as I understand, all specimens ever referred to Cryptovaranoides [13, 19] come from a single fill that is "Rhaetian, probably late Rhaetian (equivalent of Cotham Member, Lilstock Formation)" [13: pp. 2, 15].

We appreciate this comment; the recent literature, however, suggests that variable ages are implied by the biostratigraphy at the English Fissure Fills, so we have chosen to keep this as is. Also note that several isolated bones were not recovered with the holotype but were discussed by Whiteside et al. (2024). The provenance of these bones was not clearly discussed in that paper.

59-60: Why "putative"? Just to express your disagreement? I would do that in a less misleading way, for example: "and found this taxon as a crown-group squamate (squamate hereafter) in their phylogenetic analyses." - plural because [19] presented four different analyses of two matrices just in the main paper.

We have removed this word.

121-124: The entepicondylar foramen is homologous all the way down the tree to Eusthenopteron and beyond. It has been lost a quite small number of times. The ectepicondylar foramen - i.e., the "supinator" (brachioradialis) process growing distally to meet the ectepicondyle, fusing with it and thereby enclosing the foramen - goes a bit beyond Neodiapsida and also occurs in a few other amniote clades (...as well as, funnily enough, Eusthenopteron in later ontogeny, but that's independent).

We agree. However, the important note here is that the features on the humerus of Cryptovaranoides are not comparable (differ in location and morphology) to the ent- and ectepondylar foramina in other reptiles, as we discuss at length. As such, we have kept this sentence as is.

153: Yes, but you [18] mistakenly wrote "strong anterior emargination of the maxillary nasal process, which is [...] a hallmark feature of archosauromorphs" in the main text (p. 14) - and you make the same mistake again here in lines 200-206! Also, the fact [19: Figure 2a-c] remains that Cryptovaranoides did not have an antorbital fenestra, let alone an antorbital fossa surrounding it (a fossa without a fenestra only occurs in some cases of secondary loss of the fenestra, e.g., in certain ornithischian dinosaurs). Unsurprisingly, therefore, Cryptovaranoides also does not have an orbital-as-opposed-to-nasal process on its maxilla [19: Figure 2a-c].

Line 243-249 (in original manuscript) deal with the emargination of maxillary nasal process (but this does not imply a full antorbital fenestra).  We explicitly state that this feature alone "has limited utility" for supporting archosauromorph affinity.

158-173: The problem here is not that the capitellum is not preserved; from amniotes and "microsaurs" to lissamphibians and temnospondyls, capitella ossify late, and larger capitella attach to proportionately larger concave surfaces, so there is nothing wrong with "the cavity in which it sat clearly indicates a substantial condyle in life". Instead, the problem is a lack of quantification (...as has also been the case in the use of the exact same character in the debate on the origin of lissamphibians); your following sentence (lines 173-175) stands. The rest of the paragraph should be drastically shortened.

We appreciate this comment. We note that the ontogenetic variation of this feature is in part the issue with the interpretation provided by Whiteside et al. (2024). The issue is the lack of consistency on the morphology of the capitellum in that study. We are unclear on what the reviewer means by ‘quantification,’ as the character in question is binary. 

250-252: It's not going to matter here, but in any different phylogenetic context, "sphenoid" would be confusing given the sphenethmoid, orbitosphenoid, pleurosphenoid, and laterosphenoid. I actually recommend "parabasisphenoid" as used in the literature on early amniotes (fusion of the dermal parasphenoid and the endochondral basisphenoid is standard for amniotes).

We have added "(=parabasisphenoid)" on first use but retain use of sphenoid because in the squamate and archosauromorph literature, sphenoid (or basisphenoid) is used more frequently.

314-315: Vomerine teeth are, of course, standard for sarcopterygians. Practically all extant amphibians have a vomerine toothrow, for example. A shagreen of denticles on the vomer is not as widespread but still reaches into the Devonian (Tulerpeton).

We agree, but vomerine teeth are rare in lepidosaurs and archosaurs and occur only in very recent clades e.g. anguids and one stem scincoid. Their presence in amphibians is not directly relevant to the phylogenetic placement of Cryptovaranoides among reptiles.

372: Fusion was not scored as present in [13], but as unknown (as "partial" uncertainty between states 0 and 1 [19:8]), and seemingly all three options were explored in [19].

We politely disagree with the reviewer; state 1 is scored in Whiteside et al. (2024).

377-383: Together with the partially fused NHMUK PV R37378 [13: Figure 4B, C; 19: 8], this is actually an argument that Cryptovaranoides is outside but close to Unidentata. The components of the astragalus fuse so early in extant amniotes that there is just a single ossification center in the already fused cartilage, but there are Carboniferous and Permian examples of astragali with sutures in the expected places; all of the animals in question (Diadectes, Hylonomus, captorhinids) seem to be close to but outside Amniota. (And yet, the astragalus has come undone in chamaeleons, indicating the components have not been lost.) - Also, if NHMUK PV R37378 doesn't belong to a squamate close to Unidentata, what does it belong to? Except in toothless beaks, premaxillary fusion is really rare; only molgin newts come to mind (and age, tooth size, and tooth number of NHMUK PV R37378 are wholly incompatible with a salamandrid).

The relevance of the astragalus is to the current discussion is unclear as we do not mention this element in our manuscript.  We discuss the fusion in the premaxillae in response to previous comment. 

471-474: That thing is concave. (The photo is good enough that you can enlarge it to 800% before it becomes too pixelated.) It could be a foramen filled with matrix; it does not look like a grain sticking to the outside of the bone. Also, spell out that you're talking about "suc.fo" in Figure 3j.

We are also a bit confused about this comment, as we state:

“Finally, we note here that Whiteside et al. [19] appear to have labeled a small piece of matrix attached to a coracoid that they refer to †C. microlanius as the supracoroacoid [sic] foramen in their figure 3, although this labeling is inferred because only “suc, supracoroacoid [sic]” is present in their figure 3 caption.” (L. 519-522, P. 17). We cannot verify that this structure is concave, as so we keep this text as is.

476-489: [19] conceded in their section 4.1 (pp. 11-12) that the atlas pleurocentrum, though fused to the dorsal surface of the axis intercentrum as usual for amniotes and diadectomorphs, was not fused to the axis pleurocentrum.

This is correct, as we note in the MS. The issue is whether these elements are clearly identifiable.

506-510: [19:12] did identify what they considered a possible ulnar patella, illustrated it (Figure 4d), scored it as unknown, and devoted the entire section 4.4 to it.<br /> 512-523: What I find most striking is that Whiteside et al., having just discovered a new taxon, feel so certain that this is the last one and any further material from that fissure must be referable to one of the species now known from there.

We agree with these points and believe we have devoted adequate text to addressing them. Note that the reviewer does not recommend any revisions to these sections.

553: Not that it matters, but I'm surprised you didn't use TNT 1.6; it came out in 2023 and is free like all earlier versions.

We have kept this as is following the reviewer comment, and because we were interested in replicating the analyses in the previous publications that have contributed to the debate about the identity of this taxon.  For the present simple analyses both versions should perform identically, as the search algorithms for discrete characters are identical across these versions.

562: Is "01" a typo, or do you mean "0 or 1"? In that case, rather write "0/1" or "{01}".

This has been corrected to {01}

(3) Comments on nomenclature and terminology

55, 56: Delete both "...".

This has been corrected.

100: "ent- and ectepicondylar"

For clarity, we have kept the full words.

107-108: I understand that "high" is proximal and "low" is distal, but what is "the distal surface" if it is not the articular surface in the elbow joint?

This has been corrected.

120: "stem pan-lepidosaurs, and stem pan-squamates"; Lepidosauria and Squamata are crown groups that don't contain their stems

This has been corrected.

122, 123: Italics for Claudiosaurus and Delorhynchus.

This has been corrected.

130: Insert a space before "Tianyusaurus" (it's there in the original), and I recommend de-italicizing the two genus names to keep the contrast (as you did in line 162).

This has been corrected.

130, 131: Replace both "..." by "[...]", though you can just delete the second one.

This has been corrected.

174: Not a capitulum, but a grammatically even smaller (double diminutive) capitellum.

This has been corrected.

209, 224, Table 1: Both teams have consistently been doing this wrong. It's "recessus scalae tympani". The scala tympani ("ladder/staircase of the [ear]drum") isn't the recess, it's what the recess is for; therefore, the recess is named "recess of the scala tympani", and because there was no word for "of" in Classical Latin ("de" meant "off" and "about"), the genitive case was the only option. (For the same reason, the term contains "tympani", the genitive of "tympanum".)

This has been corrected.

415-425: This is a terminological nightmare. Ribs can have (and I'm not sure this is exhaustive): a) two separate processes (capitulum, tuberculum) that each bear an articulating facet, and a notch in between; b) the same, but with a non-articulating web of bone connecting the processes; c) a single uninterrupted elongate (even angled) articulating facet that articulates with the sutured or fused dia- and parapophysis; d) a single round articulating facet. Certainly, a) is bicapitate and d) is unicapitate, but for b) and c) all bets are off as to how any particular researcher is going to call them. This is a known source of chaos in phylogenetic analyses. I recommend writing a sentence or three on how the terms "unicapitate" & "bicapitate" lack fixed meanings and have caused confusion throughout tetrapod phylogenetics, and that the condition seen in Cryptovaranoides is nonetheless identical to that in archosauromorphs.

This has been added: “This confusion in part stems from the lack of a fixed meaning for uni- and bicapitate rib heads; in any case, †C. microlanius possesses a condition identical to archosauromorphs as we have shown.”  (L.475-477, P.16).

439-440: Other than in archosaurs, some squamates and Mesosaurus, in which sauropsids are dorsal intercentra absent?

We are unclear about the relevance of the question to this section. The issue at hand is that some squamate lineages possess dorsal intercentra, so the absence of dorsal intercentra cannot be considered a squamate synapomorphy without the optimization of this feature along a phylogeny (which was not accomplished by Whiteside et al.).

458: prezygapophyses.

This has been corrected.

516: "[...]".

This has been corrected.

566: synapomorphies.

This has been corrected.

587: Macrocnemus.

This has been corrected.

585: I strongly recommend either taking off and nuking the name Reptilia from orbit (like Pisces) or using it the way it is defined in Phylonyms, namely as the crown group (a subset of Neodiapsida). Either would mean replacing "neodiapsid reptiles" with "neodiapsids".

This has been corrected to “neodiapsids.”

625: Replace "inclusive clades" by "included clades", "component clades", "subclades", or "parts," for example.

This has been kept as is because “inclusive clades” is common terminology and is used extensively in, for example, the PhyloCode. 

659: Please update.

References are updated.

Fig. 8: Typo in Puercosuchus.

This has been corrected.

(4) Comments on style and spelling

You inconsistently use the past and the present tense to describe [13, 19], sometimes both in the same sentence (e.g., lines 323 vs. 325). I recommend speaking of published papers in the past tense to avoid ascribing past views and acts to people in their present state.

This has been corrected to be more consistent throughout the manuscript.

48: Remove the second comma.

This has been corrected.

91: Replace "[13] and WEA24" by "[13, 19]".

This has been corrected.

100: Commas on both sides of "in fact" or on neither

This has been corrected.

117: I recommend "the interpretation in [19]". I have nothing against the abbreviation "WEA24", but you haven't defined it, and it seems like a remnant of incomplete editing. - That said, eLife does not impose a format on such things. If you prefer, you can just bring citation by author & year back; in that case, this kind of abbreviation would make perfect sense (though it should still be explicitly defined).<br /> 129, 145: Likewise.

We have modified this [13] and [19] where necessary.

192-198: Surely this should be made part of the paragraph in lines 158-175, which has the exact same headline?

This has been corrected.

200-206: Surely this should be made part of the paragraph in lines 148-156, which has the exact same headline?

These sections deal with different issues pertaining to the analyses of Whiteside et al. (2024) and so we have kept to organization as is.

214: Delete "that".

This has been deleted.

312: "Vomer" isn't an adjective; I'd write "main vomer body" or "vomer's main body" or "main body of the vomer".

This has been corrected.

350: "figured"

This has been corrected.

400: Rather, "rearticulated" or "worked to rearticulate"? - And why "several"? Just write "two". "Several" implies larger numbers.

These issues have been corrected.

448, 500: As which? As what kind of feature? I'm aware that "as such" is fairly widely used for "therefore", but it still confuses me every time, and I have to suspect I'm not the only one. I recommend "therefore" or "for this reason" if that is what you mean.

“As such” has been deleted.

452: Adobe Reader doesn't let me check, but I think you have two spaces after "of".

This has been corrected.

514, 539, 546, 552, 588, Fig. 3, 5, 6, Table 1: "WEA24" strikes again.

This has been corrected.

515: Remove the parentheses.

This has been corrected.

531: Insert a space after the period.

This has been corrected.

532: Remove both commas and the second "that".

This has been corrected.

538: Remove the comma.

This has been kept as is because changing it would render the sentence grammatically incorrect.

545: "[...]" or, better, nothing.

This has been corrected.

547: Spaces on both sides of the dash or on neither (as in line 553).

This has been corrected.

552: Rather, "conducted a parsimony analysis".

This has been corrected.

556: Space after "[19]".

This has been corrected.

560: Comma after "narrow".

This has been corrected.

600: Comma after "above" to match the one in the preceding line - there's an insertion in the sentence that must be flanked by commas on both sides.

This has been corrected.

603: Compound adjectives like "alpha-taxonomic" need a hyphen to avoid tripping readers up.

This has been corrected.

612: Similarly, "ancestral-state reconstruction" needs one to make immediately clear it isn't a state reconstruction that is ancestral but a reconstruction of ancestral states.

This has been corrected.

613: If you want to keep this comma, you need to match it with another after "Cryptovaranoides" in line 611.

We have kept this as is, because removing this comma would render the sentence grammatically incorrect.

615: Likewise, you need a comma after "and" because "except for a few features" is an insertion. The other comma is actually optional; it depends on how much emphasis you want to place on what comes after it.

this has been added.

622: Comma after "[48, 49]".

this has been added.

672: Missing italics and two missing spaces.

This has been corrected.

678, 680-681, 693, 700-701, 734, 742, 747, 788, 797, 799, 803, 808, 810-811, 814, 817, 820, 823, 828, 841, 843: Missing italics.

This has been corrected.

683, 689: These are book chapters. Cite them accordingly.

This has been corrected.

737: Missing DOI.

No DOI is available.

793: Missing Bolosaurus major; and I'd rather cite it as "2024" than "in press", and "online early" instead of "n/a".

This has been corrected.

835: Hoffstetter, RJ?

This has been corrected.

836: Is there something missing?

This has been corrected.

839: This is the same reference as number 20 (lines 683-684), and it is miscited in a different way...!

This has been corrected.

Reviewer #2 (Recommendations for the authors):

(1) There is a brief mention of a phylogenetic analysis being re-run, but it is unclear if any modifications (changes in scoring) based on the very observations were made. Please state this explicitly.

This is explained from lines 600-622, P.20-21, in the section “Apomorphic characters not empirically obtained.”  "In order to check the characters listed by Whiteside et al. [19] (p.19) as “two diagnostic characters” and “eight synapomorphies” in support of a squamate identity for †Cryptovaranoides, we conducted a parsimony analysis of the revised version of the dataset [32] provided by Whiteside et al. [19] in TNT v 1.5 [91]. We used Whiteside et al.’s [19] own data version"

(2) Line 20: There is almost no discussion of non‑lepidosaur lepidosauromorphs. I suggest including this, as the archosauromorph‑like features reported in Cryptovaranoides appear rather plastic. Furthermore, diagnostic features of Archosauromorpha in other datasets (e.g., Ezcurra 2016 or the works of Spiekman) are notably absent (and unsampled) in Cryptovaranoides. Expanding this comparison would greatly strengthen the manuscript.

The brief discussion (although not absent) of non-lepidosaur lepidosauromorphs is largely a function of the poor fossil record of this grade. But where necessary, we do discuss these taxa. Also see our previous study (Brownstein et al. 2023) for an extensive discussion of characters relevant to archosauromorphs.

(3) Line 38: I suggest removing "Archosauromorpha" from the keywords. The authors make a compelling case that Cryptovaranoides is not a squamate, yet they do not fully test its placement within Archosauromorpha (as they acknowledge). Perhaps use "Reptilia" instead?

We have removed this keyword.

(4) Line 99: The authors' points here are well made and largely valid. The presence of the ent‑ and ectepicondylar foramina is indeed an amniote plesiomorphy and cannot confirm a squamate identity. Their absence, however, can be informative - although it is unclear whether the CT scans of the humerus are of sufficient resolution, and Figure 4 of Brownstein et al. looks hastily reconstructed (perhaps owing to limited resolution). Moreover, the foramina illustrated by Whiteside do resemble those of other reptiles, albeit possibly over‑prepared and exaggerated.

The issue with the noted figure is indeed due to poor resolution from the scans. Although we agree with the reviewer, we hesitate to talk about absence in this taxon being phylogenetically informative given the confounding influence of ontogeny.

(5) I encourage the authors to provide slice data to support the claim that the foramina are absent (which could certainly be correct!); otherwise, the assertion remains unsubstantiated.

We only have access to the mesh files of segmented bones, not the raw (reconstructed slice) data.

(6) PLEASE NOTE - because the specimen is juvenile, the apparent absence of the ectepicondylar foramen is equivocal: the supinator process develops through ontogeny and encloses this foramen (see Buffa et al. 2025 on Thadeosaurus, for example).

See above.

(7) Line 122: Italicize 'Delorhynchus'

This has been corrected.

(8) Lines 131‑132: I'd suggest deleting the final sentence; it feels a little condescending, and your argument is already persuasive.

This has been corrected.

(9) Line 129: Please note that owenettid "parareptiles" also lack this process, as do several other stem‑saurians. Its absence is therefore not diagnostic of Squamata.<br /> Also: Such plasticity is common outside the crown. Milleropsis and Younginidae develop this process during ontogeny, even though a lower temporal bar never fully forms.

We appreciate this point. See discussion later in the manuscript.

(11) Line 172: Consider adding ontogeny alongside taphonomy and preservation. A juvenile would likely have a poorly developed radial condyle, if any. Acknowledging this possibility will add some needed nuance.

This sentence has been modified, but we have not added in discussion of ontogeny here because it is not immediately relevant to refuting the argument about inference of the presence of this feature when it is not preserved.

(12) Line 177: The "septomaxilla" in Whiteside et al. (2024, Figure 1C) resembles the contralateral premaxilla in dorsal view, with the maxillary process on the left and the palatal (or vomerine) process on the right (the dorsal process appears eroded). The foramen looks like a prepalatal foramen, common to many stem and crown reptiles. Consequently, scoring the septomaxilla as absent may be premature; this bone often ossifies late. In my experience with stem‑reptile aggregations, only one of several articulated individuals may ossify this element.

We agree that presence of a late-ossifying septomaxilla cannot be ruled out, but our point remains (and in agreement with Referee) that scoring the septomaxilla as present based on the amorphous fragments is premature.

(13) Line 200: Tomography data should be shown before citing it. The posterior margin of the maxilla appears rather straight, and the maxilla itself is tall for an archosauromorph. It would be more convincing to score this feature as present only after illustrating the relevant slices - and, as you note, the trait is widespread among non‑archosauromorphs.

See above and Brownstein et al. (2023).

(14) Line 208: Well argued: how could Whiteside et al. confidently assign a disarticulated element? Their "vagus" foramen actually resembles a standard hypoglossal foramen - identical to that seen in many stem reptiles, which often have one large and one small opening.

Thank you!

(15) Line 248: Again, please illustrate this region. One cannot argue for absence without showing the slice data. Note that millerettids and procolophonians - contemporaneous with Cryptovaranoides - possess an enclosed vidian canal, so the feature is broadly distributed.

See above.

(16) Line 258: The choanal fossa is intriguing: originally created for squamate matrices, yet present (to varying degrees) in nearly every reptile I have examined. It is strongly developed in millerettids (see Jenkins et al. 2025 on Milleropsis and Milleretta) and younginids, much like in squamates - Tiago appropriately scores it as present. Thus, it may be more of a "Neodiapsida + millerettids" character. In any case, the feature likely forms an ordered cline rather than a simple binary state.

We agree and look forward to future study of this feature.

(17) Line 283: Bolosaurids are not diapsids and, per Simões, myself, and others, "Diapsida" is probably invalid, at least how it is used here. Better to say "neodiapsids" for choristoderes and "stem‑reptiles" or "sauropsids" for bolosaurids. Jenkins et al.'s placement is largely a function of misidentifying the bolosaurid stapes as the opisthotic.

We are not entirely clear on this point since bolosaurids are not mentioned in this section.

(18) Line 298: Here, you note that the CT scans are rather coarse, which makes some earlier statements about absence/presence less certain (e.g., humeral foramina). It may strengthen the paper to make fewer definitive claims where resolution limits interpretation.

We appreciate this point. However, in the case of the humeral foramina the coarseness of the scans is one reason why we question Whiteside et al. scoring of the presence of these features.

(19) Line 314: Multiple rows of vomerine teeth are standard for amniotes; lepidosauromorphs such as Paliguana and Megachirella also exhibit them (though they may not have been segmented in the latter's description). Only a few groups (e.g., varanopids, some millerettids) have a single medial row.

We appreciate this point and have added in those citations into the following added sentence: “Multiple rows of vomerine teeth are common in reptiles outside of Squamata [76]; the presence of only one row is restricted to a handful of clades, including millerettids [77,78], †Tanystropheus [49], and some [79], but not all [71,80] choristoderes.” (L. 360-363, P. 12).

(20) Line 317: This is likely a reptile plesiomorphy - present in all millerettids (e.g., Milleropsis and Milleretta per Jenkins et al.). Citing these examples would clarify that it is not uniquely squamate. Could it be secondarily lost in archosauromorphs?

We appreciate this point and have cited Jenkins et al. here. It is out of the scope of this discussion to discuss the polarity of this feature relative to Archosauromorpha.

(21) Line 336: Unfortunately, a distinct quadratojugal facet is usually absent in Neodiapsids and millerettids; where present, the quadratojugal is reduced and simply overlaps the quadrate.

We appreciate this point but feel that reviewing the distribution of this feature across all reptiles is not relevant to the text noted.