Reviewer #2 (Public Review):

A major challenge to studying the ABC transporter dynamics "in situ" is the lack of precise measurement of various structural conformers that correspond to intermediate states during the ATP-catalytic or substrate-transport cycle. The use of the conformation-specific antibody 5D3 has recently enabled structural biology to experimentally visualize a specific structural fold that corresponds to an apo and inward-facing (IF) state of ABCG2. In this study, Gyöngy et al aimed to develop a mammalian cell-based system for ABCG2 to investigate how nucleotide or drug ligands regulate the transporter's alternating nature of its inward- and outward-facing (OF) conformations. The authors exploited the nature of 5D3 to only recognize ABCG2' IF conformers and combined flow cytometry and confocal microscopy to systematically analyze the IF-OF switches in the presence of nucleotides, drug substrates, ATPase inhibitors, and a known ABCG2 inhibitor Ko143. The authors find that nucleotide binding alone is sufficient to decrease the propensity of IF conformation, as well as to drive the high-to-low drug substrate transformation, and subsequently the drug-induced ATP hydrolysis resets the transporter to the IF and 5D3-bound state. These data provide solid cell-based evidence that adds to the ongoing discussion about the allosteric regulation of ABCG2 by both nucleotide and transport substrate ligands. Most importantly, the results support several lines of functional implications that could not be fully addressed by recent high-profile cryo-EM structures of ABCG2.

Strengths:<br /> The development of the methodology is a tour-de-force effort, as well as the biggest strength of this study. The results and the experimental protocol will likely provide a significant impact on how scientists design experiments to address structural questions without using large-scale purified systems and conventional structural biology approaches.

The validity of using the GFP-tagged ABCG2 variant is supported by several lines of functional characterizations, including protein expression, 5D3 reactivity, the responsiveness of nucleotide analogs, and mitoxantrone (MX) accumulation analysis. The application of such a strategy is particularly exemplified by the systematic treatment of various nucleotide analogs, which is sufficient to establish kinetic analysis by looking into apparent nucleotide affinities. These initial works add confidence in performing experiments by using either transport substrates or ABCG2 inhibitors. It is very compelling to see candidly the drug-coupled stimulation of ATP hydrolysis, given that both ATP and ADP decrease the 5D3-bound population.

The development of fluorescence correlation spectroscopy (FCS), the first in such a study, allows the measurement of colocalization of both fluorescence-labeled transporter and transport substrates. As illustrated by MX binding to ABCG2, the data supports the notion that nucleotide binding drives substrate release from the transporters, which in this case, can be explained by the high-to-low substrate binding affinity or IF-OF conformational switch of the transporter. Moreover, such an assay will be likely used as part of a high-throughput pipeline in search of therapeutic drugs against ABCG2.

Weakness:<br /> Although the paper presents solid and compelling cell-based evidence describing the relationship between structural changes of ABCG2 and ligand bindings, the enthusiasm is slightly dampened by the fact that this study seems mostly used to support the hypotheses that were proposed by recent cryo-EM structures. It is unclear from this study whether new insight into ABCG2's working mechanism can be proposed based on the data in this manuscript.

In addition, the IF-OF switch represents the transformation of two extreme conformations in ABCG2. The authors do not address the intermediate states, such as occluded conformers, in this study, which makes one wonder whether this is a limitation of the methodology presented in this manuscript. Moreover, interdomain crosstalk is highlighted in this manuscript to address the communication between NBD and TMD. However, it is not clear how the data could say anything about the crosstalk between NBD and TMD. For example, one limitation of recording 5D3 sensitivity on WT proteins may not allow us to pinpoint how structural motifs at the NBD-TMD interface (e.g., Q-loop, triple-helix bundle, polar relay, etc) transmit signals that cause IF-OF switch. The authors do not address how the strategy described here can address such a gap.

Lastly, the authors illustrate the conformational change of 5D3 epitope in the extracellular domain (ECD) by using atomic models in the presence and absence of the antibody. The dynamic information of how the ECD transforms to non-5D3 reactive is limited through this study; for instance, what the timing is to set loose of antibodies upon nucleotide binding, or to what degree of drug binding, IF starts to transit to OF under the physiological condition. Despite this, it is worth noting that 30% of MX-ABCG2 colocalization was still observed in untreated cells, perhaps suggesting a dynamic equilibrium between substrate-bound IF and other conformers.

Reviewer #3 (Public Review):

The goal of this study was to probe the transition from the IF to OF conformations inside the cells of a multidrug ABC transporter, ABCG2. In order to do so the authors used an antibody that specifically recognized the IF state (the epitope is 'disorganized' in the OF conformation) and this tool was particularly useful to address the conformational changes of ABCG2 that take place inside the permeabilized cells, depleted or not in ATP, and complemented with different combinations of nucleotides, drugs, and inhibitors. This technique was also used to show that the drugs increase the transition from the IF to the OF state.

By using confocal microscopy, the authors showed that ATP depletion led to a majority of ABCG2 that reside in a mitoxantrone-bound IF conformation.

The fluorescence correlation spectroscopy was another powerful approach used by the authors to convincingly demonstrate that the mitoxantrone drug could bind to ABCG2 in the IF conformation only.

Overall, the experiments are sound and the main conclusions drawn by the authors are very well supported by their data. This study unravels the first steps of the catalytic cycle of ABCG2 inside the cells, from drug-binding to a high-affinity site in the IF conformation to drug release from a low-affinity site in the OF conformation. It helps us to better understand how this transporter works in an environment that is physiologically relevant.

Reviewer #1 (Public Review):

Hafez and collaborators describe the construction and analysis of a computational model of a mushroom body neuron. The anatomy derives from a combination of electron microscopy reconstructions of MBON-α3 and also from light microscopy. The physiological parameters derive from publications that measured them, in addition to the author's own electrophysiological recordings with patch-clamp.

There are two main findings. First, the dendritic arbor of MBON-α3 is electrotonically compact, meaning, individual connections from Kenyon cells will similarly elicit action potentials independently as to where, spatially, the synapses lay on the arbor. Second, in simulation, exploration of changes in the strength of Kenyon cell inputs illustrate two possible ways to alter the strength of the KC-MBON physiological connection, showing that either could account for the observed synaptic depression in the establishment of associative memories. The properties of each approach differ.

Overall, the manuscript clearly describes the journey from connectomics and electrophysiology to computational modeling and exploration of the physiological properties of a circuit in simulation.

Reviewer #2 (Public Review):

"The cellular architecture of memory modules in Drosophila supports stochastic input integration" is a classical biophysical compartmental modelling study. It takes advantage of some simple current injection protocols in a massively complex mushroom body neuron called MBON-a3 and compartmental models that simulate the electrophysiological behaviour given a detailed description of the anatomical extent of its neurites.

This work is interesting in a number of ways:

- The input structure information comes from EM data (Kenyon cells) although this is not discussed much in the paper<br /> - The paper predicts a potentially novel normalization of the throughput of KC inputs at the level of the proximal dendrite and soma<br /> - It claims a new computational principle in dendrites, this didn't become very clear to me

Problems I see:

- The current injections did not last long enough to reach steady state (e.g. Figure 1FG), and the model current injection traces have two time constants but the data only one (Figure 2DF). This does not make me very confident in the results and conclusions.<br /> - The time constant in Table 1 is much shorter than in Figure 1FG?<br /> - Related to this, the capacitance values are very low maybe this can be explained by the model's wrong assumption of tau?<br /> - That latter in turn could be because of either space clamp issues in this hugely complex cell or bad model predictions due to incomplete reconstructions, bad match between morphology and electrophysiology (both are from different datasets?), or unknown ion channels that produce non-linear behaviour during the current injections.<br /> - The PRAXIS method in NEURON seems too ad hoc. Passive properties of a neuron should probably rather be explored in parameter scans.

Questions I have:

- Computational aspects were previously addressed by e.g. Larry Abbott and Gilles Laurent (sparse coding), how do the findings here distinguish themselves from this work<br /> - What is valence information?<br /> - It seems that Martin Nawrot's work would be relevant to this work<br /> - Compactification and democratization could be related to other work like Otopalik et al 2017 eLife but also passive normalization. The equal efficiency in line 427 reminds me of dendritic/synaptic democracy and dendritic constancy<br /> - The morphology does not obviously seem compact, how unusual would it be that such a complex dendrite is so compact?<br /> - What were the advantages of using the EM circuit?<br /> - Isn't Fig 4E rather trivial if the cell is compact?

Overall, I am worried that the passive modelling study of the MBON-a3 does not provide enough evidence to explain the electrophysiological behaviour of the cell and to make accurate predictions of the cell's responses to a variety of stochastic KC inputs.

Reviewer #3 (Public Review):

This manuscript presents an analysis of the cellular integration properties of a specific mushroom body output neuron, MBON-α3, using a combination of patch clamp recordings and data from electron microscopy. The study demonstrates that the neuron is electrotonically compact permitting linear integration of synaptic input from Kenyon cells that represent odor identity.

Strengths of the manuscript:

1) The study integrates morphological data about MBON-α3 along with parameters derived from electrophysiological measurements to build a detailed model.<br /> 2) The modeling provides support for existing models of how olfactory memory is related to integration at the MBON.

Weaknesses of the manuscript:

1) The study does not provide experimental validation of the results of the computational model.<br /> 2) The conclusion of the modeling analysis is that the neuron integrates synaptic inputs almost completely linearly. All the subsequent analyses are straightforward consequences of this result.<br /> 3) The manuscript does not provide much explanation or intuition as to why this linear conclusion holds.

In general, there is a clear takeaway here, which is that the dendritic tree of MBON-α3 in the lobes is highly electrotonically compact. The authors did not provide much explanation as to why this is, and the paper would benefit from a clearer conclusion. Furthermore, I found the results of Figures 4 and 5 rather straightforward given this previous observation. I am sceptical about whether the tiny variations in, e.g. Figs. 3I and 5F-H, are meaningful biologically.

Reviewer #1 (Public Review):

The authors present a retrospective study of COVID-19 mortality within 30 days from a positive SARS-CoV-2 PCR in 1115 patients with cancer and 2851 patients without cancer. Patients were recruited from 16 different centres from 8 countries across 5 continents. Patients were recruited between January and November 2020. All patients with a positive SARS-CoV-2 PCR were included. Demographic and clinical data were collected from electronic patient records. The primary outcome was 30-day mortality. Data were retrieved from patient records and there is a significant proportion of missing data.

The authors found that age and the presence of cancer were independent risk factors of 30-day mortality. Remdesivir was associated with reduced mortality. Within cancer patients, those with haematological malignancies and lung cancer had the highest risk. Overall, the findings of this study are in line with previously published results and don't provide major new insights.

Strength:

This is a multicentric study across several countries including over 3000 patients.

Limitations<br /> 1) This is not the first cohort study in cancer patients, several large studies have addressed risk factors of mortality before (for example Kuderer et al., The Lancet, 2020 and Chaves-McGregor et al. JAMA Oncology, 2021).

2) The authors identify Remdesivir to reduce mortality in cancer patients and those without cancer. The efficacy of Remdesivir has been addressed in large prospective trials, albeit not in cancer patients.

3) Treatment of patients with COVID-19 likely varied by country but the authors haven't addressed the impact of this.

4) Given that the recruited patients were all unvaccinated, the results are likely not completely transferable to the current situation. Vaccination and current antivirals and monoclonal antibodies have reduced the risk of severe disease and death. The current omicron variant has different properties compared to earlier strains. In fact, studies have shown that mortality in cancer patients has improved since 2020 (OnCovid Study Group, JAMA Oncology, 2021).

In conclusion, the authors largely confirm findings from other studies that patients with cancer were at an increased risk of death after COVID-19 infection, especially early on in the pandemic.

Reviewer #2 (Public Review):

The paper entitled "International Multicenter Study Comparing Cancer to Non-Cancer Patients with COVID-19: Impact of Risk Factors and Treatment Modalities on Survivorship" by Raad et al. is a multi-center, international, matched cohort, with a relatively large sample size. The aim of this work is to determine independent risk factors that impact survival in the setting of "novel treatment modalities" like Remdesivir. It enrolled patients with COVID-19 and cancer and compared them to cancer-negative controls. The authors conclude that cancer increases mortality from COVID-19 and that Remdesivir can reduce all-cause mortality in a subset of patients receiving low-flow oxygen and the results support their conclusions. Overall, this paper adds to the growing body of literature that implicates cancer as a worse predictor of survival among patients with COVID-19. The use of a matched cohort makes it unique and strengthens the findings of this study. The potential weaknesses of this study are its retrospective nature and lack of data on the effect of vaccination in this population since the study was conducted prior to the introduction of vaccines.

Reviewer #1 (Public Review):

The authors aim was to determine the role of initial procalcitonin (PCT) measurements in cancer patients admitted with COVID-19 infection in reducing the intensity and duration of empiric antibiotic therapy. This was a retrospective study of all patients admitted to a single cancer center with COVID-19 infection and at least one PCT test within 72 hours of admission. The cut off PCT value to divide patients into two groups was 0.25 ng/ml (those with >= 0.25 ng/ml having a higher suspicion of bacterial infection). The study found that compared to patients with low PCT levels had shorter hospital stays, lower rate of mortality, and received less antibiotic therapy. The paper is well written, the study methods and statistics are sound, the population well characterized and large enough for valid comparison, and the results support the authors conclusions. The study provides support that PCT can be used in this special at risk population as it has been used in other COVID-19 patient populations that have been better studied. The study has limitations that the authors report: retrospective, single center study, bacterial infections may have been missed (no uniformity of cultures collected), and empiric antimicrobial therapy was at the discretion of the treating team (no standardized empiric therapy). The findings of this study may not be generalizable to other cancer patient populations and there may be other confounding variables not identified.

Reviewer #2 (Public Review):

This is a well written manuscript that provides a useful analysis on using PCT for guiding antibiotics use among cancer patients with COVID19, a very common issue with COVID 19 patients in general but more challenging in the cancer population. Analysis is relatively straight forward. Results support the claim that antibiotics for more than 72 hours may be unnecessary for cancer patients with negative cultures and PCT<0.25. This is can be useful clinically to limit unnecessary antibiotic use, however would only apply this as a very broad generalization. It would be interesting to see what outcomes are and if this applies for more specific and challenging but not uncommon clinical scenarios with cancer patients (i.e neutropenic patients, patients undergoing active therapy, ICU admission) where clinicians may favor longer use of antibiotics.

Reviewer #3 (Public Review):

In this retrospective study, the authors intend to demonstrate the utility of serum procalcitonin in reducing the use of antibacteral agents in cancer patients with COVID-19, by identifying the subset of their highly immunocompromised population where early discontinuation of antibacterial therapy would not be harmful.

This study has a large population size > 500 patients over the span of 16 months. The groups with low procalcitonin and high procalcitonin have similar baseline characteristics, which makes the subsequent comparisons valid and relevant. The authors have considered all the relevant variables that could affect the outcomes being studied, and used sound statistical methods.

This study has some limitations. It is retrospective by nature, with possibility for confounders. In addition to the limitations mentioned by the authors, the study spans the period from March 2020 to June 2021 through which our knowledge of COVID has evolved, multiple variants have emerged, immunization has become available in the later part of the study period, more therapies (antivirals, monoclonal antibodies) became available, all of which have definitely affected COVID-related mortality, and could be an important confounder here. While the authors report the level of severity of the infection, using proxies such as supplemental oxygen and ICU admission, the use of COVID-directed therapies, including immunosuppressants such as steroids and tocilizumab (which in turn can increase the risk of bacterial infections and decrease the risk of mortality) is not reported. It also seems that the management of antibacterial therapy was left at the discretion of the treating physician, which can lead to a wide variety of practices, the nature of antibacterials administered is not reported here.

The results presented here support the conclusions made by the authors, and one has to appreciate the difficulty of antimicrobial stewardship efforts in an immunocompromised population such as the one being studied here. Many of these patients have been immunosuppressed for prolonged periods of time, could have profound defects in their immune systems, and could have had multiple previous infections, sometimes with atypical presentations. These patients are typically excluded from most large clinical trials, thus retrospective studies such as this one are usually the most informative pieces of literature available to support evidence-based medicine in this special patient population. I think this study should encourage clinicians to consider the use of serum procalcitonin as one additional clue to support their pursuit of antibacterial de-escalation or discontinuation in cancer patients with COVID-19.

Reviewer #1 (Public Review):

Diehl & Redish set out to capture how cognitive and behavioral linked activity varies along the medial wall of the rodent prefrontal cortex during a complex decision-making task. They found four clusters of cells along the dorsal-ventral axis that were firing more similarly to other cells in the same cluster than cells in other clusters, suggesting there are 4 distinct subdivisions in rodent mPFC. Their detailed analysis of decision-making, reward, and evaluation showed that though some cells in each area responded to these different cognitive aspects, there was a difference in how widespread these signals were in the different subdivisions. They found more decision-related activity in the ACC, more post-decision evaluative activity in the dorsal parts of the prelimbic, and more ventral areas involved with motivational factors. They argue that the prelimbic area is actually 2 distinct areas that should be considered separately. This paper is very well analyzed and the methodological aspects regarding histological confirmation and neuronal spiking are exceptionally thorough. The task is well-studied and conclusively provides insights into multiple facets of high-level cognition. The main weakness is the unequal distribution of cells recorded in each area. Mainly, this is a problem for the ACC where substantially fewer units were recorded. This takes away some from the interpretation of ACC activity, however, most of the findings about ACC are consistent with previous reports from this lab and others. This does not take away from the success the authors achieved in characterizing the differences and similarities in functional correlates along the medial wall. The identification of two distinct subdivisions in the prelimbic area is novel and is likely to have a substantial impact on the field. At the least, the specific location within prelimbic that future studies purport to either record from, sample from, or manipulate will need to be reported so that these future findings can be correctly interpreted. This is a major shift in the field's conceptualization of this oft-studied part of the brain.

Reviewer #2 (Public Review):

In this paper, Diehl and Redish recorded simultaneously from multiple medial frontal cortical regions while rats are performing a restaurant-row task. Their results provide insights into how neurons in the different regions may represent different aspects of the decision-making process.

The strength of the study is the experimental design. The restaurant-row task is an excellent and rich paradigm for evaluating decision-making, with specific unique components that may be relatable to economic subjective choices. The other strength is the electrophysiological approach, which enables the author to simultaneously record from multiple medial frontal cortical regions. This leads to a large data set of >3,000 single units recorded during behavior. A weakness of this study is the insistence to dissect the results and assign each region to specific behaviors, while the data seem to suggest that similar signals can be observed across multiple regions, albeit to different degrees. The framework of distributed vs. gradient vs. subregions seems like a strawman idea that does not help with the interpretation of the results, whereas the actual data are already quite rich and interesting.

Reviewer #3 (Public Review):

This is an interesting study in which the authors record simultaneously from neurons along the medial bank of the rodent PFC as rats perform the restaurant row task, an economic decision-making task in which subjects are offered different reward types with a specified delay, and they need to decide whether to accept or reject the offer. The authors find functional correlates of anatomical subdivisions of the mPFC; interestingly, they find that PL perhaps should be subdivided into dorsal and ventral subregions, a finding that is consistent with some known anatomical features. They characterize the task-related responses of neurons in these different subdivisions and find that in general, the dorsal regions (ACC, dPL) encode decision-related variables, whereas the ventral regions (vPL, IL) encode more motivational variables, such as the trial number in the session and the amount of lingering time.

Strengths:<br /> - The observed dichotomy between decisional and motivational factors mapping onto dorsal and ventral aspects of mPFC is interesting and, as far as I am aware, novel.<br /> - There are a number of rich, interesting observations, such as a lack of encoding of the reward delay in the offer zone, but then encoding of that variable in the wait zone (in all areas except ACC). This is intriguing given that their previous work has suggested that the decisions made in the offer and wait zones are in some ways dissociable, implying that they might rely on distinct neural circuits.<br /> - Overall, the data and analyses are of high quality, and the results are interesting.<br /> - The finding that PL should be subdivided into two distinct subregions will be of broad interest to researchers studying the mPFC. The approach and finding will also be of interest to the growing number of groups using linear silicon (including Neuropixels) probes to record from multiple brain areas simultaneously.

Weaknesses:<br /> - The authors find that dorsal regions of mPFC, particularly ACC, encode the upcoming decision of the animal. However, the upcoming choice will be correlated with animal movements (as is often the case). Given that ACC is adjacent to the motor cortex, and more posterior parts of the cingulate have been documented to reflect particular types of movements, it would be helpful to know if these signals would be observed for movements outside of the task, or if they really reflect the upcoming decision in this behavioral context.<br /> - I think some of the statistical analyses can be strengthened. For instance, the authors correlate neural activity against a large number of behavioral variables, some of which are correlated with each other. I would encourage a regression-based approach, which takes into account the correlations between variables for error bars/significance tests for each regressor.

In general, I think the authors' claims about their data are justified.

Public Review:

In this article, the authors have taken up the substantial task of combing through thousands of published meta-analyses and systematic reviews, with the goal of identifying the subset that specifically seeks to measure the association between elapsed time ("lag-time") in various milestones of cancer diagnosis or treatment (e.g. time elapse from symptom onset to first seen by primary care physician) and cancer outcomes. Within this subset, they have identified and summarized the findings on how these lag times are related to certain cancer outcomes. For example, how much does a delay in the start of adjuvant chemotherapy after surgery for breast cancer increase the mortality rate for these patients? The overarching goal of this work is to characterize the pre-Covid-19 landscape of these relationships and thereby provide a basis for studying what impact the pandemic had on worsened outcomes for cancer patients due to treatment delays. The authors have done an excellent job in their review of systematic reviews and meta-analyses, both describing their methodology well and interpreting their findings. The immediate connection to the Covid-19 pandemic is somewhat tenuous and primarily left to the reader to determine.

Reviewer #1 (Public Review):

This paper shows how evolutionary dynamics, together with high variance species-species interactions in a generalized Lotka-Volterra framework, can stabilize the population and delay extinctions. Moreover, the stable regime is shown to correspond to the clonal interference regime from population dynamics. Thus, this work extends Robert May's seminal work on the stability of a complex system by considering the stabilizing effect of evolution.

Strengths:

- The paper is well written, the questions well-motivated and the ideas presented in a coherent and easy to understand manner. Prior literature was referenced to a sufficient degree (though of course a lot was left out). Importantly, the author is honest about the limitations of the modeling choices, not attempting to over-sell the work or to hide inconvenient details. In this sense, this paper is a good contribution to the literature since it gives the reader a clear perspective on an interesting question.

- Kudos for sharing the code in github. The code looks organized and easy to reuse.

Weaknesses:

- Interactions are assumed to be drawn from a log-normal distribution. Clearly, this does not capture true ecological interactions. It is unclear how applicable the results are to real ecosystems.

- The paper assumes saturating nutrients and states that they "do not expect that the addition of a reasonable carrying capacity will change our qualitative results". However, competition for resources can lead to loss of diversity. Moreover, ecological systems are known to respond to large changes in the carrying capacity. Therefore, it should be further elucidated if indeed the addition of a carrying capacity will destabilize the results. Especially since there appears a significant increase in the population size in the stable conditions: an increase that is not clear if it could be supported when the carrying capacity was already limiting population sizes before the increase.

- It appears in the text that "there are key differences between the model and actual bacteria-phage systems, and the model should not be interpreted as one that will directly map onto a biological scenario". I agree with this statement. However, by distancing the model from biological scenarios it makes its predictions hard to validate in a real system, leaving us with no obvious way to infer how to apply its conclusions. Indeed, both explicit examples given in lines 125-130: phase-bacteria and T-cell-antigen are not quite captured by modeling choices. I would have much preferred a specific biological system fixed in mind, then minimally modeled in a way that there is hope to directly link the modeling results to experiments. Especially since there is a wealth of available microbial population data, as well as much being generated.

- As stated, "the population fitness distribution is never able to 'settle'..." is indicative of the driven nature (driven by strong noise) of the quasi steady state as opposed to a stability that arises from the system dynamics.

Justification of claims and conclusions:

The paper is honest in reflecting the weaknesses (stated above) in the modeling generality and applicability on actual systems. This is commendable, and the claims as stated are justified but the applicability of these claims remains unclear. There are some conjectures raised in the discussion but they remain unsupported and allocated to "future work".

Reviewer #2 (Public Review):

This work by Martis illustrates, in a predator-prey or parasite-host eco-evolutionary context, the classical idea of bet hedging or biological insurance: where a single population would fluctuate and perhaps risk extinction, summing over multiple sub-populations with asynchronous dynamics (some going up while others go down) allows a stabler total abundance.

Here the sub-populations are various genotypes of one predator and one prey species, fluctuations are due to their ecological interactions, their dynamics are more asynchronous when predation is more specialized (i.e. the various predator genotypes differ more in which prey types they can eat), and mutations allow the regeneration of genotypes that have gone extinct, thus ensuring that the diversity of subpopulations is not lost (corresponding to a "clonal interference" regime with multiple coexisting genotypes).

While the general idea of bet hedging has been explored in many settings, the devil is usually in the details: for instance, sub-populations should be connected enough to allow the rescue of those going extinct, but a too strong connection would simply synchronize their temporal dynamics and lose the benefit of bet hedging. In some cases, connections between sub-populations could even be destabilizing (e.g. Turing instabilities in space).

In a recent surge of physics-inspired many-species theories, where fluctuations arise from ecological dynamics, these details are notably starting to be understood in the case of spatial bet hedging, i.e. genetically identical subpopulations in multiple patches connected by migration (see e.g. Roy et al PLoS Comp Bio 2020 or Pierce et al PNAS 2020).

In the non-spatial eco-evolutionary setting considered here, the connecting flux is one of mutations rather than migrations, and a predator genotype can in principle interact with all prey genotypes (whereas in usual spatialized models, interactions cannot occur between different patches). Another possibly important detail here is that similar genotypes do not have similar interaction phenotypes, meaning there is no risk of evolution being confined in a neighborhood of similar phenotypes. According to the author and my own cursory exploration of the relevant eco-evo literature (with which I am less familiar than pure ecology), this setting has yet to see many developments in the spirit of the many-species theories mentioned above.

These differences make this new inquiry worthwhile and I applaud the author for undertaking it. From a theoretical perspective, three results emerging from the simulations stand out in this article as potentially very interesting:<br /> - rather sharp transitions in extinction probability and strain diversity as mutation flux and predator specialization increase.<br /> - how mutation rate and interaction strength combine, notably in power-law expressions for total population abundance<br /> - the discussion of susceptibilities, i.e. how predator and prey populations respond to perturbations, as a key ingredient in understanding the previous results, in particular with counter-intuitive negative susceptibilities indicating positive feedback loops.

It is a bit unfortunate that these more novel points are only briefly explored in the main text: while they are more developed in appendices, these arguments are not always as complete, polished and distilled as they might have been in a main text, so an article focusing entirely on explaining them deeply and intuitively would have been far more exciting to me.

Finally, I will note that I am not convinced by the framing of the current manuscript as a counterpoint to Robert May's idea of destabilizing diversity - in many ways I think this is a less relevant context than that of bet hedging, and it does a worse job at showcasing what is genuinely interesting and original here; I would thus encourage readers to read this paper in the framing I propose above.

Reviewer #1 (Public Review):

This is a well-written report on one of the biggest killer diseases. The report is based on a large longitudinal cohort and uses solid analytical methodologies. Three main valuable findings are reported: association between coronary heart disease (CHD) and a polygenic risk score (PRS), a combination of multiple traditional risk factors (SCORE2), and history of Fusobacterium nucleatum infection. While the first 2 associations are not novel, they are welcome independent replications of previous findings in a novel design. A putative role of F. nucleatum and other infections in increasing CHD risk has also been reported before but remains more elusive with some suggestion that they may increase CHD risk by promoting arterial inflammation. The strength of this study is to demonstrate an independent role of this bacterium after controlling inflammation markers as well as other risk factors in a prospective study. If this finding can be confirmed, the prevalence of the bacterium (15% in the cohort) means it should be considered as another serious CHD risk factor. The authors should discuss the implications of multiple testing.

Reviewer #2 (Public Review):

Coronary heart disease (CHD) is a major form of cardiovascular disease, the first cause of mortality in the world. The etiology of CHD is multifactorial and polygenic, with atherosclerosis as the main cause of coronary stenosis and ischemic events. Evidence from basic research in small animal models and clinical trials aiming at lowering proinflammatory cytokines such as IL-1β implicated low-grade inflammation in atherosclerosis pathogenesis. Genetic studies in humans report large numbers of risk variants and genes related to macrophages, monocytes and T-lymphocytes biology supporting further immunity and inflammation in CHD risk. The study conducted by this group reports the results of the investigation of the potential association between 22 persistent or frequently recurring pathogen infections with the risk of CHD in a CoLaus|PsyCoLaus study, a prospective population-based and urban cohort of European ancestry. The authors accessed data over 12 years and assessed the association between traditional risk factors through the SCORE2 estimation of risk, a recently validated score specifically developed to predict cardiovascular risk in European countries, genetic risk scores based on GWAS findings, and seropositivity to infections. They were able to confirm the utility of the application of SCORE2 in their population, and report a significant association of the genetic score with the incidence of CHD, both traditional and genetic factors being independent predictors, which was expected. An intriguing result regards reported seropositivity with F. nucleatum to significantly predict incident CHD. This a commensal bacterium that belongs to the normal oral microbiome reported playing an important role in the development and progression of gingivitis (gum inflammation) and periodontitis (infection of the gums). There are several existing lines of evidence connecting oral infections as an independent risk factor for CHD. The results reported by this study provide support for the increased risk of CHD in oral infection through seropositivity to F. nucleatum. However, the direct clinical implications, through the recommendation to search for prior infection with this bacterium as a predicting biomarker of this disease are not on the clinical application agenda yet. The infection seropositivity was measured only at the beginning of the study, with no information on how the oral seropositivity to this pathogen may have evolved over time. And given the novelty of this association, these results need to be replicated in independent cohorts with similar designs (prospective cohorts) before recommendation to screen for seropositivity could be recommended.

Reviewer #1 (Public Review):

In 2007 it was observed that, although the central elements of galactose utilization are similar in both S. cerevisiae and C. albicans (clustered metabolic genes, transcriptional induction in the presence of galactose) the induction mechanisms were different. Until now, however, although the way the presence of galactose was sensed and this information transmitted to the induction of gene expression was well understood in S. cerevisiae, it was quite mysterious in C. albicans. This work proposes that in C. albicans, the general transcription regulator Rep1 serves as a direct galactose binding protein and that the binding of galactose to Rep1 allows it to serve as a scaffold to collect the transcriptional machinery necessary to induce the elements of the Gal regulon.

The first line of evidence for the Rep1 scaffold model is the observation that Rep1 is needed for C. albicans to both grow on galactose and to induce the genes encoding the galactose processing proteins Gal1 and Gal10. Previous candidate regulators Rtg1 and Rtg3 only blocked growth on galactose in the presence of Antimycin A, so Rep1 represents a first element specifically required for galactose growth. Further analysis of Rep1 function involved the observation that Rep1 was a member of the family of transcription factors including Ntd80, a TF that has been implicated in a variety of cellular controls. The authors investigated a specific unique domain of Rep1 by moving it to Ndt80 - the fusion protein did not allow complementation of the galactose growth defect, suggesting this domain was not critical to the Rep1 involvement in galactose growth. Further analysis of Rep1 domains by deletions showed that removal of the putative transcriptional activation domain of the protein also did not block either growth on galactose medium or galactose-mediated induction of GAL1 and GAL10 expression. The Rep1 protein was found to be constitutively bound to the promoters of GAL1 and GAL10, and not really influenced in this binding by carbon source.

To attempt to determine the connection between the apparent constitutive binding and the galactose-mediated induction of gene expression the authors investigated the relationship between sugars and the Rep1 protein. Modelling suggested a possible galactose binding pocket, binding was shown biochemically, and mutations within the presumed binding site disrupted galactose binding and protein function.

The authors next assess how the binding of galactose to Rep1 leads to gene induction because the binding to the regulated promoters seems constitutive, and the activation domain seems unimportant for protein function, and in fact, doesn't act as an activation domain in a 1 hybrid assay. They speculate protein binding and search for interacting proteins by mass spec after IP with a tagged Rep1 protein in the presence of galactose. Orf19.4959 is identified and tested. The binding data is presented as a supplementary table and includes many hits that do not appear promising candidates. Inactivation of the TF Orf19.4959 blocks growth on galactose and induction of the GAL1 and GAL10 genes, and the protein, called Cga1, does have transactivating ability in a 1 hybrid assay. The authors thus propose that galactose binding to Rep1 facilitates the binding of Cga1 and leads to the activation of gene expression for galactose metabolism.

This model is tested by immunoprecipitation assays that showed Cga1-Rep1 interaction only in the presence of galactose, and that DNA association of Cga1 to GAL promoters was galactose and Rep1 dependent. Further experiments provide a framework for Rep1 function in other pathways and suggest a candidate polyA binding motif for the Rep1 protein. The generalization of the model is proposed by noting a pattern of Rep1/Cga1 presence in other fungal species.

Reviewer #2 (Public Review):

Despite previous studies, regulation of the genes required for galactose metabolism in Candida albicans has remained murky. For example, previous work had highlighted Rtg1 and Rtg3 as the key regulator components, an interesting finding given that these factors are important for glucose not galactose regulation in S. cerevisiae. As galactose metabolism is one of the best-understood regulatory systems, the evolutionary difference in the regulation of the galactose response has the potential to teach us about regulatory evolution in general.

The authors initially sought to understand how GlcNAc signaling cross-reacted with galactose gene induction, but they quickly discovered that Reg1, the mediator of GlcNAc signaling was essential for galactose metabolism while Rtg1 and Rtg3 were not. Overturning previous work requires strong evidence, and the authors deliver with a series of growth assays, qPCR, and chromatin IP in wt and mutant backgrounds.

The authors go further to demonstrate that the factor itself interacts with galactose based on isothermal calorimetry (although it would be nice to have seen if this was specific to galactose over glucose or GlcNAc). They show glucose regulation occurs by Reg1 recruiting Cga1 in a manner independent of the activation domain of Reg1 (immunoprecipitation, reporter assays, and chromatin IP). In contrast, Reg1 activation mediated by GlcNAc requires Reg1's activation domain and employs Rgs1. Once again, the experimental evidence for these two regulatory mechanisms is strong.

Evolutionary analysis shows that this specific instantiation of this mechanism, the combination of Reg1 and Cga1, is probably restricted to the CTG clade. While the paper does not explain how these regulatory changes happen, it sets the foundation for future work to tease this apart; work that before this paper would have been unlikely to have been successful.

Reviewer #1 (Public Review):

Oxidation of some KCNQ7 channels enhances channel activity. The manuscript by Nuñez and coauthors concluded that oxidation in the S2S3 linker of these channels disrupted the interaction between S2S3 and CaM EF-hand 3 (EF3). This mechanism is Ca2+-dependent. The apo EF3 no longer interacted with S2S3, and H2O2 no longer activated the channel. Electrophysiological recordings and fluorescence and NMR measurements of CaM with isolated helices A and B (CRD) and S2S3 of the channel were performed. While the results were in general clear with good quality, how the results support the conclusion was not clearly described. The approach using isolated molecular components in the study needs further validation since some of the results seem to show major conflicts with the results and mechanisms proposed in previous studies.

1) Previous studies showed differential responses of Kv7 channels to oxidation; Kv7.2, 4, and 5 are sensitive to oxidation regulation but Kv7.1 and 3 do not change upon H2O2 treatment. These differences were attributed at least partially to the sequence differences in S2S3 among Kv7 channels (ref 10 of this manuscript). The results in this manuscript show some major differences from the previous study. First, in all experiments, no difference was observed among Kv7 channels. Second, in Fig 3-6, S2S3 from KCNQ1 was used. The rationale for using KCNQ1 S2S3 and the interpretation of results is not justified considering that KCNQ1 S2S3 has fewer Cys residues and was least affected by oxidation in the previous study.

2) In Fig 6, oxidation of S2S3 leads to a reduction of S2S3-CaM interaction, which leads to an increase of currents (Fig 1C). In Fig 4, Ca2+ loading leads to a reduced S2S3-CaM (EF3) interaction, which should also lead to an increase of currents based on Fig 6 conclusions. However, it is the EF3 mutation (destroying Ca2+ binding) that leads to the current increase (Fig 1B), contradictory to what Fig 6 data suggested.

Reviewer #2 (Public Review):

The study by Nunez et al. builds upon structural work from the MacKinnon lab and the authors' labs to characterize how Ca2+, via calmodulin, interacts with Kv7 channels to mediate redox sensitivity. Using FRET experiments to support electrophysiology, the authors demonstrate an interaction defined by calmodulin, the helixA-helixB fork, and the S2-S3 linker. The experiments are well performed and the conclusions drawn are appropriate. These experiments help further define the redox signaling for Kv7 channels. A weakness is that the model in Figure 7 seems speculative, as the data provided do not appear to explain how the VSD is engaged/disengaged from the pore. Rather, most of the data concentrate on biochemical interactions and structural interpretations (via FRET signals, etc.) of conformational changes in the presence of calcium. Further, the model as presented is not informative. The illustrations do not demonstrate successfully what the authors wish to claim, and the illustrations/models are not sufficiently supported by the data presented.

Reviewer #3 (Public Review):

KV7 channels play an important role in setting the resting membrane potential of neurons. As such, modulation by reactive oxygen species is an important and physiologically relevant form of channel regulation. Here, the authors propose a mechanism for this modulation in which ROS disrupts the interaction between the S2S3 loop of the channel and CaM, resulting in an overall enhancement of channel activity. The authors propose that this S2S3/CaM interaction is selectively mediated through CaM EF3, and is dependent on Ca2+. The results are supported by patch-clamp data, as well as NMR measurements and a FRET-based binding assay. The paper contains a considerable amount of data that point towards the conclusion.

The authors conclude that the EF3 of CaM is 'by itself sufficient and necessary for the oxidative response of KV7 channel complex and for gating the calcium responsive domain of KV7 channels." This is a very strong conclusion, and while much of the data points towards an important role for EF3, it is difficult to conclude that it is sufficient and necessary. The sparse description of the experiments makes the interpretation of the results a bit challenging. Based on the description provided, some of the results appear contradictory, limiting the conclusions drawn by the authors.

Reviewer #1 (Public Review):

Observations made on histological patterns of SCC tumor invasion prompt the authors to investigate the seemingly broad distribution of invasion strategies employed by SCC tumor cells in tissue. Using computational modelling and testing the arising predictions in two experimental models of SCC invasion, the authors conclude matrix proteolysis and cell-cell junctions to play key roles in determining invasion strand width and cell adhesion strength to be a minor contributor.

Strengths of the study:<br /> - The authors acknowledge the complexity of invasion patterns employed by SCC tumor cells in tissue and provide new insight into the underlying complex cellular processes.<br /> - The approach of combining computational simulations and testing their predictions experimentally with two models is powerful.

Weaknesses of the study:<br /> - Cell proliferation (affected by proteolysis and cell-cell junctions) is indicated as a key contributor to the generation of broad strand invasion. However, proliferation is not investigated using the same experimental models used to investigate invasion and is not included as a parameter in the computational models.<br /> - The outcomes of their KO strategies on the cell-matrix and cell-cell adhesion are not fully demonstrated.

Reviewer #2 (Public Review):

Kato, Jenkins, et al. investigates cell-intrinsic and environmental determinants of diverse modes of collective cancer cell invasion in mucosal squamous cell carcinoma (muSCC). To explore this large parameter space, the authors develop a Cellular Potts model recapitulating two distinct in vitro muSCC - cancer-associated fibroblast (CAF) co-culture models: an organotypic platform containing an air/extracellular matrix (ECM) interface and a spheroid model mimicking dermal invasion and confinement by 3D ECM. Integrating between in silico predictions and quantitative assessment of the two experimental platforms, the authors make several interesting observations regarding determinants of the mode of collective SCC invasion. Of these, the most significant include the ability of SCCs to invade with deletion of β1 integrin in their organotypic model although invasion phenotype is altered, and identification of a synergistic dependence on cell-cell adhesion and matrix proteolysis for controlling strand width and growth within the invading cohort. Cell-cell adhesions are essential for maintaining supracellular actomyosin coupling to coordinate the invading cohort, while matrix proteolysis is necessary for creating physical space that supports both invasion and cell growth within confined space.

Overall, despite some concerns regarding support for specific claims, alternative considerations, and clarity in presentation, this study is rigorous and of high quality, and should serve as an important technical and conceptual resource that provides new insight into multicellular coordination in SCC invasion. More broadly, it illustrates the utility of coupling computational models with advanced 3D cell culture platforms to parse multifactorial control over complex forms of tissue morphogenesis.

Reviewer #3 (Public Review):

This study by Kato et. al used a combination of computational modeling, in vitro experimentation, and confirmatory in vivo mouse work to define what influences collective cell invasion in squamous cell carcinoma (SCC). Looking at a multitude of parameters, the authors found that cancer cell-cancer cell contacts and matrix degradation work cooperatively in SCC invasion.

The authors provide a rigorous and systematic approach to querying the importance of the parameters tested; first setting their hypothesis computationally followed by in vitro experimentation in two different cancer cell culturing methods (organotypic and spheroid). Importantly, the experimental data convincingly confirmed the computational predictions, lending credence to their methodology. This is a major strength of the manuscript and will be beneficial to the field with regard to investigating invasion in other cancer types.

Additionally, the varied parameters tested (cancer cell-cell adhesion, cancer cell-matrix adhesion, cancer cell-fibroblast adhesion, fibroblast-matrix adhesion, cell-intrinsic motility, matrix displacement, and matrix proteolysis) were thoughtful and rooted in the literature. However, though considerate of the role the extracellular matrix (ECM) may play (via interrogating cancer cell-matrix adhesions as parameters), the characteristics of the matrix itself (e.g. stiffness, alignment) were not investigated. These attributes have been previously shown to affect collective cell invasion. Indeed, while investigating the contributions of matrix proteolysis on invasion, the authors found a parabolic relationship where both too much and too little matrix negatively impacted the ability of SCC cells to invade. Moreover, it is unclear what the role of fibroblast-matrix adhesions was to this system, though it was originally stated as a tested parameter.

Reviewer #1 (Public Review):

This manuscript provides an in-depth analysis of the advantages and potential pitfalls of the application of Granger Causality (GC) to calcium imaging data, especially regarding various types of pre-processing. The key strength of the manuscript is the rigor and thoroughness of the authors' approach, and it is very clear how one would go about replicating their work. On the other hand, it is not from the results how well one should trust the results of GC for an unknown system, as many results rely on having some specialized knowledge about the measurements beforehand.

Strengths:

- Understanding how to measure causality is a key problem in modern science, and with the increasing abundance of wide-field calcium imaging, understanding how to assess information flow between neurons from these data is of wide interest and importance.

- I was impressed by the rigor and explicitness of the authors' approach. In papers like this, there is the temptation to sweep problems under the rug and highlight the successes. Here, the authors present, in a clearly organized format, the effects of various methods and analysis decisions. Moreover, the methods are described in a manner such that they could be (relatively) easily implemented by the reader.

- In general, the approach of using the GC value of the F-statistics and then normalizing by a null model is an appealing method that has a lot of intuitive and quantitative value.

Weaknesses:

- It's not clear to me what lessons are specific to the system they are studying and which ones are to be taken as more general lessons. Certainly, dealing with slow calcium dynamics, motion artifacts, and smoothing, are general problems in calcium imaging, but I found myself puzzled a bit about how to decide which neurons are "strange" without a lot of system-specific knowledge. This seems to be a rather important effect, and having a bit more guidance in the discussion would be useful.

- Somewhat related, I'm not entirely sure what results I should take home from the hindbrain analysis. It is clear that there is a more-or-less global signal modulating all neural activity, but this is a common occurrence in population recordings (often, one subtracts this off via PCA or another means before proceeding). Is the general lack of causal links (via the MVGC at least) a generic phenomenon in recurrent networks, or is there something more system-specific here? Accordingly, it might be interesting to run a recurrent neural network simulation with similar properties to the hindbrain (and perhaps with correlated driving) to see what GC/MVGC would predict. Is there any hope of these methods finding information flow in recurrent networks, or should we restrict the method to networks where we expect the primary mode of information transmission to be feedforward?

Reviewer #2 (Public Review):

The authors consider the application of Granger causality (GC) analysis to calcium imaging data and identify several challenges therein and provide methodological approaches to address them. In particular, they consider case studies involving fluorescence recordings from the motoneurons in embryonic zebrafish and the brainstem and hindbrain of larval zebrafish to demonstrate the utility of the proposed solutions in removing the spurious links that the naive GC identifies.

The paper is well-written and the results on the chosen case studies are compelling. However, the proposed work would benefit from discussing the contributions of this work in the context of existing and relevant literature and clarifying some of the methodological points that require more rigorous treatment. I have the following comments:

Major comments:

1) I would like to point out recent literature that adapts the classical GC for both electrophysiology data and calcium imaging data:

[1] A. Sheikhattar et al., "Extracting Neuronal Functional Network Dynamics via Adaptive Granger Causality Analysis", PNAS, Vol. 115, No. 17, E3869-E3878, 2018.

[2] N. A. Francis et al., "Small Networks Encode Decision-Making in Primary Auditory Cortex", Neuron, Vol. 97, No. 4, 2018.

[3] N. A. Francis et al., "Sequential Transmission of Task-Relevant Information in Cortical Neuronal Networks", Cell Reports, Vol. 39, No. 9, 110878, 2022.

In reference [1], a variation of GC based on GLM log-likelihoods is proposed that addresses the issues of non-linearity, non-stationarity, and non-Gaussianity of electrophysiology data. In [2] and [3], a variation of GC using sparse multi-variate models is introduced with application to calcium imaging data. In particular, all three references use the sparse estimation of the MVAR parameters in order to mitigate overfitting and also use corrections for multiple comparisons that also reduce the number of spurious links (see my related comments below). I suggest discussing these relevant references in the introduction (paragraphs 2 and 3) and discussion.

2) A major issue of GC applied to calcium imaging data is that the trials are typically limited in duration, which results in overfitting of the MVAR parameters when using least squares (See references [2] and [3] above, for example). The authors mention on page 4 that they use least squares to estimate the parameters. However, for the networks of ~10 neurons considered in this work, stationary trials of a long enough duration are required to estimate the parameters correctly. I suggest that the authors discuss this point and explicitly mention the trial durations and test whether the trial durations suffice for stable estimation of the MVAR parameters (this can be done by repeating some of the results on the synthetic data and using different trial lengths and then assessing the consistency of the detected GC links).

3) The definition of the "knee" of the average GC values as a function of the lag L needs to be a bit more formalized. In Fig. 2H using the synthetic data, the "knee" effect is more clear, but in the real data shown in Fig. 2I, the knee is not obvious, given that the confidence intervals are quite wide. Is there a way to quantify the "knee" by comparing the average GC values as well as their confidence bounds along the lag axis?

4) While the measures of W_{IC} and W_{RC} form suitable guiding principles for the pipeline presented in this work, it would be helpful if the authors discuss how such measures can be used for other applications of GC to calcium imaging data in which a priori information regarding the left/right symmetry or the rostrocaudal flow of information is missing.

5) Removing the "strange" neurons discussed in Section C5 is definitely an important pre-processing step in applying GC. However, the criterion for identifying the strange neurons seems a bit ad hoc and unclear. Could this be done by clustering the neurons into several categories (based on their time courses) and then removing a "strange" cluster? Please clarify.

6) Another key element of existing GC methods applied to large-scale networks is dealing with the issue of multiple comparisons: for instance, in Figures 2, 3, 4, 6, 7, and 8, it seems like all arrows corresponding to all possible links are shown, where the colormap indicates the GC value. However, when performing multiple statistical tests, many of these links can be removed by a correction such as the Benjamini-Hochberg procedure. It seems that the authors did not consider any correction of multiple comparisons; I suggest doing so and adding this to your pipeline.

7) The authors use TV denoising and also mention that it is a global operator, and changes the values of a time series at time t based on both the past and future values of the process. As such, it is not clear how TV denoising could affect the "causal" relations of the time series. In particular, TV denoising would significantly change the \Gamma_{ii} coefficients in Eq. (8). Is it possible to apply a version of TV denoising that only uses the information from the past to denoise the process at time t? In other words, using a "filter" as opposed to a "smoother". Please clarify.

8) The idea of using an adaptive threshold as in Section C8 is interesting; but this problem was previously considered in [30] (in the manuscript) and reference [1] above, in which new test statistics based on log-likelihoods are used that have well-known asymptotic null distributions (i.e., chi-square distributions). In particular, reference [1] above identifies and applies the required rescaling for the asymptotic null distributional assumptions to hold. I suggest discussing your work regarding the adaptive thresholds in the context of these existing results.

9) Related to the previous comment, given that the authors use a shuffling procedure to obtain the null, it is not clear why fitting the F-distribution parametrically and using its quantiles for testing would provide further benefits. In fact, as shown in Figure S9B, the rescaled F-distribution does not fully match the empirical null distribution, so it may be worth using the empirical null to obtain the non-parametric quantiles for testing. Please clarify.

10) In Figure 5C, the values of W_IC for the MV cases seem to be more than 1, whereas by definition they should be less than or equal to 1. Please clarify.

11) Is there evidence that the lateralized and rostrocaudal connectivity of the motoneurons occurs at the time-scale of ~750 ms? Given that this time scale is long enough for multiple synapses, it could be the case that some contralateral and non-rostrocaudal connections could be "real", as they reflect multi-hop synaptic connections. Please clarify.

12) While it is useful to see the comparison of the BV and MV cases shown in Figs. 1 and 2, given extensive evidence in the GC literature on the shortcomings of the BV version of GC, it seems unnecessary to report the BV results in Figs. 3 onward. I suggest discussing the shortcoming of the BV case when presenting figures 1 and 2 and removing the BV results from the subsequent results.

Reviewer #3 (Public Review):

This manuscript provides a helpful and transparent guide on the application of granger-causality (GC) to calcium datasets. This is a useful entry point toward understanding the suitability and limitations of GC to neural data. However, it is not entirely convincing that the variations of GC analysis provided in this manuscript can be effectively applied to large-scale calcium datasets without prior knowledge of the underlying circuit, especially when such networks are likely to contain redundancy and recurrent links.

I would like to acknowledge that, at the outset, I held an unfavorable prior belief toward GC, for reasons that are well addressed in this manuscript, including the dangers of applying spectral GC to nonlinear networks, as well as a variety of pathologies that can undermine naive GC.

The manuscript has been helpful, both for its effective presentation of both bivariate GC and its multivariate extension, as well as the practical considerations that are essential to applying it to real-life data. It was particularly helpful to see a treatment of the challenges and their possible resolutions. I commend the authors for their transparency - they should certainly be rewarded rather than punished for their transparency.

Major<br /> 1. Redundant signals: throughout the brain, it's expected that a population of neurons can encode the same information. It's unclear how GC (both the original and the modified versions) can handle this redundancy. Given how pervasive redundant signals are in the brain, this should be addressed in both simulation and experimental data. For example, in one of the manuscript's simulated networks, replace one neuron with 10 copies of it, each with identical inputs and outputs but with the weights scaled by 1/10. Such a network is functionally equivalent to the original but may pose some challenges for the various versions of GC. I believe this issue also accounts for the MVGC results in the hindbrain dataset. It might be more appropriate to apply GC to groups of neurons (as indeed the authors cited), instead of applying it at the single-cell level with redundant signals.<br /> 2. Similarly, there is recurrent connectivity throughout the brain. The current manuscript appears to assume feedforward networks. Is the idea that GC cannot be applied to recurrent networks? If so, this needs to be clearly stated. If the authors believe that GC can recover casual links even in the presence of recurrent connectivity, this needs to be demonstrated.<br /> 3. Both BVGC and MVGC appear to be extremely sensitive to any outlier signals. The most worrying aspect is that the authors developed their corrections and pipelines with the benefit of knowing the structure of the underlying system, whereas in the case where GC would be most useful, the user would be unable to rely on prior knowledge of the underlying structure. For instance, the motion artifact in Fig 3a-c was a helpful example of a vulnerability of naive GC, but one could easily imagine scenarios involving an unmeasured disturbance (e.g. the table is bumped) causing a similar artifact, but if the experimenter is unaware of such unmeasured disturbances then they will not be included in Z, and hence can result in the detection of widespread spurious links.<br /> There is a circularity here that's concerning. It seems that one already needs to have the answer (e.g. circuit connectivity) in order to clean up the data sufficiently for BVGC or MVGC to work effectively. Perhaps the authors would be interested in incorporating ideas from the systems identification literature, which can include the estimation of unmeasured disturbances, perhaps in conjunction with L1 regularization on the GC links. This is certainly out of scope for the present work, but it would be worth acknowledging the difficulties of unmeasured disturbances and deferring a general solution to future work. Similar considerations apply to a common unmeasured neuronal input (e.g. from a brain region not included in the field of view of the imaging).<br /> 4. Interpretation - would it be correct to state that BVGC identifies plausible causal links, while MVGC identifies a plausible system-level model? I think these interpretations, carefully stated, might provide a helpful way of thinking about the two GC approaches. Taking the results of the paper together, neither BVGC nor MVGC is definitive - BVGC may overestimate the true number of causal links but MVGC is prone to a winner-take-all phenomenon that may represent just one of many plausible system-level models that can account for the observed data. This should be more clearly stated in the manuscript.<br /> 5. "correlation completely misses the structure" - links are signed, so they should be shown with "bwr" colormap, with zero mapped to white (i.e. v_min is blue, 0 is white, v_max is red, |v_min| = |v_max|, this is natively supported in PyPlot and can be trivially implemented or downloaded in MATLAB). It is misleading that correlation appears to miss certain links marked in black, until one realizes that these links are inhibitory. It would substantially aid clarity and consistency if all panels followed this signed "bwr" convention. I think the emphasis for the GC panels is on whether links are detected, rather than the weight of the link, so I would suggest indicating detected inhibitory links as -1 (blue) and detected excitatory links as +1 (red), and link not detected as 0 (white).

Reviewer #1 (Public Review):

For membrane transporters, the factors that define transport cycle state equilibria and kinetics remains a major question. In contrast to ion channels where electrophysiological single-channel recordings reveal transitions between states, this has not been possible for slower transport proteins and so this information must be extracted from bulk transport behavior. However, recent single-molecule microscopy studies, such as FRET, have provided a new way of identifying transitions between conformational ensembles and connecting this to transport behaviors. However, the resolution of FRET can be limiting in that it requires multiple labeling with large fluorophores that have their own freedom to move, thus reducing the ability to detect small conformational changes. In the present study, Zhou et al. address this by using a different single-molecule approach of polarization microscopy, and investigate the small conformational changes associated with the AdiC arginine/agmatine antiporter from the APC super-family of transport proteins. Here, they anchor bis-TMR-maleimide onto helix 6, a part of the protein that has been identified to change orientation in the different crystal structures of AdiC and other APC homologues in inward, outward and occluded states. By "fixing" the protein onto microscopy slides, they are able to detect the change in polarization angles of the emitted fluorescence and map that onto relative changes in helix 6 orientation. Analyzing these data, they propose a model of four states that exchange in equilibrium, with and without the substrate, setting the stage for quantifying equilibrium constants and kinetics for a detailed mapping of the transport cycle, presented in an accompanying article.

This is certainly a cutting-edge approach that offers the potential to resolve the equilibrium reactions between small conformational changes and thus has the potential to push forward the mechanistic and quantitative investigation of membrane transport. However, at this point the studies require further validation on several levels. This includes an independent investigation of whether the protein being studied (i.e. with all tags, mutations, labeling, nanodisc solubilzation) confers the same substrate binding and transport behavior that has been reported previously, and is being used as comparison data here. In addition, there is some concern that the anchoring of the protein may bias conformational equilibria in some way and so it would be worthwhile to map out if this effect is limiting by changing linker lengths, within a range where it is still possible to resolve changes in polarization angles. Finally, the results are very dependent on the post-processing of the single-molecule trajectories that include changepoint analysis, averaging and clustering algorithms, yet there is little data provided to examine the robustness of each of these steps in the ultimate determination of the four-state model. While the observation that some of the states identified show a linkage to the arginine substrate, further validation along the lines mentioned above are required before a full analysis of the transport cycle is rationalized.

Reviewer #2 (Public Review):

The antiporter AdiC is a member of the amino-acid and polyamine organocation (APC) transporter superfamily. It imports the single-charged arginine (Arg+) and exports the double-charged agmatine (Agm2+). Thus, it increases the intracellular pH, helping some pathogenic enterobacteria survive in acidic environments. The APC transporters are known to sample 4 major conformations in the transport cycle. Monitoring the conformational transitions is important for understanding the transport mechanism, but methods detecting multi-state conformational changes are very limited. The authors use high-resolution polarization microscopy to resolve 4 different states in substrate-free (Apo) or substrate-bound conditions. This work further demonstrates the power of fluorescence polarization microscopy in studying protein dynamics. The authors introduced an interesting normalization step in data processing to average results obtained for different protein particles. However, the 4 states could be identified from single traces and the normalization from trace to trace could not be done without the pre-identified states on single traces. Thus, the improvement provided by the normalization compared to the published work (NSMB 2019a, 2019b, 2019c) is relatively limited.

Reviewer #3 (Public Review):

In this work, Zhou et al. employed the polarization microscope (PM) method to track the orientations of helix 6a in the bacterial amino-acid transporter AdiC. It is very impressive that the authors were able to optimize the technique to achieve an overall resolution of 5{degree sign} for detecting changes in the inclination and rotation angles (𝜃 and 𝜓). However, I am deeply concerned about how the authors linked PM-detected conformational states to the structural states obtained using crystallography. Overall, I think it was an overstatement that the work resolved the equilibrium conditions for the major states in AdiC's transport cycle, and I urge the others to be more transparent with the readers about the limitations of their technique and be more thorough in considering alternative interpretations.

Reviewer #4 (Public Review):

In this paper, Zhou et al. propose a polarization microscope for measuring the emission polarization of bifunctional rhodamine molecules attached to AdiC transporters. The polarization is used to resolve the orientation of the fluorophores, which allows the authors to successfully resolve the four conformations of AdiC at a temporal resolution of tens of milliseconds. The measured orientation for each conformation is validated with the results using crystallography.

Overall, I believe the paper is well written and demonstrates a great application for orientation imaging using polarized microscopes. Detailed experimental procedures, calibrations, and mathematical frameworks are included. I have the following recommendations to improve the manuscript.

1) On page 20, the authors note that they set a threshold to filter out molecules whose total intensity varies during the measurements. The statement that "while fluorescence intensity is expected to vary among different polarization directions, the total intensity should be essentially invariant" is not true. Since the authors use TIRF illumination to excite the molecules, the excitation polarization component along the tilting direction (e.g., along the y-axis) of the excitation is 0, i.e., molecules oriented along that direction (e.g., y-oriented) will be excited less effectively compared to other orientations.

2) Could the authors provide more details regarding how the clusters are ranked? The authors note that C1-C4 are "ranked according to the values of both angles". It is not clear to me how this is done. Also, what is the range of the measured theta_L and phi_L? And how is the warping of the spherical coordinates handled in the ranking process, e.g., a change from 350 deg to 10 deg is +20 deg or -340 deg.

3) Is the k-means clustering also based on the distance in the Cartesian space, similar to the state identification?

Reviewer #1 (Public Review):

This interesting manuscript from the Perozo and Faraldo-Gomez labs investigates the molecular mechanisms underlying the activation of the mechanosensitive ion channel MscS. The authors use a clever combination of cryoEM, coarse-grained (CG) and all-atom (AA) molecular dynamics simulations to determine the first (putatively) open conformation of the WT MscS channel and to show that this channel induces profound deformations of the membrane in the closed but not in the open state. Strikingly, MD simulations reveal that, contrary to what was previously assumed, lipids occupying cavities near the closed pore (hook lipids) come from the outer rather than inner leaflets. On pore opening, the membrane adopts a more relaxed conformation where the lipids contacting the protein are in less strained and tilted conformations. The authors thus propose a mechanism for sensing tension where the equilibrium between the open and closed conformations of the channel is dictated by differences in the membrane morphology in the two states rather than by the association and dissociation of individual lipids with the protein.

Major<br /> The observations on the hook lipids are critical and should be documented better. Based on previous work, it had been proposed that the hook lipids are associated with the inner leaflet and that they leave upon (partial) channel opening. In contrast, the present MD simulations indicate these lipids are associated with the outer leaflet and that their association to the channel persists on opening. These critical observations need to be documented better.<br /> i. Do the authors observe hook lipids in the cryoEM structure of the open channel? If yes, data should be shown. If no, then the discrepancy between MD and EM should be explicitly addressed.<br /> ii. Please show the comparison of the position and coordination of the hook lipids in MD simulations and in the closed (and/or open) structures.<br /> iii. The authors acknowledge that the volume of the cavity where the hook lipids are located decreases on channel opening. How does this not affect the association of the hook lipids with the protein?<br /> iv. Past work revealed several lipids in MscS structures near these cavities besides the hook lipids, and their ordered dissociation from the channel was proposed to be important for gating. Do the simulations show lipids in these cavities?<br /> v. Does the occupancy of the hook lipids in MD simulations change between the open and closed conformations? This should be analyzed.<br /> vi. Is the occupancy of other lipids in the nearby cavity altered upon channel opening?<br /> vii. Is the exchange of lipids near Ile150 affected by the conformational change?

I am a bit confused by the claim that "The comparison clearly highlights the reduction in the width of the transmembrane span of the channel upon opening, and how this changed is well matched by the thickness of the corresponding lipid nanodiscs (approximately from 38 to 23 Å)."<br /> i. How was the nanodisc membrane thickness determined? This should be described.<br /> ii. I do not see a ~15A change in the vertical length of the channel protein or of the nanodisc. While the panels in Fig.2 clearly show a vertical compression of the membrane, it appears that the ~15 A claim might be overstated. Adding a panel with measurements would be helpful to quantify this claim. If this is difficult on the membrane, maybe measurements could be performed on the protein.<br /> iii. What happens to the N-terminal cap structure in the open state? What are the rearrangements that allow the extracellular ends of the TM1 to disassemble the cap.

The data shown in Fig. 6 is cryptic and should be explained better in the main text. As it stands there is a cursory mention in pg. 12 and not much else.<br /> i. It would be helpful if the authors showed the position of Ile150 in the structure.<br /> ii. Does the total number of lipids in proximity of Ile150 change over time? Or the fold change represents ~1:1 exchange of lipids in the pocket?<br /> iii. I am confused by the difference in the maximum possible fold-change in unique lipids, does this reflect the difference in total number of lipids in each leaflet in each system? If so, I am a bit confused as to why there is a ~30% difference in the AA simulations whereas the values are nearly identical for the CG one.<br /> iv. Is it possible to quantify the residence time of the lipids in the pocket of each subunit?

The authors state on Pg. 21 "Nevertheless, we question the prevailing view that density signals of this kind are evidence of regulatory lipid binding sites; that is, we do not concur with the assumption that lipids regulate the gating equilibrium of MscS just like an agonist or antagonist would for a ligand-gated receptor-channel." I am a bit confused by this statement. In principle, binding and unbinding of modulatory ligands can happen on relatively fast time scales, so the observation that in MD simulations lipids exchange on a faster time scale than that of channel gating is not sufficient to make this inference. Indeed, there is ample evidence from other channels (i.e. Trp channels, HCN channels etc) where visualization of similar signals led to the identification of modulatory lipid binding sites. Thus, while I do not necessarily disagree with the authors, I would encourage them to tone down the general portion of the statement.

Reviewer #2 (Public Review):

The manuscript by Park et al. reports a new structure of the mechanosensitive channel MscS of E. coli in the open state and the results of extensive coarse grained and atomistic molecular dynamics (MD) simulations of MscS and the related channel MSL1 of plant mitochondria in presumed closed and open states. The major new finding is that in the closed state, the lipid bilayer contacting the channel is severely distorted. In the open state, this distortion is not present. The MD simulations forming the basis of this finding have been carefully executed and the finding is interesting and relevant for the understanding of channel mechanosensation. The MD simulations are ideally suited to probe the lipid interactions of the channel in a state-dependent manner and to identify possible membrane distortions. However there are some issues that should be addressed.

1) Are the structures stable in the membrane also without the weak restraints on the dihedral angles? Continuing at least one of the atomistic simulations without restraints for about 1 microsecond in a tension-free membrane would address a possible concern that the severe membrane distortion could go away by a more extensive relaxation of the channel structure.

2) Does the observed effect occur also in membranes with physiologically relevant PE lipids? Performing a simulation with a lipid mix closer to that in E. coli (and thus high in PE) would address a possible concern that the observed effect is not physiologically relevant.

3) Please include a figure showing that the lipid positions in the MD simulations match the lipid densities in the cryo-EM maps.

4) Is the reported mobility of helices TM2-TM3 of MSL1, as deduced from a comparison of different cryo-EM structures (ref 18), sufficient to impact the lipid organisation?

5) Did the initial lipid configuration in atomistic MD simulations already contain the deformations of the inner leaflet, or did these form spontaneously both in coarse-grained and atomistic simulations?

6) Did the earlier MD simulations of the closed-state structure 6PWN of MscL give any indications on the membrane deformation?

7) Are there distinct interactions between the headgroups of distorted inner-leaflet lipids with charged amino acids? If so, are these amino acids conserved?

Reviewer #3 (Public Review):

This paper combines experimental structures with careful molecular dynamics to address a crucially important topic in cellular biology - how are mechanosensitive ion channels gated by the membrane? There are many flavors of mechanosensitive proteins, and here the authors study MscS from e. coli and the eukaryotic homolog MSL1 from Arabidopsis. The key finding is that the closed states of both channels induce high curvature in the inner leaflet due to the membrane protruding into the cytoplasm to lipidate exposed hydrophobic patches on the protein. The open state structures exhibit far less membrane deformation. Moreover, comparing the open and closed state structures reveals that the membrane-protein surface area is not significantly different in the two states - hence all of the mathematical models to date (and many experimental models too) that posit that tension-induced gating is driven by expansion of the in-plane area of the protein must be revised. Instead, the authors convincingly argue that the role of tension is to increase the energy of the protein-membrane system in the closed state (with its large membrane deformations) compared to the flat-membrane open state. Forgive me for not going on more about the structures that have been solved here, and how they are likely more representative of the native open state than previously solved structures - I agree with the authors' assertions, and they represent a major step forward in elucidating the full gating transition in both bacterial and eukaryotic systems. This is an important discovery, and it would have been impossible without the structure and simulation coming together. Future work attempting to quantify the energy of the membrane deformations, protein free energy difference between the channels in open and closed states, and the role of tension will be essential but outside the scope of what the authors were trying to do here.

Reviewer #1 (Public Review):

The authors focused on linking physiological data on theta phase precession and spike-timing-dependent plasticity to the more abstract successor representation used in reinforcement learning models of spatial behavior. The model is presented clearly and effectively shows biological mechanisms for learning the successor representation. Thus, it provides an important step toward developing mathematical models that can be used to understand the function of neural circuits for guiding spatial memory behavior.

However, as often happens in the Reinforcement Learning (RL) literature, there is a lack of attention to non-RL models, even though these might be more effective at modeling both hippocampal physiology and its role in behavior. There should be some discussion of the relationship to these other models, without assuming that the successor representation is the only way to model the role of the hippocampus in guiding spatial memory function.

1. Page 1- "coincides with the time window of STDP" - This model shows effectively how theta phase precession allows spikes to fall within the window of spike-timing-dependent synaptic plasticity to form successor representations. However, this combination of precession and STDP has been used in many previous models to allow the storage of sequences useful for guiding behavior (e.g. Jensen and Lisman, Learning and Memory, 1996; Koene, Gorchetchnikov, Cannon, Hasselmo, Neural Networks, 2003). These previous models should be cited here as earlier models using STDP and phase precession to store sequences. They should discuss in terms of what is the advantage of an RL successor representation versus the types of associative sequence coding in these previous models.

2. On this same point, in the introduction, the successor representation is presented as a model that forms representations of space independent of reward. However, this independence of spatial associations and reward has been a feature of most hippocampal models, that then guide behavior based on interactions between a reward representation and the spatial representation (e.g. Redish and Touretzky, Neural Comp. 1998; Burgess, Donnett, Jeffery, O'Keefe, Phil Trans, 1997; Koene et al. Neural Networks 2003; Hasselmo and Eichenbaum, Neural Networks 2005; Erdem and Hasselmo, Eur. J. Neurosci. 2012). The successor representation should not be presented as if it is the only model that ever separated spatial representations and reward. There should be some discussion of what (if any) advantages the successor representation has over these other modeling frameworks (other than connecting to a large body of RL researchers who never read about non-RL hippocampal models). To my knowledge, the successor representation has not been explicitly tested on all the behaviors addressed in these earlier models.

3. Related to this, successes of the successor representation are presented as showing the backward expansion of place cells. But this was modeled at the start by Mehta and colleagues using STDP-type mechanisms during sequence encoding, so why was the successor representation necessary for that? I don't want to turn this into a review paper comparing hippocampal models, but the body of previous models of the role of the hippocampus in behavior warrants at least a paragraph in each of the introduction and discussion sections. In particular, it should not be somehow assumed that the successor representation is the best model, but instead, there should be some comparison with other models and discussion about whether the successor representation resembles or differs from those earlier models.

4. The text seems to interchangeably use the term "successor representation" and "TD trained network" but I think it would be more accurate to contrast the new STDP trained network with a network trained by Temporal Difference learning because one could argue that both of them are creating a successor representation.

Reviewer #2 (Public Review):

The authors present a set of simulations that show how hippocampal theta sequences may be combined with spike time-dependent plasticity to learn a predictive map - the successor representation - in a biologically plausible manner. This study addresses an important question in the field: how might hippocampal theta sequences be combined with STDP to learn predictive maps? The conclusions are interesting and thought-provoking. However, there were a number of issues that made it hard to judge whether the conclusions of the study are justified. These concerns mainly surround the biological plausibility of the model and parameter settings, the lack of any mathematical analysis of the model, and the lack of direct quantitative comparison of the findings to experimental data.

While the model uses broadly realistic biological elements to learn the successor representation, there remain a number of important concerns with regard to the biological plausibility of the model. For example, the model assumes that each CA3 cell connects to exactly 1 CA1 cell throughout the whole learning process so that each CA1 cell simply inherits the activity of a single CA3 cell. Moreover, neurons in the model interact directly via their firing rate, yet produce spikes that are used only for the weight updates. Certain model parameters also appeared to be unrealistic, for example, the model combined very wide place fields with slow running speeds. This leaves open the question as to whether the proposed learning mechanism would function correctly in more realistic parameter settings. Simulations were performed for a fixed running speed, thereby omitting various potentially important effects of running speed on the phase precession and firing rate of place cells. Indeed, the phase precession of CA1 place cells was not shown or discussed, so it is unclear as to whether CA1 cells produce realistic patterns of phase precession in the model.

The fact that a successor-like representation emerges in the model is an interesting result and is likely to be of substantial interest to those working at the intersection between neuroscience and artificial intelligence. However, because no theoretical analysis of the model was performed, it remains unclear why this interesting correspondence emerges. Was it a coincidence? When will it generalise? These questions are best answered by mathematical analysis of the model (or a reduced form of it).

Several aspects of the model are qualitatively consistent with experimental data. For example, CA1 place fields clustered around doorways and were elongated along walls. While these findings are important and provide some support for the model, considerable work is required to draw a firm correspondence between the model and experimental data. Thus, without a quantitative comparison of the place field maps in experimental data and the model, it is hard to draw strong conclusions from these findings.

Overall, this study promises to make an important contribution to the field, and will likely be read with interest by those working in the fields of both neuroscience and artificial intelligence. However, given the above caveats, further work is required to establish the biological plausibility of the model, develop a theoretical understanding of the proposed learning process, and establish a quantitative comparison of the findings to experimental data.

Reviewer #1 (Public Review):

In this manuscript, the authors have assembled a reference transcriptome of the whole head of Loligo vulgaris and used it to perform single cell transcriptomics. With about 20,000 cells, they identify 32 clusters corresponding to a few identifiable cell types - neurons, stem cells, sensory cells, and epidermis. They use select marker genes from these clusters and perform HCR in situs on Loligo heads to describe these cell types. Their in situs describe a region similar to the lateral lip seen in other cephalopods where neural progenitors are found and from where neurons migrate into the brain.

Reviewer #2 (Public Review):

In 'Molecular characterization of cell types in the squid Loligo vulgaris', the authors study profile cell types of the squid brain, using single cell RNAseq and FISH for anatomical localization. They reveal many different cell types, some of which have correspondences in other organisms and some of which reflect cephalopod-specific innovations. The current study is one of 4 recent preprints (Styfhals et al. 2022, Songco-Casey et al. 2022, Gavriouchkina et al. 2022) profiling cephalopod tissues using scRNAseq and FISH-based anatomical localization. Together these studies begin to reveal the cellular complexity of these fascinating animals.

Reviewer #3 (Public Review):

In this manuscript, the authors leverage new single-cell sequencing data to unravel cell type diversity in the head of Loligo vulgaris hatchlings. This analysis recovers 33 clusters and the authors describe the cell type populations with HCR in situ hybridization. This work provides an important next step in describing neural and sensory cells in an understudied class of invertebrates that goes beyond traditional morphological characterization.

Reviewer #1 (Public Review):

This work aimed at investigating how a BMI decoding performance is impacted by changing the conditions under which a motor task is performed. They recorded motor cortical activity using multielectrode arrays in two monkeys executing a finger flexion and extension task in four conditions: normal (no load, neutral wrist position), loaded (manipulandum attached to springs or rubber bands to resist flexion), wrist (no load, flexed wrist position) or both (loaded and flexed wrist). They found, as expected, that BMI decoders trained and tested on data sets collected during the same conditions performed better at predicting kinematics and muscle activity than others trained and tested across conditions. They also report that the performance of monkeys a BMI task involving the online control of a virtual hand was almost unaffected by changing either the actual manipulandum conditions as above or switching between decoders trained from data collected under different conditions. As for the neuronal activity, they found a mix of changes across task contexts. Interestingly, a principal component analysis revealed that activity in each context falls within well-aligned manifolds, and that the context-dependent variance in neuronal activity strongly correlated to amplitude of muscle activity.

Strengths:

The current study expands on previous findings about BMI decoders generalizability and contributes scientifically in at least three important ways.

First, their results are obtained from monkeys performing a fine finger control task with up to two degrees of freedom. This provides a powerful setting to investigate fine motor control of the hand in primates. The authors use the accuracy of BMI decoders between data sets as a measure of stationarity in the neurons-to-fingers mapping, which provides a reliable assessment. They show that changes in wrist angle or finger load affect the relationship between cortical neurons and otherwise identical movements. Interestingly, this result hold up for both kinematics and muscle activity predictions, albeit being stronger for the latter.

Second, their results confirming that neuronal activity recorded during different task conditions lies effectively within a common manifold is interesting. It supports prior observations, but in the specific context of finger movements.

Third, the dPCA results provide interesting and perhaps unexpected information about the fact that amplitude of muscle activity (or force) is clearly present in the motor cortical activity. This is possibly one of the most interesting findings because extracting a component from neural activity that can related robustly to muscle activity across context would provide great benefits to the development of BMIs for functional electrical stimulation.<br /> Overall, the analyses are well designed and the interpretation of the results is sound.

Weaknesses:

I found the discussion about the possible reasons why offline decoders are more sensitive to context than online decoders very interesting. Nonetheless, as the authors recognize, the possibility that the BMI itself causes a change in context, "in the plant", limits their interpretation. It could mean for the monkeys to switch from one suboptimal decoder to another, causing a ceiling effect occluding generalization errors.

Overall, several new and original results were obtained through these experiments and analyses. Nonetheless, I found it difficult to extract a clear unique and strong take-home message. The study comes short of proposing a new way to improve BMIs generalizability or precisely identifying factors that influence decoders generalizability.

Reviewer #2 (Public Review):

The authors motivate this study by the medical need to develop brain-machine interfaces (BMIs) to restore lost arm and hand function, for example through functional electrical stimulation. More specifically, they are interested in developing BMI decoding algorithms that work across a variety of "contexts" that a BMI user would encounter out in the real world, for example having their hand in different postures and manipulating a variety of objects. They note that in different contexts, the motor cortex neural activity patterns that produce the desired muscle outputs may change (including neurons' specific relationship to different muscles' activations), which could render a static decoder trained in a different context inaccurate.

To test whether this potential challenge is indeed the case, this study tested BMI control of virtual (on-screen) fingers by two rhesus macaques trained to perform 1 or 2 degree-of-freedom non-grasping tasks either by moving their fingers, or just controlling the virtual finger kinematics with neural activity. The key experimental manipulations were context shifts in the form of springs on the fingers or flexion of the wrist (or both). BMI performance was then evaluated when these context changes were present, which builds on this group's previous demonstration of accurate finger BMI without any context shifts.

The study convincingly shows the aforementioned context shifts do cause large changes in measured firing rates. When neural decoding accuracy (for both muscle and position/velocity) is evaluated across these context changes, reconstruction accuracy is substantially impaired. The headline finding, however, is that that despite this, BMI performance is, on aggregate, not substantially reduced. Although: it is noteworthy that in a second experiment paradigm where the decoder was trained on the spring or wrist-manipulated context and tested in a normal context, there were quite large performance reductions in several datasets as quantified by multiple performance measures; this asymmetry in the results is not really explored much further.

The changes in neural activity due to context shifts appear to be relatively modest in magnitude and can be fit well as simple linear shifts (in the neural state space), and the authors posit that this would make it feasible (in future work) to find context-invariant neural readouts that would result in more robust muscle activity decoders.

An additional novel contribution of this study is showing that these motor cortical signals support quite accurately decode muscle activations during non-prehensile finger movements (and also that the EMG decoding was more negatively affected by context shifts than kinematics decoding); previous work decoded finger kinematics but not these kinetics. Note that this was demonstrated with just one of the two monkeys (the second did not have muscle recordings).

This is a rigorous study, its main results are well-supported, and it does not make major claims beyond what the data support. One of its limitations is that while the eventual motivating goal is to show that decoders are robust across a variety of tasks of daily living, only two specific types of context shifts are tested here, and they are relatively simple and potentially do not result in as strong a neural change as could be encountered in real-world context shifts. This is by no means a major flaw (simplifying experimental preparations are a standard and prudent way to make progress). But the study could point this out a bit more prominently that their results do not preclude that more challenging context shifts will be encountered by BMI users, and this study in its current form does not indicate how strong a perturbation the tested context shifts are relative to the full possible range of hand movement context shifts that would be encountered during human daily living activities.

A second limitation is that while the discrepancy between large offline decoding performance reduction and small online performance reduction are attributed to rapid sensorimotor adaptation, this process is not directly examined in any detail. Third, the assessment of how neural dynamics change in a way that preserves the overall shape of the dynamics is rather qualitative rather than quantitative, and that this implementation of a more context-agnostic finger BMI is left for future work.

Reviewer #3 (Public Review):

In this manuscript the authors ask whether finger movements in non-human primates can be predicted from neural activity recorded from the primary motor cortex. This question is driven by an ultimate goal of using neural decoding to create brain-computer interfaces that can restore upper limb function using prosthetics or functional electrical stimulation systems. More specifically, since functional use of the hand (real or prosthetic) will ultimately require generating very different grasp forces for different objects, these experiments use a constant set of finger kinematics, but introduce different force requirements for the finger muscles using several different techniques. Under these different conditions (contexts), the study examines how population neural activity changed and uses decoder analyses to look at how these different contexts affect offline predictions of muscle forces and finger kinematics, as well as the animals' ability to use different decoders to control 1 or 2-DOF online. In general, the study found that when linear models were trained on one context from offline data, they did not generalize well to the other context. However, when performance was tested online (monkeys controlling a virtual hand in real time using neural activity related to movement of their own hands) with a ReFIT Kalman filter, the animals were able to complete the task effectively, even with a decoder trained without the springs or wrist perturbation. The authors show data to support the idea that neural activity was constrained to the same manifold in the different contexts, which enabled the animals to rapidly change their behavior to achieve the task goals, compared to the more complex requirement of having to learn entirely new patterns of neural activity. This work takes studies that have been conducted for upper-limb movements and extends them to include hand grasp, which is important for creating decoders for brain-computer interfaces. Finally, the authors show using dPCA can extract features during changes in context that may be related to the activity of specific muscles that would allow for improved decoders.

Strengths:

The issue of hand control, and how it compares to arm control, is an important question to tackle in sensorimotor control and in the development of brain-computer interfaces. Interestingly, the experiments use two very different ways of changing the muscle force requirements for achieving the same finger movements; springs attached to a manipulandum and changes in wrist posture. Using both paradigms the decoder analysis clearly shows that linear models trained without any manipulation do not predict muscle forces or finger kinematics well, clearly illustrating the limitations of common linear decoders to generalize to scenarios that might encompass real grasping activities that require forceful interactions. Using a well-described real-time decoder (ReFIT Kalman Filter), the authors show that this performance decrease observed offline is easily overcome in online testing. The metrics used to make these claims are well-described, and the likely explanations for these findings are described well. A particular strength of this manuscript is that, at least for these relatively simple movements and contexts, a component of neural activity (identified using dPCA) is identified that is significantly modulated by the task context in a way that sensibly represents the changes in muscle activity that would be required to complete the task in the new contexts.

Weaknesses:

The differences between exemplar data sets and comprehensively tested contexts was difficult to follow. There are many references to how many datasets or trials were used for a particular experiment, but overall, this is fragmented across the manuscript. As a result, it is difficult to assess how generalizable the results of the manuscript were across time or animal, or whether day-to-day variations, or the different data collection schedules had an effect.

The introduction allocates a lot of space to discussing the concepts of generating (computing) movements as opposed to representing movements and relates this to ideas of neural dynamics. The distinction between these as described in the introduction is not very clear, nor is it clear what specific hypothesis this leads to for these experiments. Further, this line of thinking is not returned to in the discussion, so the contribution of these experiments to ideas raised in the introduction are unclear.

The complexity of the control that was possible in this task (1 or 2 DOF finger flexion/extension) was low. Further, the manipulations that were used to control context were simple and static. Both these factors likely contribute to the finding that there was little change in the principal angles of the high-variance principal components. While this is not a criticism of the specific results presented here, the simplicity of the task and contexts, contrasted with the complexity of hand control more generally, especially for even moderately dexterous movements, makes it unclear how well the finding of stable manifolds will scale. On a related point, it is unclear whether the feature, identified using dPCA, that could account for changes in muscle activity, could be robustly captured in more realistic behaviors. It is stated that future work is needed, but at this point, the value of identifying this feature is highly speculative.

The maintained control in online BMI trials could also be explained by another factor, which I don't think was explicitly described by either of the two suggestions. Prism goggle experiments introduce a visual shift can be learned quickly, and some BCI experiments have introduced simple rotations in the decoder output (e.g. Chase et. al. 2012, J Neurophys). This latter case is likely similar in concept to in-manifold perturbations. Regardless, the performance can be rapidly rescued by simply re-aiming, which is a simple behavioral adaptation. In a 1DOF or 2DOF control case like used in these experiments, with constant visual feedback on performance, the change in context could likely be rapidly learned by the animals, maybe even within a single trial. In other words, the high performance in the online case may be a consequence of the relatively simple task demands, and the simple biomechanical solution to this problem (push harder). What is the expectation that the results seen in these experiments would be relevant to more realistic situations that require grasp and interaction?

Some of the figures were difficult to read and the captions contained some minor incorrect information. The primary purpose of some of the figures was not immediately clear from the caption. For example, the bar plots in Figures 5 and 6 were very small and difficult to read. This also made distinguishing the data from the two different animals challenging.

There is no specific quantification of the data in Figures 4D and 5D. In Figure 4D it seems apparent that the vast majority of the points are below the unity line. But, it remains unclear, particularly in Figure 5D whether the correlations between the two contexts truly are different or not in a way that would allow conclusive statements.

Reviewer #1 (Public Review):

Chakrabarti et al study inner hair cell synapses using electron tomography of tissue rapidly frozen after optogenetic stimulation. Surprisingly, they find a nearly complete absence of docked vesicles at rest and after stimulation, but upon stimulation vesicles rapidly associate with the ribbon. Interestingly, no changes in vesicle size were found along or near the ribbon. This would have indicated a process of compound fusion prior to plasma membrane fusion, as proposed for retinal bipolar cell ribbons. This lack of compound fusion is used to argue against MVR at the IHC synapse. However, that is only one form of MVR. Another form, coordinated and rapid fusion of multiple docked vesicles at the bottom of the ribbon, is not ruled out. Therefore, I agree that the data set provides good evidence for rapid replenishment of the ribbon-associated vesicles, but I do not find the evidence against MVR convincing. The work provides fundamental insight into the mechanisms of sensory synapses.

Reviewer #2 (Public Review):

Chakrabarti et al. aimed to investigate exocytosis from ribbon synapses of cochlear inner hair cells with high-resolution electron microscopy with tomography. Current methods to capture the ultrastructure of the dynamics of synaptic vesicle release in IHCs rely on the application of potassium for stimulation, which constrains temporal resolution to minutes rather than the millisecond resolution required to analyse synaptic transmission. Here the authors implemented a high-pressure freezing method relying on optogenetics for stimulation (Opto-HPF), granting them both high spatial and temporal resolutions. They provide an extremely well-detailed and rigorously controlled description of the method, falling in line with previously use of such "Opto-HPF" studies. They successfully applied Opto-HPF to IHCs and had several findings at this highly specialised ribbon synapse. They observed a stimulation-dependent accumulation of docked synaptic vesicles at IHC active-zones, and a stimulation-dependent reduction in the distance of non-docked vesicles to the active zone membrane; while the total number of ribbon-associated vesicles remained unchanged. Finally, they did not observe increases in diameter of synaptic vesicles proximal to the active zone, or other potential correlates to compound fusion - a potential mode of multivesicular release. The conclusions of the paper are mostly well supported by data, but some aspects of their findings and pitfalls of the methods should be better discussed.

Strengths:

While now a few different groups have used "Opto-HPF" methods (also referred to as "Flash and Freeze) in different ways and synapses, the current study implemented the method with rigorous controls in a novel way to specifically apply to cochlear IHCs - a different sample preparation than neuronal cultures, brain slices or C. elegans, the sample preparations used so far. The analysis of exocytosis dynamics of IHCs with electron microscopy with stimulation has been limited to being done with the application of potassium, which is not physiological. While much has been learned from these methods, they lacked time resolution. With Opto-HPF the authors were successfully able to investigate synaptic transmission with millisecond precision, with electron tomography analysis of active zones. I have no overall questions regarding the methodology as they were very thoroughly described. The authors also employed electrophysiology with optogenetics to characterise the optical simulation parameters and provided a well described analysis of the results with different pulse durations and irradiance - which is crucial for Opto-HPF.

Further, the authors did a superb job in providing several tables with data and information across all mouse lines used, experimental conditions, and statistical tests, including source code for the diverse analysis performed. The figures are overall clear and the manuscript was well written. Such a clear representation of data makes it easier to review the manuscript.

Weaknesses:

There are two main points that I think need to be better discussed by the authors.

The first refers to the pitfalls of using optogenetics to analyse synaptic transmission. While ChR2 provides better time resolution than potassium application, one cannot discard the possibility that calcium influx through ChR2 alters neurotransmitter release. This important limitation of the technique should be properly acknowledged by the authors and the consequences discussed, specifically in the context in which they applied it: a single sustained pulse of light of ~20ms (ShortStim) and of ~50ms (LongStim). While longer, sustained stimulation is characteristic for IHCs, these are quite long pulses as far as optogenetics and potential consequences to intrinsic or synaptic properties.

The second refers to the finding that the authors did not observe evidence of compound fusion (or homotypic fusion) in their data. This is an interesting finding in the context of multivesicular release in general, as well as specifically for IHCs. While the authors discussed the potential for "kiss-and-run" and/or "kiss-and-stay", it would be valuable if they could discuss their findings further in the context of the field for multivesicular release. For example, the evidence in support of the potential of multiple independent release events. Further, as far as such function-structure optical-quick-freezing methods, it is not unusual to not capture fusion events (so-called omega-shapes or vesicles with fusion pores); this is largely because these are very fast events (less than 10 ms), and not easily captured with optical stimulation.

Reviewer #3 (Public Review):

Precise methods were developed to validate the expression of channelrhodopsin in inner hair cells of the Organ of Corti, to quantify the relationship between blue light irradiance and auditory nerve fiber depolarization, to control light stimulation within the chamber of a high-pressure freezing device, and to measure with good precision the delay between stimulation and freezing of the specimen. These methods represent a clear advance over previous experimental designs used to study this synaptic system and are an initial application of rapid high-pressure freezing with freeze substitution, followed by high-resolution electron tomography (ET), to sensory cells that operate via graded potentials.

Short-duration stimuli were used to assess the redistribution of vesicles among pools at hair cell ribbon synapses. The number of vesicles linked to the synaptic ribbon did not change, but vesicles redistributed within the membrane-proximal pool to docked locations. No evidence was found for vesicle-to-vesicle fusion prior to vesicle fusion to the membrane, which is an important, ongoing question for this synapse type. The data for quantifying numbers of vesicles in membrane-tethered, non-tethered, and docked vesicle pools are compelling and important. These quantifications would benefit from additional presentation of raw images so that the reader can better assess their generality and variability across synaptic sites.

The images shown for each of the two control and two experimental (stimulated) preparation classes should be more representative. Variation in synaptic cleft dimensions and numbers of ribbon-associated and membrane-proximal vesicles do not track the averaged data. Since the preparation has novel stimulus features, additional images (as the authors employed in previous publications) exhibiting tethered vesicles, non-tethered vesicles, docked vesicles, several sections through individual ribbons, and the segmentation of these structures, will provide greater confidence that the data reflect the images.

The introduction raises questions about the length of membrane tethers in relation to vesicle movement toward the active zone, but this topic was not addressed in the manuscript. Seemingly quantification of this metric, and the number of tethers especially for vesicles near the membrane, is straightforward. The topic of EPSC amplitude as representing unitary events due to variation in vesicle volume, size of the fusion pore, or vesicle-vesicle fusion was partially addressed. Membrane fusion events were not evident in the few images shown, but these presumably occurred and could be quantified. Likewise, sites of membrane retrieval could also be marked. These analyses will broaden the scope of the presentation, but also contribute to a more complete story.

Overall, the methodology forms the basis for future studies by this group and others to investigate rapid changes in synaptic vesicle distribution at this synapse.

Reviewer #4 (Public Review):

This manuscript investigates the process of neurotransmitter release from hair cell synapses using electron microscopy of tissue rapidly frozen after optogenetic stimulation. The primary finding is that in the absence of a stimulus very few vesicles appear docked at the membrane, but upon stimulation vesicles rapidly associate with the membrane. In contrast, the number of vesicles associated with the ribbon and within 50 nm of the membrane remains unchanged. Additionally, the authors find no changes in vesicle size that might be predicted if vesicles fuse to one-another prior to fusing with the membrane. The paper claims that these findings argue for rapid replenishment and against a mechanism of multi-vesicular release, but neither argument is that convincing. Nonetheless, the work is of high quality, the results are intriguing, and will be of interest to the field.

1) The abstract states that their results "argue against synchronized multiquantal release". While I might agree that the lack of larger structures is suggestive that homotypic fusion may not be common, this is far from an argument against any mechanisms of multi-quantal release. At least one definition of synchronized multiquantal release posits that multiple vesicles are fusing at the same time through some coordinated mechanism. Given that they do not report evidence of fusion itself, I fail to see how these results inform us one way or the other.

2) The complete lack of docked vesicles in the absence of a stimulus followed by their appearance with a stimulus is a fascinating result. However, since there are no docked vesicles prior to a stimulus, it is really unclear what these docked vesicles represent - clearly not the RRP. Are these vesicles that are fusing or recently fused or are they ones preparing to fuse? It is fine that it is unknown, but it complicates their interpretation that the vesicles are "rapidly replenished". How does one replenish a pool of docked vesicles that didn't exist prior to the stimulus?

Reviewer #1 (Public Review):

Hu et al. present findings that extend the understanding of the cellular and synaptic basis of fast network oscillations in the sensory cortex. They developed the ex vivo model system to study synaptic mechanisms of ultrafast (>400Hz) network oscillation ("ripplets") elicited in layer 4 (L4) of the barrel cortex in the mouse brain slice by optogenetically activating thalamocortical axon terminals at L4, which mimic the thalamic transmission of somatosensory information to the cortex. This model allowed them to reproduce extracellular ripplet oscillations in the slice preparation and investigate the temporal relationship of cellular and synaptic response in fast-spiking (FS) inhibitory interneurons and regular spiking (RS) with extracellular ripplet oscillations to common excitatory inputs at these cells. FS cells show precisely timed firing of spike bursts at ripplet frequency, and these spikes are highly synchronized with neighboring FS cells. Moreover, the phase-locked temporal relationship between the ripplets and responses of FS and RS cells, although different phases, to thalamocortical activation are found to closely coincide with EPSCs in RS cells, which suggests that common excitatory inputs to FS and RS cells and their synaptic connectivity are essential to generate reverberating network activity as ripplet oscillations. Additionally, they show that spikes of FS cells in layer 5 (L5) reduced in the slice with a cut between L4 and L5, proposing that recurrent excitation from L4 excitatory cells induced by thalamocortical optogenetic stimulation is necessary to drive FS spike bursts in layer 5 (L5).

Overall, this study helps extend our knowledge of the synaptic mechanisms of ultrafast oscillations in the sensory cortex. However, it would have been nice if the authors had utilized various methodologies and systems.

Although the overall findings are interesting, the conclusion of the study could have been strengthened according to the following points:

1. The authors investigate the temporal relationship between ripplets and FS and RS cells' response elicited by optogenetic activation of TC axon terminals, which is mainly supported by phase-locked responses of FS and RS cells with local ripplets oscillations to optogenetic activation. They also show highly synchronized FS-FS firing by eliminating electrical gap-junction and inhibitory synaptic connections to this synchrony. Based on these findings, the authors suggest that common excitatory inputs to FS and RS cells in L4 would be essential to generate these local ripplets. However, it interferes with the ability to follow the logical flow for biding other findings of phase-locking responses of FS and RS cells in ripplet oscillations in L4.

2. The authors suggest that the optogenetic activation of TC axon terminal elicits local ripplet oscillations via synchronized spike burst of FS inhibitory interneurons and alternating EPSC-IPSC of RS cells in phase-locked with ripplets in L4 barrel cortex, which would be generated by following common excitatory inputs from the local circuits to these cells at the ripple frequency. Thus they intend to investigate the source of these excitatory inputs at this local network of L4 by suppressing the firing of L4 RS cells. However, they show FS spike bursts in L5B, instead of L4, due to the technical limitations of their experimental setup, as described in the manuscript. Although L5 FS spike bursts decrease after cutting the L4/L5 boundary, supposedly inhibiting excitatory input from L4 as depicted in Fig 6D in the author's manuscript, the interpretation of data seems overly extended because it does not necessarily represent cellular and synaptic activities which are phase-locked with the ripplets observed in L4.

3. Authors suggested a circuit model. It would be recommended that the authors try to perform in silico analysis using the suggested model to explore the function of thalamocortical axons on the fast-spiking and regular-spiking neurons to support their circuit model.

Reviewer #2 (Public Review):

This manuscript studied potential cellular mechanisms that generate ultrafast oscillations (250-600Hz) in the cortex. These oscillations correlate with sensory stimulation and might be relevant for the perception of relevant sensory inputs. The authors combined ex-vivo whole-cell patch-clamp recordings, local field potential (LFP) recordings, and optogenetic stimulation of thalamocortical afferents. In a technical tour de force, they recorded pairs of fast-spiking (FS)-FS and FS-regular-spiking (RS) neurons in the cortex and correlated their activity with the LFP signal.

Optogenetic activation of thalamic afferents generated ripple-like extracellular waveforms in the cortex, which the authors referred to as ripplets. The timing of the peaks and troughs within these ripplets was consistent across slices and animals. Activation of thalamic inputs induced precisely timed FS spike bursts and RS spikes, which were phase-locked to the ripplet oscillation. The authors described the sequences of RS and FS neuron discharge and how they phase-locked to the ripplet, providing a model for the cellular mechanism generating the ripplet.

The manuscript is well-written and guides the reader step by step into the detailed analysis of the timing of ripplets and cellular discharges. The authors appropriately cite the known literature about ultrafast oscillations and carefully compare the novel ripplets to the well-known hippocampal ripples. The methods used (ex-vivo patch-clamp and LFP) were appropriate to study the cellular mechanisms underlying the ripplets.

Overall, this manuscript develops means for studying the role of cortical ultrafast oscillations and proposes a coherent model for the cellular mechanism underlying these cortical ultrafast oscillations.

Reviewer #3 (Public Review):

In this study, Hu et al. aimed to identify the neuronal basis of ultrafast network oscillations in S1 layer 4 and 5 evoked by the optogenetic activation of thalamocortical afferents in vitro. Although earlier in vivo demonstration of this short-lived (~25 ms) oscillation is sparse and its significance in detecting salient stimuli is not known the available publications clearly show that the phenomenon is consistently present in the sensory systems of several species including humans.

In this study using optogenetic activation of thalamocortical (TC) fibers as a proxy for a strong sensory stimulus the in vitro model accurately captures the in vivo phenomenon. The authors measure the features of oscillatory LFP signals together with the intracellular activity of fast-spiking (FS) interneurons in layer 4 and 5 as well as in layer 4 regular spiking (RS) cells. They accurately measure the coherence of intra- and extracellular activity and convincingly demonstrate the synchronous firing of FS cells and antiphase firing of RS and FS cells relative to the field oscillation.

Major points:

1) The authors conclude the FS cell network has a primary role in setting the frequency of the oscillation. While these data are highly plausible and entirely consistent with the literature only correlational not causal results are shown thus direct demonstration of the critical role of GABAergic mechanisms is missing.

2) The authors put a strong emphasis on the role of RS-RS interactions in maintaining the oscillation once it was launched by a TC activity. Its direct demonstration, however, is not presented. The alternative scenario is that TC excitation provides a tonic excitatory background drive (or envelope) for interacting FS cells which then impose ultrafast, synchronized IPSPs on RS cells. Similar to the RS-RS drive in this scenario RS cells can also only fire in the "windows of opportunity" which explains their antiphase activity relative to FS cells, but RS cells themselves do not participate in the maintenance of oscillation. Distinguishing between these two scenarios is critical to assess the potential impact of ultrafast oscillation in sensory transmission. If TC inputs are critical the magnitude of thalamic activity will set the threshold for the oscillation if RS-RS interactions are important intracortical operation will build up the activity in a graded manner.

Earlier theoretical studies (e.g Brunel and Wang, 2003; Geisler et al., 2005) strongly suggested that even in the case of the much slower hippocampal ripples (below 200 Hz) phasic activation of local excitatory cells cannot operate at these frequencies. Indeed, rise time, propagation, and integration of EPSPs can likely not take place in the millisecond (or submillisecond) range required for efficient RS-RS interactions. The alternative scenario (tonic excitatory background coupled with FS-FS interactions) on the other hand has been clearly demonstrated in the case of the CA3 ripples in the hippocampus (Schlingloff et al., 2014. J.Nsci).

When the properties of the ultrafast oscillation were tested as various stimulation strengths (Figure 2) weaker stimulation resulted in less precise timing. If TC input is indeed required only to launch the oscillation not to maintain it, this is not expected since once a critical number of RS cells were involved to start the activity their rhythmicity should no longer depend on the magnitude of the initial input. On the other hand, if the entire transient oscillation depends on TC excitation weaker input would result in less precise firing.

3) The experiments indicating the spread of phasic activity from L4 RS to L5 FS cells can not be accepted as fully conclusive. The horizontal cut not only severed the L4 RS to L5 FS connections but also many TC inputs to the L5 FS apical dendrites as well as the axons of L4 FS cells to L5 FS cells both of which can be pivotal in the translaminar spread.

Reviewer #1 (Public Review):

The authors had previously developed a method of determining conformational free energy differences between the alternative DFG-in and DFG-out conformational states of kinases using an energy function based on a Potts model. They did this because direct estimates of this free energy change from molecular simulations, while possible in principle, would in practice be hard to do with sufficient accuracy to be useful for such a large conformational transition. Potts model energies have been shown to be correlated with overall protein stability, so it is reasonable that dividing the contacts into DFG-in and DFG-out sets should allow the estimation of a free energy difference between conformational states. In this work they examine the differences between Tyrosine Kinases (TKs) and Serine/Threonine Kinases (STKs) more closely, finding that the model predicts a small free energy change for converting DFG-in to DFG-out for TKs but a significant unfavorable free energy cost to converting to DFG-out for the STKs. The most insightful part of the paper comes in its analysis of how this conformational change may contribute to the overall binding free energies. Calculating binding free energies for Type II inhibitors (which bind DFG-out) by alchemical methods neglects the contribution from any unfavorable conformational change ("reorganization energy") required to adopt the DFG-out conformation. Thus comparing this calculated binding free energy with the total binding free energy estimated from experiment allows an estimate of the conformational reorganization energy. It is found that this estimate is nicely correlated with the free energy change for conformational rearrangement estimated from the Potts model analysis. Thus an important contribution to Type II inhibitor binding is this conformational transition. The different contributions to Type II binding are analyzed in detail by further dissecting the Potts model.

Reviewer #2 (Public Review):

This paper focuses on an important topic. It explores how the activation loop conformations affect the type II inhibitor binding in Tyr and Ser/Thr kinases. The comprehensive computational results agree with the available experimental data. It is a remarkably comprehensive, high quality paper.

Reviewer #3 (Public Review):

Tyrosine kinases (TKs) belong to a relatively small family of protein kinases that are a product of later evolution and play a critical role in the regulation of cell behavior in multicellular organisms. Major differences between TKs and Serine/threonine kinases (STKs) are very well known, however, it is still unclear if there are specific sequence signatures that favor a specific inactive conformation of TKs that can be exploited for efficient drug design. The authors used Potts Hamiltonian models (PHMs) along with other computational methods to tackle this problem. The are two main weaknesses of this approach. First, it relies on multiple sequence alignment that requires a large set of related sequences and can't be applied to smaller families. Second, it requires a relatively large number of structures that have similar inactive structures. Although all active kinases have very similar structures, their inactive structures are very diverse. However, there are several groups of inactive conformations that share a high level of similarity. The authors study one of them, the so-called "DFG-out" conformation, and present a set of convincing results that define several key residues that favor this conformation. They demonstrated the strength of the PHMs approach that allows the detection of critical contacts that are specific for certain conformations. These results can be used to predict the "DFG-out" conformation of a TK even if its structure is not known or predict the effects of mutations in a TK if they involve some of the critical residues. In general, the paper presents a set of solid results that will facilitate the development of highly specific inhibitors for TKs.

Reviewer #1 (Public Review):

This paper investigates potential mechanisms underlying the generation of hippocampal theta and gamma rhythms using a combination of several modeling approaches. The authors perform new simulation experiments on the existing large-scale biophysical network model previously published by Bezaire et al. Guided by their analysis of this detailed model, they also develop a strongly reduced, rate-based network model, which allows them to run a much larger number of simulations and systematically explore the effects of varying several key parameters. The combined results from these two in silico approaches allow them to predict which cell types and connections in the hippocampus might be involved in the generation and coupling of theta and gamma oscillations.

In my view, several aspects of the general methodology are exemplary. In the current work as well as several earlier papers, the authors are re-using a large-scale network model that was originally developed in a different laboratory (Bezaire et al., 2016) and that still represents the state-of-the-art in detailed hippocampal modeling. Such model reuse is quite rare in computational neuroscience, which is rather unfortunate given the amount of time and effort required to build and share such a complex model. Very often, and also, in this case, the original publication that describes a detailed model provides only limited validation and analysis of model behavior, and the re-use of the same model in later studies represents a great opportunity to further examine and validate the model.

Combining detailed and simplified models can also be a powerful approach, especially when the correspondence between the two is carefully established. Matching results from the two models, in this case, allow strong arguments about key mechanisms of biological phenomena, where the simplified model allows the identification and characterization of necessary and sufficient components, while the detailed model can firmly anchor the models and their predictions to experimental data.

On the other hand, I have several major concerns about the implementation of these approaches and the interpretation of the results in the current study. First of all, the detailed model of Bezaire et al. is considered strictly equivalent, in all of its relevant details, to biological reality, and no attempt is made to verify or even discuss the validity of this assumption, even when particular details of the model are apparently critical for the results presented. I see this as a fundamental limitation of the current work - the fact that the Bezaire et al. model is the best one we have at the moment does not automatically make it correct in all its details, and features of the model that are essential for the new results certainly deserve careful scrutiny (preferably via detailed comparison with experimental data).

An important case in point is the strength of the interactions between specific neuronal populations. This is represented by different quantities in the detailed and simplified model, but the starting point is always the synaptic weight (conductance) values given by Bezaire et al. (2016), also listed in Tables 2 and 3 of the current manuscript. Looking at these parameters, one can identify a handful of connections whose conductance values are much higher than those of the other connections, and also more than an order of magnitude higher (50-100 nS) than commonly estimated values for cortical synapses (normally less than about 5 nS, except for a few very special types of synapse such as the hippocampal mossy fibers). Not surprisingly, several of these connections (such as the pyramidal cell to pyramidal cell connections, and the CCK+BC to PV+BC connections) were found to be critical for the generation and control of theta and gamma oscillations in the model. Given their importance for the conclusions of the paper, it would be essential to double-check the validity of these parameter values. In this context, it is worth noting that, unlike the anatomical parameters (cell numbers and connectivity) that had been carefully calculated and discussed in Bezaire and Soltesz (2013), biophysical parameters (the densities of neuronal membrane conductances and synaptic conductances) in Bezaire et al. (2016) were obtained by relatively simple (partly manual) fitting procedures whose reliability and robustness are mostly unknown. Specifically for synaptic parameters in CA1, a more systematic review and calculation were recently carried out by Ecker et al. (2020); their estimates for the synaptic conductances in question are typically much lower than those of Bezaire et al. (2016) and appear to be more in line with widely accepted values for cortical (hippocampal) synapses.

Furthermore, some key details concerning the construction of the simplified rate model are unclear in the current manuscript. The process of selecting cell types and connections for inclusion in the rate model is described, and the criteria are mostly clear, although the results are likely to be heavily affected by the problems discussed above, and I do not understand why the strength of external input was included among the selection criteria for cell types (especially if the model is meant to capture the internal dynamics of the isolated CA1 region). However, the main issue is that it remains unclear how the parameters of the rate model (the 24 parameters in Table 4) were obtained. The authors simply state that they "found a set of parameters that give rise to theta-gamma rhythms," and no further explanation is provided. Ideally, the parameters of the rate model should be derived systematically from the detailed biophysical model so that the two models are linked as strongly as possible; but even if this was not the case, the methods used to set these parameters should be described in detail.

An important inaccuracy in the presentation of the results concerns the suggested coupling of theta and gamma oscillations in the models. Although the authors show that theta and gamma oscillations can be simultaneously present in the network under certain conditions, actual coupling of the two rhythms (e.g., in the form of phase-amplitude coupling) is not systematically characterized, and it is therefore not clear under what conditions real coupling is present in the two models (although a probable example can be seen in Figure 1C(ii)).

The Discussion of the paper states that gamma oscillations in the model(s) are generated via a pure interneuronal (ING) mechanism. This is an interesting claim; however, I could not find any findings in the Results section that directly support this conclusion.

Finally, although the authors write that they can "envisage designing experiments to directly test predictions" from their modeling work, no such experimental predictions are explicitly identified in the current manuscript.

Reviewer #2 (Public Review):

The goal of this study is to find a minimal model that produces both theta and gamma rhythms in the hippocampus CA1, based on the full-scale model (FSM) of Bezaire et al, 2016. The FSM here is treated as equivalent to biological data. This seems to be a second part of a study that the same authors published in 2021, and is extensively cited here. The study reduces the FSM to a neural rate model with 4 neurons, which is capable of producing both rhythms. This model is then simulated and its parameter dependencies are explored.

The authors succeed in producing a rate model, based on 4 neuron types, that captures the essence of the two rhythms. This model is then analyzed at a descriptive level to claim that the synapse from one interneuron type (CCK) to another (PV+) is more effective than its reciprocal counterpart (PV+ to CCK synapse) to control theta rhythm frequency.

The results fall short on several fronts:<br /> The conclusions rely exclusively on the assumption that the FSM is in fact able to faithfully reflect the biological circuits involved, not just in its output, but in response to a variety of perturbations. Although the authors mention and discuss this assumption, in the end, the reader is left with a (reduced) model of a (complex) model, but no real analysis based on this reduction. In fact, the reduced model is treated in a manner that could have been done with the full one. Thus the significance of the work is greatly reduced not by what the authors do, but by what they fail to do, which is to properly analyze their own reduced model. Consequently, the impact of this study on the field is minimal.<br /> Related to the first point, throughout the manuscript, multiple descriptive findings, based on the authors' observations of the model output, are presented as causal relationships. Even the main finding of the study (that one synapse has a larger effect on theta than another) is not quantified, but just simply left as a judgment call by the authors and reader of comparing slopes on graphs.

Reviewer #3 (Public Review):

While full-scale and minimal models are available for CA1 hippocampus and both exhibiting theta and gamma rhythms, it is not fully clear how inhibitory cells contribute to rhythm generation in the hippocampus. This paper aims to address this question by proposing a middle ground - a reduced model of the full-scale model. The reduced model is derived by selecting neural types for which ablations show that these are essential for theta and gamma rhythms. A study of the reduced model proposes particular inhibitory cell types (CCK+BC cells) that play a key role in inhibitory control mechanisms of theta rhythms and theta-gamma coupling rhythms.

Strengths:<br /> The paper identifies neural types contributing to theta-gamma rhythms, models them, and provides analysis that derives control diagrams and identifies CCK+BC cells as key inhibitory cells in rhythm generation. The paper is clearly written and approaches are well described. Simulation data is well depicted to support the methodology.

Weaknesses:<br /> The derivation methodology of the reduced model is hypotheses based, i.e. it is based on the selection of cell types and showing that these need to be included by ablation simulations. Then the reduced model is fitted. While this approach has merit, it could "miss" cell types or not capture the particular balance between all types. In particular, it is not known what is the "error" by considering the reduced model. As a result, the control plots (Fig. 5 and 6) might be deformed or very different. An additional weakness is that while the study predicts control diagrams and identifies CCK+BC cell types as key controllers, experimental data to validate these predictions is not provided. This weakness is admissible, in my opinion, since these recordings are not easy to obtain and the paper focuses on computational investigation rather than computationally guided experiments.

Reviewer #1 (Public Review):

This study focuses on the role of polo like kinase 1 (PLK-1) during oocyte meiosis. In mammalian oocytes, Plk1 localizes to chromosomes and spindle poles, and there is evidence that it is required for nuclear envelope breakdown, spindle formation, chromosome segregation, and polar body extrusion. However, how Plk1 is targeted to its various locations and how it performs these functions is not well understood. This study uses C. elegans oocytes as a model to explore PLK-1 function during meiosis. They take advantage of an analogue-sensitive allele of plk-1, which enabled them to bypass nuclear envelope breakdown defects that occur following PLK-1 RNAi. This allowed them to dissect later roles of PLK-1 in oocytes, demonstrating that depletion causes defects in spindle organization, chromosome congression, segregation, and polar body extrusion. Moreover, the authors defined mechanisms by which PLK-1 is targeted to chromosomes, showing that CENP-C (HCP-4) is required for localization to chromosome arms and that BUB-1 is required for targeting to the midbivalent region. Finally, they demonstrate that upon removal of PLK-1 from both domains, there are severe meiotic defects. These findings are interesting. However, there is a need for additional analysis to better support some of their conclusions, and to aid in interpretation of particular phenotypes. Specific comments are below.

- For many important claims of the paper, a single representative image is shown but the n is not noted. This is an issue throughout the paper for much of the localization analysis (e.g. Figure 1B, 1C, 1D, 2A, 2B, 3A, 3B, 3C, etc.); in cases like this, numbers should be included to increase the rigor of the presented data. How many images or movies were analyzed that looked like the one shown? For linescans, were they done only on one image? How many independent experiments were done, etc?.

- In the abstract, it is stated that PLK-1 plays a role in spindle assembly/stability (this is also stated elsewhere, e.g. line 101). This phrasing implies that the authors have demonstrated roles in both spindle assembly and stability. However, to distinguish between these roles, they would have to show that removal of PLK-1 before spindle assembly causes defects, and also that removal of PLK-1 from pre-formed spindles causes collapse. I don't think it is necessary to do this, as the spindle roles of PLK-1 are not a focus of the paper. However, the language should be altered so that it does not imply that the paper has demonstrated roles in both. A good place to do this would be in the section from lines 144-147, where they first discuss the spindle defects. It would be straightforward to explain that their approach does not distinguish between spindle assembly and stability, and that PLK-1 could have a role in either or both.

- It is stated that there is kinetochore localization of PLK-1 (and I do see some dim cup-like localization in images after PLK-1 is removed from the chromosome arms via HCP-4 RNAi). However, this cup-like localization is not clear in most wild-type images (e.g. Figure 1B, 1D, 2A, 3A, etc.). Although I recognize that the chromatin staining might be obscuring kinetochore localization, if PLK-1 was truly a kinetochore protein I would also expect it to localize to filaments within the spindle (as many other kinetochore proteins do), especially since the authors state that BUB-1 targets PLK-1 to the kinetochore (and BUB-1 is in the filaments). In fact, the only images where it looks like PLK-1 may be localized to filaments are in Figure 4C and 6A, when HCP-4 has been depleted (though I don't know if this generally true across all HCP-4 RNAi images). For me, this calls into question the conclusion that PLK-1 truly is on the kinetochore in wild type conditions - could it be that PLK-1 only localizes to the kinetochore (and to the filaments) when HCP-4 is depleted? The authors need to resolve this issue and provide better evidence that PLK-1 normally localizes to the kinetochore, if they want to make this claim. Additionally, the observation that PLK-1 is not on the kinetochore filaments (in wild type conditions) should be addressed in the text somewhere - do the authors think that this is a special type of kinetochore protein that does not localize to the filaments?

- The authors should provide a control experiment, treating wild-type worms with 10uM 3-IB-PP1. This would be important to ensure that the spindle defects seen at this concentration in the plk-1as strain are not non-specific effects of the inhibitor. There is a control in Figure 1 - figure supplement 3 using 1uM 3-IB-PP1 but didn't see a control for 10uM (the concentration at which spindle defects are observed).

- In Figure 2F, the gels for BUB-1+PLK-1 look different in the presence and absence of phosphorylation by Cdk1 - for these data, I agree with the authors that it looks as if the complex elutes at a higher volume if BUB-1 is not phosphorylated (lines 200-204). However, Figure 2G has a repeat of the condition with phosphorylated BUB-1, and in this panel, the complex appears to elute at a higher volume than it did on the gel in panel F. The gel in panel G looks much more similar to the unphosphorylated condition in panel F. The authors need to explain this discrepancy (i.e., Is there a reason why the gels cannot be compared between panels? How reproducible are these data?). Ideally, the authors would include a repeat of the unphosphorylated BUB-1 + PLK-1 condition in panel G, done at the same time as the conditions shown in that panel, to avoid the impression that their results may not be reproducible.

- The authors would need to provide convincing evidence that co-depletion of BUB-1 and HCP-4 delocalizes PLK-1 from the chromosomes entirely, and that this co-depletion condition is more severe than either single depletion alone. Additionally, the bub-1T527A and hcp-4T163A alleles are nice tools to, in theory, more specifically delocalize PLK-1 from the midbivalent and chromosome arms, respectively, to explore the functions of chromosome-associated PLK-1. However, I think the authors cannot rule out the possibility that other proteins are also being depleted from the midbivalent and/or chromosome arms in their conditions, and that this delocalization may contribute to the phenotypes observed. For example, hcp-4 depletion was recently shown to delocalize KLP-19 from the chromosome arms (Horton et.al. 2022), so in the experiment shown in Figure 6E (HCP-4 RNAi in the bub-1 mutant), PLK-1 was likely not the only protein missing from the chromosome arms. Therefore, understanding if other proteins are absent from these domains (in the bub-1T527A and hcp-4T16A3 mutants) would help the reader understand and interpret the presented phenotypes (and how specific they are to PLK-1 loss). Consequently, I think that to better understand the co-depletion analysis presented in Figure 6 (and Figure 6 supplement 1), the authors should analyze other midbivalent and chromosome arm proteins, to determine if any are also delocalized (e.g. SUMO, KLP-19, MCAK, etc.). Additionally, instead of performing a combination of mutant and RNAi analysis (i.e. HCP-4 RNAi in the bub-1 mutant (Figure 6) and BUB-1 RNAi in the hcp-4 mutant (Figure 6 figure supplement 1)), it would be more powerful to generate a double mutant - this has a higher chance of being a more specific depletion condition.

Reviewer #2 (Public Review):

In this manuscript, Taylor et al. analyzed the role of the Polo-like kinase PLK-1 during female meiosis in the C. elegans oocyte. By temporally inhibiting an analogue-sensitive PLK-1 mutant (bypassing the PLK-1 requirement for nuclear envelope breakdown) they demonstrate that PLK-1 is involved in meiotic spindle assembly and/or stability, chromosome alignment and polar body extrusion. Consistent with its role in these processes, the authors demonstrate that PLK-1 localizes to multiple regions of the meiotic spindle: the spindle poles, chromosome arms, kinetochores and midbivalent region between the homologous chromosomes during meiosis I. They further dissected the mechanism recruiting PLK-1 to these structures and showed that CENP-CHCP-4 recruits PLK-1 to the chromosome arms while BUB-1 recruits PLK-1 to the midbivalent and kinetochores. The interaction between PLK-1 and its partners is mediated by phosphorylation of a Polo-docking site (consensus STP) in BUB-1 and CENP-CHCP-4. Finally, the authors show that both PLK-1 recruitment pathways are critically required for PLK-1 function in female meiosis.

This fundamental work substantially advances our understanding of PLK-1 function during female meiosis.<br /> Overall, the data presented are of very high quality and support the major conclusions of the paper with one or two exceptions.

Reviewer #3 (Public Review):

This is a very well written manuscript which addresses the role of the mitotic kinase Polo like kinase 1 in meiosis using the C. elegans fertilized oocyite as a model system. The authors show that PLK-1 localizes at different locations on meiotic spindles and chromosomes and identify the mechanisms required for the different localization patterns. Finally, the authors show which pool of PLK-1 is required for the different functions of PLK-1 in meiosis, using the power of genetics via CRISPR.

The strengths of the manuscript are the temporal inhibition of PLK-1 to study the meiotic roles of this kinase, the identification of the mechanisms that control PLK-1 localization and how this is regulated (phosphorylation) and the combination of cell biology and biochemstry.

This work will be of high interest to both the Polo like kinase and the meiotic communities.

Reviewer #1 (Public Review):

This manuscript presents information that will be of great interest to yeast geneticists - standard gene deletions can lead to misleading phenotypes due to effects on adjacent genes. The experiments carefully document this in one case, for the DBP1 gene, and present additional evidence that it can occur at additional genes. An improved version of the standard gene replacement cassette is described, with evidence that it functions in an improved fashion, insulated from affecting adjacent genes.

Reviewer #2 (Public Review):

The impact of the work will be for yeast researchers in the clear and careful presentation of a case study wherein phenotypes might be ascribed to the knockout of a particular gene but instead derive from effects on a neighboring gene. In this case, a transcript expressed from within or adjacent to a knockout of DBP1 by a selectable marker towards the adjacent gene MRP51 interferes with the adjacent gene's normal transcription start sites. Furthermore, although neighboring MRP51 ORF is present on the longer mRNA isoform that is generated, it is not efficiently translated. The authors expand on this phenotypic observation to demonstrate that a substantial fraction of selectable marker insertions can generate transcription adjacent to or within and going away from, selectable markers.

The strengths of the work are that the derivation of the observed phenotypes for the dpb1∆ alleles is clearly and carefully elucidated and the creation of new selectable marker cassettes that overcome the potential for cryptic transcript emanation from or near to the selectable markers. This is valuable for the community as a clear demonstration of how only the exact right experiments might detect underlying mechanisms for potentially misattributed phenotypes and that many times these experiments may not be performed. While understandable in terms of how the experiments likely played out, the manuscript seems in between biology and tool development, as the biology in question was related to a gene that is not the focus of this lab. The tool development is likely to be useful but potentially non-optimal. The mechanism for interference identified in this example case (via a long undecoded transcript isoform (LUTI) has already been described for other loci and in a number of species, including in work from the Brar lab. The concept of marker interference with neighboring genes has also been increasingly appreciated by a number of other studies.

Reviewer #1 (Public Review):

This study analyzes the detailed chemical mechanics of the formation of a physiologically important protein multimer. The primary strengths of the study are careful analyses of two distinct methods, CG-MALS a direct measure of multimerization, and environment-sensitive tryptophan fluorescence, that each indicates that Ca2+ activation of the C-lobe alone can change the physical interaction with an SK2 C-terminal peptide. An intriguing finding is that while either the N- or C-lobes alone can interact with the C-terminal peptide, only with full-length CaM can the SK C-terminal peptide be bound by two CaM molecules simultaneously. This study also clearly demonstrates that Ca2+ activation of the N-lobe triggers binding to the SK2 C-terminal peptide. Methods descriptions are thorough and excellent. Discussion of relevance to structures and function are nuanced and free of presumptions. The weaknesses of this manuscript are that the physiological implications of these findings are not clear: CaM interacts with regions of SK channels besides the C-terminal peptide studied here, and no evidence is provided here that C-lobe calcium binding alters channel opening. Overall, the evidence for conformational changes of the complex due to Ca2+ binding to the C-lobe alone is very strong, and physiological importance seems likely. The interpretation of data in this manuscript is mostly cautious and logically crystalline, with alternative interpretations discussed at many junctures.

Reviewer #2 (Public Review):

Activation of SK channels by calcium through calmodulin (CaM) is physiologically important in tuning membrane excitability. Understanding the molecular mechanism of SK activation has therefore been a high priority in ion channel biophysics and calcium signaling. The prevailing view is that the C-terminal lobe of CaM serves as an immobile Ca2+-independent tether while the N-lobe acts as a sensor whose binding activates the channel. In the present study, the authors undertake extensive biophysical/biochemical analysis of CaM interaction with SK channel peptide and rigorous electrophysiological experiments to show that Ca2+ does bind to the C-lobe of CaM and this potentially evokes conformational changes that may be relevant for channel gating. Beyond SK channels, the approach and findings here may bear important implications for an expanding number of ion channels and membrane proteins that are regulated by CaM.

A strength of the study is that the electrophysiological recordings are innovative and of high quality. Given that CaM is ubiquitous in nearly all eukaryotes, dissecting the effects of mutants particularly on individual lobes is technically challenging, as endogenous CaM can overwhelm low-affinity mutants. The excised patch approach developed here provides a powerful methodology to dissect fundamental mechanisms underlying CaM action. I imagine this could be adaptable for studying other ion channels. Armed with this strategy authors show that both N- and C-lobe of CaM are essential for maximal activation of SK channels. This revises the current model and may have physiological importance.

The major weakness is that nearly all biochemical inferences are made from analysis of isolated peptides that do not necessarily recapitulate their arrangement in an intact channel. While the use of MALS provides new evidence of the potentially complex conformational arrangement of CaM on the C-terminal SK peptide (SKp), it is not fully clear that these complexes correspond to functionally relevant states. Lastly, perhaps as a consequence of these ambiguities, the overarching model or mechanism is not fully clear.

Reviewer #3 (Public Review):

Halling et. al. probe the mechanism whereby calmodulin (CaM) mediates SK channel activity in response to calcium. CaM regulation of SK channels is a critical modulator of membrane excitability yet despite numerous structural and functional studies significant gaps in our understanding of how each lobe participates in this regulation remain. In particular, while Ca2+ binding to the N-lobe of CaM has a clear functional effect on the channel, the C-lobe of CaM does not appear to participate beyond a tethering role, and structural studies have indicated that the C-lobe of CaM may not bind Ca2+ in the context of the SK channel. This study pairs functional and protein binding data to bridge this gap in mechanistic understanding, demonstrating that both lobes of CaM are likely Ca2+ sensitive in the context of SK channels and that both lobes of CaM are required for channel activation by Ca2+.

Strengths:<br /> The molecular underpinnings of CaM-SK regulation are of significant interest and the paper addresses a major gap in knowledge. The pairing of functional data with protein binding provides a platform to bridge the static structural results with channel function. The data is robust, and the experiments are carefully done and appear to be of high quality.<br /> The use of multiple mutant CaMs and electrophysiological studies using a rescue effect in pulled patches to enable a more quantified evaluation of the functional impact of each lobe of CaM provides a compelling assessment of the contribution of each lobe of CaM to channel activation. The calibration of the patch data by application of WT CaM is innovative and provides precise internal control, making the conclusions drawn from these experiments clear. This data fully supports the conclusion that both lobes of CaM are required for channel activation.

Weaknesses:<br /> The paper focuses heavily on the results of multi-angle light scattering experiments, which demonstrate that a peptide derived from the C-terminus of the SK channel can bind to CaM in multiple stochiometric configurations. However, it is not clear if these complexes are functionally relevant in the full channel, making interpretation challenging.

Reviewer #1 (Public Review):

The overarching hypothesis is that cadherin adhesion molecules specify the code that enables the premotor brainstem breathing circuits to innervate the phrenic motor neurons that control the primary breathing muscle, the diaphragm. The authors show that multiple type 1 and 2 cadherins (N-, 6, 9, 10) are expressed by phrenic motor neurons and are necessary for motor neuron development and breathing, and complementarily, that adhesion signaling in medullary breathing circuits are required for normal breathing. The presented data support a model whereby combinations of redundant adhesion molecules create a code to wire the breathing circuit.

Strengths:<br /> 1) The authors first use a complex, rigorous genetic approach to eliminate N, 6, 9, 10 cadherins from motor neurons and discover using whole body plethysmography that neonates do not breath.<br /> 2) Then, the authors provide a thorough description of the anatomy of the mutant motor neurons and discover that the number of motor neurons decreases, the soma anatomical positions and dendritic arborization shift, and there is decreased innervation of the diaphragm breathing muscle.<br /> 3) That Cdh9 medullary expressing neurons are premotor to Cdh9 expressing phrenic motor neurons.<br /> 4) Cadherin signaling is required for normal breathing.

Weaknesses: The main conclusion that ablation of the cadherin code decreases synaptic connectivity between the rVRG and phrenic motor neurons is never directly shown. This can only be inferred by the data.<br /> 1) Conclusion that the connectivity between rVRG premotor and phrenic nerve motor neurons is "weaker". This conclusion is inferred from several experiments but is never directly demonstrated. Alternative interpretations of the decreased amplitude of the in vitro phrenic nerve burst is that the rootlet contains fewer axons (as predicted by the fewer motor neurons in S3 and innervation of the diaphragm S2). Additionally, the intrinsic electrophysiological properties of the motor neurons might be different. To show this decisively, the authors could use electrophysiological recordings of phrenic motor neurons to directly measure a change in synaptic input (for example, mEPSPs or EPSPs after optogenetic stimulation of rVRG axon terminals). Without a direct measurement, the synaptic connectivity can only be inferred.<br /> 2) Conclusion that the small phenic nerve burst size in Dbx1 deleted cadherin signaling is due to less synaptic input to the motor neurons. Dbx1 is expressed in multiple compartments of the medullary breathing control circuit, like the breathing rhythm generator (preBötC). The smaller burst size could be due to altered activity between preBötC neurons to create a full burst, the transmission of this burst from the preBötC to the rVRG, etc.<br /> 3) In vitro burst size. The authors use 4 bursts from each animal to calculate the average burst size. How were the bursts chosen? Why did the authors use so few bursts? What is the variability of burst size within each animal? What parameters are used to define a burst? This analysis and the level of detail in the figure legend/methods section is inadequate to rigorously establish the conclusion that burst size is altered in the various genotypes.<br /> 4) The authors state that the in vitro frequency in figure 4 is inaccurate, but then the in vitro frequency is used to claim the preBötC is not impacted in Dbx1 mutants (conclusion section "respiratory motor circuit anatomy and assembly"). To directly assess this conclusion, the bursting frequency of the in vitro preBötC rhythm should be measured.<br /> 5) The burst size in picrotoxin/strychnine is used to conclude that the motor neurons intrinsic physiology is not impacted. The bursts are described, and examples are shown, but this is never quantified across many bursts within in a single recordings nor in multiple animals of each genotype.

Reviewer #2 (Public Review):

This is an extremely thorough investigation of the role of cadherins in generating a functional motor circuit. The work represents a major step forward in the field as it addresses several outstanding questions and verifies anatomical data with functional outcomes. First, the data show that a combination of type I (N) and type II (6, 9, 10) cadherins is needed to generate normal connectivity and function. This is novel as prior work has suggested that the two types do not work collaboratively to generate circuits. Second, the data show that cell body position (in this case) is modulated by N-cadherin but in a manner that is independent from the impact of N-cadherin on connectivity. While position and connectivity have been shown to be separable in some cases, the data support that N-cadherin plays important but separate roles toward both actions, and type II cadherins, mainly in connectivity. These findings also underscore that cadherin roles reported for hippocampus, retina, and spinal cord motor neuron pools are not generalizable across circuits. Third, while the data show that type I and type II cadherins are required for VRG to phrenic motor neuron connectivity, they also show that there are some outcomes controlled only by N-cadherin. Finally, the data reveal much about a very poorly understood and essential circuit. The approaches are sound and range from the standard (in situ, immuno, diI, breathing measurements) to the difficult (rabies-based tracing) to the impressive (challenging ephys preps, and some painstaking mouse crosses), and they incorporated strong and creative strategies for comparison and quantification. Minor questions do not detract from a really impressive piece of work.

Reviewer #3 (Public Review):

Vagnozzi et al. analyze the role of cadherins in respiratory circuit development. The authors previously identified a combinatorial cadherin code that defines phrenic motor neurons (Vagnozzi et al., eLife 2020). Here they find that combined loss of type I N-cadherin and type II cadherins 6, 9 and 10 results in respiratory failure and reduction in phrenic motor neuron bursting activity. Furthermore, diaphragm innervation, phrenic motor neuron (MN) number, cell body position as well as dendrite orientation are all impaired in mice lacking N-cadherin and cadherins 6, 9, 10. Analysis of different genotypes indicates that phrenic MN cell body position is regulated by N-cadherin, but that dendrite orientation is regulated by the combinatorial action of N-cadherin and cadherins 6, 9, and 10. They subsequently determine that cadherin signaling in presynaptic interneurons is required for phrenic MN bursting activity. Together, the results indicate that cadherins are essential for respiratory circuit function and suggest that a combinatorial cadherin code regulates wiring specificity in this circuit.

The manuscript is well presented with clear figures and text. My comments below mainly revolve around the interpretation of some of the findings and the correlation between phenotypes in NMNΔ6910-/- mice and βγ-catDbx1Δ mice in light of specific cadherin expression patterns and connectivity between rVRG and prenic MNs.

Major points<br /> 1. Page 8: 'In addition, NMNΔ and NMNΔ6910-/- mice showed a similar decrease in phrenic MN numbers, likely from the loss of trophic support due to the decrease in diaphragm innervation (Figure S3c).' This statement should be corrected: phrenic MN number in NMNΔ mice does not differ from controls, in contrast to NMNΔ6910-/- mice (Fig. S3). Similarly, diaphragm innervation is not significantly different from controls in NMNΔ (Fig. S2). Alternatively, these observations could be strengthened by increasing the number of mice analyzed to determine whether there is a significant reduction in PMN number and diaphragm innervation in NMNΔ mice.<br /> 2. A similar comment relates to the interpretation of the dendritic phenotype in NMNΔ and NMNΔ6910-/- mice (Fig. 3m): the authors conclude 'When directly comparing NMNΔ and NMNΔ6910-/- mice, NMNΔ6910-/- mice had a more severe loss of dorsolateral dendrites and a more significant increase in ventral dendrites (Figure 3l-m).' (page 9). The loss of dorsolateral dendrites in NMNΔ6910-/- mice indeed differs significantly from control mice, and is more severe than in NMNΔ mice, which do not differ significantly from controls. For ventral dendrites however, the increase compared to controls is significant for both NMNΔ and NMNΔ6910-/- mice, and the two genotypes do not appear to differ from each other. This suggests cooperative action of N-cadherin and cadherin 6,9,10 for dorsolateral dendrites, but suggests that N-cad is more important for ventral dendrites. This should be phrased more clearly.<br /> 3. Related comment, page 10: 'Furthermore, the fact that phrenic MNs maintain their normal activity pattern in NMNΔ mice suggests that neither cell body position nor phrenic MN numbers significantly contribute to phrenic MN output.' This should be rephrased, phrenic MN number does not differ from control in NMNΔ mice (Fig. S2c).<br /> 4. The authors conclude that spinal network activity in control and NMNΔ6910-/- mice does not differ (page 10, Fig. 4f). It is difficult to judge this from the example trace in 4f. How is this concluded from the figure and can this be quantified?<br /> 5. RphiGT mice: please explain the genetic strategy better in Results section or Methods, do these mice also express the TVA receptor in a Cre-dependent manner? Crossing with the Cdh9:iCre line will then result in expression of TVA and G protein in phrenic motor neurons and presynaptic rVRG neurons in the brainstem, as well as additional Cdh9-expressing neuronal populations. How can the authors be sure that they are looking at monosynaptically connected neurons?<br /> 6. The authors use a Dbx1-cre strategy to inactivate cadherin signaling in multiple brainstem neuronal populations and perform analysis of burst activity in phrenic nerves. Based on the similarity in phenotype with NMNΔ6910-/- mice it is concluded that cadherin function is required in both phrenic MNs and Dbx1-derived interneurons. However, this manipulation can affect many populations including the preBötC, and the impact of this manipulation on rVRG and phrenic motor neurons (neuron number, cell body position, dendrite orientation, diaphragm innervation etc) is not described, although a model is presented in Fig. 7. These parameters should be analyzed to interpret the functional phenotype.<br /> 7. Additional evidence is needed to support the model that a selective loss of excitatory rVRG to phrenic motor neuron connectivity underlies the reduced bursting activity phenotype in NMNΔ6910-/- mice, for instance by labeling the connections from rVRG to phrenic MNs and quantifying connectivity.

Reviewer #1 (Public Review):

This paper describes the results of a MEG study where participants listened to classical MIDI music. The authors then use lagged linear regression (with 5-fold cross-validation) to predict the response of the MEG signal using (1) note onsets (2) several additional acoustic features (3) a measure of note surprise computed from one of several models. The authors find that the surprise regressors predict additional variance above and beyond that already predicted by the other note onset and acoustic features (the "baseline" model), which serves as a replication of a recent study by Di Liberto.

They compute note surprisal using four models (1) a hand-crafted Bayesian model designed to reflect some of the dominant statistical properties of Western music (Temperley) (2) an n-gram model trained on one musical piece (IDyOM stm) (3) an n-gram model trained on a much larger corpus (IDyOM ltm) (4) a transformer DNN trained on a mix of polyphonic and monophonic music (MT). For each model, they train the model using varying amounts of context.

They find that the transformer model (MT) and long-term n-gram model (IDyOM stm) give the best neural prediction accuracy, both of which give ~3% improvement in predicted correlation values relative to their baseline model. In addition, they find that for all models, the prediction scores are maximal for contexts of ~2-7 notes. These neural results do not appear to reflect the overall accuracy of the models tested since the short-term n-gram model outperforms the long-term n-gram model and the music transformer's accuracy improves substantially with additional context beyond 7 notes. The authors replicate all these findings in a separate EEG experiment from the Di Liberto paper.

Overall, this is a clean, nicely-conducted study. However, the conclusions do not follow from the results for two main reasons:

1. Different features of natural stimuli are almost always correlated with each other to some extent, and as a consequence, a feature (e.g., surprise) can predict the neural response even if it doesn't drive that response. The standard approach to dealing with this problem, taken here, is to test if a feature improves the prediction accuracy of a model above and beyond that of a baseline model (using cross-validation to avoid over-fitting). If the feature improves prediction accuracy, then one can conclude that the feature contributes additional, unique variance. However, there are two key problems: (1) the space of possible features to control for is vast, and there will almost always be uncontrolled-for features (2) the relationship between the relevant control features and the neural response could be nonlinear. As a consequence, if some new feature (here surprise) contributes a little bit of additional variance, this could easily reflect additional un-controlled features or some nonlinear relationship that was not captured by the linear model. This problem becomes more acute the smaller the effect size since even a small inaccuracy in the control model could explain the resulting finding. This problem is not specific to this study but is a problem nonetheless.

2. The authors make a distinction between "Gestalt-like principles" and "statistical learning" but they never define was is meant by this distinction. The Temperley model encodes a variety of important statistics of Western music, including statistics such as keys that are unlikely to reflect generic Gestalt principles. The Temperley model builds in some additional structure such as the notion of a key, which the n-gram and transformer models must learn from scratch. In general, the models being compared differ in so many ways that it is hard to conclude much about what is driving the observed differences in prediction accuracy, particularly given the small effect sizes. The context manipulation is more controlled, and the fact that neural prediction accuracy dissociates from the model performance is potentially interesting. However, I am not confident that the authors have a good neural index of surprise for the reasons described above, and this limits the conclusions that can be drawn from this manipulation.

Reviewer #2 (Public Review):

This manuscript focuses on the basis of musical expectations/predictions, both in terms of the basis of the rules by which these are generated, and the neural signatures of surprise elicited by violation of these predictions.

Expectation generation models directly compared were gestalt-like, n-gram, and a recently-developed Music Transformer model. Both shorter and longer temporal windows of sampling were also compared, with striking differences in performance between models.

Surprise (defined as per convention as negative log prior probability of the current note) responses were assessed in the form of evoked response time series, recorded separately with both MEG and EEG (the latter in a previously recorded freely available dataset). M/EEG data correlated best with surprise derived from musical models that emphasised long-term learned experiences over short-term statistical regularities for rule learning. Conversely, the best performance was obtained when models were applied to only the most recent few notes, rather than longer stimulus histories.

Uncertainty was also computed as an independent variable, defined as entropy, and equivalent to the expected surprise of the upcoming note (sum of the probability of each value times surprise associated with that note value). Uncertainty did not improve predictive performance on M/EEG data, so was judged not to have distinct neural correlates in this study.

The paradigm used was listening to naturalistic musical melodies.

A time-resolved multiple regression analysis was used, incorporating a number of binary and continuous variables to capture note onsets, contextual factors, and outlier events, in addition to the statistical regressors of interest derived from the compared models.

Regression data were subjected to non-parametric spatiotemporal cluster analysis, with weights from significant clusters projected into scalp space as planar gradiometers and into source space as two equivalent current dipoles per cluster

General comments:

The research questions are sound, with a clear precedent of similar positive findings, but numerous unanswered questions and unexplored avenues

I think there are at least two good reasons to study this kind of statistical response with music: firstly that it is relevant to the music itself; secondly, because the statistical rules of music are at least partially separable from lower-level processes such as neural adaptation.

Whilst some of the underlying theory and implementation of the musical theory are beyond my expertise, the choice, implementation, fitting, and comparison of statistical models of music seem robust and meticulous.

The MEG and EEG data processing is also in line with accepted best practice and meticulously performed.

The manuscript is very well-written and free from grammatical or other minor errors.

The discussion strikes a brilliant balance of clearly laying out the interim conclusions and advances, whilst being open about caveats and limitations.

Overall, the manuscript presents a range of highly interesting findings which will appeal to a broad audience, based on rigorous experimental work, meticulous analysis, and fair and clear reporting.

Reviewer #3 (Public Review):

The authors compare the ability of several models of musical predictions in their accuracy and in their ability to explain neural data from MEG and EEG experiments. The results allow both methodological advancements by introducing models that represent advancements over the current state of the art and theoretical advancements to infer the effects of long and short-term exposure on prediction. The results are clear and the interpretation is for the most part well reasoned.

At the same time, there are important aspects to consider. First, the authors may overstate the advancement of the Music Transformer with the present stimuli, as its increase in performance requires a considerably longer context than the other models. Secondly, the Baseline model, to which the other models are compared, does not contain any pitch information on which these models operate. As such, it's unclear if the advancements of these models come from being based on new information or the operations it performs on this information as claimed. Lastly, the source analysis yields some surprising results that don't fit with previous literature. For example, the authors show that onsets to notes are encoded in Broca's area, whereas it should be expected more likely in the primary auditory cortex. While this issue is not discussed by the authors, it may put the rest of the source analysis into question.

While these issues are serious ones, the work still makes important advancements for the field and I commend the authors on a remarkably clear and straightforward text advancing the modeling of predictions in continuous sequences.

Reviewer #1 (Public Review):

The authors used data from extracellular recordings in mouse piriform cortex (PCx) by Bolding & Franks (2018), they examined the strength, timing, and coherence of gamma oscillations with respiration in awake mice. During "spontaneous" activity (i.e. without odor or light stimulation), they observed a large peak in gamma that was driven by respiration and aligned with the spiking of FBIs. TeLC, which blocks synaptic output from principal cells onto other principal cells and FBIs, abolishes gamma. Beta oscillations are evoked while gamma oscillations are induced. Odors strongly affect beta in PCx but have minimal (duration but not amplitude) effects on gamma. Unlike gamma, strong, odor-evoked beta oscillations are observed in TeLC. Using PCA, the authors found a small subset of neurons that conveyed most of the information about the odor (winner cells). Loser cells were more phase-locked to gamma, which matched the time course of inhibition. Odor decoding accuracy closely follows the time course of gamma power.

I think this is an interesting study that uses a publicly available dataset to good effect and advances the field elegantly, especially by selectively analyzing activity in identified principal neurons versus inhibitory interneurons, and by making use of defined circuit perturbations to causally test some of their hypotheses.

Major:

- The authors show odor-specificity at the time of the gamma peak and imply that the gamma coupling is important for odor coding. Is this because gamma oscillations are important or because gamma is strongest when activity in PCx is strongest (i.e. both excitatory and inhibitory activity, which would cancel each other in the population PSTH, which peaks earlier)? To make this claim, the authors could show that odor decoding accuracy - with a small (~10 ms sliding window) - oscillates at approx. gamma frequencies. As is, Fig. 5 just shows that cells respond at slightly different times in the sniff cycle. What time window was used for computing the Odor Specificity Index? Put another way, is it meaningful that decoding is most accurate when gamma oscillations are strongest, or is this just a reflection of total population activity, i.e., when activity is greatest there is more gamma power, and odor decoding accuracy is best?

- The authors say, "assembly recruitment would depend on excitatory-excitatory interactions among winner cells occurring simultaneously during gamma activity." Can the authors test this prediction by examining the TeLC recordings, in which excitatory-excitatory connections are abolished?

- The authors show that gamma oscillations are abolished in the TeLC condition and use this to claim that gamma arises in the PCx. However, PCx neurons also project back to the OB, where they form excitatory connections onto granule cells. Fukunaga et al (2012) showed that granule cells are essential for generating gamma oscillations in the bulb. Can the authors be sure that gamma is generated in the PCx, per se, rather than generated in the bulb by centrifugal inputs from the PCx, and then inherited from the bulb by the PCx?

Reviewer #2 (Public Review):

This is a very interesting paper, in which the authors describe how respiration-driven gamma oscillations in the piriform cortex are generated. Using a published data set, they find evidence for a feedback loop between local principal cells and feedback interneurons (FBIs) as the main driver of respiration-driven gamma. Interestingly, odour-evoked gamma bursts coincide with the emergence of neuronal assemblies that activate when a given odour is presented. The results argue in favour of a winner-take-all mechanism of assembly generation that has previously been suggested on theoretical grounds.

The article is well-written and the claims are justified by the data. Overall, the manuscript provides novel key insights into the generation of gamma oscillations and a potential link to the encoding of sensory input by cell assemblies. I have only minor suggestions for additional analyses that could further strengthen the manuscript:

1. The authors' analysis of firing rates of FFIs and FBIs combined with TeLC experiments make a compelling case for respiration-driven gamma being generated in a pyramidal cell-FBI feedback mechanism. This conclusion could be further strengthened by analyzing the gamma phase-coupling of the three neuronal populations investigated. One would expect strong coupling for FBIs but not FFIs (assuming that enough spikes of these populations could be sampled during the respiration-triggered gamma bursts). An additional analysis to strengthen this conclusion could be to extract FBI- and FFI spike-triggered gamma-filtered signals. One might expect an increase in gamma amplitude following FBI but not FFI spiking (see e.g., Pubmed ID 26890123).

2. The authors utilize the neurons' weight in the first PC to assign them to odour-related assemblies. This method convincingly extracts an assembly for each odour (when odours are used individually), and these seem to be virtually non-overlapping. It would be informative to test whether a similar clear separation of the individual assemblies could be achieved by running the analysis on all odours simultaneously, perhaps by employing a procedure of assembly extraction that allows to deal with overlapping assembly membership better than a pure PCA approach (as used for instance in the work cited on page 11, including the authors' previous work)? I do not doubt the validity of the authors' approach here at all, but the suggested additional analysis might allow the authors to increase their confidence that individual neurons contribute mostly to an assembly related to a single odour.

3. Do the authors observe a slow drift in assembly membership as predicted from previous work showing slowly changing odour responses of principal neurons (Schoonover et al., 2021)? This could perhaps be quantified by looking at the expression strengths of assemblies at individual odour presentations or by running the PCA separately on the first and last third of the odour presentations to test whether the same neurons are still 'winners'.

4. Does the winner-take-all scenario involve the recruitment of specific sets of FBIs during the activation of the individual odour-selective assemblies? The authors could address this by testing whether the rate of FBIs changes differently with the activation of the extracted assemblies.

5. Given the dependence on local gamma oscillations, one might expect that odour-selective assemblies do not emerge in the TeLC-expressing hemisphere. This could be directly tested in the existing data set.

Reviewer #1 (Public Review):

In this manuscript, the authors built logistic regression prediction models for linear growth faltering using demographic, socioeconomic, and clinical variables, with the objective of developing a clinical prediction rule that could be applied by healthcare workers to identify and treat high-risk children. A model with 2 variables selected by random forest variable importance performed similarly to a model with 10 variables. Age and HAZ at baseline were selected for the 2-variable model, consistent with existing literature. The authors externally validated the 2-variable model and found similar discriminative ability. Based on typical rule-of-thumb cutoffs, model performance was moderate (AUCs of ~0.65-0.75, depending on model specification); models may still be useful in practice, but this should be further discussed by the authors.

Strengths:

Linear growth faltering is a pressing issue with broad, negative impacts on the health, development, and well-being of children worldwide. In this work, the authors applied clearly explained, thoughtful approaches to variable selection, model specification, and model validation, with large, multi-country cohorts used for training and external validation. Appropriate datasets for external validation can be challenging to find, but the MAL-ED data used here is well-suited to the task, with similar predictor and outcome measurements to the GEMS training data. The well-characterized studies allowed the authors to explore a wide range of potential predictors for stunting, including socioeconomic factors, antibiotic use, and diarrheal etiology.

Weaknesses:

This work would benefit from additional discussion around the clinical relevance of the results. For example, what is the current standard of care for prevention of stunting, and how much would this model improve the status quo? Is specificity of 0.47 in the context of sensitivity of 0.80 an acceptable tradeoff with regards to the interventions that would be used? More discussion around these points is necessary to support the authors' conclusions that these models could potentially be used to support clinical decisions and target resources.

In addition to the external validation, further investigation of model performance in key subpopulations would strengthen the importance and applicability of the work. For example, performance of prediction models may vary widely by setting; it would be valuable to show that the model has similar performance in each country. Another key sensitivity analysis would be to show consistent model performance by HAZ at baseline. The authors note that stunting may be challenging to reverse (p.20), and many of the children are already below the typical cutoff of HAZ<-2 at baseline; it would be valuable to show model performance among the subgroup of children for whom treatment would be most beneficial.

Reviewer #2 (Public Review):

The manuscript documents a thorough and well-validated clinical prediction model for risk of severe child linear growth faltering after diarrheal disease episodes, using data from multiple studies and countries. They identified a parsimonious model of child age and current size with relatively good predictive accuracy. However, I don't believe the prediction rule should be used in it's current form due to the outcome used the danger of missing treating children who require nutritional supplementation.

The outcome used for prediction in a binary indicatory for a decrease in height-for-age Z-score >= 0.5. A child who fails to gain height by future measurements is of concern, but this outcome also misses children who are already experiencing growth failure, and is vulnerable to regression to the mean effect. The two most important predictors were age and current size, with current size having a positive association with risk of growth faltering. As mentioned in the discussion, there is "the possibility that children need to have high enough HAZ in order to have the potential to falter." Additionally, there may be children with erroneously high height measurements at the first measurement, so that the HAZ change >= 0.5 associated with high baseline HAZ is from measurement-error regression to the mean. I recommend also predicting absolute HAZ (or stunting status) as a secondary outcome and comparing if the important predictors change.

In its current form, the results and conclusions from the results have problematic implications for the treatment of child malnutrition. The conclusion states: "In settings with high mortality and morbidity in early childhood, such tools could represent a cost-effective way to target resources towards those who need it most." If the current CPR was used in a resource-constrained setting, it would recommend that larger children should be prioritized for nutritional supplementation over already stunted children who may have reached their growth faltering floor. In addition, with a sensitivity of 80%, the tool would miss treating a large number of children who would experience growth faltering. The results of the clinical prediction tool need to be presented with care in how it could be used to prioritize treatment without missing treating children who would benefit from nutritional supplementation. Including absolute HAZ as an outcome will help, along with additional discussion of how the CPR fits alongside current treatment recommendations. For example, does this rule indicate treating children who aren't currently treated, or are there children who don't need treatment given current guidelines and the created CPR.

In sum, this is a thorough, well done, clearly explained exercise in creating a clinical prediction tool for predicting child risk of future growth faltering. The writing and motivation is clear, and the methods have applicability far beyond the specific use-case.

Reviewer #1 (Public Review):

The current study uses microbiology, biochemistry, microscopy, and viral vectors to establish a role for prefrontal cortex expression of the immediate early gene NPAS4 in sucrose preference and dendritic spine morphology in the mouse social defeat stress model. The experimental designs are appropriate and the hypotheses addressed are interesting. The paper is generally very well-written and the figures are clear. Most of the statistical analyses are appropriate, and they are reported in clear and useful tables. Thus, the general potential for the studies is quite high. The authors conclusively show that NPAS4 is induced in mPFC in response to social defeat stress and that NPAS4 is important for stress-induced changes in mPFC dendritic spine number. However, some of the key data regarding reward motivation are difficult to properly interpret and do not convincingly demonstrate a behavioral result of NPAS4 knockdown in mPFC. Moreover, the spine morphology and sequencing analyses lack depth. Most importantly, although the authors explore the effects of reducing NPAS4 expression in mPFC, they do not explore the effects of increasing NPAS4 expression or function, and thus the studies seem incomplete and cannot be fully interpreted.

Reviewer #2 (Public Review):

The authors investigate whether neuronal activity-regulated transcription factor 4 (NPAS4) in the medial prefrontal cortex (mPFC) is involved in stress-induced effects on neuronal spine synapse density (as a proxy for synaptic activity) and reward behaviors. A major strength of the manuscript is that NPAS4 is shown to be necessary for stress-induced reward deficits and pyramidal neuron spine density. In addition, whole transcriptome analysis of NPAS4 target genes identify a number of genes previously found to be regulated in the postmortem brain of humans with MDD, providing translational relevance to these studies. A weakness is that studies were only performed in male mice so its unclear how generalizable these effects are to females. Despite this, the work will likely impact the field of neuropsychiatry by providing novel information about the molecular and cellular mechanisms in mPFC responsible for stress-induced effects on spines synapses and reward behaviors.

Reviewer #3 (Public Review):

Hughes et al. report a role for the transcription factor NPAS4 in mediating chronic stress-induced reward-related behavioral changes, but not other depression-like behaviors. The authors find that NPAS4 is transiently upregulated in Camk2a+ PFC neurons following a single bout or repeated social defeat stress, and that knocking down PFC Npas4 prevents anhedonia. Presentation of linked individual data for social interaction/avoidance measures with/without interaction partners (Fig2C, E) is commended - all CSDS papers should show data this way. Npas4 also appears to mediate the known effect of stress on spines in PFC, providing novel mechanistic insight into this phenomenon. Npas4 knockdown altered baseline transcription in PFC, which overlapped with other stress and MDD-associated transcriptional changes and modules. However, stress-induced changes in transcription with knockdown remain unknown. A major drawback is that only male mice were used, although this is discussed to some extent. Results are presented with appropriate context and references to the literature. Conclusions are appropriate.

Additional context: Given NPAS4's role as an immediate early gene, it will be important for future work to elucidate whether IEG knockdown generally dampens transcriptional response to stress/other salient experiences. Nevertheless, the authors do show several pieces of evidence that Npas4 knockdown does not simply make mice less sensitive to stress and/or produce deficits in threat/fear-related learning and memory which is an important piece of this puzzle.

Reviewer #1 (Public Review):

The study by Osei-Owusu and colleagues addresses the mechanism of desensitization of the proton-activated chloride (PAC) channel. In three recent milestone papers, the authors have cloned the channel, identified its cryo-EM structure under high-pH and low-pH conditions, and addressed the mechanism of its pH-dependent activation. Interestingly, despite dramatic rearrangements in the TM domain, both the high- and the low-pH structures showed a closed pore, suggesting that the latter might represent an inactivated state. In the current study, the authors show that prolonged exposure of PAC to an acidic extracellular solution causes inactivation which is rapidly reversible at high pH. They further show that four mutations (H98R, E107R, D109R, H250R) that are predicted to disrupt interactions that stabilize the low-pH structure reduce PAC inactivation. On the other hand, two mutations that accelerate inactivation (D91R, E94R) are predicted to stabilize the low-pH structure based on MD simulations. The work thus functionally supports the earlier hypothesis that the low-pH cryo-EM structure indeed represents an inactivated state. Moreover, it identifies several key titratable residues that are involved in this process.

The choice of the tested residues is based on strong structural evidence, and the electrophysiological data largely seem to support the conclusions, even though the analysis is not always rigorous. (Time constants seem unreliable as they are extracted from decay time courses that are too short to be reliably fitted, but comparisons of the simple parameter "fractional surviving current after 30 s" seem convincing enough.) Some of the mechanistic conclusions are largely based on MD simulations which I am not qualified to assess.

Reviewer #2 (Public Review):

In this paper, Osei-Owusu uses a combination of electrophysiology, structure-guided mutagenesis, and molecular dynamics to understand the desensitization of the proton-activated chloride channel (PAC). They show the extent and rate of desensitization is pH-dependent with lower pH promoting faster and more complete desensitization. They identify multiple residues with important roles in desensitization in two clusters at the extracellular end of TM1 and at the interface between the transmembrane and extracellular domains. Together with previously determined structures, the authors offer a model in which interactions between these residues play key roles in stabilizing the desensitized over the open conformation. This work provides important molecular insight into molecular mechanisms underlying the function of this widely expressed ion channel.

Reviewer #3 (Public Review):

The manuscript by the Qiu and Lu labs investigates the mechanism of desensitization of the acid-activated Cl- channel, PAC. These trimeric channels reside in the plasma membrane of cells as well as in organelles and play important roles in human physiology. PAC channels, like many other ion channels, undergo a process known as desensitization, where the channel adopts a non-conductive conformation in the presence of a prolonged physiological stimulus. For PAC the molecular mechanisms regulating this process are not well understood. Here the authors use a combination of electrophysiological recordings and MD simulations to identify several acidic residues and a conserved histidine side chain as important players in PAC desensitization. The results are overall interesting and clearly indicate a role for these residues in this process. However, there are several weaknesses in the experimental design, inconsistencies between the mutagenesis data and the MD results, as well as in the interpretation of the data. For these reasons I do not think the authors have made a convincing mechanistic case.

Major weaknesses:<br /> The underlying assumption in the interpretation of all the data is that the mutations stabilize or destabilize the desensitized conformation of the channel. However, none of the functional measurements provide direct evidence supporting this key assumption. Without direct evidence supporting the notion that the mutations specifically impact the rate of recovery from desensitization, I do not think the authors have made a convincing mechanistic case.

Overall, the agreement between the MD simulations, functional data, and interpretation are often weak and some issues should be acknowledged and addressed.<br /> For example:<br /> 1) The experimental data suggests that H98, E107, and D109 play analogous roles in PAC desensitization. However, the MD simulations suggest that the H98-D109 interaction energy is ~4 times larger than that of H98-E107. This should lead to a much greater effect of the D109 mutation. How is this rationalized?<br /> 2) The experimental data shows that E94 plays a key role in desensitization and the authors argue that this is due to the interactions of this residue with the β10-11 linker. However, the MD simulations show that these interactions happen for a small fraction, ~10%, of the time and with interaction energies comparable to those of the H98-E107-D109 cluster. It is not clear how these sparse and transient interactions can play such a critical role in desensitization. Also, if the interaction energies are of the same sign, how come one set of mutants favors desensitization and one does not?

The authors' MD analysis critically depends on assumptions on the protonation states of multiple residues, that are often located in close proximity to each other. In the methods, the authors state they use PropKa to estimate the pKa of residues and assigned the protonation states based on this. I have several questions about this procedure:<br /> - What pH was considered in the simulations? I imagine pH 4.0 to match that of the electrophysiological experiments.<br /> - Was the propKa analysis run considering how choices in the protonation state of neighboring residues affect the pKa of the other residues? This is critical because the interaction energies will greatly depend on the protonation state chosen.<br /> - Was the pKa for the mutant constructs re-evaluated? For example, does having a Gln or Arg in place of a His affect the pKa of nearby acidic residues?<br /> - H98R and Q have the same functional effect. The MD partially rationalizes the effect of H98R, however, it is not clear how Q would have the same effect as R on the interaction energies.<br /> - Are 600 ns sufficient to evaluate sampling of the different conformations?

Reviewer #1 (Public Review):

The authors suggest that there is a long-term periodicity of individual antibody response to influenza A (H3N2). The interesting periodicity may be surely appeared. Though the authors assume that the periodicity is driven by pre-existing antibody responses, the authors could provide more supportive data and discuss some possibilities.

1. The authors can investigate whether the periodicity reflects an epidemic/invasion record of A(H2N3) within Guangzhou or the surrounding city, e.g., the numbers of flu-infected people yearly can be referred to.

2. The authors can consider whether the participants are recently/previously vaccinated and/or infected with flu. The remaining antibodies may reflect a long memory but may show a recent activation.

3. The strains inducing high HI titers may have similar mutations and may be reactive to the same antibodies. What are the mutation frequencies among 21 A(H3N2) strains?

Reviewer #2 (Public Review):

This is a well-thought-out, clearly exposed article. It builds upon the platform of 'original antigenic sin' (OAS), a notion first developed from studying individuals infected with influenza. According to OAS, the initial infection will set the dominant immune response targets (antigens) that immune cells will recognize, such that infection with a related strain will cause a strong response focused mainly against the initially infecting strain, that then goes on to protect against the new-infecting strain. This study builds off this idea, showing that as strains become increasingly antigenically distant as inferred by the time between strain appearance, the cross-protection can drop to a point where it needs to be invigorated with a potentially new response. The potential biological mechanisms behind this aren't discussed, but a model is built that conveys the potential for 'relative risk' of an individual over the course of the life, based essentially on when one was born.

The basic premise was to measure from serum influenza haemagglutinin-inhibition (HI) titers of 21 strains of influenza A (H3N2) - related strains causing disease at various times over a period of some 40 years- from a diverse set of ≈800 participants of various ages, at two time points, spaced 2 yr apart. The authors then calculated the HI titer for the 21 strains for each individual. From this, each participant's age, their age at the time of a strain's development, and when a strain emerged were used to assess whether there was periodicity to immune responses by performing a splined Fourier transform for each individual and then examining the composite pattern across time for HI titers. The authors propose that on average there is a 24-year periodicity to immune responses to influenza strains, such that after the initial infection, cross-reactivity reduces to the point where it may be less meaningful for protection over around 24-year, and suggests activation of a 'new' immune response might be required to control the more distant strain involved in the response at that time. The periodicity was longer than would be predicted if age were not a factor involved in the HI titer patterns across time. Further, variability in the periodicity was shown to involve broad cross-reactivity between strains and narrow cross-reactivity in more highly-related (closer in time) strains, individual HI titer, and periodic population fluctuations. In the literature, viral strains are estimated to mutate to the point of losing 50% cross-reactivity with a T1/2 of approximately 2.5 yr, which would make the inferred lifespan plausible but perhaps surprisingly long, implying there are immune feedback parameters that influence periodicity. The authors also use an independent cohort of approximately 150 individuals from a separate, published, study to validate some findings revealed in the primary data set.

Strengths: Overall, the study is well executed and the patterns that are visually apparent in Figure 1A (the 'raw' data) are built on to inform a model of the potential breadth of cross-reactivity in a given individual at any given time after birth, integrated with the influenza strains to which they are most likely to have been first exposed. It is a complex thing to make sense of data involving many individuals who could be infected or vaccinated at any and variable points in time over the course of their life, but the authors derive a model that probabilistically accounts for possible infection events, so controls for this nicely, or at least to a degree that is practicable.

Questions related to the main limitation: The level of math in this paper makes it hard for a basic biologist to critique the approach, but the argued points are intriguing. Foremost, in the final part of the paper the authors move from building a model to testing its potential to predict HI titers in the final quarter strains of the study period, placing individuals into one of four phases: I) early increasing to high titer response, II) waning response phase where they are returning back to the average population-level response against a strain, III) sub-par response against a strain and then reinitiation of HI titers in phase IV. Pleasingly this shows a good correlation between individuals' ages and their predicted phase. However, while the fit predicts phase well in Fig 4C and 4D, it looks to perform less adequately in Fig 4B.

Q1: Why is this?

Another point for consideration is that the time between samplings (2010-2012) is comparatively short, given a 24-yr predicted periodicity. Q2: What would happen to the predictions if the periodicity were 35-yr or 6-yr? Would the model fail to call individuals accurately in these cases?

Q3: Similarly, if the samples were taken further apart, would the model still be effective at predicting phase?

Reviewer #1 (Public Review):

While the circuits underlying the computation of directional motion information in the fly brain are very well described, much less is known about the neurons serving the detection of objects. In a previous publication from the same lab, it has been shown that flies perform body saccades to track a moving object during flight. In the current paper, Frighetto and Frye provide evidence that T3 cells, a population of neurons within the optic lobes, are involved in this task. First, they performed 2-photon Calcium imaging from T3 cells to show that these cells respond to moving bars, which they later use in behavioural experiments. They then silenced T3 cells using genetic tools and tested the behavior of these flies in response to a rotating bar using two different setups. In one, the flies are fixed and bilateral changes in wing stroke amplitude are used as a measure for turning, in the other, flies are magnetically tethered such that they can rotate around the vertical body axis. Silencing T3 cells leads to the abolishment of the steering response induced by object position using a bar that is defined by its motion relative to the surround, but leaves the response to object motion intact. In the magnetically tethered flies, it reduces the number of saccades and thus leads to an impairment of bar-tracking behavior. In another set of experiments they optogenetically activated the whole population of T3 neurons (which supposedly impairs their normal function), which leads to an increase in the number of saccades after the activation (when the light stimulus used to activate the cells is turned off). Silencing the neurons necessary for detection of local motion, T4 and T5 cells, in contrast reduces responses elicited by object motion rather than position, but also has an impact on object tracking saccades. The authors provide a simple model, where speed-dependent signals from multiple T3 cells are integrated and trigger a saccade, when a threshold is reached.

The data generally support the conclusion that T3 cells play a role in detecting bar position and in controlling saccades in response to rotating bars. However, there are some inconsistencies in the data that are not sufficiently explored and discussed.

1. In a previous paper from the lab (Keleş et al., 2020), it was shown that T3 cells respond preferentially to small objects, whereas here they robustly respond to elongated bars and even large-field gratings. This discrepancy is not discussed.

2. In a previous paper, the authors showed that integrated positional error rather than bar position is used to elicit bar-tracking saccades and that saccade amplitude is relatively stereotyped. However, here they show, that T3 cells respond much more strongly to a slowly moving stimulus (18{degree sign}/s) rather than to the fast moving stimuli used for the behavioral experiments (> 90{degree sign}/s). This response property plays an important role for the model they propose. My general concern here is that the findings might not be generalizable to slower moving bars, where more precise, position-dependent responses could play a larger role, and that these fast moving bar stimuli represent an extreme situation, where the flies cannot accurately track bar position any more.

3. The claim that T3 cells are tuned to stimulus velocity is not supported by the data in my view. For the bar stimuli, the authors only tested speeds of 18{degree sign}/s and above 90{degree sign}/s, but nothing in between. For the grating motion there seems to be an influence of temporal frequency for the same stimulus velocity (see e.g. Fig.1_1), but this is not quantified.

4. The results from the optogenetic activation experiments are hard to interpret, as it is unclear how a prolonged activation of all T3 cells would affect the downstream circuitry. It is not clear that this experiment is equivalent to a "loss-of-function perturbation" of T3 cells as the authors claim in the text.

Reviewer #2 (Public Review):

In their manuscript titled "Feature detecting columnar neurons mediate object tracking saccades in Drosophila", Frighetto & Frye study the effect manipulating T3 neurons has on tethered flight saccades. The authors first characterize the responses of T3 neurons to simple visual stimuli, and then manipulate T3 cells (with both Kir2.1 and CsCrimson) and study the effects on the fly's tethered flight behavior, focusing on different types of sharp turns (saccades). Finally, the authors suggest an integrate and fire model to explain how an array of T3-like neurons can produce some of the recorded behavior.

The authors study the elementary, yet challenging, computation of object discrimination. They hone in on a cell type that most likely plays an important role in the circuit. However, the authors do not sufficiently clarify the framework in which they conceptualize T3's role in object discrimination, neither when discussing it in the introduction/discussion nor when explaining experimental results. The authors present the work in comparison to T4/T5 cells. However, T4/T5 cells have been shown to be both local motion detectors and the main cell types to compute motion in the fly's eye. Downstream neurons integrate over these local units to detect different patterns of global and local motion (Authors should cite Krapp 1996 Nature). Are the authors suggesting that T3 neurons perform a similar function only as local object detectors? That is a bold claim that will need to be supported with more experimental results and reconciled with previous results. We already know of other Lobula Columnar neurons (LCs) that respond to different sizes, some even smaller than the optimal T3 stimulus (e.g. Klapoetke 2022 Neuron) and we know of LCs that respond to small objects that do not receive major inputs from T3 cells (e.g. Hindmarsh 2021 Nature).

These differences between T4/T5 cells and T3s also make interpreting the experimental manipulations more challenging. When hyperpolarizing T4/T5 or 'blinding' them with CsCrimson activation, the visual motion circuit is severely disrupted. However, the same cannot be said about inactivating/blinding T3 neurons and the object detection circuit (if it is indeed a single circuit). The authors are justified in deducing a connection between blocking T3 neurons and a reduction in bar tracking, but generalizing the results to object detection requires more experiments and clarifications.

When framing the manuscript in the object detection framework, previous results regarding the definition of an object should also be addressed. Maimon Curr. Biol. 2008 and work from their own lab (Mongeau, 2019) have already shown that tethered flies respond differently to bars and small objects (fixating on the former while anti-fixating on the latter). Previous work has also shown that T3 neurons respond strongly to small objects and suppress responses to long bars (Tanaka Curr. Biol. 2020). Since all the behavioral experiments in the current manuscript and all the visual stimuli are full arena-length bars, it is impossible to tell whether the T3 results generalize to small objects and even how to reconcile the stronger response to small objects with the role ascribed to T3 cells in generating behavioral responses to long bars.

Finally, the authors propose a model for a hypothetical neuron downstream of T3 that would integrate over several T3s and generate saccades. However, given the current knowledge level in the fly vision field, the model should either be grounded more in actual circuit connectivity or produce testable predictions that would guide further research.

The authors should decide whether they would like to address these concerns with more specific experiments that would shed light on the role T3 has to play under different conditions and different definitions of a visual object, or whether they would prefer to limit the scope of their claims.

Reviewer #3 (Public Review):

In free flight, flies largely change their course direction through rapid body turns termed saccades. Given how important these turns are in determining their overall behavior and navigation, it is important to understand the neural circuits that drive the timing of triggering these saccades, as well as their amplitude. In this paper the authors leverage the powerful genetic tools available in the fruit fly, Drosophila, to address this question by performing physiology experiments as well as behavioral experiments with inactivation and activation perturbations.

The authors make three primary conclusions based on their experiments: (1) the feature detecting visual pathway (T3) is responsible for triggering saccades in response to moving objects, but not widefield motion, (2) the pathway primarily responsible for wide field motion encoding (T4/T5) is responsible for triggering saccades in response to widefield motion, and (3) the T4/T5 pathways is responsible for controlling the amplitude of both object and widefield motion triggered saccades.

The authors go on to show that using calcium imaging data of T3 activity it is possible to predict under what conditions flies will initiate a saccade when presented with objects moving at different speeds, resulting in a parsimonious model for how saccades are triggered.

Together, the imaging, behavior, and modeling provide compelling evidence for claims 1 and 2, however, the evidence and modeling for point 3 - the amplitude of the saccades - is lacking. The statistical analysis does not go into sufficient detail in comparing across different cases, and in particular, there is little mention of the effect sizes, which appear to be quite small (this is primarily in reference to 3F and 4E). The data suggest that both the T3 and T4/T5 pathways contribute to saccade amplitude, instead of T4/T5 being the only or primary drivers.

Reviewer #1 (Public Review):

This work provides a new general framework for estimating missing data on cervical cancer epidemiology, including sexual behavior, HPV prevalence, and cervical cancer incidence. These data are useful to determine impact projections of cervical cancer prevention. The authors suggest a three-step approach: 1) a clustering method applied on registries with an intermediate level of data availability to cluster cervical cancer incidence based on a Poisson-regression-based CEM algorithm, 2) a classification method applied on registries with a low level of data availability to classify cervical cancer incidence based on a Random Forest, 3) a projection method applied on missing data based on the mean of available data. The authors use India as a case study to implement this new methodology. Results indicate that two patterns of cervical cancer incidence are identified in India (high and low incidence), classifying all Indian states with missing data to a low incidence. From this classification, missing data is approximated using the mean of the available data within each cluster.

A strength of this approach is that this methodology can be applied to regions with missing data, although a minimum set of information is needed. This makes it possible to have individual data for each unit in the region.

One of the weaknesses of this methodology is the need for a minimum set of epidemiological data to enable impact projections. It is true that when epidemiological cervical cancer data is not available, authors mentioned that general indicators (e.g., human development index, geography) can be used but projections will be probably less realistic. As observed with other techniques, countries with fewer resources have less data available and cannot benefit from these types of techniques to have more adequate guidelines.

Imputation of missing data is always a challenging issue. The technique proposed in this manuscript is an interesting new approach to missing data imputation that could be applied with a minimum set of available data. However, we must focus on obtaining reliable data from each region of the world to help local health authorities implement better preventive measures for the local population.

Reviewer #2 (Public Review):

The burden of cervical cancer worldwide is well recognized. While prevention strategies, including vaccination against human papillomavirus (HPV), cervical cancer screening, and pre-cancer treatment, can reduce the burden of cervical cancer, access to these measures is still limited, especially in low- and middle-income countries. Since the impact of prevention strategies is heavily dependent on the disease's burden on a particular population, we need to know the latter to assess the impact of these context-specific prevention strategies.

However, epidemiological data on cervical cancer are not always available for all geographical areas. This paper uses India as a case study to propose a framework called "Footprinting" to comprehensively evaluate the burden of cervical cancer. The authors applied a three-step analytical strategy to impute cervical cancer epidemiological data in states where this information was unavailable using data from cervical cancer incidence, HPV prevalence, and sexual behaviour from other regions. The findings suggest a high and low incidence of cervical cancer incidence in different parts of India; all Indian states with missing data were classified as low incidence.

The proposed analytical strategy presents an important solution for imputing data from geographic areas of a country where data are missing.

One conceptual limitation of this work is the lack of explanation or evidence that sexual behaviour can be used to approximate cervical cancer and/or HPV rates. Also, full information on the three main indicators is only available in two states. This is used to impute the values for the other states. Moreover, the available data used in this study also present some limitations; for example, cervical cancer incidence data were from 2012 to 2016, while sex behaviour data were from 2006. This large gap is likely to have a significant cohort effect, especially given changes in sexual norms in Western countries over the last few decades, which may have gradually influenced other countries, especially in this age of the internet and social media. Finally, it would be interesting to validate this methodology to confirm its utility.

The proposed framework's strength is difficult to evaluate because the steps and justification for the model variables were not clearly presented, nor were the models validated. Based on the authors' interpretation of the framework findings, this framework may help extrapolate data from one country to another. I'm curious as to whether this framework could be applied across states and countries.

Reviewer #1 (Public Review):

The manuscript by Masschelin et al. describes how Vitamin B2 deficiency affects body composition, energy expenditure, and glucose metabolism. B2 deficient mice have lower O2 consumption, and locomotor activity, with no difference in food intake. These mice also have lower liver FAD levels, which is expected given that B2 is a necessary cofactor for this coenzyme. Additionally, these mice have lower blood glucose levels following pyruvate injection, implying a lower capacity for gluconeogenesis. Using PPAR KO mice, they show that this effect on pyruvate tolerance is due to PPARα activation, though there is still a minor difference between wild-type and KO mice. Importantly, they show that fenofibrate PPARagonism can improve glucose output following pyruvate injection in the absence of B2. The authors also perform robust metabolomics in each experimental condition and phenotype of the mouse well.

1. The authors have yet to explore other explanations of differences in glucose metabolism under B2D +/-Fenofibrate. The canonical targets of PPARα are involved in fatty acid oxidation, ketogenesis, and VLDL/HDL metabolism, in addition to gluconeogenesis (Bougarne et al. 2018). Gluconeogenesis is more of a fasting response due to CREB, FOXO1/PGC1activation rather than PPAR. In response to B2D, the PPARα KO mice have increased plasma TGs, which may suggest a difference in VLDL TG secretion (Suppl. S3). Perhaps lipid metabolism is more directly affected, and changes in glucose metabolism are secondary to that of triglyceride metabolism. Regarding ketogenesis, the fenofibrate+ B2D fed mice have decreased plasma beta-hydroxybutyrate, suggesting decreased ketogenesis, which is a more canonical PPARα pathway (Suppl. S3). Testing each of these processes would help control that this mechanism is specific to gluconeogenesis and not secondary to something else.

2. Is the effect on ISR dependent on PPARα? Is the mechanism of Fenofibrate on the liver, or on another cell type? In Figure 1, the authors state that Riboflavin deficiency alters body composition and energy expenditure, and then focuses on the liver. However, FAD levels are also increased in the heart and kidneys in addition to the liver. These tissues also respond to PPARα agonism, in addition to the muscle which plays a role in regulating glucose metabolism (B2D mice also have a higher lean mass (Fig 1e)). Additionally, the authors haven't shown specifically if the effects of fenofibrate on electron transport and the ISR are dependent on the presence of PPARα (Figure 5, 6).

Reviewer #2 (Public Review):

The objective of this work by Masschelin et al. is to investigate the physiological relevance of flavin adenine dinucleotide (FAD). In particular, FAD supports the activity of flavoproteins involved in the production of cellular energy. Mutations in genes encoding flavoproteins often are associated with inborn errors of metabolism (IEMs), thus the clinical interest in investigating in more depth the physiological role of FAD. In this study, the authors first subjected male mice to a vitamin B12 deficient diet (B2D), demonstrating that loss of B12 replicates the phenotypes often observed with IEMs, including loss of body weight, hypoglycemia, and fatty liver. Using a combination of metabolomic phenotyping, transcriptomic analyses, and pharmacology (treatment with fenofibrate, a PPARa agonist), the authors then reach the general conclusion that activation of the nuclear receptor PPARa can rescue the B2D phenotypes, thus revealing that PPARa directly controls the metabolic responses to FAD availability. Although the phenotypic analysis of the mice subjected to B2D increases our knowledge of the physiological impact of depleting the FAD pools on global energy metabolism, not all conclusions and statements made by the authors are totally supported by the data. In particular, the study is overall too descriptive and lacks mechanistic insights. While PPARa is likely an important player in the metabolic response to FAD availability, the molecular details on how FAD controls the activity of PPARa either directly or indirectly are entirely missing. Therefore, the authors are encouraged to directly assess whether B2D directly influences PPARa activity on the genes identified in the study, perform rescue experiments in the liver of PPARa KO mice and explore the possibility that other factors (including nuclear receptors) also participate in the response to B12 deficiency and diminished FAD pools.

Reviewer #1 (Public Review):

In their manuscript, Haenelt et al. investigated the structure-function relationship for cortical columns in the in vivo human brain. The example they used is the thick stripe - pale stripe - thin stripe organisation of secondary visual cortex (V2).

The specific strength of the current study lies in the combination of cutting edge imaging protocols for quantitative measurements of myelin-related signals (qMRI) together with functional activation, both at submillimeter resolution at high field (7T). This allowed the visualisation of the stripy organisation of V2 with regards to colour (thin stripes) and binocular disparity (thick stripes) as well as myelination in individual human subjects. The main results suggest higher myelination for the pale stripe regions. This is in line with some earlier studies, across primate species, but not with others.

One potential issue is that the high myelination signal is associated with the compartment in V2 (pale stripes) which was not functionally defined itself but by the absence of specific functional activations. No difference was reported between those stripes that were defined functionally. Other explanations for the differential pattern of a qMRI signals, e.g. ROI distribution for presumed pale stripes is not evenly distributed (more foveal), ROIs with low activations due to some other factor show higher myelin-related signals, cannot be excluded based on the analysis presented.

Another theoretical and practical issue is the question of "ground truth" for the non-invasive qMRI measures, as the authors - as their starting point - roundly dismiss direct histological tissue studies as conflicting, rather than take a critical look at the merit of the conflicting study results and provide a best hypothesis. If so, they need to explain better how they calibrate their non-invasive MR measurements of myelin.

While this paper makes an important contribution to the question of the association of specific myelination patterns defining the columnar architecture in V2, it is not entirely clear whether the authors can fully resolve it with the data presented.

The highly sophisticated methods and detailed analysis show that high resolution investigation of structure-function relationship of the columnar organisation in human visual cortex are feasible and reliable. V2 stripe patterns can be visualised structurally (with quantitative myelin-related measurements) and functionally (based on functional selectivity, which is of considerable importance for the field. The results indicate that in humans, the pale or inter strip regions might be associated with high patterns of myelination.

Reviewer #2 (Public Review):

This is a nice study that uses cutting-edge MRI measurements in the context of a carefully designed visual experiment. The data would seem to be of high quality and in general, the approach is promising for opening up avenues for non-invasive measurements of cortical myelination.

Unfortunately, this particular study seems to fall into an unhappy middle ground in terms of the conclusions that can be drawn: the relaxometry measures lack the specificity to be considered "ground truth", while the authors claim that the literature lacks consensus regarding the structures that are being studied. The authors propose that their results resolve whether or not stripes differ in their patterns of myelination, but R1 lacks the specificity to do this. While myelin is a primary driver of relaxation times in cortex, relaxometry cannot be considered to be specific to myelin. It is possible that the small observed changes in R1 are driven by myelin, but they could also reflect other tissue constituents, particularly given the small observed effect sizes. If the literature was clear on the pattern of myelination across stripes, this study could confirm that R1 measurements are sensitive to and consistent with this pattern. But the authors present the work as resolving the question of how myelination differs between stripes, which over-reaches what is possible with this method. As it stands, the measured differences in R1 between functionally-defined cortical regions are interesting, but require further validation (e.g., using invasive myelin staining).

Moreover, the results make clear that R1 differences are not sufficiently strong to provide an independent measure of this structure (e.g., for segmentation of stripe). As such, one would still require fMRI to localise stripes, making it unclear what role R1 measures would play in future studies.

The Introduction concludes with the statement that "Whereas recent studies have explored cortical myelination ... using non-quantitative, weighted MR images... we showed for the first time myelination differences using MRI on a quantitative basis". As written, this sentence implies that others have demonstrated that simpler non-quantitative imaging can achieve the same aims as qMRI. Simply showing that a given method is able to achieve an aim would not be sufficient: the authors should demonstrate that this constitutes an important advance.

The study includes a very small number of participants (n=4). The advantage of non-invasive in-vivo measurements, despite the fact that they are indirect measures, should be that one can study a reasonable number of subjects. So this low n seems to undermine that point. I rarely suggest additional data collection, but I do feel that a few more subjects would shore up the study's impact.

The paper overstates what can be concluded in a number of places. For example, the paper suggests that R1 and R2* are highly-specific to myelin in a number of places. For example, on p7 the text reads" "We tested whether different stripe types are differentially myelinated by comparing R1 and R2*..." Relaxation times lack the specificity to definitively attribute these changes purely to myelin. Similarly, on p11: "Our study showed that pale stripes which exhibit lower oxidative metabolic activity according to staining with CO are stronger myelinated than surrounding gray matter in V2." This implies that the study directly links CO staining to myelination. In addition to using non-specific estimates of myelination, the study does not actually measure CO.

I'm confused by the analysis in Figure 5. I can appreciate why the authors are keen to present a "tripartite" analysis (thick, thin, and pale stripes). But I find the gray curves confusing. As I understand it, the gray curves as generated include both the stripe of interest (red or blue plots) and the pale stripes. Why not just generate a three-way classification? Generating these plots in effect has already required hard classification of thin and thick stripes, so it is odd to create the gray plots, which mix two types of stripes. Alternatively, could you explicitly model the partial volume for a given cortical location (e.g., under the assumption that partial volume of thick and thin strips is indicated by the z-score) for the corresponding functional contrast? One could then estimate the relaxation times as a simple weighted sum of stripe-wise R1 or R2.

Reviewer #3 (Public Review):

Haenelt et al. used sub-mm resolution fMRI and quantitative R1 and R2*imaging in humans to investigate the relationship between putative myelin densities and functional responses confined to different mesoscale sub-compartments of area V2. Specifically, they presented color and disparity-varying stimuli, which are known to preferentially activate thick and thin V2 stripes in human and nonhuman primates. Based on these color and disparity signals, they created ROIs corresponding to the color-biased thin stripes, disparity-biased thick stripes, and the third non-thick non-thin compartment, putatively corresponding to the pale (or inter) stripes. Comparison of the R1 values across these functionally defined V2 sub-compartments revealed lower R1 values in both the color-biased thin and disparity-biased thick stripes relative to the putative pale stripes. The interpretation is that myelin densities in pale stripes is higher than in the two other V2 compartments, which corroborates previous studies using post-mortem Gallyas staining (myelin) in primates (yet not other histological studies using other markers for myelin density). The authors conclude that the combination of high-resolution high-sensitive quantitative and functional MRI enables studies whereby the relationship between anatomical and functional properties can be investigated in-vivo.

This study builds upon previous studies of the authors who now combined forces to merge their respective skills in mesoscale functional imaging on the one hand and quantitative MRI on the other hand. The distinction between color- and disparity-biased thin and thick stripes has been previously shown by Nasr, Polimeni and Tootell, yet it is the combination with R1 and R2* imaging that is unique in this study. Dumoulin et al. previously used T1/T2 ratios instead of R1 and R2*values to investigate exactly the same question. Surprisingly, that previous study led to the opposite conclusion, as they showed that pale stripes contain lower myelin densities compared to thick and thin stripes -possibly due to the use of other functional markers in their attempt to differentiate between thin and thick stripes, as also discussed in the present manuscript. The only other study, to the best of my knowledge, that used MRI techniques to separate the three stripe compartments in are V2, was a macaque study, also using the T1/T2 ratio as a surrogate for myelin densities. That monkey study yielded basically the same results as the current study by Haenelt and colleagues: pale stripes are more myelinated than the thick and inter stripes.

Hence the present study aids to resolve existing and important controversies in both the histology and (f)MRI literature. It needs to be kept in mind, however, that all the MRI measures used so far are a 'proxy' for determining myelin densities, hence the final ground-truth will have to come from a combination of functional studies with (novel?) histological methods to determine exactly myelin densities, which can then be used to compare with functional properties segregating the three V2 compartments.<br /> Given the prior discrepancies between histological studies and between different MRI studies, and given the intrinsic importance to link function to fine-grained structural properties, the present study is potentially of great importance for the neuroimaging field -despite the relative small number of participating subjects. The experiments are well performed using state-of-the-art equipment, the analyses are well-done and the writing is excellent showing the scholastic skills of the authors. In addition, the authors discuss and exclude a number of alternative explanations for their results, which is highly informative for the reader.

Reviewer #1 (Public Review):

This manuscript attempts to disclose new insights into barrel cortex cell class-dependent and cell depth-dependent membrane potential (Vm) dynamics during active whisker sensing. The results highlight similarities but also specific differences between different types of cortical neurons. The approach used is very effective and direct: somatosensory stimulation is performed in awake animals without anesthesia, the neurons are recorded with intracellular whole cell patch clamp recording that can provide fast responses with high resolution, and the identification of various neuron types is achieved by using mice expressing genetically defined selective fluorescent markers. The results support the main conclusions. The work is an extension of previous, similar work performed by this group, However, most previous Vm studies in the mouse barrel cortex during behavior have largely focused on superficial neurons located in the upper ~300 μm of the neocortex since these are more easily targeted through two-photon microscopy. In this study, the authors extend current knowledge by investigating Vm dynamics across a greater range of depths including two-photon targeted whole-cell recordings across the upper ~600 μm of the neocortex. I believe that this manuscript uses a demanding, but excellent approach that will be useful to other researchers in the field. The manuscript is likely to be influential.

Reviewer #2 (Public Review):

This paper is a technical tour de force and provides interesting results. This group has indeed contributed to the understanding of membrane potential and firing dynamics of different cortical neuron subclasses during sensation in various previous papers. Yet, the paper falls short in providing a cohesive conclusion and interpretation of their results on pyramidal neurons, PV, SST, and VIP cells in response to free whisking and active touch at different cortical depths. The authors clearly claim that this manuscript aims to extend the current knowledge by investigating Vm dynamics of pyramidal neurons and various GABAergic subtypes across a greater range of cortical depths. The major shortcoming of this paper is indeed a lack of a clear conclusion or picture of how different cortical neuron types are engaged by different states. Overall, I struggle to find a novel message emerging from the present manuscript that hasn't already been described by the same lab. And this is a pity, as the experiments are of the highest quality and the data is definitely hard-won.

Reviewer #3 (Public Review):

The manuscript by Kirtani et al. describes intracellular recordings from barrel cortex neurons identified under 2p microscopy in vivo during whisking. The major strengths of this work are that it is a technical feat and represents a unique dataset. It is a building block for future studies. The major weakness however is that it is a purely descriptive and observational study. There are no experimental manipulations, nor are there attempts to integrate the observations into a larger framework. As a result, there are no mechanistic or functional insights from this study. There is some speculation and discussion about how these results might fit into other studies of circuit connectivity or computational modeling, however, but this is relatively limited.

Reviewer #1 (Public Review):

Proton Activated Chloride (PAC) channels have been recently identified as important contributors to endosomal acidification, and their activity in the plasma membrane increases under certain pathological conditions and can induce cellular death. There is very limited information on the pharmacology of these ion channels. By recording from endogenous PAC channels stimulated with an acidic extracellular solution in HEK293 cells using the patch-clamp technique, this study finds that PAC channels are inhibited physiological concentrations of the soluble short-chain PIP2 analog dic-8-PIP2. Inhibition is quantified for several PIP2-related lipids with different number of headgroup phosphates and shorter or longer acyl chains, and it is found that an acyl chain with more than 8 carbons and a negative headgroup charge are both required for robust inhibition. Importantly, inhibition appears to result from PIP2 incorporated into the outer membrane leaflet, as treatment of the inner leaflet with PIP2 or poly-lysine to either increase or decrease PIP2, respectively, did not have any effect on channel activity, as opposed to when the lipid is extracellularly applied. A structure of the channel in the presence of PIP2 was obtained using single-particle cryo-electron microscopy - the structure resembles a previously observed conformation for PAC channels that likely represents a non-conducting desensitized state, and it contains densities with a shape that is consistent with a bound PIP2 molecule in the outer leaflet. Mutations to alanine based on the channel-lipid interactions observed in the structure were all found to disrupt inhibition of PAC channels by PIP2, consistent with the location of the lipid binding site proposed in the study. By comparing the amino acid sequence of human PAC channels with those of other species, it is found that the proposed lipid binding site is highly conserved except in zebrafish. Notably, zebrafish PAC channels are less susceptible to inhibition by PIP2, and mutation of residues at the binding site to those present in the human channel increases inhibition, consistent with the proposed location of the binding site for PIP2. Finally, it is found that the kinetics of inhibition by PIP2 are positively correlated with the degree of channel activation and also with the kinetics of desensitization, suggesting that PIP2 binds more favorably to the desensitized state of the channel whereas it does not bind to the closed state, providing a possible mechanism for the inhibition.

Results are clearly reported and findings are generally robust. One concern is that most of the electrophysiological characterization of the inhibition of PAC channels by PIP2 lipids was done using endogenously expressed channels. It is unclear why this was done because mutant channels are studied in a PAC KO cell line that could have been used for all experiments. The effects of acidic pH and acidic pH + PIP2 in cells that do not express PAC channels is therefore not shown, but would be important to establish that the measured effects of the lipid are specific to PAC channels.

Another concern for the study is related to the uncertainty in establishing that the bound lipid is indeed PIP2. Although the mutagenesis results are all consistent with the proposed binding site, it remains a possibility that the mutations affect PIP2 inhibition indirectly by e.g. changing the rate of channel desensitization, which was not measured for any of the mutants on Figure 3E. There is not additional analysis performed to determine whether other types of lipids could occupy the density that is proposed to represent PIP2. Although this might be difficult because no density for the headgroup of the lipid was observed.

A final caveat of the study is that effects of PIP2 on the extracellular leaflet might be non-physiological. If this were the case, however, the identification of a non-physiological binding site that favors desensitization might still be beneficial for drug design in the context of this channel.

Reviewer #2 (Public Review):

Proton-activated chloride channel (PAC or ASOR) is a newly discovered anion channel which has a broad tissue expression and is implicated in important physiological processes, such as regulation of endosomal acidification and macropinocytosis. PAC is also implicated in pathological conditions related to acidosis on the plasma membrane. Since its discovery and initial characterization, several structures were solved in resting, activated and desensitized states, revealing an overall channel architecture and its mechanism of action. However, little is known about modulation of PAC channel by endogenous molecules. In the present manuscript, the authors sought to explore the modulation of PAC by lipids, particularly by PIP2, as this lipid is known to modulate numerous unrelated membrane proteins.

The major strength of the manuscript is the variety of approaches which the authors implement to characterize the mechanism of modulation of PAC by PIP2. Firstly, the authors demonstrate that PIP2 inhibits PAC channel if applied extracellularly. Furthermore, the authors demonstrate that PIP2 acts on the activated/poised towards desensitization, and not on the resting state of the channel. To explore the effect further, the authors tested various PIP molecules, varying in the number of phosphates in the inositol headgroup, and the length of acyl chains. The inhibition of PAC was more potent with the increase of the number of phosphates, and with the lengthening of acyl chains. The lipid chain without inositol, or the inositol without acyl chains, were not as potent in inhibiting PAC. The authors conclude that inositol headgroup together with acyl chains of at least 8 carbons in length are both required to potently inhibit PAC.

To investigate the potential PIP2 binding site, the authors proceeded to solve the structure of PAC in complex with PIP2. Surprisingly, a density representing a putative PIP2 molecule is found on the extracellular side of the protein. This is a rather unusual finding, given that PIP2 is mostly localized to the inner leaflet of the plasma membrane. To further confirm the binding of PIP2 molecule to this site, the authors mutate the residues interacting with PIP2 molecule in their structure, and observe the decrease in inhibition of the channel by PIP2. Furthermore, the authors observe that these residues are not conserved in all PAC homologs. D. rerio PAC channel does not have these residues and is not inhibited by PIP2 as potently as the human homolog (hPAC). Introducing equivalent residues in D. rerio PAC channel endowed it with modulation by PIP2, similar to hPAC, further strengthening the conclusion that the identified site indeed binds PIP2.

Overall, the authors succeeded in identifying and characterizing an endogenous molecule with the potential to modulate PAC channel. The present study is the first case of identifying a modulator, characterizing its binding site and mechanism of action on PAC channel. This opens new exciting avenues for structure-guided drug design for this newly-discovered ion channel. However, the localization of the PIP2 binding site to the outer membrane leaflet is quite unexpected, and it is unclear if PAC could be modulated by PIP2 in a physiological context and whether this would be mediated by another lipid transporter. The work will be of interest to ion channel field and a broader membrane protein community with the emphasis on lipid modulation of membrane proteins.

Reviewer #3 (Public Review):

This compelling manuscript by Mihaljević et al. describes an unusual regulatory mechanism for the proton-activated channel (PAC) where phosphatidylinositol (4,5)-biphosphate (PI(4,5)P2) inhibits the channel by direct interaction with a binding pocket in its extracellular/lumenal domain. This conclusion is supported by electrophysiology data collected on endogenously expressed channels in a human cell line. The authors support their finding with a structural model of acyl groups determined by cryo-electron microscopy. The core experimental design is sound and the data support the narrow conclusions of the paper.

This manuscript must consider the biological context of PI(4,5)P2 and the relevance of this interaction. Previous studies have documented that PI(4,5)P2 exists on the outer leaflet of the plasma membrane, but as a minor component relative to the overall levels of membrane PI(4,5)P2. The same applies for endosomes, where PIPs are enriched on the cytosolic membrane. The inositol headgroup is unresolved in the structural model of PI(4,5)P2-bound PAC, indicating that this interaction is nonspecific for PI(4,5)P2. This brings up the question as to whether PI(4,5)P2 is the relevant endogenous antagonist for PAC or whether it is a proxy for another ligand that has yet to be determined.

Reviewer #1 (Public Review):

In this study, Luo, Han, and Yin et al. conduct a fecal microbiota transplant from MSTN KO pigs exhibiting hypertrophy to recipient antibiotic-depleted B6 mice. The microbiota transplants successfully transferred muscle hypertrophy phenotypes to the mice. Aspects of the pig gut microbiome were recapitulated in the recipient mice, including a higher abundance of Romboutsia and valeric acid. The authors then demonstrate that 5 weeks of daily gavage of valerate, but not isobutyrate or water, was sufficient to increase type IIb myofiber growth and GA muscle mass, and protect mice against dexamethasone-induced muscle atrophy. Taken together, these data neatly demonstrate that genetic disruption of the myostatin gene results in a microbiome-dependent increase in valeric acid, which in turn results in significantly altered skeletal muscle growth.

Reviewer #2 (Public Review):

I would like to congratulate the authors for testing the hypothesis that the gut microbiome from animals that lack myostatin is sufficient to improve muscle-related measures (except treadmill running time). Subsequent experiments should examine if the identified bacteria are sufficient, on their own, to impact muscle, which may open the field to muscle-improving probiotics. Alternatively, data for the SCFA, valerate, may foster approaches aimed at improving muscle with SCFA supplementation. RCTs are needed to test these hypotheses.

Strengths include a translational approach, including findings in pigs, in colonized mice, and in cells.

Weaknesses include the need to normalize muscle-related measures to body weight. Is muscle mass increased, for example, when divided by body weight? If not it would argue against the role of fecal transplantation in increasing muscle mass from myostatin KO pigs.

The authors achieved their aims, and the results support their conclusions.

Reviewer #3 (Public Review):

The link between gut microbiota and maintenance of skeletal muscle mass was demonstrated in previous publications (including Lahiri et al., 2019), which also revealed that supplementing germ-free mice with a cocktail of short-chain fatty acids (SCFAs) could rescue the decreased skeletal muscle mass of germ-free mice. Increased MSTN expression in skeletal muscle causes sarcopenia (Cho et al., 2022). Moreover, the idea that Myostatin (MSTN) changes the composition of intestinal microorganisms is not novel (Pei et al., 2021 and Wen et al., 2022). In this manuscript, Quan et al. showed that knockout of MSTN in pigs affected the composition of gut microbes and that fecal microbiota transplantation (FMT) from MSTN KO pigs into mice caused hypertrophy of the GP muscle via activation of the Akt/mTOR pathway and increased presence of fast type IIb fibers. This effect was attributed to MSTN KO FMT-derived valeric acid, a SCFA, which when administered alone could recapitulate the phenotype of mice that were subjected to MSTN KO FMT. While the phenotypic results of this study are convincing, it lacks novelty in that the mechanisms that are studied were previously known. Instead, it would be interesting to explore how exactly does MSTN affect the composition of gut microbiota. This question was only briefly addressed (the authors showed that MSTN KO leads to changes in intestinal structure), however, a causal relationship was not established. Also, it is unclear how the mechanism of action of valeric acid is any different from the cocktail of acetic acid, butyric acid, or propanoic acid that was previously used. Therefore, overall, this study scores lowly in uniqueness. Nevertheless, the link of gut microbiota to MSTN is interesting and should be pursued by the authors in greater detail.

Reviewer #1 (Public Review):

The authors have used eye image data to create an aging clock of the retina in data from eyePACS with validation in the UK Biobank. They show that the clock predicts mortality independently of chronological age and that it is correlated with phenotypic age. Moreover, a GWAS is conducted in the UK Biobank, which identifies novel genetic loci and a top site located in the ALKAL2 region that is functionally validated in a drosophila model. Overall, the study is interesting with sound methodology and is a nice contribution to the field providing a GWAS summary statistic of the eye clock useful for follow-up analyses.

Reviewer #2 (Public Review):

This paper reports a novel measure of biological age derived from machine-learning analysis of retinal imaging data with chronological age as the criterion measure. The resulting algorithm is impressive. Not only can the retinal image data accurately predict chronological age in the training data and record changes over short time intervals, but it also proves accurate in independent test data and appears to contain information related to mortality risk. In addition, the authors report a GWAS of the new measure.

I would like to see a bit more validation data in the UKB - how does EyeAge relate to (a) tests of visual acuity - e.g. does it explain aging-related differences? (b) measures of morbidity and disability - e.g. how is EyeAge Accel associated with at least some of the counts of chronic diseases, self-reported physical limitations, tests of physical performance, measures of fluid intelligence?

But overall, this is a very strong report of an exciting new biomarker of aging. It was unclear to me whether the algorithm to compute the measure would be publicly available. The authors should clarify.

Reviewer #1 (Public Review):

This work leverages single-cell RNA-sequencing to probe changes in various immune functionings within the ovary in aging. The data provided is the most comprehensive of ovarian immune cells at the resolution of single-cell transcriptomics to-date and will be valuable to other researchers. The authors explore four distinct immune functionings:

- The authors identify macrophages and a unique CD3+CD8-CD4- T-cell subpopulation that change in abundance with aging. While these are interesting findings that align with flow cytometry results, the lack of batch correction and application of single-cell differential abundance tools limit the strength of the claims. The authors also do not further probe gene expression changes specific to these populations.

- The authors also analyze changes in global gene expression across cell types using an enrichment analysis; Figure 3B specifically is an excellent visualization summarizing potential global and cell-type specific changes in gene expression programs during aging.

- The authors infer differences in cell-cell communication mediated by various chemokines and cytokines. In this analysis, they claim a decreased inflammatory response due to aging. Here, the global decrease in gene expression in many cell types is not accounted for. Visualizations and quantitative analyses could benefit from existing, specialized in cell-cell communication tools.

- A discussion of changes to the expression of SASP receptors on immune cell types.

While both the data and biology presented are quite interesting, this study is perhaps too wide in breadth such that no individual result is extensively and rigorously explored.

Reviewer #2 (Public Review):

The paper by Ben Yaakov et al. describe a single cell analysis of the mammalian ovary in young, adult and old mice. In comparison with previous studies that used single cell RNAseq to characterize the heterogeneity of cell types in the ovary, this study focuses only on immune cells resulting in much better coverage to characterize the changes that these cells undergo as a function of age. The paper provides a useful dataset and informative data analysis with interesting findings including the increases in DNT cells in the ovary of old mice. Some discussion on how the presented results might be related to reduced fertility with age would be good to tie the results back to the original questions with which the authors start their paper.

Reviewer #3 (Public Review):

This work studied age-related alterations in the ovarian immune cells in mice using single-cell RNA sequencing and flow cytometry. Based on gene expression profiles, the authors identified cell clusters corresponding to immune cell populations in mouse ovaries and compared their abundance in aged compared to adult animals. The authors identified two parallel immune processes in aging ovaries: a decrease in proportions of myeloid cells such as macrophages and neutrophils accompanied by an increase in proportions of CD3+ T cells. The latter cell population was increased in abundance due to an expansion of CD3+ cells that do not express CD4 and CD8, referred to as "double-negative T cells." These immune alterations were identified by single-cell RNA sequencing using small numbers of mice, and the authors partially validated the data using flow cytometry analysis in larger groups of animals. In addition, based on the gene expression data, they predicted which signaling pathways were altered in the aged immune cells and analyzed putative changes in the chemokine and cytokine networks, pointing at potential crosstalk of immune cell populations with senescent cells in aging ovaries.

The combination of single-cell RNA sequencing and flow cytometry used by the authors is a robust and unbiased approach to characterize immune cell alterations in aging ovaries. Overall, the data and analyses presented in this study reveal profound modifications of the immune system in the aging reproductive system in mice. Additional computational approaches predicting cell-cell communications affected by aging in the ovaries presented in this study can extend our understanding of the aging immune system. However, most of the conclusions from single-cell RNA sequencing results are not confirmed using additional approaches, including a more detailed flow cytometry analysis of ovarian immune cell subsets and functional validations of the predicted biological processes affected by aging.

The presented data do not specify whether the identified changes in the ovarian immune system are specific to aging ovaries or reflect a common alteration of the aging immune system in mice. Recently, several papers unbiasedly identified immune alterations associated with aging in different tissues using single-cell RNA sequencing and flow cytometry techniques (e.g., Almanzar et al., Nature 2019; Kimmel et al., Genome Res 2019; Mogilenko et al., Immunity 2021). This study does not compare the findings with previous single-cell-based results from different tissues and does not clearly state if the immune aging in the ovaries is paralleled by similar alterations in immune cell subsets in other tissues in mice.

The authors show that the CD4- CD8- double-negative T cell subset is profoundly increased in abundance in aging ovaries. However, the population of double-negative T cells is not sufficiently characterized in the study. For example, it is unclear if similar cells can be found in aged tissues other than the ovaries. Moreover, using single-cell RNA sequencing, the authors show that the double-negative T cells co-express Trbc2 (TCRb) and Tcrgc2 (TCRg) genes, but the flow cytometry analysis of TCRg/d expression on these cells is not presented. The authors speculate that the double-negative T cells might have a regulatory function. However, a recent paper identified a population of pro-inflammatory T cells that co-express TCRab and TCRgd in mice and humans (including CD4- CD8- double-negative cells) (Edwards et al., J Ex Med 2020), suggesting that the double-negative T cells might be pro-inflammatory. It remains unclear if the double-negative T cell subset is unique to aging ovaries or phenotypically similar to the previously characterized double-negative TCRab+ and TCRgd+ cells.

The authors identified multiple transcriptional changes in genes encoding cytokines and chemokines, reflecting their decreased expression in aged ovarian immune cells. This observation is interesting because it contradicts the basic assumption of enhanced inflammation in old tissues. However, the presented findings are limited by the single-cell RNA sequencing level of evidence and are not supported or exemplified by an orthogonal analysis showing similar changes at the protein levels.

The authors claim that aging affects the recognition of senescent cells by ovarian immune cells. This exciting statement is based only on the single-cell RNA sequencing data in immune cells. The interaction between the immune cells and senescent cells in the ovaries involving the discussed pathways is not validated at protein levels in this study.

Reviewer #1 (Public Review):

This study presents an implementation of single-particle tomography within the Bayesian framework of the Relion software package. Similar to previously proposed strategies, the approach leverages single-particle analysis tools and tomographic geometric constraints to improve map resolution. Results on the EMPIAR-10164 benchmark dataset appear to match the performance of previous methods, but no maps were made available or deposited, and no direct comparisons with previous results are shown. Consistent with previously published strategies that use 2D projections instead of sub-volumes, the approach performs favorably in terms of resolution when compared to traditional subvolume averaging.

Strengths

- Use of a Bayesian framework for image refinement and reconstruction requires less parameter tweaking.<br /> - By making the new implementation accessible through a GUI already familiar to many SPA users, this tool will make SPT easier to use.<br /> - The implementation of 3D classification could be potentially beneficial to study sample heterogeneity in situ.<br /> - In cases where high resolution can be achieved (better than 3A), the approach has the potential to correct for higher-order optical aberrations.<br /> - Using two cryo-ET datasets, resolution improvements are shown over traditional subvolume averaging (as implemented in the AV3 Matlab suite of programs [Forster et al., 2007] and Dynamo suite [Castano-Diez, 2012]).

Weaknesses

- The approach recapitulates previously proposed strategies for SPT refinement that use raw tomographic projections instead of sub-volumes to improve resolution. Strategies that leverage the increased SNR of average structures to optimize particle pose and deformation, tilt-series alignment, and CTF refinement, were proposed and validated in earlier studies [1,24].<br /> - Compared to end-to-end pipelines for tomography data analysis such as EMAN2 and Dynamo, this approach only implements the subtomogram averaging step, while still relying on external tools for initial tilt-series alignment, CTF estimation, and particle picking.<br /> - In terms of performance, the HIV-1 Gag maps obtained from the benchmark dataset EMPIAR-10164 do not represent an improvement in resolution over previous methods.<br /> - No validation is provided to support the claim that the tool can correct for higher-order optical aberrations of the microscope from cryo-ET data.<br /> - No results are provided to validate the 3D classification routines to study heterogeneity, and no experiments are shown to support the claim that the new approach is more accurate than previous sub-volume classification strategies that compensate for the missing wedge (such as the approach implemented in the earlier version of Relion [4]).

Overall, this implementation of SPT would be a valuable resource for the cryo-ET community.

Reviewer #2 (Public Review):

Zivanov et al. present a new approach for multi-particle averaging from cryo-electron tomography data. They propose that refining directly against 2D tilt series images instead of the traditional reconstructed 3D subtomograms would simplify and improve structure determination. This would represent the experimental data more faithfully than traditional subtomogram averaging and circumvents the need for missing wedge correction. The authors describe a data structure termed 'pseudosubtomograms' where the tilt images are represented as their Fourier transform pre-multiplied with the CTF, accompanied by an array describing how often each 3D-voxel has been observed and the sum of the squared CTF. They then present a new regularized likelihood target function for cryo-ET particle alignment which uses the pseudosubtomograms data structure. This approach is implemented within the general RELION refinement framework and allows for the use of pseudosubtomograms for 3D classification, initial model generation, and 3D refinement.

The authors also introduce methods for refining optical and geometrical parameters in the tilt series taking advantage of the average map obtained after 3D refinement. This allows for more accurate tilt series alignment, per-particle motion tracking, and calculation of per-particle CTF. They propose that iteratively refining these parameters, extracting new pseudosubtomograms, and realigning the particles should lead to more accurate structure determination. The methods are validated using three different datasets, and the authors show that the iterative refinement within their framework increases the resolution of the 3D reconstruction and that the resulting maps are resolved to the same or better resolution than previously published methods.

The introduction of a more direct representation of the 2D tilt series images is a novel approach to subtomogram averaging, and the authors show that it is as good or better than current approaches. Comparing the subtomogram average to the tilt series to correct for optical and geometrical parameters of the data has already been implemented in the program M. Here, the authors show that their algorithms can reach the same resolution as M for the HIV immature capsid, but discuss that M might be superior at very high resolution, as it models beam-induced rotation of particles. Nevertheless, the new approaches are implemented in a single framework - the popular open-source software package RELION - thereby greatly facilitating their accessibility to uses. This is a very welcome contribution and development in the field.

Reviewer #1 (Public Review):

The authors present a strong set of experiments to uncover what type of role non-mutant stromal cells might be playing in the development of VM and AST, two vascular lesions that share some similarities.

Questions about experimental design.

1) For quantification of gene expression in VM and AST specimens in Figure 2, the methods say qPCR data were normalized to housekeeping genes, but it would be helpful to normalize to endothelial content. It might be that increased TGFa is due to increased endothelium.

2) The mutant allelic frequency for the HUVEC-PIK3CA WT versus HUVEC-PIK3CA H1047R should be provided. This is critically needed for the interpretation of the results.

3) From Figure 5, it appears that the human primary fibroblasts are not required for the mutant ECs to form perfused vessels (panel H). Is it possible that TGFa from the ECs is sufficient to drive vascular malformation?

Reviewer #2 (Public Review):

In this manuscript, Ilmonen H. et al explored potential crosstalk between endothelial cells and fibroblasts in a context of sporadic vascular malformation (venous malformation and angiomatoses of soft tissue). With a high level of evidence, they found that mutated endothelial cells secrete TGFA that will activate surrounding fibroblasts, leading in turn to VEGFA secretion that will stimulate endothelial cell sprouting and vascular malformation development.

Experiments are well-designed and support their hypothesis.

Some controls are missing, particularly in Fig. 2. Indeed, it is mandatory to provide data from healthy skin biopsies (that are available in many laboratories): TGFa, CD31, P-EGFR staining.

Reviewer #1 (Public Review):

The current study melds computational and docking methods with functional measurements in a systematic approach: first, they analyze the mechanism of inhibitor binding to EAAT2; second, they mutate ASCT to resemble EAAT and show that the general binding pocket and inhibition mechanism are conserved; third, they perform an in silico screen to identify compounds that bind to the WT ASCT binding pocket; fourth, they perform electrophysiological assays showing that this novel compound allosterically modulates ASCT function. This is a complete and comprehensive study with extensive experimental support for the major conclusions. The authors identify an allosteric ASCT inhibitor, and although only partial inhibition is achieved, this study serves as proof-of-concept that this site can be targeted in diverse SLC-1 transporters as an allosteric inhibitory site.

Reviewer #2 (Public Review):

This study set out to explore the nature of a previously described non-competitive and selective inhibitor of the human glutamate transporter, EAAT1 and to explore if this mechanism was conserved across the glutamate transporter family. The non-competitive nature of UCHPH-101 inhibition of EAAT1 has previously been demonstrated with both functional analysis and structures of EAAT1. Here, the authors use detailed electrophysiology analysis to confirm this mechanism of inhibition and to demonstrate that the inhibitor slows the steps of the transport cycle associated with substrate translocation, rather than substrate or sodium ion binding. These findings agree with previous studies that have shown that the compound binds at the interface of the transport and scaffold domains in EAAT1, two domains that are required to move relative to each other for the transport process to occur. UCPH-101 also prevents the transporter from entering an anion-conducting state, which agrees with a recent structure and MD simulations of EAAT1 that demonstrate movements of the transport domain relative to the scaffold domain are required for the EAAT1 to move into the anion-conducting state and support the mechanism of UCPH-101 inhibition confirmed in this study (PMID: 35192345; PMID: 33597752).

While UCPH-101 has been shown to be selective for EAAT1 over other human glutamate transporter subtypes (notably EAAT2 and EAAT3), Dong et al., show that this inhibitor can also reduce transport by another member of the SLC1A family, a neutral amino acid exchanger, ASCT2. Using MD simulations and functional analysis, they show that UCPH-101 acts as a partial, low-affinity inhibitor of ASCT2 and identify two amino acid residues in the binding site that appear to be responsible for the different affinities for EAAT1 and ASCT2. Indeed, when these two residues are changed to the corresponding residues in EAAT1, UCPH-101 becomes a full inhibitor of ASCT2 with an increased affinity.

ASCT2 is a neutral amino acid transporter that can transport glutamine and it is known to be upregulated in several cancers. Thus, finding new compounds and novel ways to inhibit ASCT2 is worthy of investigation. In the last section of this study, the authors conduct a virtual screen of 3.8 million compounds to identify other compounds that could bind to this allosteric site in ASCT2. One compound was identified, and while it had relative low affinity it provides the basis for further exploration of this site.

Reviewer #3 (Public Review):

Using whole-cell patch-clamp measurements, the authors nicely elaborate the competitive inhibition mechanism of UCPH-101 on EAAT1, concluding that it blocks conformational changes during transmembrane translocation, without inhibiting Na+/glutamate binding. The authors demonstrate that UCPH-101 binds to ASCT2 with strongly reduced affinity. Informed by sequence comparison between EAAT1 and ASCT2, the authors identify a pair of mutations, which makes the putative allosteric-binding pocket (which has been identified by crystallography earlier) in ASCT2 more similar to EAAT1 and restores the inhibitory effect of UCPH-101 in ASCT2. Overall, the electrophysiological experiments appear sound and convincing.

Furthermore, using virtual screening against the UCPH-101 binding pocket in ASCT2, the authors identified a novel (non-UCPH-101-like) compound #302 that they experimentally demonstrate to also inhibit ASCT-2. However, the study lacks a detailed investigation of the inhibition mechanism of this compound and it remains unclear if #302 also mediates allosteric inhibition as the authors propose. Furthermore, the study lacks any experimental verification of the assumed binding site of #302.

In addition, the study includes molecular-dynamics (MD) simulations on interactions of UCPH-101 with EAAT1 and ASCT2. These simulations intend to support the interpretations of the electrophysiological experiments, i.e., relatively tight interactions of UCPH-101 with EAAT1 and weaker binding to ASCT2, which can be restored using two point-mutations in ASCT-2. Unfortunately, this is a relatively weak part of the study. Due to the lack of any convergence analysis, the statistical significance of the drawn conclusions remains unclear. Furthermore, since it is not reported how UCPH-101 has been parameterized, the chemical accuracy of these models is unclear.

Reviewer #1 (Public Review):

Using simultaneous EEG-fMRI, authors asked whether neurovascular coupling is already functional in preterm-born neonates, in whom the underlying physiological mechanisms may still be immature at several levels. The question is very interesting and has implications for the study of brain function development as well as early brain injuries. The manuscript reports a correlation between the "mean duration of EEG microstates" and "fMRI BOLD signal-change", through which authors suggest that such a relationship between the EEG activity and BOLD signal highlights the functionality of neurovascular coupling already at the preterm period. The methodology is interesting, but more (not extensive) analysis is required to support the main conclusion and explain the results.

1. The main finding of the study in support of the conclusion comes from relating the inter-individual variability between EEG microstate duration and fMRI BOLD signal change. Given the few subjects (n=13), small even for neuroimaging in infants, studying effects based on inter-individual variability needs to be done with extra care. It is thus important to check whether interindividual variability can be observed for/accounted for by more basic effects in this population :

- The age range is relatively large (age at scan 31 PMA to 36 PMA - but also the age at birth: 29 to 35 weeks) for the number of included infants. Given the intense age-related changes in brain development at this period, it is important to take this factor into account and study it and to have them perhaps explicitly addressed in the manuscript: a ) Does the duration of EEG microstates depend on the age of the infants? b ) Does the time-to-peak in BOLD decrease with increasing age (Arichi et al., 2012)? c) and eventually does the relation between microstate duration and BOLD signal change holds once controlled for their common dependency on the age (i.e. Once partial correlations are used)?

2. The mean/std for the number of epochs per infant can be detailed more. What was the minimum number of epochs? Did such variability in the number of epochs impact microstate properties such as global explained variance/duration? How variable GEV was across infants and would that relate to the variability in duration?

3. Given that sensory-driven changes in microstates follow a sequential pattern (Hu et al., NeuroImage, 2014), could some "microstate syntax" characterize the underlying brain dynamics during stimulation processing in these neonates? Studying the presence of such syntax could be a way to show structured sensorimotor processing, and to further help quantify the inter-individual variability.

4. Is the sleep state monitored (from the EEG signal itself for example)? Given that the sleep state affects EEG activity and in particular EEG microstate properties in newborns (Khazaei et al., Brain topography, 2021), is there a way to rule out that the variability in microstate duration/BOLD signal change is not due to vigilance states?

5. Some of the conclusions/discussion points could be more cautiously stated and developed. On page 7: "However, our results imply that immature neurovascular coupling may not have a significant role in the pathophysiology of cerebral tissue injuries typically seen in preterm born infants (Volpe, 2009); and even that clinical interventions for perinatal brain injury could account for, accommodate, or capitalize on the presence of neurovascular coupling in the preterm human brain to minimize the severity of the injury and its long-term consequences." With an age range covering very preterm infants to late preterm period, generalizing such a conclusion could be potentially misleading for younger infants for example (fNIRS work in younger preterms does not support neurovascular coupling - Nourhashemi et al, Human brain mapping, 2020). As a group - on average - such a pattern may be reported, but the number of infants at each age does not allow us to draw a conclusion about the developmental stage at which such coupling is truly in place. These points could be more directly discussed with regards to the previous literature.

Reviewer #2 (Public Review):

In this work, the authors aim "to assess whether the relationship between neural activity and hemodynamic responses is present" "before the time of normal birth". In other words, they aim at showing that neurovascular coupling is present before term-equivalent gestational age. They use simultaneous EEG and fMRI in preterm infants presented with tactile stimuli.

Neuroimaging methods and stimulation methods are sound and rely on previously published works from the same group using neonatal MRI during somatosensory stimulation. The novelty resides in the use of simultaneous EEG to measure neuronal activity simultaneously with BOLD.

Methodological weaknesses are related to:

- Participant selection and characterization: there is a large variability in gestational age at birth, from very preterm (29 weeks) to late preterm (35 weeks) infants, which produces a large variability in chronological age at measurement (2 to 26 days). Considering how physiology and brain structure change dramatically with these factors, such variability seems an important bias. As stated in the introduction "In the time leading up to full-term human birth, rapid maturational changes are taking place across nearly all of the components which both relate to and occur within the neurovascular coupling cascade". There may be an effective neurovascular coupling in a neonate born at 35 weeks and tested at 2 days, and a very atypical or ineffective neurovascular coupling in an infant born at 29 weeks and tested after a month of intensive care, invasive respiratory support, and medication. This bias is also present in EEG analysis since "microstate basis vectors were derived from periods within the grand average signal that were topographically consistent across trials/subjects": any variability due to prematurity/NICU time is lost with this process.

- Not accounting for sleep states. During sleep, preterm infants alternate between slow and agitated sleep states, the pattern of state cycles changing with gestational age. Although the authors used EEG, they do not report looking for sleep states. Sleep state changes during stimulation would likely affect strongly EEG microstates sequence, duration, and power, as well as BOLD amplitude and distribution (ipsi vs. contralateral). This would be easy to verify and would allow a deeper understanding of the data, such as the variability of EEG and BOLD responses in each participant and among participants.

The main issue with the manuscript is the discrepancy between the stated aims ("to assess whether the relationship between neural activity and hemodynamic responses is present") and the literature available on the topic, on one hand, and between the stated aims and the actual work that was performed and discussed in the manuscript, on the other hand.

Aims vs. literature: The presence of a neurovascular coupling before term-equivalent gestational age has already been shown years ago, including by this group. For example, in: Arichi, T., et al. (2010). Somatosensory cortical activation identified by functional MRI in preterm and term infants. NeuroImage, 49(3), 2063-2071, where the following sentence begins the Conclusion "This is the first description of well-localised somatosensory cortical activation in the premature brain using a fully automated and programmable passive motor stimulus. Predominately positive BOLD signal change during stimulation was seen".

Or in:

Arichi, T., et al. (2012). Development of BOLD signal hemodynamic responses in the human brain. NeuroImage, 63, 663-673.

And by other groups using fMRI:

Heep, A., Scheef, L., Jankowski, J., Born, M., Zimmermann, N., Sival, D., et al. (2009). Functional magnetic resonance imaging of the sensorimotor system in preterm infants. Pediatrics, 123(1), 294-300.

Other examples of neurovascular coupling before term can be found with auditory-evoked BOLD responses in fetuses:

Jardri, R., et al. (2008). Fetal cortical activation to sound at 33 weeks of gestation: a functional MRI study. NeuroImage, 42(1), 10-18.<br /> but also, with various types of stimuli using fNIRS, for example:<br /> Mahmoudzadeh, M., et al. (2013). Syllabic discrimination in premature human infants prior to complete formation of cortical layers. Proceedings of the National Academy of Sciences, 110(12), 4846-4851.<br /> And:<br /> Roche-Labarbe, N., et al. (2014). Somatosensory evoked changes in cerebral oxygen consumption measured non-invasively in premature neonates. NeuroImage, 85, 1-8.<br /> Including simultaneous EEG and fNIRS :<br /> Roche-Labarbe, N. et al., 2007. Coupled oxygenation oscillation measured by NIRS and intermittent cerebral activation on EEG in premature infants. NeuroImage, 36(3), pp.718-727.

Be it in the Introduction or the Discussion, the authors only consider MRI literature whereas neurovascular coupling has been described and used for cognitive studies in premature neonates using fNIRS. There is no reason to restrict oneself to one technology when discussing fundamental physiological or cognitive processes.

Aims vs. actual work: The work that was actually performed is to measure EEG microstates' duration and power following tactile stimulation and to compare BOLD amplitude with these measures. The question being answered is whether the relationship that exists between microstates duration and BOLD amplitude in adults can also be observed in preterm infants. This in itself is an interesting purpose and should be stated as such in the Abstract and Introduction.

The Introduction is short and lacking in essential information. A review of microstates, what they are and what they mean, and how they are described in premature infants (particularly sensory-evoked microstates), is necessary. Previous studies of neurovascular coupling in preterm infants using evoked potentials, or no EEG at all when measuring the hemodynamic (fMRI or fNIRS) response associated with sensory stimuli. The introduction should argue why microstates would be more meaningful than SEP for EEG-fMRI studies, and what relationship with hemodynamics is expected based on previous studies with older participants. A comprehensive review of neurovascular coupling in preterm neonates, including non-MRI studies, is also necessary. The sentence "Here we test the hypothesis that despite the apparent immaturity of the underlying physiology, neurovascular coupling is functional before the normal time of birth." should be replaced by something along the lines of "Here we test whether the relationship between EEG microstates and neurovascular response is similar in premature infants with adults". Then the experimental contribution will make sense and the Discussion can focus on what it entails for understanding neurovascular coupling that amplitude is related to the duration, not power, of EEG microstates.

A Discussion (distinct from the Results) of the scientific and clinical relevance is currently lacking and it is difficult to assess the significance of the experimental contribution. An interesting discussion of microstates in the preterm brain is presented, but because the topic of microstates' relevance in neonates was not mentioned in the Introduction, it is confusing to read results such as "the observed composite progression of microstates indicates that the preterm brain is already capable of multi-level local sensory elaboration in the primary sensorimotor cortices." that does not correspond to any previously formulated hypothesis.

In the results, the authors should report if microstate duration varies among repeated identical stimuli in each child. The authors may look at this variability in terms of gestational age at birth (for example, in the participants who were born the earliest and have stayed the longest in the NICU, are microstates durations after a stimulus more variable than in the late-preterm participants?). The method for microstate analysis does not give clear information to the reader unfamiliar with Ragu other than the fact that one duration value was calculated for each participant. However, it would be informative to see some sort of dispersion range for both Mean BOLD and microstate duration values. It would be interesting to regress this information with gestational age at birth (or chronological age at scan) and sleep state.

After these changes have been made, I expect that the authors may find a more relevant title for their manuscript. "Neurophysiological basis of hemodynamic responses" does not give a precise idea of the experimental findings. Similarly, the abstract should be adjusted by removing sentences like "These results suggest that effective neurovascular coupling is present in the human brain even before the normal time of birth", a long-known fact, and detailing instead "a complex relationship between EEG and fMRI signals underpinned by patterns of activity across distinct neural ensembles."

Details of the stimulation sequence are unclear:

- Why were stimuli varying in duration from 7.5 to 10.5 seconds? The results report "the median BOLD hemodynamic response peaked at 14 seconds after stimulus onset": was it calculated regardless of stimulus duration? It is unlikely that the peak was reached after the same delay for 7 and 10 s stim. Was this accounted for in the MRI analysis?<br /> - There was a maximum of 24 epochs per participant, but how many epochs were kept for each participant after artifact rejection? How were distributed the 76 epochs remaining for analysis, among the participants?

Reviewer #3 (Public Review):

This significant EEG-fMRI study highlights the functionality of the neurovascular coupling in response to somatosensory stimuli in the somatosensory cortices of premature neonates. The methods here developed are highly compelling and go beyond the current state of the art. This neurovascular adaptation is described together with an analysis of the relationship between changes in microstate cortical activity and the hemodynamic activities that suppose an already well-organized hierarchical processing of sensory information.

Strengths:

Analyzing simultaneously the changes in microstates (EEG) and BOLD signal (fMRI) in relation to somatosensory stimuli in preterm neonates allowed to demonstrate a correlation between the duration of the microstates and the amplitude of the BOLD response in premature neonates.<br /> The procedure for recording simultaneously EEG and fMRI in preterm neonates is a real challenge that has been very well conducted in terms of methodology.

Weaknesses:

Although the paper does have strengths in principle, the weaknesses of the paper are that the authors did not discuss the changes in neurovascular coupling in response to spontaneous bursts of activities or external stimuli in preterm neonates using other modalities such as fetal MEG or simultaneous EEG-fNIRS. While it can be easily understandable that the number of preterm neonates is small, the age range is wide and as discussed by the authors changes in EEG activities are important during the last trimester of gestation.<br /> The sleep stage is not reported but authors might present raw data of the microstates (around 30 secs). In addition, the lack of discussion about the effect of discontinuity which is a characteristic of EEG in premature neonates

Reviewer #1 (Public Review):

This manuscript reports a series of studies that evaluate the role of long descending propriospinal neurons arising in the cervical spinal cord that project axons to the lumbar spinal cord in locomotor function recovery after spinal cord injury. The experiment uses several different evaluations of gait including BBB, ladder rung walk tests, and kinematics to compare walking before and after synaptic silencing of long descending propriospinal neurons projecting axons to L2. The data reveal that silencing of these neurons mildly improves walking function. The experiments are carefully described and well-controlled. The use of several different methods to evaluate locomotor function is a strength as is a well-thought-out approach to synaptic silencing. The data support the conclusions proposed by the authors. There are caveats to be considered in interpreting the results which are thoughtfully and thoroughly articulated in the discussion.

Reviewer #2 (Public Review):

The manuscript by Shepard et al. expands on prior recent publications by the group which demonstrated that silencing long ascending propriospinal neurons (LAPNs) disrupts left-right coordination in certain contexts in uninjured rats but improves locomotion following thoracic contusion. Here, the same reversible silencing strategy is used but instead targeted to the long descending propriospinal neurons (LDPNs). Interlimb coordination and several other locomotor metrics are examined in both uninjured and SCI conditions. The effects of LDPN silencing were quite similar to those of LAPN silencing with a few notable differences. In intact rats, the deficits were observed following silencing on both high and low-friction surfaces. The effects are stronger during the second Dox administration than during the first in intact, and possibly the opposite after SCI. Also, the reversal of deficits by silencing after SCI was more modest.

The major strengths of the study are the methodology and research design employed. The reversible silencing of a specific population of neurons identified by the locations of their somata and terminals is powerful. This also allowed for comparisons of pre-/post-silencing in the same subject both in uninjured and SCI conditions. The primary shortcoming of the study is the lack of histological analysis to demonstrate the degree of loss and/or whether there is any selectivity or bias towards functional subclasses of neurons that are shown to be LDPNs, even at the level of ipsilateral/contralateral and transmitter phenotype.

The presented data support the major conclusions of the study. It is interesting that silencing the LDPNs or the LAPNs, disrupting communication in either direction, has similar effects and that these effects are predominantly related to cross-cord coordination at each girdle. Additionally, the long propriospinal neurons, LDPNs in particular, are thought to be potential targets for relays and adaptive plasticity after spinal cord injury. However, their silencing after SCI leads to locomotor improvements rather than exacerbated of dysfunction. Whether this is due to an imbalance of spared projection neurons, maladaptive plasticity/sprouting, or other mechanism is of interest for future studies targeting spared projections to enhance functional recovery.

Reviewer #3 (Public Review):

This study aims at determining the contribution of propriospinal neurons projecting from cervical to lumbar segments to the coordination of inter-limb coordination. In addition, the impact of silencing these neurons on motor parameters affected by spinal cord injury was assessed. While the study contains many important data describing the contribution of these propriospinal neurons, there is little information about the underlying circuit mechanisms.

Public Review:

The study of Choi and collaborators provide novel information about the microstructural morphology and the crystallographic structure of palaeognathid eggshells.<br /> In terms of format and structure, the work is well organized and the extinction of each section is appropriate. All figures, both those from the main text and Supplementary Information, are of good-quality, informative, and useful, facilitating the understanding of the text. The bibliography is very updated, and all essential references are mentioned.

One of the strongest points of the work, in my opinion, is the designee of the study itself, which included specimens from all living palaeognathid birds and several extinct taxa from a large range of lineages.

The methodology used for analysing the crystallographic nature of the studied specimens (EBSD) is appropriate for the goals of the study. The phylogenetic approaches are also right, which are based on the most recent studies about the phylogenetic relationship of ratites.

Despite their complexity, the results are well presented, being relatively easy to understand for a person not versed in the subject. In fact, the ways in which they are described give them the potential to be used as a guideline to anyone interested in eggshell microstructure.

The discussion of the results seems consistent with the data obtained. Despite the phylogenetic relationship between some palaeognathid taxa remains partially instable, authors present different plausible scenario to explain the variability of the eggshell microstructure within a single monophyletic lineage (homology vs homoplasy). In fact, the homoplastic scenario is, perhaps, the most shocking one to me. In part, it is because it intrinsically suggests that all phylogenetic studies based on eggshell morphological features, and conducted during the last 20 years, are potentially artefacts, and they do not represent real phylogenetic relationships. Far from being a criticism, this interpretation has massive implications, especially for those studies where the taxonomic attribution of a fossil egg is based on phylogenetic results (i.e. Montanoolithus, Cairanoolithus).

Although I do not find negative arguments for any special section of the study, I have a question regarding Triprismatoolithu stephensis:

As mentioned in the text, Triprismatoolithu is analysed by the authors, and several pictures are provided in Fig.S12 alongside a brief description in de Supplementary Tex4. But it seems that it is not included in any of the phylogenetic analyses or figures. Why?

If the specimen has no implication for any of the main analyses, there is no need to be considered as "studied material".

Reviewer #1 (Public Review):

Zeng and colleagues investigated the neural underpinnings of visual-vestibular recalibration. Specifically, they measured changes in three monkeys' perception of unisensory heading cues as well as associated changes in neuronal responses to these cues in three different cortical areas following prolonged exposure to systematic visual-vestibular discrepancies. Behavioral responses in a motion direction discrimination task indicate unisensory perceptual shifts in opposite directions that account for the cross-modal discrepancy the monkeys were exposed to. Neuronal firing patterns, related to motion discrimination judgments by means of neurometric functions indicated analogous shifts in neuronal tuning in areas MSTd and PIVC. In contrast, in area VIP tuning for visual heading stimuli shifted in the same direction as tuning for vestibular stimuli and thus in contradiction to the observed perceptual shifts.

The shifts observed in MSTd and PIVC fit nicely with existing theories and results regarding cross-modal recalibration and substitute claims that activity in these areas might underlie perceptual decisions. The shift of visual tuning in VIP is surprising and will certainly spark further investigation.

Overall the results are really interesting, yet, the manuscript in its current form needs revisions along two dimensions, 1) data analysis and 2) writing.

Reviewer #2 (Public Review):

The manuscript by Zeng and colleagues aims to investigate how neural representations of sensory cues in two modalities (visual and vestibular) change when conflicts are introduced between the cues. The manuscript convincingly demonstrates that this recalibration process differs between areas MSTd (a multisensory region), where sensory responses recalibrated differently for visual and vestibular cues, following each modality's conflict, and area VIP ( a higher-level region), where responses follow the vestibular cue. More limited insights are present for area PIVC, where visual responses are limited.

The analyses generally support the conclusions of the authors, but I have two major suggestions to strengthen the statistical robustness of the manuscript:

1) The analysis about the lack of visual recalibration in area PIVC would have been more convincing if the authors had used Bayesian statistics instead of regular t tests. In this way it would have been possible to estimate if the lack of visual recalibration in this area, for those few neurons that show visual tuning, can be taken as evidence for the absence of an effect or not. In the absence of this additional analysis, it is in fact difficult to properly interpret the results about area PIVC. Is PIVC more in line with MSTd, in view of the lack of visual responses? Or is there actually no visual recalibration, in contrast to both MSTd and VIP?

2) For all statistical analyses, multi-level statistics would have been more appropriate than simple t-tests. In fact, since recordings come from few subjects, which in turn have relatively few recording sessions, there is a risk that the results are influenced by one subject and do not represent the full population. Admittedly, this is unlikely in view of the apparently large effect size and low p values. Nonetheless, a more appropriate statistical analysis would make the results more robust and convincing.

Once these issues are addressed, I believe that the manuscript would provide relevant evidence supporting the hypothesis that multisensory processing in the cortex is an area-specific phenomenon, and that effects observed in one area cannot be simply expected to operate elsewhere. This will therefore elucidate the mechanisms of multimodal plasticity.

Reviewer #3 (Public Review):

This study documents an empirical investigation of a fundamental brain process: adaptation to systematic cross-sensory discrepancies. The question is important, the experiment is carefully designed, and the results are striking. Following an unsupervised recalibration block, perceptual judgments of self-motion on the basis of visual and vestibular cues are systematically altered. These behavioral effects are mirrored by changes in the response properties of single neurons in areas MSTd and PIVC (provided that neurons in these areas exhibited selectivity for the sensory cue). Remarkably, neurons in downstream area VIP adjust their response properties in a very different manner, seemingly exclusively reflecting vestibular recalibration (which is opposite in direction to visual perceptual shifts). In the former two areas, the neural-behavior association follows the stimulus dynamics. In VIP, this association remains high beyond the life span of the stimulus. VIP typically exhibits strong choice signals. These decreased in strength after recalibration (an effect unique to area VIP). Together, these findings further dissociate VIP's functional role from that of MSTd and PIVC, without however, fully revealing what that role may be. These results offer a novel perspective on the neural basis of cross-sensory recalibration and will inspire future modeling studies of the neural basis of perception of self-motion.

Reviewer #1 (Public Review):

The manuscript by Coates et al. from the Brown lab adds a fascinating and colorful set of tiles to the growing mosaic of small molecule control of the sterol pathway through strategic employment of different parts of the proteostasis pathway. Dr. Brown is an active and creative leader in this field, and this story brings some new and surprising twists to our understanding of the ways that metabolites, and potentially other small molecules, can alter protein processing and life cycle as part of normal cellular function or pathophysiological states. The data are convincing and thorough, and do a great job of revealing many mechanistic aspects of the intriguing observation that hypoxia changes SM processing and activity by altering its degradative fate. The contributing parts of the whole process include altered MARCH 6 E3 ligase activity, new metabolite-ligand regulators (squalene), and ligand-dependent escape from the proteasome to allow the production of a novel form of SM that is freed from the normal regulation of the full-length protein caused by cholesterol, as the authors have previously described. I particularly appreciate three aspects of this study.

First, they test a lot of hypotheses to gain a very full understanding of the gears that are turning to make this hypoxia response machine run. Importantly, these studies also rule out some oxygen sensing mechanisms that work in other contexts, like proline hydroxylation. Second, the authors go to great lengths to integrate the action of the moving parts in a quantitative way, to ascertain if the effects are explained by the coordinated separate changes that are occurring when hypoxia is imposed. And third, the work includes a very well-thought-out set of ideas about why this sort of response is occurring, both in normal cells experiencing either transient or long-term hypoxia, as well as in cancer cells that seem to prefer this form of truncated and alternatively regulated SM.

There is a growing interest in studying and harnessing small molecules to alter and affect protein stability, and these studies add weight to the idea that there are many evolved mechanisms that can teach us lessons both about foundational biology, and new approaches to drug discovery. These beautiful studies will be an important addition to the literature and will be read and referenced by many.

Reviewer #2 (Public Review):

Previous work from the Brown lab showed that SM undergoes proteasomal dependent processing of its N-terminal regulatory region to generate truncSM, which retains catalytic activity. In this manuscript, Hudson and colleagues show that the generation of truncSM correlates with hypoxic conditions. This process appears to be independent of the transcription factor HIF1α and proline hydroxylation. Instead, their data suggest that hypoxia-induced truncSm results from 1) upregulation of the E3 Ub ligase MARCH; 2) accumulation of squalene, the substrate for SM. Finally, the authors have linked these observations to pathologies, such as hypoxic endometrial cancer tissues, arguing that overactive truncSM may contribute to the growth and survival of malignant cells. Overall, this paper provides some interesting concepts on the regulation of the cholesterol biosynthesis pathway upon low oxygen levels. However, the functional consequences of truncSM accumulation under hypoxia have not been addressed.

Another important open question is the role of squalene in promoting truncSM. Any additional information to address these issues would significantly strengthen this study. The analysis and some of the data on the relative abundance of SM and truncSM could also be improved.

Reviewer #3 (Public Review):

Results of this manuscript provide a new link between oxygen sensing and cholesterol synthesis. In previous studies, this group showed that the cholesterol synthetic enzyme squalene monooxygenase (SM) is subjected to partial proteasomal degradation, which leads to the production of a truncated, constitutively active enzyme. In this study, the authors provide evidence for the physiological significance of SM truncation. In a series of experiments, the authors show that subjecting cells to hypoxia (oxygen deprivation) induces truncation of SM. The synthesis of cholesterol requires 11 molecules of oxygen and SM is the first oxygen-dependent enzyme in the cholesterol-committed branch of the pathway. Evidence is presented that hypoxia causes squalene, the substrate of SM, to accumulate, which results in the enzyme's truncation. In addition, hypoxia stabilizes MARCHF6, the E3 ligase required for sterol-dependent ubiquitination and degradation of SM. Finally, the authors provide an experiment showing that truncation of SM correlates with hypoxia in endometrial cancer tissues.

Overall, the data presented in this manuscript are compelling for the most part. Hypoxia-induced truncation of SM and MARCHF6 is very clear according to the presented results. The specificity of SM-induced truncation is strong; both direct addition and inhibitor studies are presented. The major strength of this manuscript is that it provides the physiological relevance for the authors' previous finding that squalene accumulation leads to truncation of SM. However, there are a few issues that should be addressed to improve the interpretation of the data presented. The manner in which quantified immunoblots are presented is very confusing and difficult to interpret. This is evident in experiments in several figures. For example, it is difficult to determine the role of ubiquitination (Figure 2D) and MARCHF6 (Figure 2E) in the generation of truncated SM. The authors should present quantified data of all lanes of the immunoblots to reduce confusion.

The other important finding of this manuscript is that hypoxia stabilizes MARCHF6. This is supported by the results of Fig. 3A; however, the result of Figure 3B is not clear. A new band appears upon inhibition of VCP and MG-132 seems to reduce protein expression. These results could be removed from the manuscript without impacting the conclusions drawn. Finally, the results shown in Figure 5 showing that truncation of SM correlates with hypoxia in endometrial cancer tissues are a little preliminary. Multiple bands are detected in SM immunoblots, which interferes with interpretation. This experiment could be removed and speculated upon in the discussion.

Reviewer #1 (Public Review):

Using two openly available multi-task fMRI datasets, the authors decompose thalamic activity into a smaller set of components. They show that voxels with higher loadings on the main components (high task hub property) also have a high participation coefficient as derived from resting state data. Cortical activity patterns can be predicted to some degree from thalamic activity patterns, and generally better than from a number of other cortical areas. This prediction relies mainly on the voxels with high task hub scores. The results are valuable and methodological generally solid, with some aspects being incomplete.

1. The finding that thalamic activity exhibits a low dimension structure is in my opinion less of a finding, but rather an assumption that motivates the use of dimensionality reduction techniques. When the authors ask (line 101) "whether thalamic task activity exhibits similar low dimensional structure", what is the alternative hypothesis? I think it is a foregone conclusion that with a restricted number of tasks, and the intrinsic smoothness of fMRI activity data, there are always K<<N components that capture 50,75, 90% of the variance. If you had measured the spiking of the entire population of thalamic neurons or increased the threshold to 99%, the structure of activity would be more high dimensional. So I believe you can either frame this as an assumption going in, or you build carefully an alternative hypothesis of what a "high-dimensional" structure would look like. Generating activity data i.i.d would be the simplest case, but given that both signal and measurement noise in fMRI are reasonably smooth, this would be a VERY trivial null hypothesis.

2. The measure of "task hub" properties that is central to the paper would need to be much better explained and justified. You motivate the measure to be designed to find voxels that are "more flexibly recruited by multiple thalamic activity components", but it is not clear to me at this point that the measure defined on line 634 does this. First, sum_n w_i^2 is constrained to be the variance of the voxel across tasks, correct? Would sum_n abs(w) be higher when the weights are distributed across components? Given that each w is weighted by the variance (eigenvalue) of the component across the thalamus, would the score not be maximal if the voxel only loaded on the most important eigenvector, rather than being involved in a number of components? Also, the measure is clearly not rotational invariant - so would this result change after some rotation PCA solution? Some toy examples and further demonstrations that show why this measure makes sense (and what it really captures) would be essential. The same holds for the participation index for the resting state analysis.<br /> 3. For the activity flow analysis, the null models (which need to be explained better) appear weak (i.e. no differences across tasks?), and it is no small wonder that the thalamus does significantly better. The Pearson correlations are not overwhelmingly impressive either. To give the reader a feel for how good/bad the prediction actually is, it would be essential that the authors would report noise ceilings - i.e. based on the reliability of the cortical activity patterns and thalamic activity patterns, what correlation would the best model achieve (see King et al., 2022, BioRxiv, as an example).<br /> 4. Overall it has not been made clear what the RDM analysis adds to the prediction of the actual activity patterns. If you predicted the activity patterns themselves up to the noise ceiling, you would also hit the RDM correctly. The opposite is not the case, you could predict the correct RDM, but not the spatial location of the activity. However, the two prediction performances are never related to each other and it remains unclear what is learned from the latter (less specific) analysis.

Reviewer #2 (Public Review):

This study investigates how thalamic functional MRI activations change across subjects performing many cognitive tasks. The results reveal localised regions in anterior, medial (and potentially posterior) portions of the thalamus that co-activate most consistently across multiple tasks. The authors then try to link these task hubs to cortical association cortices, first by showing that association cortices are most connected to thalamic task hubs. Second, by showing that thalamic activations can predict

The findings are important, mainly because thalamic fMRI activations are largely ignored by the current literature. The major strengths of the study lie in examining thalamic activations under many cognitive tasks and replicating results across two independent datasets.

The findings of thalamic hubs are compelling. However, this current version of the manuscript could be strengthened by providing better links with the wider literature (e.g. with thalamic resting-state networks). The study also falls short in properly quantifying the similarity of findings across the two independent datasets. The subtle discrepancies between the results of the two datasets throughout the manuscript could point to finer-grained fractionations of the identified thalamic hubs. The least compelling set of results (though not necessarily wrong) is the thalamic prediction of cortical activations. This is because the functional connectivity (FC) matrix used to link the thalamus and cortex was derived from the same data after regressing out task-related variance. However, this process might not be clean enough. A stronger test would utilize an FC matrix derived from an independent dataset.

Reviewer #1 (Public Review):

The authors serendipitously discovered that silencing Reln+ stellate neurons from medial entorhinal cortex layer II (mEC2) transiently by hyperpolarizing them causes them to degenerate. They replicate this result with two different tools to hyperpolarize these neurons, as well as with a tool to inhibit synaptic vesicle release at mECII axon terminals. They gain mechanistic insight into the degeneration process by performing a careful time course of axon morphological changes and caspase activation: somatic hyperpolarization causes axon retraction bulbs, while inhibition of glutamate release causes axon fragmentation. Crucially, they find that, unlike mEC2 neurons, neighboring Wfs1+ pyramidal cells or parasubicular cells do not degenerate when silenced in similar ways.

The vulnerability of mEC2 to inactivity is particularly compelling because the authors use three different tools to demonstrate it, two that hyperpolarize neurons (ivermectin-mediated activation of the modified glycine receptor alpha subunit, expressed transgenically; and Kir2.1 overexpression using AAV stereotaxic injection), and one that inhibits synaptic vesicle release at mEC2 terminals (Tetanus toxin overexpression using AAV stereotaxic injection). Each of these tools has its flaws but taken together the findings are very convincing. A few pieces of evidence that the various tools are achieving exactly what the authors say they are achieving are missing. But again, the convergence of the data between the three tools compensates for this to some extent.

I found the significance of the findings really fundamental and the writing of the paper absolutely remarkable - beautifully structured, crystal clear in its articulations and its implications. This paradigm has the potential to reveal crucial biology about plasticity in the adult, and about degeneration and vulnerability mechanisms. Vulnerability is such an important topic common to most neurodegenerative diseases, with absolutely no hints, until now, of what could render some cells more prone to degeneration, and immense potential for the discovery of central disease mechanisms. Even if degeneration relies here on the overexpression of an exogenous protein, it does not rely on the overexpression of a pathological protein directly associated with neurodegeneration, or on the invalidation of an essential protein. There is nothing trivial about the degeneration phenotype observed here, which makes the observations absolutely fascinating. What's more the authors show here evidence for the Grail of vulnerability: the side-by-side comparison of two similar/neighboring cell types treated in the same way, only one of which undergoing degeneration (Reln+ EC2 neurons Wfs1+ EC2 and parasubiculum neurons). The vulnerable cell type here also happens to be the very cell type that is most vulnerable to degeneration in Alzheimer's disease.

These findings are of major importance for a few different reasons:<br /> - Neuronal excitability is clearly an early event occurring in the EC of incipient Alzheimer's disease. This study suggests that the silencing of certain cells by Alzheimer's lesions might contribute to their degeneration.<br /> - A competition-based mechanism for the survival or degeneration of axons and neurons from EC2, is known to operate during development until the end of critical periods. This study suggests that EC2 neurons, which might be particular for their need to be plastic into adulthood, might use these mechanisms as well.<br /> - Again, they establish a paradigm for the mechanistic study of comparative vulnerability between cell types that can be investigated further to understand the molecular underpinnings of degeneration.

Reviewer #2 (Public Review):

One major enigma in neurodegeneration is why it tends to start many times in the entorhinal cortex. This paper tries to address this issue, by showing the vulnerability of reelin-positive entorhinal cells to inactivation, thus leading to the compelling idea that neurodegenerative processes are initiated by prolonged brain inactivity in specific brain regions. The paper is straightforward and performs a whole set of experiments to demonstrate the specificity of the effect on these cells, trying to partially decipher the underlying mechanisms which lead to the vulnerability of these specific cells.

The paper performs a series of tests on these cells. First, the chemogenetic silencing of layer 2 entorhinal neurons causes cell death and axonal degeneration. Second, this effect is specific to entorhinal neurons and spares other regions. Third, the effect seems to be mediated by synaptic silencing, in addition to general neuronal inactivity, and finally - the effect seems to be governed by neuronal competition and not by a general non-specific change in neuronal activity levels.

I think the paper is a great first step. In the future, more work will be needed in order to better understand the causes of this vulnerability and to connect this work to the cascade of neurodegeneration leading to the known phenomena associated with AD.

Reviewer #1 (Public Review):

By studying the effect of Treg depletion in a CD8+ T cell-dependent diabetes model the group around Ondrej Stepanek described that in the absence of Treg cells antigen-specific CD8+ OT-I T cells show an activated phenotype and accelerate the development of diabetes in mice. These cells - termed KILR cells - express CD8+ effector and NK cell gene signatures and are identified as CD49d- KLRK1+ CD127+ CD8+ T cells. The authors suggest that the generation of these cells is dependent on TCR stimulation and IL-2 signals, either provided due to the absence of Treg cells or by injection of IL-2 complexed to specific anti-IL-2 mAbs. In vivo, these cells show improved target cell killing properties, while the authors report improved anti-tumor responses of combination treatments with doxorubicin combined with IL-2/JES6 complexes. Finally, the authors identified a similar human subset in publicly available scRNAseq datasets, supporting the translational potential of their findings.

The conclusions are mostly well supported, except for the following two considerations:

1) From Fig. 4A and B it is not conclusively shown, that Tregs limit IL-2 necessary for the expansion of OT-I cells and subsequent induction of diabetes. An IL-2 depletion experiment (e.g. with combined injection of the S4B6 and JES6-1 antibodies) would further strengthen this claim. Along these lines, the authors claim "IL-2Rα expression on T cells can be induced by antigen stimulation or by IL-2 itself in a positive feedback loop [20]. Accordingly, downregulation of IL-2Rα in OT-I T cells in the presence of Tregs might be a consequence of the limited availability of IL-2.". The cited reference 20 did observe CD25 upregulation by IL-2 on T cells but the observed effect might only be caused by upregulation of CD25 on Treg cells, which increases the MFI for the whole T cell population. Did the authors observe significant upregulation of CD25 on effector CD4+ and CD8+ T cells in their experiments with IL-2/S4B6 or IL-2/JES6 treatment?

2) The anti-tumor efficacy of KILR cells is intriguing but currently, it is unclear if it is indeed mediated by KILR cells. Have KILR cells been identified by flow cytometry in the BCL1 and B16F10 models treated with doxorubicin and IL-2/JES6? Were specific KILR cell depletion studies conducted, e.g. with an anti-KLRK1 depleting antibody? Additional experiments addressing these questions would be desirable to further support the authors' claims.

Reviewer #2 (Public Review):

In this study, the authors determine the superior cell killing abilities of KLRK1+ IL7R+ (KILR) CD8+ effector T cells in experimental diabetes and tumor mouse model. They also provide evidence that Tregs suppress the formation of this previously uncharacterized subset of CD8+ effector T cells by limiting IL-2.

Strength and Limitation

This study focuses on the relationship between Tregs and CD8+ T cells. They used different experimental diabetes mouse models to reveal that Tregs suppress the CD8+ effector T cells by limiting IL-2. They also found a unique subset of KLRK1+ IL7R+ (KILR) CD8+ effector T cells with superior cell killing abilities through single-cell sequencing, but killing abilities could be inhibited by Tregs. They also tested their theory in in vivo tumor model. The data, in general, support the conclusions; however, some issues need to be fully addressed, as detailed below.

1. This study used the concentration of urine glucose as the standard for diabetes ({greater than or equal to} 1000 mg/dl for two consecutive days). However, multiple reasons may lead to a high level of urine glucose. As a type I diabetes mouse model, authors could use immunohistological analysis of islet to show the proportion of T cells and islet cells in islet, which can display the geographic distribution of immune cells, severity and histology structure of damaged pancreas islet directly. If possible, different subsets of immune cells, especially CD4 vs CD8+ cells should be stained for their location.

2. This article shows that KILR effector CD8+ T cells have strong cytotoxic properties. However, they do not describe the potential proliferation ability vs apoptosis of this subset from islets.

3. Figure 7 shows that the antitumor efficacy of IL-2 depends on CD8+ T cells. But in this part, there is no data to show the change of KLRK1+ IL7R+ CD8+ effector T cells in tumor tissue. Therefore, the article needs to add more data to verify that IL-2 enhances antitumor ability via KLRK1+ IL7R+ CD8+ effector T cells.

4. It is unclear why the authors chose Dox to combine with IL-2/JES6. The authors should provide a more rational introduction to bridge such a combination. Authors should also explain the reason why there is no antitumor effect of IL-2/JES6 treatment alone.

Reviewer #1 (Public Review):

The authors are trying to determine how time is valued by humans relative to energy expenditure during non-steady-state walking - this paper proposes a new cost function in an optimal control framework to predict features of walking bouts that start and stop at rest. This paper's innovation is the addition of a term proportional to the duration of the walking bout in addition to the conventional energetic term. Simulations are used to predict how this additional term affects optimal trajectories, and human subjects experiments are conducted to compare with simulation predictions.

I think the paper's key strengths are its simulation and experimental studies, which I regard as cleverly-conceived and well-executed. I think the paper's key weakness is the connection between these two studies, which I regard as tenuous for reasons I will now discuss in detail.

The Title asserts that "humans dynamically optimize walking speed to save energy and time". Directly substantiating this claim would require independently manipulating the (purported) energy and time cost of walking for human subjects, but these manipulations are not undertaken in the present study. What the Results actually report are two findings:<br /> 1. (simulation) minimizing a linear combination of energy and time in an optimal control problem involving an inverted-pendulum model of walking bouts that (i) start and stop at rest and (ii) walk at constant speed yields a gently-rounded speed-vs-time profile (Fig 2A);<br /> 2. (experiment) human subject walking bouts that started and stopped at rest had self-similar speed-vs-time profiles at several bout lengths after normalizing by the average duration and peak speed of each subject's bouts (Fig 4B).<br /> If the paper established a strong connection between (1.) and (2.), e.g. if speed-vs-time trajectories from the simulation predicted experimental results significantly better than other plausible models (such as the 'steady min-COT' and 'steady accel' models whose trajectories are shown in Fig 2A), this finding could be regarded as providing indirect evidence in support of the claim in the paper's Title. Personally, I would regard this reasoning as rather weak evidence - it would be more accurate to assert 'brief human walking bouts look like trajectories of an inverted-pendulum model that minimize a linear combination of energy and time' (of course this phrasing is too wordy to serve as a replacement Title -- I am just trying to convey what assertion I think can be directly substantiated by the evidence in the paper). But unfortunately, the connection between (1.) and (2.) is only discussed qualitatively, and the other plausible models introduced in the Results are not revisited in the Discussion. To my naive eye, the representative 'steady min-COT' trace in Fig 2A seems like a real contender with the 'Energy-Time' trace for explaining the experimental results in Fig 4, but this candidate is rejected at the end of the third-to-last paragraph in the 'Model Predictions' subsection of Results based on the vague rationale that is never revisited.

An additional limitation of the approach not discussed in the manuscript is that a fixed step length was prescribed in the simulations. The 'Optimal control formulation' subsection in the Methods summarizes the results of a sensitivity analysis conducted by varying the fixed step length, but all results reported here impose a constant-step-length constraint on the optimal control problem. Although this is a reasonable modeling simplification for steady-state walking, it is less well-motivated for the walking bouts considered here that start and stop at rest. For instance, the representative trial from a human subject in Figure 8 clearly shows initiation and termination steps that differ in length from the intermediate steps (visually discernable via the slope of the dashed line interpolating the black dots). Presumably different trajectories would be produced by the model if the constant-step-length constraint were removed. It is unclear whether this change would significantly alter predictions from either the 'Energy-Time' or 'steady min-COT' model candidates, and I imagine that this change would entail substantial work that may be out of scope for the present paper, but I think it is important to discuss this limitation.

With my concerns about the paper's framing and through-line noted as above, I want to emphasize that I regard the computational and empirical work reported here to be top-notch and potentially influential. In particular, the experimental study's use of inexpensive wearable sensors (as opposed to more conventional camera-based motion capture) is an excellent demonstration of efficient study design that other researchers may find instructive. To maximize potential impact, I encourage the authors to release their data, simulations, and details about their experimental apparatus (the first two I regard as essential for reproducibility - the third a selfless act of service to the scientific community).

I think the most important point to emphasize is that the bulk of prior work on human walking has focused on steady-state movement - not because of the real-world relevance (since one study reports 50% of walking bouts in daily life are < 16 steps as summarized in Fig 1B), but rather because steady walking is a convenient behavior to study in the laboratory. Significantly, this paper advances both our theoretical and empirical understanding of the characteristics of non-steady-state walking.

It is also significant to note the relationship between this study, where time was incorporated as an additive term in the cost of walking, with previous studies that incorporated time in a multiplicative discount in the cost of eye and arm movements. There is an emerging consensus that time plays a key role in the generation of movement across the body - future studies will discern whether and when additive or multiplicative effects dominate.

Reviewer #2 (Public Review):

This paper provides a novel approach to quantifying the tradeoff between energetic optimality during walking and the valuation of time to travel a given distance. Specifically, the authors investigated the relationships between walking speed trajectories, distance traveled, and the valuation of (completion) time. Time has been proposed as a potential factor influencing movement speed, but less is understood about how individuals balance energetic optimality and time constraints during walking. The authors used a simple, sagittal-plane walking model to test competing hypotheses about how individuals optimize gait speed from gait initiation to gait termination. Their approach extends literature in the space by identifying optimal gaits for shorter, partially non-steady speed walking bouts.

The authors successfully evaluated three competing walking objectives (constant acceleration, minimum cost of transport at steady speed, and the energy-time objective), showing that the energy-time objective best matched experimental data in able-bodied adults. Although other candidate objectives may exist, the paper's findings provide a likely-generalizable explanation of how able-bodied humans select movement strategies that encompass studies of steady-speed walking.

Overall, this paper provides a foundation for future studies testing the validity of the energy-time hypothesis for human gait speed selection in able-bodied and patient populations. Extensions of this work to patient populations may explain differences in walking speed during clinical assessments and provide insight into how individual differences in time valuation impact performance on assessments. For example, understanding whether physical capacity or time valuation (or something comparable) better explains individual differences in walking speed may suggest distinct approaches for improving walking speed.

Strengths:<br /> The authors presented a compelling rationale for the tradeoffs between energetic optimality and time and their results provide strong support for a majority of their conclusions. In particular, significant reductions in the variance of experimental speed trajectories provides good support for the scaling of speeds across individuals and the plausibility of the energy-time hypothesis. Comparison to theoretical (model-based) reductions across difference time valuation (cT) parameters would further enhance confidence in the practical significance of the variance reductions. Further, while additional work is needed to determine the range of "normal" valuations of time, the authors present experimental ranges that appear reasonable and are well explained. The computational and analytical methods are rigorous and are supported by the literature. Overall, the paper's conclusions are consistent with experimental and computational results.

The introduction of a model-based analytical approach to quantify the effects of time valuation of walking could generalize to test other cost functions, populations, or locomotion modes. Further, models of varying complexity could be implemented to test more individualized estimates of metabolic cost, ranging from 3D dynamic walking models (Faraji et al., Scientific Reports, 2018) or physiologically-detailed models (Falisse et al., Journal of The Royal Society Interface. 2019). The relatively simple set of analyses used in this paper is consistent with prior literature and should generalize across applications and populations.

The authors justified simplifications in the analysis and addressed major limitations of the paper, such as using a fixed step length in model predictions, using a 2D model, and basing energy estimates on the mechanical work of a simple model. It is unlikely that the paper's conclusions would change given additional model complexity. For example, a 3D walking model would need to control frontal plane stability. However, in able-bodied adults, valuation of frontal-plane stability during normal walking would not likely alter the overall shape of the predicted speed profiles.

Weaknesses:<br /> The primary weakness of this work is that alternative objectives may provide similar speed profiles and thus be plausible objectives for human movement. For example, the authors tested an objective minimizing the steady-speed cost of transport. This cost function is consistent with the literature, but (as predicted) unlikely to explain acceleration and deceleration during gait. An objective more comparable to the energy-time hypothesis would be to minimize the net energy cost over the entire bout, including accelerations and decelerations. This may produce results similar to the energy-time hypothesis. However, a more complex model that incorporates non-mechanical costs (e.g., cost of body weight support) may be needed to test such objectives. Therefore, the energy-time hypothesis should be considered in the context of a simple model that may be incapable of testing certain alternative hypotheses.

An experimental design involving an intervention to perturb the valuation of time would provide stronger support for time being a critical factor influencing gait speed trajectories. The authors noted this limitation as an area of future work.

While the results are compelling, the limited sample size and description of participants limit the obvious generalizability of the results. Older adults tend to have higher metabolic costs of walking than younger adults, which may alter the predicted time-energy relationships (Mian OS, et al., Acta physiologica. 2006). As noted in the introduction, differences in walking speeds have been observed in different living environments. General information on where participants lived (city, small town, etc...) may provide readers with insight into the generalizability of the paper's conclusions. Additionally, the experimental results figures show group-level trends, but individual-specific trends and the existence of exceptional cases are unclear.

The authors' interpretation of clinical utility is vague and should be interpreted with caution. A simple pendulum-based walking model is unlikely to generalize to patient populations, whose gait energetics may involve greater positive and negative mechanical work (Farris et al., 2015; Holt et al., 2000). Additionally, the proposed analytical framework based on mechanical work as a proxy for the metabolic cost may not generalize to patient populations who have heterogeneous musculotendon properties and increased co-contraction (e.g., children with cerebral palsy; Ries et al., 2018). Consequently, the valuation of time for an individual could be incorrectly estimated if the estimates of metabolic cost were inaccurate. Therefore, as the authors noted for their able-bodied participants, more precise measures of metabolic rates will be critical for translating this work into clinical settings.

Reviewer #1 (Public Review):

Tomanek and Guet describe the results of an evolution experiment where they allowed the bacterium E. coli to adapt to various concentrations of galactose as an additional carbon source. These conditions impose different degrees of demand for the galK enzyme, whose expression level depends on the promoter sequence and on the number of copies of the galK locus. Given that the initial promoter is random and weak, both amplifications of the locus and mutations in the promoter are expected to be adaptive. The experimental strains of E. coli were equipped with a fluorescent reporter system designed to discriminate between these two types of mutations. Furthermore, two strains, IS+ and IS-, were engineered with high and low rates of duplication around the galK locus, respectively. The main result is that at higher concentrations of galactose, where the demand for galK is high, E. coli adapts by acquiring combinations of both types of adaptive mutations, amplifications, and promoter mutations. In contrast, at low concentrations of galactose, where the demand for galK is low but not zero, E. coli appear to adapt by acquiring either an amplification or a promoter mutation but not their combinations. The observation of apparent interference between the acquisition of these two types of mutations is interesting and novel. The authors provide an intuitive explanation for it: when one mutation is sufficient to achieve the optimal expression of the gene, the mutation that is acquired first makes the other mutation obsolete, i.e., there is negative epistasis (possibly even sign epistasis) between these mutations, in the sense that the second mutation is much less adaptive (or possibly even deleterious) in the presence of the first one, in the low-demand environment. The authors discuss the possible implications of this finding for our understanding of molecular evolution and propose a new Amplification Hindrance hypothesis. This hypothesis states that, since amplifications occur at much higher rates than individual point mutations, they can slow down or even prevent sequence divergence. The amplification hindrance hypothesis stands in contrast to the Innovation-Amplification-Divergence hypothesis which is currently the default paradigm and states that amplifications generally accelerate sequence divergence.

STRENGTHS:

The authors designed a powerful reporter system that allows them to monitor the evolutionary dynamics of amplifications and promoter mutations. They ask an important question: how do early evolutionary dynamics of adaptation to environments with different demands for gene expression look like? The phenotypic data they present looks very interesting and shows the existence of interference between amplifications and point mutations in low-demand but not in high-demand conditions. The Amplification Hindrance hypothesis is a novel and useful intellectual contribution to the field.

WEAKNESSES:

In my opinion, the main weakness of the paper is that, while the interference between amplifications and point mutations in the low-demand condition clearly happens (most convincingly shown in Figure 5), its causes remain unclear. In particular, the authors claim that this interference is caused by negative epistasis, but the possibility of clonal interference without epistasis has not been decisively ruled out. The authors mention clonal interference tangentially in the Discussion, but they do not seriously address this alternative explanation. Yet, understanding the cause of this phenomenon is important because clonal interference and negative epistasis have different implications for long-term evolution.

The authors' main hypothesis is that, in the low-demand conditions, expression-increasing point mutations in the promoter provide much lower fitness benefits (or even incur fitness costs) in strains with galK amplifications compared to the ancestral strain without amplifications. The most direct way to test this hypothesis would be to measure the fitness effects of a point mutation in genetic backgrounds with and without amplifications in conditions with low and high demand for galK. This decisive experiment has unfortunately not been done. Instead, the authors construct an indirect argument, whose essence is as follows.

They show that, over the course of the experiment in the low-demand environment, the IS+ populations have acquired fewer point mutations than IS- populations (Figure 5). In addition, the phenotypic data in Figures 2 and 4 demonstrate that IS+ mutations in the low-demand environment contain three phenotypic classes of cells: ancestral, YFP+ and YFP+CFP+. The YFP+ clones are shown to have only one or two promoter mutations. The YFP+CFP+ cells must have duplications, and it is likely (although not quite certain, see below) that they do not have any promoter mutations. These data demonstrate quite convincingly that, whenever adaptation by duplications is possible, the rate at which point mutations segregate and accumulate declines. These data are consistent with the authors' hypothesis based on negative epistasis. However, they also seem to be consistent with the idea that amplifications and point mutations exhibit clonal interference without negative epistasis.

It may be possible to construct an argument against this alternative hypothesis based on the comparison between different environments, but such an argument would have to take into account the fact that clonal interference depends not only on the rates of mutations (which are presumably the same in all environments) but also on their fitness effects which vary across the environment. Another possibility to argue against clonal interference might be by carrying out simulations, although this approach also seems challenging without knowing some key population genetic parameters. The most direct way to resolve this ambiguity would be to demonstrate negative epistasis as discussed above.

Another, less critical but still important, issue mentioned above concerns the authors' claim that the YFP+CFP+ cells have only duplications but no promoter mutations (e.g., LL. 276-277). This is certainly consistent with intuition since these cells have an increased level of both YFP and CFP relative to the ancestor. However, as far as I can tell, there is no evidence to support this claim directly. My understanding is that the authors base this claim on the fact that YFP+CFP+ cells form a cluster of points on the YFP vs CFP plots that is distinct from the cluster of "mixed" cells, which are shown to have both an amplification and a promoter point mutation (Figure 3). But it is still logically possible that the YFP+CFP+ cells have an amplification and a promoter mutation other than the one found in the "mixed" cells (e.g., weaker). The most direct way to show that YFP+CFP+ cells have no promoter mutations would be to sequence a few of them. Another possibility would be to calibrate the YFP/CFP fluorescence measurements against galK copy number.

Reviewer #2 (Public Review):

This paper by Tomanek and Guet investigates the evolutionary dynamics of the very earliest steps in the process of evolution through gene duplication and divergence. They use a cleverly designed experimental system where they can tune the benefit of mutations that cause increased expression of a gene, and where they have reporter genes that can be used to distinguish between promoter up mutations and (most) gene duplications.

The major conclusion is that the dynamics of adaptive gene duplications and adaptive point mutations can be very different in different conditions - In "low demand" conditions, where a single mutation (duplication or snp) is enough to achieve the maximum (for that environment) fitness improvement duplications and promoter mutations acts with negative epistasis and become mutually exclusive. Contrary to previous literature that discusses evolution by duplication - divergence, duplications can thus act to prevent or slow down divergence.

The strengths of the paper: The genetic system is simple but cleverly designed. Using a gene (galK) that made it possible to tune the benefit of increased expression (by varying the amounts of galactose in the growth medium) made it possible to make observations that others have missed.

Possible weakness, which this paper has in common with much of the literature on evolution by duplication-divergence: Duplications are very often very unstable and are lost at rates that exceed their rate of formation. This means that in the absence of selection duplications are usually lost very quickly unless selected for, and all experiments and conclusions are based on stable conditions with a continuous selection that may not reflect a natural situation.

The aims of the paper were achieved and the presented data support the conclusions nicely.

This paper provides evidence that evolution by gene duplication is more complex than how it is usually described. Even if two mutations (e.g. gene duplication and promoter mutations) have additive or positive epistasis on a measurable quantity (be it enzyme kinetics, gene expression levels, or some other observable trait) the mutations could show negative (or even sign?) epistasis on the fitness of an organism. Hopefully, this paper will serve as a reminder of this even outside of the duplication-divergence field.

Reviewer #3 (Public Review):

The goal of this study was to determine the conditions in which adaptive copy-number mutations interfere with point mutations. One of the strengths of this study is its experimental design. The authors engineered a genetic reporter system to 'easily' distinguish between the two types of mutations: copy-number and point mutations. Thus, this system allows capturing mutations that appear 'de novo' during the evolution experiment and could be broadly used to study early duplication events. This system is also powerful given that gene expression demand can be tuned, allowing determining the conditions in which the Amplification Hindrance hypothesis holds. Finally, by combining measures of single-cell fluorescence and sequencing of the promoter region, the authors give more support to their conclusions (e.g., confirming the presence/absence of mutations).

An additional strength of this study is the use of three additional random promoter sequences. Even if the evolutionary dynamics for one of the promoters differed from the original promoter, the authors propose that this is due to the promoter mimicking a low expression demand. Thus, the use of three additional random promoter sequences strengthens their conclusion that negative epistasis between copy-number and point mutations occurs in low gene expression demand environments.

Overall, the methods and analyses are sound, and the conclusion that gene amplification hinders the fixation of adaptive mutations is correctly supported by the data. These findings have the potential to have broad implications for our understanding of the adaptive process in bacteria given that it provides a new mechanism for rapid adaptation that does not require de novo point mutations.

Reviewer #1 (Public Review):

Dectin-1 is a known C-type lectin receptor that has a role in recognizing pathogen glycans, particularly beta-glucan. Haji et al. present evidence that Dectin-1 also has an endogenous ligand, another C-type lectin that is enriched on platelets called CLEC-2. Using a Dectin-1 reporter line, they identify human platelets as a source for the ligand, raising a panel of mAbs to human platelets and identify one (6D11) that prevents signalling and use this to identify CLEC2 as the Dectin-1 receptor by immunopurification. The authors go on to further characterise this interaction by showing that it occurs between the human but not mouse orthologues, showing that it is the stalk region of human Dectin-1 that contains the ligand which consists of sialylated core 1 O-linked glycans displayed on Thr 105 and adjacent amino acids. Finally, the authors over-express human Dectin-1 in mice within the context of a null background for another CLEC-2 ligand (podoplanin) and show that this rescues embryonic lethality. They show that Dectin-1-dependent CLEC-2 signalling is not sufficient to induce platelet aggregation but can rescue perinatal lethality in podoplanin-deficient mice.

There is a large amount of work in this manuscript and the experiments are well controlled and the results unequivocal and usually verified by several independent techniques. The paper is very well-written making the volume of data accessible. There is novelty here in that this unusually finds that 2 C-type lectin receptors directly interact and the biochemical characterisation of the glycan binding determinants is very well performed. While the authors show that the interaction occurs between the human but not mice orthologues (because mouse Dectin-1 lacks the critical EDxxT motif that is necessary for displaying the o-linked glycan in the correct context) they were able to investigate the functional role of the interaction by generating mice that overexpressed human Dectin-1 in a podoplanin-deficient background. Normally, podoplanin-deficient mice die at birth due to defects in lymphatic vessel development, but this lethality is rescued by overexpression of human Dectin-1. This xeno-overexpression data can be difficult to interpret but suggests that while human Dectin-1 signalling to mouse platelet CLEC-2 is insufficient to drive platelet activation and thrombus formation, it is sufficient to rescue some aspects of platelet function involved in lymph vessel development. They conclude that human Dectin-1 (which is broadly expressed on different tissues) provides a basal level of tonic signalling through platelet-expressed CLEC-2 to establish platelet signalling thresholds.

I thought the manuscript was comprehensive in its approach, well written, and that the experimental data supported the conclusions.

Reviewer #2 (Public Review):

C-type lectin receptors are well-known for their pathogen recognition and their immunoregulatory properties. However, most C-type lectins also engage host-derived ligands. While many microbial targets have been identified, the characterization of endogenous ligands has so far lagged behind. In this paper, Haji et al. identified human Dectin-1 as a bonafide self-ligand for the platelet-specific C-type lectin receptor CLEC-2.

Strengths:<br /> Haji et al. actually identified the first glycan-dependent C-type lectin - C-type lectin interaction, resulting in a 2-way activation cascade downstream of both the Dectin-1 and CLEC-2 receptors. They performed a highly detailed molecular characterization, revealing both the interacting domains with Dectin-1 as well as the interacting glycan sialylated core 1 ligand. Moreover, the authors provide proof of the functional relevance of the Dectin-1 - CLEC-2 interaction in a mouse model deficient for the CLEC-2 ligand podoplanin, demonstrating that human Dectin-1 can rescue the phenotype observed in these podoplanin KO mice.

Limitations:<br /> The main limitation of this work is the use of Dectin-1 and CLEC-2 transfectants. Glycosylation patterns in transfected 2B4 cells (a T cell line) might not mimic the natural glycosylation pattern on Dectin-1 in vivo. A follow-up study should address which human Dectin-1 positive immune cell subsets are recognized by human CLEC-2 and how human Dectin-1 glycosylation is regulated during immune cell activation and differentiation.<br /> In addition, Dectin-1 polymorphisms have been identified in the human population, which strongly decreases Dectin-1 expression. Yet, these individuals mainly suffer from fungal infections and so far have not been shown to have lymphatic defects. This leaves the actual in vivo role of the human Dectin-1 - CLEC-2 interaction yet to be resolved.

Reviewer #1 (Public Review):

The authors tackle an interesting problem: how do ant colonies regulate foraging in response to their collective hunger? In previous work, the authors related the colony's response to individual ants sensing their own food levels and its temporal dynamics. Looking more carefully at the spatial dynamics of ants, the authors now find that foragers tend to move toward the depth of the nest when their food load is high and toward the nest exit when it is low. This is an elegant and computationally inexpensive set of rules that explains the spatiotemporal dynamics of the system.

Overall, the paper is written clearly, the methods are sound, and I agree with the interpretation of the results.

I do have a few comments and suggestions:

1) How exactly are the inward outward directions defined? Is it simply, away, or towards the entrance? It is not clear from the text, and since this system is not symmetric (cubic with entrance at one of the corners) the authors should clarify.

2) To analyze the biased random walk analysis of the ants, the authors "coarse-grained" the steps as being "inwards" "outwards" and "stay". It's not clear how this level of granulation is justified. Since the authors have access to the actual trajectories and all trophallaxis events, why not just calculate the actual turning angles between consecutive steps the ants take? This would give an actual assessment of both the bias and the noise imposed on the random walks, which the authors could then use directly in their models.

3) It would be important to better connect the author's previous mechanism (relating the colony's response to individual ants sensing their own food levels and its temporal dynamics) to the new mechanism (spatial-temporal dynamics). Are they mutually exclusive? It would be useful to elaborate on this in the Discussion.

4) It would be useful to add a few supplementary movies from the experiments, showing ants moving toward the entrance with low food loads, and moving away from the entrance with high food loads.

Reviewer #2 (Public Review):

Because individuals in most colonies of eusocial insects (i.e., ants, social bees, social wasps, and termites) cannot directly reproduce, theory suggests that natural selection will shape the behavior and physiology of such individuals to be hyper-sensitive to the needs of their colony. In the context of foraging, an individual should make decisions of how often to search for new food based on the "hunger" of the colony that she belongs to. In fact, in previously published work, the authors of this manuscript have confirmed empirically that the frequency of foraging events for individual workers in colonies of _Camponotus sanctus_ carpenter ants is correlated with the amount of food stored within the collection of ants within the nest -- as the colony "satiated" (i.e., the communal stomach of the average nest ant became full), the foraging frequency would decrease (and vice versa). In that work, the authors showed that an individual's decision to leave a nest to return to foraging was predictable from her own communal stomach ("crop") level and how quickly it was being depleted by nest ants receiving it. From that observation, the authors previously suggested that a cognitive process within each individual ant could monitor these two internal variables (crop level and rate of change) and lead an ant to make a decision as to if and when to leave a nest. In the current work, the authors suggest an alternative mechanism that exports the discrete decision making into the nest cavity itself and only requires an individual forager to adjust her movement pattern based on her current level of crop load. In particular, they use computational and mathematical models to show that spatiotemporal statistics similar to real ants emerge when hypothetical modeled foragers move deeper into a nest when their crop level is above a certain threshold and instead move toward the nest exit when their crop level is below that threshold (leaving the nest when randomly encountering it). This simple crop-based rule does not require estimation of depletion rate nor require an ant to deliberate over when to exit. Foragers in "hungry" colonies have shallow penetration in their nests before turning around and quickly returning to foraging while foragers in "satiated" colonies have deeper penetration and may remain in their nests for long periods of time. This proposed mechanism provides the adaptive foraging patterns observed in real carpenter ants with significantly reduced assumptions about individual cognitive abilities when compared to previous mechanistic explanations of this behavior. Broadly speaking, it (combined with other recent work from these authors and others) helps to demonstrate proof of concept of cognitive hypotheses that are embodied in the physical environment around the individual apparently making the decision.

The movement rule proposed by the authors is elegantly simple and produces trajectories that are, at least to the human eye, a good match to the stereotyped trajectories from real ant colonies in terms of their directionality and duration, and the length of these trajectories is modulated by colony hunger-state in exactly the same way as the real ant trajectories. Although the authors do not provide statistics on multiple runs of the simulation (they provide examples of single runs), they do complement their simulation work with both deterministic and stochastic models of statistics of the modeled paths and show that those statistics have the same qualitative relationship to colony hunger-state as the statistics of the real ants. Consequently, the paper provides a compelling argument via the use of multiple types of models for a novel behavioral rule that answers an important question in collective decision making in confined physical spaces.

Much of the authors' argument rests on trajectories and statistics generated from a two-dimensional computational simulation that may be overly simplistic. The computational model simulates a single forager (as opposed to multiple foragers) arriving to a nest that is partitioned into a grid of squares with an immobile ant in the center of every square. Foragers move in discrete steps from square to square, with the guarantee of an interaction in each step. This "grid world" model of ant nest movement is significantly different than the experience of real foraging ants returning to the nest, and the authors even admit that deviations between the empirical data and the computational model may be due to nest-ant clumping and interaction sparsity in the paths of real ants. Continuous-motion agent-based models are commonly used to investigate collective-motion hypotheses, and so the choice of a grid world model instead seems notable and weakens the authors' arguments. Furthermore, whereas the deterministic mathematical model of grid-world forager trajectories seems too simplistic, the stochastic model buried in the appendix that is meant to validate the deterministic model's results seems to have some potential flaws and is itself not validated experimentally against replicated simulation data. Instead of perfecting these models, the authors could have bolstered their arguments using more familiar approaches from statistical mechanics that might help explain the likely depth an ant "diffuses" into such a nest. In the current form of the manuscript, the mathematical models do not add much beyond the simulation models (and the lack of replication of the simulated data may make some readers wonder if the example trajectories are representative).

There are also a few questionable parameters that the authors have chosen in their model, likely for analytical tractability. For example, the authors assume that at each interaction between a forager and a nest ant, the forager offloads enough food to fill 15% of the crop space remaining in the receiving ant. One can assume that this parameter is something like the 63.21% associated with an exponential time constant or may be based on empirical measurements of transfer in real ants, but the actual justification is not completely clear from the manuscript. Because the mathematical models make predictions that depend upon these parameters, their existence (and plausible values) is itself an important assumption that needs to be defended for the argument to be truly compelling.

Beyond these methodological issues, the behavioral model described by the authors assumes that ants are able to choose a direction toward their nest's entrance at any time. This within-nest path-integration ability does not seem cognitively inexpensive, which narrows the cognitive distance between the behavioral model they propose here and the one they proposed in their prior work and weakens the argument for the relevance of this new model. The authors failed to place their work within the context of other simple cue-based motion-switching behaviors discussed in the literature for other taxa - such as "running" and "tumbling" in E. coli bacteria - but if they had, they might have envisioned an alternative crop-based motion rule that would have the same effect as their current rule (i.e., movement toward the entrance on low crop state) without having to assert that the ant moves directly back toward the entrance.

Focusing on the explanatory power of this model specifically for (some) ants, the authors do not address how to empirically reconcile the ambiguity between the more cognitive mechanisms proposed in their previous work (where ants "decide" to exit a nest) and the current proposal (where the nest cavity "decides" when the ant will exit). For this new hypothesis to be useful, it must be empirically discriminable from the previous hypothesis. At first glance, it is difficult to imagine an experiment that would lead to different predicted behavior from the two different hypotheses. In other words, at the moment, it seems impossible to tell whether the "ant decide" or the "nest decide" model is a better predictor of real ant behavior/cognitive architectures. The lack of discriminability becomes even more problematic when considering that the current version of the model actually increases some cognitive demands by assuming (as described above) that ants keep track of the position of the entrance over the trajectory within the nest.

The arguments in the current form of this manuscript could be strengthened by adding realism, connections to related literature in collective motion and motion ecology, and more general models from statistical mechanics, and it is important for the authors to identify potential ways to empirically discriminate between the model introduced here and the behavioral model suggested in their prior work. That said, the salient features of the basic crop-cue-based two-motion-primitive model proposed by the authors are elegant and novel and help to further demonstrate how cognition can be embodied in the physical spaces it is embedded within. The authors focus on a particular example in ants, but it is easy to imagine extending the same model to a variety of other scales and application spaces. For example, there may be microbiological examples of coordination among collectives where individuals face even more stringent cognitive constraints. Moreover, the same methods might be used to build artificial swarms in engineering contexts that allocate to tasks based on demand without significant communication or sensing requirements. Even in industrial organization, there may be ways to use methods like these to ensure an emergent adaptive re-allocation of human workers to tasks based on need. In general, this manuscript provides a new example of how spatiotemporal properties of decision making long thought to be associated with cognitive processes endogenous to individuals can be alternatively generated by simple cue-based behaviors interacting in a non-trivial environment. This is a relatively new perspective that may be useful in both the analysis of natural systems as well as the design of artificially intelligent systems. With the right framing, the example from this manuscript could be very useful not only to ant biologists but to scientists and engineers interested in collective decision making more broadly.

Reviewer #3 (Public Review):

This work adds to our understanding of the many diverse ways that different species of social insects organize the regulation of foraging behavior. This work compares model results with data previously collected on Camponotus sanctus, an ant species that collects nectar. Unlike other species in which foragers collect prey, seeds or other items that they do not ingest, in nectar-feeding species such as this one, the foragers drink nectar and then must unload it by regurgitating to other workers at the nest. This work presents a model that suggests that, like honey bees who also collect nectar, a C. sanctus forager's decision to exit the nest on its next trip depends on when it can unload the nectar, which is linked to the amount of nectar currently held by other workers.

Reviewer #1 (Public Review):

This paper addresses an important issue of how humans select foot placement when running on uneven terrain. The authors examine empirical and model-based data to determine whether runners specifically opt for more level locations on such surfaces. The manuscript suggests potential strategies of how humans mitigate fore-aft impulses during foot contact so as to maintain stability in response to changes in terrain.

Overall, the manuscript provides additional insight into lower-limb mechanics of runners on natural surfaces. The model-based analysis of lower limb compliance is especially useful in the context of stability and understanding human motor control. In addition, participant foot placement analysis using empirical and statistical models is very compelling and provides insight into how much planning occurs during running on uneven terrain. However, there are a few concerns of note:

- A central motivation of the study appears to be that past research did not incorporate height and slope variations when studying gait on uneven terrain. Although some of the past work cited by the authors does focus on step-like terrain, the (Voloshina, Ferris 2015) study had both height and slope variations. In that particular study, the terrain consisted of blocks that were smaller than the dimensions of the average human foot. This means that, during each foot-flat phase, the foot had to span at least two blocks of different heights, placing it at a slope in the fore-aft direction. Similarly, the columns of that terrain layout provided slope variations in the medio-lateral direction. Referring to this surface as "step-like" is inaccurate and potentially misleading. Considering that the terrains in the present study and the study from 2015 likely cause very similar types of perturbations to the runner, the motivation behind the current study is not strongly validated. The authors should consider re-evaluating their results in the context of past studies.

- A primary outcome of the study is the analysis of empirical and model-based fore-aft impulses experienced by runners during foot contact. The authors suggest that this measure is related to stability but do not provide extensive explanations. It would be helpful to include additional background information on how impulse analyses have been used previously and why they are particularly fitting in this context.

- The study evaluates participants running back and forth on a 24m track for up to 10 minutes. This means that participants had to perform many turns during each trial. The authors present metabolic energy expenditure data but do not address how these data may be skewed due to the large instances of directional changes. Measuring metabolic data during such tasks is generally noise-prone, potentially leading to an inaccurate representation of energy expenditure. Considering that this is a comparative study and participants had to perform such turns for each terrain trial, this issue could be minor. However, the authors are encouraged to provide more detail on experimental protocol. Addressing whether or not participants stepped off the terrain to switch directions and providing insight into how this experimental approach could potentially affect outcomes could be particularly helpful.

Reviewer #2 (Public Review):

The authors' paper extends their earlier work based on a 2D model of running stability while negotiating sloped terrain of random variable height, extending from a traditional point mass-spring model (SLIP) but with a moment of inertia about the CoM ([19], Dhawale et al. Roy Soc Open Sci 2019). In this study the authors carry out an experimental study of human subjects running over an experimentally-created undulating terrain surface (0.6 m wide x 24 m long) with a known 3D topography, in which they combine a 3D kinematics analysis of foot movement trajectory and placement relative to the terrain topography and in relation to body CoM (hip) movement; with measurements of ground reaction forces to estimate foot-substrate impulses over a subregion of the terrain, and measurements of the runners' metabolic energetics via a portable runner-carried gas analyzer system.

The authors' findings are generally supported by their results, showing that runners do not appear to rely on visual guidance to select foot placement on undulating terrain (this based on computational Monte Carlo simulations of foot placement probabilities favoring level terrain surfaces) and likely achieve stability while running largely by means of limb joint compliance that passively adjusts to variable foot-ground impulses (based on ground reaction force estimates and a collisional multi-segment limb joint model for which joint compliance was varied). As a result, the authors found no significant increase in the metabolic cost of uneven terrain versus level surface running.

However, whereas the authors motivate their study by its relevance to the evolution of human running ability and persistence hunting, which requires running over uneven natural terrain, a weakness is that their in-depth analysis is heavily focused on the mechanics and resulting energetics of running over undulating terrain in the context of foot placement strategies for maintaining stability and whether this depends on visual guidance of foot placement relative to the terrain. The authors claim surprise (Discussion, l.191-192) that the runners do not appear to rely on visual information about unevenness to guide their footsteps. However, based on the nature of their sloped undulating surface, their results were unsurprising to this reviewer.

The authors' study was also motivated to examine the effect of sloped surfaces on running biomechanics, as previous studies have examined step-like terrain comprised if piecewise level blocks or step height transitions, which the authors (correctly) note represent obstacle negotiation rather than how runners may be challenged by undulating sloping terrain. The authors argue (l. 5-6) that a combination of height and slope variations like a natural undulating terrain will be more challenging than one that involves only step height transitions. However, the basis for this statement is not clear. And, indeed, the results the authors find for humans running over a sloped, undulating terrain (height range ~ 40 mm) shows that a sloped, undulating terrain does not actually present a significant challenge, given that it appears to require little or no visual guidance of foot placement and no significant increase in metabolic energy use. To the contrary, this reviewer would argue that obstacle avoidance is the more challenging feature of natural terrains that must be successfully negotiated, which is a common experience for trail runners. The reviewer, therefore, fully agrees with the authors' conclusion (l. 259-261) "Our data thus suggests that terrain-guided foot placement strategies are not required for stability on gently undulating terrain [compared with obstacle avoidance on more complex terrain]".

The principal novelty and value of the authors' study is the analysis of fore-aft impulse and the role of limb joint compliance for adjusting to changes in fore-aft impulse to favor running stability. The authors' paper suffers from overstating the broader relevance of its findings and by merging methods and discussion with the results that it reports. The methods, themselves, are detailed and thorough in their description, and the authors' modeling approaches appear sound, sophisticated and appropriate for the analyses of foot placement strategies and limb compliance in relation to collisional impulse.

Repeatedly in the Results section, however, these methods are summarized when reporting a result (based on the method) and discussion points are mentioned. Specifically:

l. 66-96 This starting section does not present results per se, but a summary description of experimental methods an analytical approach. Actual results findings are not presented until l. 97.

l. 111-117: This summarizes analytical methods; not results per se.

l. 135-145: Summary of methods/analytical approach continues to be blended in with results in these sections.

l. 169 - Comparison of limb retraction rate on uneven vs level terrain of human subjects here with running birds is fine for discussion but not results per se.

l. 170-171: This is a discussion point, not a result.

l. 187-189: Again, discussion not a result.

A final concern is whether and how the requirement that runners repeatedly decelerate, turn and reaccelerate to run back and forth over the 24 m long uneven and level terrains at 3 m/s affects the metabolic measurements? Running at 3 m/s indicates 8 sec to traverse the runway length and, if adding another second for turning to reverse direction and run back = 9 s, this would indicate for a 8 to 10 min metabolic running trial ~53 to 67 turns per trial. Presumably, these would have an effect on running cost.

Reviewer #1 (Public Review):

This is a fascinating paper that takes up an important question with a creative and new approach. We have a few suggestions that we hope are constructive for the authors.

1. One nagging concern is that the category structure in the CNN reflects the category structure baked into color space. Several groups (e.g. Regier, Zaslavsky, et al) have argued that color category structure emerges and evolves from the structure of the color space itself. Other groups have argued that the color category structure recovered with, say, the Munsell space may partially be attributed to variation in saturation across the space (Witzel). How can one show that these properties of the space are not the root cause of the structure recovered by the CNN, independent of the role of the CNN in object recognition?

2. In Figure 1, it could be useful to illustrate the central observation by showing a single example, as in Figure 1 B, C, where the trained color is not in the center of the color category. In other words, if the category structure is immune to the training set, then it should be possible to set up a very unlikely set of training stimuli (ones that are as far away from the center of the color category while still being categorized most of the time as the color category). This is related to what is in E, but is distinctive for two reasons: first, it is a post hoc test of the hypothesis recovered in the data-driven way by E; and second, it would provide an illustration of the key observation, that the category boundaries do not correspond to the median distance between training colors. Figure 5 begins to show something of this sort of a test, but it is bound up with the other control related to shape. Similarly, if the claim is that there are six (or seven?) color categories, regardless of the number of colors used to train the data, it would be helpful to show the result of one iteration of the training that uses say 4 colors for training and another iteration of the training that uses say 9 colors for training. The text asserts that Figure 2 reflects training on a range of color categories (from 4 to 9) but doesn't break them out. This is an issue because the average across these iterations could simply be heavily biased by training on one specific number of categories (e.g. the number used in Figure 1). These considerations also prompt the query: how did you pick 4 and 9 as the limits for the tests? Why not 2 and 20? (the largest range of basic color categories that could plausibly be recovered in the set of all languages)?

3. Regarding the transition points in Figure 2A, indicated by red dots: how strong (transition count) and reliable (consistent across iterations) are these points? The one between red and orange seems especially willfully placed.

4. Figure 2E and Figure 5B are useful tests of the extent to which the categorical structure recovered by the CNNs shifts with the colors used to train the classifier, and it certainly looks like there is some invariance in category boundaries with respect to the specific colors uses to train the classifier, an important and interesting result. But these analyses do not actually address the claim implied by the analyses: that the performance of the CNN matches human performance. The color categories recovered with the CNN are not perfectly invariant, as the authors point out. The analyses presented in the paper (e.g. Figure 2E) tests whether there is as much shift in the boundaries as there is stasis, but that's not quite the test if the goal is to link the categorical behavior of the CNN with human behavior. To evaluate the results, it would be helpful to know what would be expected based on human performance.

5. The paper takes up a test of color categorization invariant to luminance. There are arguments in the literature that hue and luminance cannot be decoupled-that luminance is essential to how color is encoded and to color categorization. Some discussion of this might help the reader who has followed this literature. Related, the argument that "neighboring colors in HSV will be neighboring colors in the RGB space" is not persuasive. Surely this is true of any color space?

6. The paper would benefit from an analysis and discussion of the images used to originally train the CNN. Presumably, there are a large number of images that depict man-made artificially coloured objects. To what extent do the present results reflect statistical patterns in the way the images were created, and/or the colors of the things depicted? How do results on color categorization that derive from images (e.g. trained with neural networks, as in Rosenthal et al and presently) differ (or not) from results that derive from natural scenes (as in Yendrikhovskij?).

7. It could be quite instructive to analyze what's going on in the errors in the output of the classifiers, as e.g. in Figure 1E. There are some interesting effects at the crossover points, where the two green categories seem to split and swap, the cyan band (hue % 20) emerges between orange and green, and the pink/purple boundary seems to have a large number of green/blue results. What is happening here?

8. The second experiment using an evolutionary algorithm to test the location of the color boundaries is potentially valuable, but it is weakened because it pre-determines the number of categories. It would be more powerful if the experiment could recover both the number and location of the categories based on the "categorization principle" (colors within a category are harder to tell apart than colors across a color category boundary). This should be possible by a sensible sampling of the parameter space, even in a very large parameter space.

9. Finally, the paper sets itself up as taking "a different approach by evaluating whether color categorization could be a side effect of learning object recognition", as distinct from the approach of studying "communicative concepts". But these approaches are intimately related. The central observation in Gibson et al. is not the discovery of warm-vs-cool categories (these as the most basic color categories have been known for centuries), but rather the relationship of these categories to the color statistics of objects-those parts of the scene that we care about enough to label. This idea, that color categories reflect the uses to which we put our color-vision system, is extended in Rosenthal et al., where the structure of color space itself is understood in terms of categorizing objects versus backgrounds (u') and the most basic object categorization distinction, animate versus inanimate (v'). The introduction argues, rightly in our view, that "A link between color categories and objects would be able to bridge the discrepancy between models that rely on communicative concepts to incorporate the varying usefulness of color, on the one hand, and the experimental findings laid out in this paragraph on the other". This is precisely the link forged by the observation that the warm-cool category distinction in color naming correlates with object-color statistics (Gibson, 2017; see also Rosenthal et al., 2018). The argument in Gibson and Rosenthal is that color categorization structure emerges because of the color statistics of the world, specifically the color statistics of the parts of the world that we label as objects, which is the same approach adopted by the present work. The use of CNNs is a clever and powerful test of the success of this approach.

Reviewer #2 (Public Review):

Vries et al. investigated the mechanism of the color categorical perception and tried to answer the question of whether it develops universally or it is relative to local communication. So they investigated whether a categorical representation of color emerges from a Convolution Neural Network (CNN) that is trained to perform an object recognition task. The results indicate that the CNN has a categorical representation of color, which suggests that the color categorical perception might emerge from the object recognition.

In general, I think the results are interesting. They performed a psychophysical experiment with the CNN, which shows the border of color category was largely invariant to the training colors. Also, further experiments with the evolution algorithm and other experiments confirm this.

However, I think the approaches to address this question are not straightforward. All of the approaches in the paper rely on the retraining of the last layer. I was hoping they would provide more direct evidence to support their claim. Also, if they can show the color categorical information revealed by the CNN is similar to the human's color perception, that would help to strengthen their claim.

Reviewer #3 (Public Review):

This paper investigates the emergence of color categories as a result of acquiring object recognition. The authors find that color categorization is an emergent property of a Convolutional Neural Network (CNN) trained with ImageNet for object recognition. In short, they find CNN, precisely a ResNET, can represent color in a categorical manner. They also show the categories obtained through the model are meaningful for more complex images and tasks. Analyzing how deep neural networks represent color categories is an under-studied but important problem in cognition and the authors did an excellent job presenting their analysis and results. The finding reveals features of deep neural networks in color processing and can also guide future theoretical and empirical work in high-level color vision. The method can be used to investigate other questions in high-level vision.

Strength:

The current modeling results support the immediate conclusion that color categories can emerge from learning object recognition. The method is novel and the result is intriguing. Most of the analysis is clear and the paper is easy to follow. Extensive experiments are done with the model and convincing results are presented.

Weakness:

The main weakness of the paper is the scope. In many places in the paper, the authors write that the results support several unsolved issues in biological color processing and color categorization. I am not convinced how the results, purely obtained from modeling CNN, connect to the biological color processing as the authors speculated in many places in the article including Introduction and Discussion. To support these claims, psychophysical data or experimenting with published psychophysical data are needed.

Specifically, I find the following speculations not immediately supported by the results from this paper.

First, I am not sure about the connection the author draws between the emergence of color categories from CNN (findings in this paper) with the debate of Universalists and Relativists, and support that "categories can emerge independent of language development". The fact that output layers of CNN trained on object recognition can cluster color into categories does not mean the color categories used in humans are formed before they have language. Even though the network isn't explicitly trained with color names, the CNN has been trained with object labels. Aren't the object labels part of language acquisition?

Second, the authors wrote "The current findings can explain why the general development of categories is so similar across languages: If color categorization is a side effect of acquiring basic visual skills (given relatively similar circumstances across the globe) color categories are expected to shape in a similar fashion throughout many cultures". There are no explicit measurements of how different cultures would agree on these color categories. The current results only support that CNN trained on object recognition can discover limited color categories. It doesn't say anything about human color categorization across cultures.

Third, in the Discussion, the authors wrote "they can explain why the emergence of color categories over cultures broadly follows a universal pattern". How can a CNN trained with ImageNet explain broad cultures? Even though ImageNet contains common objects labeled mostly by people from western countries, they do not represent a diversity of cultures. The current results suggest a relationship between object recognition and color categorization. But this relationship may vary from culture to culture.

Finally, it would be great if the authors can experiment with network architectures other than ResNET. An alternative model trained on different image datasets can answer the question of under what circumstance color categories emerge from pre-trained models.

Reviewer #1 (Public Review):

This paper uses 2 cohorts from the UK and links a) risk factors associated with low antibody levels after vaccination and b) risk factors for infection. The paper makes the important point that following the third vaccination, risk factors associated with low antibody response after the first vaccination, are less likely to lead to sub-protective levels. This highlights the importance of obtaining a booster shot. Though it is not a primary finding of the paper, the observed discordance between self-reported infection and anti-nucleocapsid positivity is an important finding. While these findings are potentially useful, the presentation of the data is somewhat unfocused, and the message is presented in a diffuse fashion. Moreover, certain key components of the analysis such as the assay threshold and timing of samples after the 2nd vaccine are a bit confusing and require clarification. The use of univariate analyses can be misleading. Finally, the relevance of the findings relative to our current stage of the pandemic with multiple new VOCs requires a clearer explication.

Reviewer #2 (Public Review):

In this study, the authors collected and analyzed blood samples from >9,000 participants from two cross-sectional cohort studies in the UK. The ALSPAC cohort only collected data during April and May 2021, whereas the TwinsUK cohort collected data during April and May 2021 and November 2021 to January 2022. They measured anti-Nucleocapsid and anti-Spike antibodies using the collected blood samples. They investigated the variation in antibody levels and risk factors for lower antibody levels following each round of SARS-CoV-2 vaccination across a wide range of socio-demographic, SARS-CoV-2 infection and vaccination, and health factors. Alongside the descriptive analysis, the authors performed some multivariable regression analysis.

Peer review report

Title: Burden of HCoV infection in children hospitalized with lower respiratory infection in Cape Town, South Africa

version: 1

Referee: Jessica Price

institution: University of Witwatersrand

email: Jessica.Price@wits.ac.za

ORCID iD: 0000-0002-4020-6850

General assessment

This manuscript it well written, with a clear description of the methods, results and discussion of findings. I think that with minor corrections it would be ready for publication.

Essential revisions that are required to verify the manuscript

Methods:

Study procedure:

1) Please add a reference to the parent study which details the full methods of the parent study.

2) In the third paragraph of this section the authors refer to children young than 18 months with a positive HIV Elisa as being confirmed to have HIV infection. Please double check this - I think it is supposed to be that these children were categorised as HIV- exposed with a confirmatory HIV PCR preformed to determine HIV infection.

Results:

3) The current phrasing of the results suggests that there were no refusals amongst those eligible to participate. Is that accurate?

4) In table 1: HIV status “exposed by negative” – is that children under 18 months or under 6 months? If only those less than 6 months please explained where children between 6-18 months are categorised.

Discussion:

5) 16 patients were found to have different human coronaviruses on the IS and NP samples. Please discuss the implications of this finding. Which would you act on? Does this bring into question the validity of the two methods if they are detecting different viruses in the same patient. I would understand if one method detected additional viruses but to have completely different viruses across the two specimens on the same patient is potentially problematic.

6) The authors note that the study group only incudes hospitalised patients and therefore cannot comment on community transmission/burden. However there have been many community-based surveillance programmes to track respiratory virus burdens and transmission patterns in SA (including work by Sharon Cohen, NICD) – it would be helpful to the readers if the authors could review some of these publications and comment on relevant similarities or differences. (most recent of these publications can be found here: https://crdm.nicd.ac.za/projects/phirst-c/)

7) Please review and redraft this paragraph 5 in the discussion – starting “compared to RSV and other respiratory viruses…”. I could not follow what the authors are trying to say in this paragraph.

8) Please add in a limitations/recommendations section.

Conclusion:

9) The final paragraph of the conclusion raises some interesting questions but does not fit as part of the conclusion. I suggest moving this to be included as part of the discussion, and more fully discussing the questions raised regarding the value of testing for disease if they do not cause severe disease, nor change treatment strategies.

Other suggestions to improve the manuscript

• Stylistic preferences in the introduction: avoid using “for example” when describing work referenced. Either just add the reference number, or use phrasing such as “as shown by (author name) who found …” etc.

• Methods - Statistics: Typo in the last paragraph – please confirm if Stata 13 or stata 16, and add in the necessary stata package reference.

Decision

Requires revisions: The manuscript contains objective errors or fundamental flaws that must be addressed and/or major revisions are suggested.

Reviewer #1 (Public Review):

The core conclusions in this manuscript are well supported. First, the genomic data clear support a recent origin of the Baltic and Arctic ecotypes from an Atlantic-like ancestor, without major bottlenecks and with gene exchange at least on the one sampled sympatric location but probably also more widely. The genome scans strongly suggest the involvement of multiple loci in divergence. This is supported by the quantitative trait locus analysis but this support is relatively weak because the number of markers used was small and markers were rather unevenly distributed on the genetic map, leaving little power to detect QTL in some areas. An analysis of the power of the QTL analysis is lacking and there might also be an issue about whether the measured trait truly captures rhythmicity.

The presence of segregating variants across all sampled populations for the loci that appear to underlie local adaptation, plus the absence of strong sweep signatures, is good evidence that adaptation to the Baltic environment was based on standing genetic variation. This is supported by the known presence of local adaptation to tidal regimes within the Atlantic ecotype, providing a mechanism for the maintenance of standing variation. Clustering of putatively adaptive loci in one region of chromosome 1 seems clear, although it is not formally compared to a random distribution.

The authors make an interesting point that it is only when many genes are involved that Gene Ontology enrichment is expected to be informative. Here, they also had clear a priori expectations which are neatly fulfilled, further suggesting that differentiated loci are good candidates for a role in adaptation.

The broader context provided for the analysis of the fascinating marine midge system is brief and could usefully be expanded. It should make clear that, despite some focus on major gene effects that can be assigned to individual loci, there is widespread evidence for polygenic adaptation. Even where structural variants are known to have major effects, many other regions of the genome typically contribute to local adaptation. It would also be helpful to refer to theoretical expectations regarding distributions of effect sizes and clustering of locally-adaptive loci, with or without gene flow.

Reviewer #2 (Public Review):

Fuhrman et al. explore a fascinating system to study the evolution and genetic architecture of ecological adaptation in marine midges. They use a number of approaches including analyses of whole genome sequences and QTL mapping to explore population structure and the loci associated with the timing and mode of reproduction. I have some concerns about the analyses and interpretations which I outline below.

1) My primary concern is in the design and interpretation of the QTL analysis. The QTL approach used here has low power, both due to the sample size and the number of markers used (it looks like ~8 per chromosome). The authors use an analysis of the sex determining locus as a "control" but because of the complete heritability of this trait in most systems it is more of a straw man to me. The authors conclude that the architecture of the trait is polygenic based on this, but we are missing key information to evaluate this.

2) There are some issues with the presentation and interpretation of the population genetic analyses. Many assumptions are made about whether introgression or ILS occurred and there are statements that are not accurate about it being "impossible" to distinguish between these scenarios.

3) Some of the analyses associated with ecological adaptation that follow on the QTL results struck me as ad hoc and with the potential to lead to spurious results. I am not familiar with the BayPass approach but since it is the approach that explicitly accounts for population structure it seems the one that would be most appropriate for the authors to focus on in a revised manuscript. The use of phylogenetic windows that associate with ecotype is concerning to me as given the level of ILS and gene flow that appears to be present in this system is would be very challenging to distinguish signal from noise.

4) There were issues with the GO analysis that should be addressed. Because the gene universe used for GO enrichment is a subset of the full gene set, GO enrichment results will be biased. This will mostly lead to false positives (i.e. overrepresentation of a GO category due to evaluating a subset of genes that fall in that category).

Reviewer #1 (Public Review):

The authors define regulatory networks across 77 tissue contexts using software they have previously published (PECA2, Duren et al. 2020). Each regulatory network is a set of nodes (transcription factors (TF), target genes (TG), and regulatory elements (RE)) and edges (regulatory scores connecting the nodes). For each context, the authors define context-specific REs, as those that do not overlap REs from any of the other 76 contexts, and context-specific regulatory networks as the collection of TFs, TGs, and REs connected to at least one context-specific RE. This approach essentially creates annotations that are aggregated across genes, elements, and specific contexts. For each tissue, the authors use linkage disequilibrium score regression (LDSC) to calculate enrichment for complex trait heritability within the set of all REs from the corresponding context-specific regulatory network. Heritability enrichments in context-specific regulatory network REs are compared with heritability enrichments in regions defined using other approaches.

Reviewer #2 (Public Review):

In this manuscript the authors develop a method, SpecVar, to perform heritability estimation from regulatory networks derived from gene expression and chromatin accessibility data. They apply this approach to public datasets available in ENCODE and Roadmap Epigenomics consortia as well as GWAS phenotype associations in UK Biobank. It promises to be a powerful method to interpret mechanisms from genetic associations. Below are some strengths and weaknesses of the paper.

Strengths

- The method performs heritability enrichment on two major genomic data types: gene expression and chromatin accessibility.<br /> - This method leverages gene regulatory networks to perform the heritability estimation, which may better capture complex disease architecture.<br /> - The authors perform an extensive comparison to other LDSC-based approaches using different tissue datasets.

Weaknesses<br /> - This approach may represent a modest advance over existing LDSC methods when looking at other complex traits.<br /> - The authors only compare with LDSC using different functional annotations as input, which may not be appropriate. A more broad comparison with other heritability methods would be helpful.<br /> - The method seems to be applied to "paired" data, but this is still bulk profiles not paired single-cell RNA/ATAC data.

The authors successfully applied a regulatory network approach to improving the heritability estimation of complex traits by using both gene expression and chromatin accessibility data. While the results could be further strengthened by comparing them to other network and non-network-based methods, it provides important insight into a few traits beyond the standard LDSC model with different functional annotations.

Given that this method is based on the widely used LDSC approach it should be broadly applied in the field. However, the authors should consider adapting this to single-cell data as well as admixed human population genetic data.

Reviewer #3 (Public Review):

Identifying the critical tissues and cell types in which genetic variants exert their effects on complex traits is an important question that has attracted increasing attention. Feng et al propose a new method, SpecVar, to first construct context-specific regulatory networks by integrating tissue-specific chromatin states and gene expression data, and then run stratified LD score regression (LDSC) to test if the constructed regulatory network in tissue is significantly associated with the trait, measured by a statistic called trait relevance score in this study. They apply their method to 6 traits for which there exists prior evidence on the most relevant tissues in the literature, and then further apply to 206 traits in the UK Biobank. They find that compared to LDSC using other sources of information to define context-specific annotations, their method can "improve heritability enrichment", "accurately detect relevant tissues", helps to "interpret SNPs" identified from GWAS, and "better reveals shared heritability and regulations of phenotypes" between traits. However, I think it requires more work to understand where exactly the benefits come from and the statistical properties of their proposed test statistic (e.g., how to perform hypothesis tests with their relevance score and whether the false positive rate is under control). In addition, it's not clear to me what they can conclude about the shared heritability (which means genetic correlation) by comparing their relevance score correlation across tissues to the phenotypic correlation between traits.

They show that SpecVar gives much higher heritability enrichment than the other methods in the trait-relevant tissues (Fig. 2). The fold enrichment from SpecVar is extremely high, e.g., more than 600x in the right lobe of the liver for LDL. First, I think a standard error should be given so that the significance of the differences can be assessed. Second, it is very rare (hence suspicious) to observe such a huge enrichment. Since SpecVar is based on LDSC, the same methodology that other methods in comparison depend on, the differences to the other methods must come from the set of SNPs annotated for each tissue. I think it is important to understand the difference between the SpecVar annotated SNPs and those from other methods. For example, is the extra heritability enrichment mainly from the SpecVar-specific annotation or from the intersection narrowed down by SpecVar?

They propose to use the relevance score (R score) to prioritise trait-relevant tissues. In Fig. 3, they show tissue-trait pairs with the highest R scores, and from there they prioritise several tissues for each trait (Table 1). I can see that some tissue has an outstanding R score, however, it is not clear to me where they draw the line to declare a positive result. The threshold doesn't seem to be even consistent across traits. For example, for LDL, only the right lobe of the liver is identified although other tissues have R scores greater than 100, whereas, for EA, Ammor's horn and adrenal gland are identified although their R scores are apparently smaller than 100. It seems to me they use some subjective criteria to pick the results. It leads to a serious question on how to apply their R score in a hypothesis test: how to measure the uncertainty of their R score? What significance threshold should be used? Whether the false positive rate is under control? Without knowing these statistical properties, readers won't be able to use this method with confidence in their own research.

Another related comment to the above is to investigate false positive associations, they should show the results for all tissues tested to see if SpecVar tends to give higher R scores even in tissues that are not relevant to the trait. It would also be useful to include some negative control traits, such as height for brain tissues.

Fig. 3 shows that tissues prioritised by LDSC-SAP and LDSC-SEG seem to make less sense than those from SpecVar. However, some of the results are not consistent with the LDSC-SEG paper (Finucane et al 2018). For example, LDL was significantly associated with the liver in Finucane et al (Fig. 2), but not in this study. How to explain the difference?

The authors highlight an example where SpecVar facilitates the interpretation of GWAS signals near FOXC2. They find GWAS-significant SNPs located in a CNCC-specific RE downstream of FOXC2 and reason these SNPs affect brain shape by regulating the expression of FOXC2. I think more work can be done to consolidate the conclusion. For example, if the GWAS signals are colocalised with the eQTL for FOXC2 in the brain. Also, note that the top GWAS signal is actually on the left of the CNCC-specific RE (Fig. 4b). A deeper investigation should be warranted.

They show that SpecVar's relevance score correlation across tissues can better approximate phenotypic correlation between traits. However, the estimation of the phenotypic correlation between traits is neither very interesting nor a thing difficult to do (it can be directly estimated from GWAS summary statistics). A more interesting question is to which extent the observed phenotypic correlation is due to common genetic factors acting in the shared tissues/cell types/pathways/regulatory networks between traits. Note that in their Abstract, they use words "depict shared heritability and regulations" but I don't seem to see results supporting that.

Line 396-402: "For example, ... heritability could select most relevant tissues ... but failed to get correct tissues for other phenotypes ... P-value could obtain correct tissues for CP ... but failed to get correct tissues for ... SpecVar could prioritize correct relevant tissues for all the six phenotypes." Honestly, I find hard to judge which tissues are "correct" or "incorrect" for a trait in real life. It would be more straightforward to compare methods using simulation where we know which tissues are causal.

Reviewer #1 (Public Review):

Yuan et al. propose that the magnesium transporter unextended (uex) controls Drosophila sleep via Mg2+ efflux, Ca2+-dependent CREB signaling, and a CNK-ERK pathway. UEX protein levels display daily oscillations in fly heads, whereas UEX-depleted flies show long sleep with low levels of Ca2+ and synaptic plasticity. Transgenic expression of wild-type UEX or the mammalian uex homolog CNNM1 possibly rescues the uex mutant sleep, supporting their evolutionary conservation. UEX forms a protein complex with the ERK signaling suppressor CNK, and UEX depletion appears to de-repress ERK activation. As expected, pan-neuronal CNK depletion phenocopies uex mutant sleep. Taken together, the authors suggest a novel mechanism whereby magnesium shapes animal sleep through the specific MAPK pathway. Overall, the authors reported quite a few exciting phenotypes in UEX-depleted flies using a range of analyses (e.g., behaviors, gene expression, Mg2+/neural imaging, and biochemistry). However, evidence for their causal link to sleep regulation is missing, and some key conclusions remain justified by more rigorous analyses. It is noteworthy that Wu et al. have previously demonstrated the Mg2+ efflux transporter activity of UEX and mapped its memory-enhancing function to a specific group of adult neurons in the fly brain (https://pubmed.ncbi.nlm.nih.gov/33242000/). Given the intimate interactions between sleep and memory, these findings may have broad implications in sleep-relevant physiology and disorders. However, a number of important control studies and statistical assessments are not included, but are necessary to conclusively interpret the data.

Reviewer #2 (Public Review):

The authors described the analysis of a magnesium transporter UEX as a sleep-regulating gene in Drosophila melanogaster. They also proposed the UEX regulates sleep through its downstream Ca2+-dependent CREB signaling and a CNK-dependent ERK pathway. The involvement of UEX in sleep regulation is novel and potentially interesting, but the data presented in the manuscript does not fully support the conclusions the authors proposed. Most of the data are derived from elav-GAL4, which is a non-specific pan-neuronal GAL4 driver. Since as the authors described, UEX functions to alter sleep in various brain regions, the relationship between UEX and other molecules in Ca2+-dependent CREB signaling and a CNK-dependent ERK pathway may be indirect in the sleep-regulating pathway, which means it may involve multiple regions of the brain using different pathways, and the sleep phenotype is the summation of different functions of UEX.

Reviewer #3 (Public Review):

In this manuscript, Yuan et al. examined the relationship between a magnesium transporter and sleep behavior. They find that the knockdown of a magnesium efflux transporter (uex) in neurons increases bout length of inactivity and recovery activity of the flies with neuronal knockdown of uex with a human homolog CNNM1. The authors suggest a model in which Mg2+ promotes sleep through the inhibition of Ca2+ levels that are wake-promoting in the mushroom body and PDF+ neurons. Overall, the idea explored here that ion homeostasis in the neurons contributes to behavior is an area that is timely and interesting to the neuroscience community. The transgenic lines of human CNNMs could be a useful tool for scientists studying metal transport and ion homeostasis in flies. Unfortunately, the results of the experiments do not entirely support the authors' conclusions.

The authors fall short of showing that the increased inactivity is sleep behavior as Mg2+ changes in neurons could be affecting the mobility of the fly. To validate that the increased inactivity is sleep, the authors should have used a combination of negative geotaxis, arousal threshold, or multibeam/video monitoring. Another characteristic of sleep is the presence of compensatory rebound following sleep deprivation. Here, when the authors sleep deprive the flies with uex knockdown, the flies do not have increased rebound sleep over control flies. Together the current data suggest that the increased inactivity may not be sleep and more evidence to the contrary should be shown.

In Fig 1, the authors show that there is a huge developmental effect on rest:activity rhythms when using the elav-gal4>uex RNAi compared to the inducible elav-geneswitch > uex RNAi, but in Fig 2, the authors use gal4 drivers rather than an inducible system. Use of an inducible system such as geneswitch, AGES, or TARGET is important to rule out developmental effects. Again, in Fig 4, the authors use the gal4 rather than the geneswitch for knocking down the other magnesium channels/transporters so it is unclear whether any sleep increase may be due to the role magnesium plays in development. In Fig 6 elav-gal4 was also used instead of GS. According to previously published work on UEX in fly neurons (Wu et al. eLife 2020 PMID: 33242000), UEX is primarily in the mushroom body and much lower expression in the PI or PDF+ neurons of the adult brains, further suggesting that sleep increases in the PDF+ and PI gal4s driving uex RNAi may be developmental.

From this work, the authors suggest that Mg2+ is sleep-promoting, and in the absence of uex efflux transporter to remove the Mg2+, Mg2+ increases to the point of inhibiting Ca2+, a wake-promoting signal; however, not all the Mg2+ transporters assayed efflux out Mg2+, but rather regulate the influx of Mg2+ into the cells. If a channel regulating Mg2+ influx is inhibited, the prediction would be that Mg2+ would be decreased and thus the flies should sleep less. But in Fig 4H that was not the case. All the Mg2+ transports/channel RNAi lines increased sleep. The authors do not reconcile this data with their proposed model. It is possible the Mg2+ transporter RNAi lines result in increased Mg2+ in the relevant neuronal subgroup in which case Mg2+ levels should be measured in the RNAi lines.

Reviewer #1 (Public Review):

Modified rabies viral vectors allow high throughput mapping of neuronal circuits with cell type specificity. However, lack of standardization in this field limits extrapolation of useful information, beyond the identity of anatomically connected regions, such as differential input densities and connectivity motifs. in this manuscript, Tran-Van-Minh et al., attempt to develop a statistical approach which will allow consolidation of new, as well as previously-acquired datasets, to yield biologically significant insights into the logic underlying rabies vectors' expansion from single starter cells.

This question the authors address is of high importance and the presentation of this manuscript is timely, as rabies tracing experiments increase at an exponential rate. The authors provide a largely complete description of the main pitfalls and caveats in current analysis approaches and common misconceptions in the interpretation of results. In addition, they correctly diagnose the potential of this methodology to extract pertinent information from such experiments and provide the reader with useful tips on how to better design, analyze and interpret them.

While such a paper is undoubtedly called for, and has the potential to substantially improve circuit mapping experiments, with little cost to the experimenter, there are a few critical flaws in their premises, which will limit a widespread adoption of their analysis approach. While the authors discuss caveats in design and interpretation of results, they do not implement these suggestions into their own experiments, such as the need for accurate estimation of starter cells and automated cell counting which is not biased by strong signal coming from dendritic arbors/axons in densely-labeled regions. While the authors claim that the reduction in residuals and relative conformity across experiments following log-transformation of n(i) and n(s) shows that their approach is robust, the large degree of variability across experiments, the datasets of some are biologically implausible, showing a decrease in n(i) as a function of n(s), suggests that this transformation disposes of useful information, which might help detect anomalies in data acquisition. The fact that the most rigorously-acquired dataset which was presented by the Allen Institute (BRAIN Initiative Cell Census Network, Nature, 2021) is the only one which, does not comply with the transformation, attests further to this caveat. This is probably the reason why the estimated number of individual neurons retrogradely labeled by a single starter cell (~1400) is more than an order of magnitude higher than any previously-published estimation, including those which include tracing from single-cell, and is highly unrealistic.

In addition, the model selected by the authors to fit the various datasets does not take into proper account saturation of n(i), as all proposed functions are growing functions. Here, the most suitable model which should be used to describe the nature of RV expansion is a cumulative distribution function, while the rest are either private cases (e.g. linear fit given low n(s) or zero divergence) or are biologically unrealistic (e.g. exponential fit). Again here, the only dataset which appears to fit this function the best comes from the BRAIN Initiative Cell Census Network (Nature, 2021).

In conclusion, while such work is called for and has the potential of becoming a staple in the connectivity toolbox, many of the premises presented here will need to be significantly adjusted, before the approach could be put into widespread use.

Reviewer #2 (Public Review):

Appropriate brains functions require the precise wiring of vast numbers of synaptic connections in broadly distributed neural circuits. Monosynaptic retrograde rabies virus tracing has become a common approach in neuroscience to assay presynaptic inputs into a given postsynaptic region. However, quantification and interpretation of rabies tracing data is confounded by the lack of uniform and appropriate measuring approaches across different studies and laboratories, as well as the lack of knowledge of the trans-synaptic transfer properties of different rabies viruses in various brain regions.

The current study comprehensively applies mathematical approaches to an example rabies tracing dataset in layer 5 of mouse visual cortex, as well as previously published datasets, to propose more standardized methodologies for rabies data analysis and interpretation. The major strength of the study is the rigorous and unbiased mathematical approaches applied to their data and a range of previously published studies in the field. Inclusion of representative image data would be helpful for readers and would further strengthen the study. Given the ubiquitous use of rabies virus tracing in the field, yet lack of insight into this crucial aspect of its use, this will provide a useful resource for the neuroscience community.

Reviewer #1 (Public Review):

The manuscript by Chen and colleagues contains a number of important findings. First, they showed with careful imaging, cell cycle characterization, and models that replisomes of fast-growing E. coli form factories and super factories (factories involving cousin replisomes), they estimated that factories split after the replication of 1/3 of the chromosome. Second, they used a replication fork block to monitor the interdependence of replisomes of factories. Third, they show that orphaned replisomes replicate slowly and frequently require a repair process to complete the replication cycle. This is the first characterization of an advantage of replicating bacterial genomes in a factory rather than as independent replisomes. This is an important discovery that simultaneously confirms the existence of DNA factories, a much-debated topic, and reveals one of their functions. The existence and characterization of factories were mostly addressed through imaging methods; here the combination of imaging with molecular genomics assays is a real asset for the manuscript.

Reviewer #2 (Public Review):

In most organisms, DNA replication is restricted to a relatively few cytologic structures termed replication factories. Studies indicate that such factories contain multiple replication forks. Although these observations suggest that replication fork colocalization has functional significance, the biological rationale for replication factories has remained elusive. To address this issue, the current study utilizes E. coli, a bacterium with a circular chromosome that replicates its DNA bidirectionally from a single origin of replication. During the first half of an E. coli DNA replication cycle, these two forks spatially co-localize into a single "factory." The experimental plan of this study is to block one of the two replication forks at various informative genomic locations and see if such blocks affect the progression and efficiency of the non-blocked fork. Using this approach, the authors find that blocking the progression of one fork at an early point in replication slows the progression of the corresponding unblocked fork and considerably increases its probability of replication fork collapse. This study considerably advances the field by demonstrating for the first time a possible biological purpose behind the replication factory - that factory formation in some yet unknown manner helps coordinate and stabilize bidirectionally oriented replication forks.

Although others have tried to study replication factories using similar experimental logic, this well-written study by Chen et. al. examines the problem with higher sensitivity and resolution using a very elegant and synergistic approach that combines 3-dimensional microscopy, deep DNA sequencing, and old-fashion cell biology with a series of carefully engineered E. coli strains containing a conditional replication fork block in different informative genomic locations. These approaches in combination allow one to make a direct experimental correlation between cytologically defined replication factories (3D fluorescent imaging of labelled replication factors with image deconvolution), and fork progression via an analysis of copy number (genomics). Their experimental approach and accompanying analysis pipeline will be of general interest to the research community.

In addition to a very careful analysis of factory formation that helps resolve several previous discrepancies on this subject, the authors used this approach to show that blocking one replication fork early in DNA replication coordinately decreases both the rate of fork progression and the level of fork stability in the unblocked sister fork. This conclusion is supported by their genomic analysis that shows the velocity of the unblocked fork slows when the other fork is blocked. To further elucidate this observation, the authors examined the likelihood that elevated replication fork collapse contributed to the decreased fork rate. As the restart of a collapsed replication fork depends upon genetic recombination, the role of recombination in fork progression in this situation was examined. Two questions were asked in this system: 1) Is the progression of the unblocked fork specifically reduced in the absence of genetic recombination (with a mutation in RecB)? and 2) Using chromatin IP, does this slow fork specifically recruit binding of a catalytically-dead Holliday-junction resolvase (RuvC)? The results from both experiments strongly support the conclusion that replication factories in some yet unknown manner are needed to stabilize the bidirectionally orientated replication forks. Although this strong conclusion indicates that the unblocked fork specifically creates DNA lesions, this approach does not unambiguously distinguish between damage resulting directly from fork collapse and damage caused by other aspects of defective DNA replication.

Reviewer #1 (Public Review):

In this study, the authors investigate multi-attribute economic decisions in humans. More specifically, they investigate the impact of adding a decoy option to a binary choice, the effect of which has been debated in the previous literature with recent studies finding effects in different directions. By re-analyzing large datasets from some of these studies and by applying computational modeling to the data, the authors propose that a subjective encoding rule comparing attributes separately, and not computed as one multiplicative value, explains participants' behavior on different levels of granularity. Context effects were well captured by a selective integration model. The work has many positive features, including praising open science, the analysis of the potential confounds of previously used designs, and both quantitative and qualitative comparisons of several well-justified models.

Reviewer #2 (Public Review):

Summary

This manuscript re-examines a distractor effect of decoy options on risky choice reported in previous research by re-analyzing data from previously published experiments that reported these effects. The previous studies reported that adding an unavailable decoy option to a choice set consisting of two available risky choices increased the discriminability between the two available risky choices, especially when the expected value difference between the two available risky options was small, by increasing the expected value of the unavailable distractor. The authors argue convincingly that the distractor effect is an artifact of two other confounding factors: one is that there is a covariance between the distractor's expected value and the subjective utility difference between the two targets; the second is that the expected value of the distractor alternative could covary with its relative position in the reward-probability space, and its relative position in the multi-attribute space could induce a well-known context effect. The first alternative explanation was established by comparing binary choice with and without the distractor present and finding the same effect in binary choice without any distractor present. The second was established by showing that the distractor effect was most pronounced when it was close to the higher-value target in the multi-attribute space, inadvertently producing a previously well-known attraction effect. These results clarify the role that an unavailable distractor plays in decisions between two risk alternatives.

Evaluation

This is a very comprehensive and somewhat complex manuscript. It does a good job of detective work to get at the bottom of the distractor effect reported in previous articles (including this journal). It essentially contains two main sections. The first section is designed to establish the conclusion that the distractor effect is an artifact of a confounding variable, the additive utility difference between the two available choices, and generalized linear model analyses were used to make this point. The second section is designed to show that the distractor effect also covaries with a well-known context effect called the attraction effect, and they use mathematical modeling of choice and response time to understand this part. Different hypotheses about how the risk information was integrated tested by varying how the drift rate was calculated in a racing drift diffusion model for choice and response time. In particular, they contrasted a divisive expected value type of integration hypothesis with a selective attention type of additive utility hypothesis. They concluded from these mathematical modeling analyses that an additive utility model for integrating the risk information was used in these experiments to evaluate the risky gambles.

Strengths the manuscript makes a very compelling case for the conclusion that the distractor effect was confounded with the additive utility difference between the available alternatives. This was achieved comparing the binary choice results, with and without the distractor, and finding little or no difference between these two conditions. The manuscript is also commendable for its rigorous mathematical modeling of the context effect of the distractor on the binary choices when the distractor was present.

One weakness is that the contribution is somewhat narrowly focused with respect to the phenomenon that it addresses - the distractor effect in risky choice. However, I do think it is important for understanding this particular phenomenon. The other main weakness is the complexity of the manuscript. The manuscript is very long with numerous detailed statistical analyses and computational modeling analyses. Generally speaking, the authors did a good job describing and summarizing all these analyses, and they made effective use of figures to illustrate the ideas and conclusions. However, there are several spots that are somewhat difficult to follow (see specific comments), and the reader is pressed to think pretty hard and fairly long and with a lot of effort to absorb all the points.

One other major concern I have regards the conclusion that the participants in these studies use an additive rather than a multiplicative rule to integrate the risk information. The additive rule is problematic in general because it fails to predict the reversal in the effect of probability on payoffs when the payoffs change sign. More specifically, increasing the probability of winning increases the probability of choosing an option when the payoff is positive, but the effect reverses when the payoff is negative. One needs to impose some pretty ad hoc assumptions to make the additive model account for this fundamental interaction between probability and payoff. Of course, the experiments reported here did not include negative payoffs, and so didn't run into this problem. In fact, when the payoffs are positive, it is possible to transform the multiplicative model to an additive model by a log transform. This transformation is only possible for the simple type of gamble investigated in this manuscript - a single amount to win with some probability of winning, otherwise win or lose nothing. If the gambles involved more than one outcome, then the theorist needs to deal with a sum of products and the log transform is no longer possible. For these reasons I am very skeptical about the general application of a summation rule for probability and value in risk choice. The authors do address this issue to some extent. They point out the abundance of other research supporting a multiplicative rule, and they speculate that the additive rule may have occurred within the restrictions of this special situation. The latter discussion is a good start, but I suggest that the authors discuss this fundamental issue in more depth.

Reviewer #3 (Public Review):

The authors re-analyze published datasets of value-based decision making with and without unavailable distractors, i.e., with ternary and binary choices. By setting the accuracy of binary choices as baseline, they show that a phantom distractor effect appears even without the presence of distractor. This result suggests that distractor effects could be partially explained by target-related covariation. They test how reward and probability are integrated under their datasets. The additive model wins over the multiplicative model in predicting both true and phantom distractor effects in binary choices. Then they test how multiple alternatives interfere with each other in ternary choices. They find that the model with the assumption of rank dominance wins over normalization models. They also replicate the correlation between individual-level decision noise and distractor-related parameters, which implies distractor effects can be emergent properties from a normative decision policy.

I see three strengths of this work.

First, the highlight of this work is that they explore the integration of the multi-attribute and multi-alternative information by bridging distinct distractor effects and providing a unified explanation. The result has a potential impact on a neuroscience topic that attracts a lot of attention in recent years-how the brain represents multiple features and items (e.g. Rigotti, Nature, 2013; Flesch et al., Neuron, 2022; Fusi et al., Curr. Opin. Neurobiol., 2016).

Second, the results of the trial-by-trial baseline approach warn that, due to the complexity of multi-attribute and multi-alternative problem, the studies of the effect should be designed and analyzed with care to prevent possible confounding factors from high dimensionality.

Third, besides static models that can only account for accuracy, the authors implement a dynamic accumulator frame to test all hypotheses. The dynamic accumulator models take into account both accuracy and reaction time. This approach strengthens their model comparison.

Overall, I think this paper is an impressive piece of work that clarifies the true effect of distractors by well-designed analysis and provides a model that bridges distinct distractor effects. Their analysis supports their claims.

Reviewer #1 (Public Review):

This manuscript provides the first cellular analysis of how neuronal activity in axons (in this case the optic nerve) regulates the diameter of nearby blood vessels and hence the energy supply to neuronal axons and their associated cells. This is an important subject because, in a variety of neurological disorders, there is damage to the white matter that may result from a lack of sufficient energy supply, and this paper will stimulate work on this important subject.

Axonal spiking is suggested to release glutamate which activates NMDA receptors on myelin-making oligodendrocytes wrapped around the axons: the oligodendrocytes - either directly or indirectly via astrocytes - then generate prostaglandin E2 which relaxes pericytes on capillaries, thus decreasing the resistance of the vascular bed and (presumably) increasing blood flow in the nerve.

Strengths of the paper

The paper identifies some important characteristics of axon-vascular coupling, notably its slow temporal development and long-lasting nature, the involvement of PgE2 in an oxygen-dependent manner, and a role for NMDARs. Rigorous criteria (constriction and dilation of capillaries by pharmacological agents) are used to select functioning pericytes for analysis.

Weaknesses of the paper

The study focuses exclusively on pericytes. It would have been interesting to assess whether arteriolar SMCs also contribute to regulating blood flow.

The slow (~10 minutes) time scale of the responses seen is remarkable when compared to grey matter neurovascular coupling which occurs in seconds. The authors suggest that this reflects movement of messengers along the nerve, but the action potential moves so rapidly down the nerve that it will activate the release of vasodilators essentially at the same time everywhere along the nerve, and there will be no concentration (or voltage) gradient to drive such a movement of messengers (unless there is a spatially localized unique site in the vascular bed where propagating responses are generated). Thus it remains unclear why the responses are so slow.

The activity-evoked dilation is thought to reflect PgE2 release at what is probably a physiological O2 concentration, but not at the hyperoxic 95% O2 used in most of the experiments. The involvement of NMDA receptors is apparently only shown (using OL-specific receptor subunit deletion) at the unphysiological high O2 level. This raises two questions: are NMDA receptors also involved in the response at low O2 (as schematized in Fig 6), and what is the messenger downstream of NMDARs at high O2 (NO would be an obvious candidate, previously shown to contribute to NVC in the grey matter).

Reviewer #2 (Public Review):

This paper describes a new concept of "axe-vascular coupling" whereby action potential traffic along white matter axons induces vasodilation in the mouse optic nerve. This is an initial report dissecting some of the mechanisms that are undoubtedly complex as in gray matter NVC. I like the novel AVC concept.

Some minor corrections and suggestions:

1) p3: "The cerebral white matter (WM) in the adult brain is particularly vulnerable to cerebrovascular diseases such as ischemia":this may be misleading since WM is actually far less vulnerable to ischemia than gray matter

2) p4-5: "The ON exhibited a median of 175.8 pc/mm2 {plus minus} 35.7 pc/mm2, more than twice the number of pericytes observed in the corpus callosum [...] and lower than cortex ": this seems incorrect, the density in cortex is not significantly different than ON

3) p5: what is the unit 'pc'? (A cellI I presume but please define at first use)

4) p7 : "To evaluate if pericytes have and retain their contractile properties, we applied the vasoconstrictor U46619 (100 nM) for 15 min followed by acetylcholine (ACh - 100 μM) as a vasodilator": if they saw an effect, how would the authors know these were mediated my pericytes and not smooth muscle cells?

5) in Fig. 3i there is a sharp step after U466... application: is this an artifact or evidence of a delayed constriction? Could a clearer trace be shown that does not confuse?

6) Fig. 4I: what does "20% CAP (norm)" mean? Why not just mV for the y-axis? Also what pulse width was used for stimulation?

7) Fig.5: it would be good to show both the CAPs (at various frequencies) and the vasorespones at 95% vs 20% O2. In particular, are the ONs able to sustain conduction at the higher frequencies (showing overlays as in 4I), and if not, could this at least partially account for the different responses at the two O2 levels?

8) Fig. 6G,H is somewhat misleading as it implies no change in AVC, at odds with 6E. Suggest some clearer labeling to reduce confusion surrounding this very important point.

9) P18 authors state radius of a MON is 150um but on p4 they say "150 μm - 200 μm thickness", pls clarify.

10) p19-20: as part of their second messenger speculation authors may also want to include NO that has been shown to induce important effects in WM. Indeed, testing the tat uncoupling peptides could be interesting to see of oligodendroglial NMDARs have a similar singling arrangement with NOS as do neurons. This may have important implications for WM neuroprotective strategies in stroke that have typically focused on gray matter mechanisms.

Reviewer #3 (Public Review):

This study used the ex vivo optic nerve preparation from adult mice to examine the organization of blood vessels and the mechanisms or neurovascular coupling (NVC). Strengths of the study include the benefits of the isolated preparation, which allows visualization of vessels and pericytes with high resolution and control over axonal activity and the extracellular environment, and the elegant analyses performed. Imaging at high resolution is critical, because vessel diameter changes can be small and slow to develop. The authors leverage this preparation to define the organization of blood vessels and pericytes in the nerve. They then examine the extent of NVC, showing that some aspects appear to be distinct. In particular, dilation does not present rapidly (over minutes) during axon stimulation, but rather emerges after the stimulation, increasing progressively over tens of minutes. It is similarly dependent on oligodendrocyte NMDARs and prostaglandin E4 receptors, but the latter only appears to be engaged during low oxygen conditions. There are several notable limitations of these studies. Less is known about NVC in the intact optic nerve, so it is unclear how well this preparation mimics the in vivo environment. All studies of NVC were performed in the presence of U46619 (an agonist of prostaglandin H2 receptors) to pre-constrict the vessels, which may interfere with NVC. The degree of vessel change was small and slow to develop, and the magnitude and timecourse of the dilation were not closely linked to the stimulation frequency, raising concerns about tissue stability and cell viability. Finally, the studies examined the role of oligodendrocyte NMDARs in NVC using a conditional gene knockout strategy to inactive the NR1 subunit in these cells. To control for possible developmental effects, additional studies could be performed using acute application of NMDAR antagonists, as this preparation contains only neuronal axons, and a further analysis of vessel structure and pericyte organization should be performed using the methodologies developed to characterize their properties in control nerves. Importantly, extracellular stimulation of the nerve, which triggers near simultaneous activation of axons may not mimic activity patterns in these nerves that occur during vision.

Reviewer #1 (Public Review):

The goal of this work is to study whether sleep and wake regulate cerebellar structural plasticity. Scanning electron microscopy and 3D reconstruction were used to characterize structural changes in spines and synapses in the mouse cerebellar cortex. The net number and size of synapses did not change between sleep and wake. However, the number of small portion of spines ("naked" spines without synapses) increased during sleep, and the number of branched spines (contains more than one synapse) increased during wake.<br /> The methodological approach (scanning electron microscopy) is laborious but adequate. However, it cannot follow the same synapses longitudinally, and live imaging, even only in superficial regions, is an ideal approach to achieve the goals.

The results did not find changes in the total number of spines or synapses during sleep and wake. Structural changes were found in small number of spines lacking a synapse. Whether these changes are functionally important is unclear. The results support circuit-specific, rather than global, effect of sleep on synaptic plasticity.

The role of sleep range from cellular maintenance to memory consolidation. Multiple evidence suggests that sleep regulates synaptic plasticity. Although this work did not attempt to understand the functional role of sleep in regulating learning and memory, it discovered that sleep promotes synaptic pruning in specific circuits of the cerebellum. Future similar anatomical studies in additional brain regions, including excitatory and inhibitory circuits, combine with physiological and behavioral assays are expected to provide insights to the role of sleep in regulating synaptic plasticity, learning and memory.

Reviewer #2 (Public Review):

This study used serial block face scanning electron in the mouse posterior vermis. The analysis showed that Purkinje cell "naked" spines, are ~5% of all spines after wakefulness but grow to ~10% of all spines after sleep. Additional analysis revealed that the observed sleep-wake difference is best explained by a change in the number of "branched" synapses, Branched spines are proposed to convert to single spines during sleep. It is speculated that sleep promotes the pruning of branched synapses. This is a beautiful study that must have taken considerable effort in addition to expertise. No such data exist in the literature and the observations are interesting in light of the prior data from cortex published by the same group.

Major critique:

• The abstract and in particular the second half is very difficult to follow and should be rewritten. It might be easier to follow if the authors compare to previous work in cortex<br /> • The figures are very well done. However, I am missing a model diagram explaining the model proposed for changes in naked spine during the sleep-wake cycle and the proposed functional consequences.<br /> • The authors have previously studied the effect of sleep on the ultrastructure of glial cells, astrocytes and oligoes. This might be a separate study, but it would be of interest to discuss the role of Bergmann glial cells in synaptic plasticity. One major difference is that Bergmann glia express AMPA receptors, unlike cortical astrocytes and these are important for the proximity of astrocytic processes to synapses.

Reviewer #1 (Public Review):

Zebrafish have become a well-established model organism for studies of damage and repair in hair cell sensory organs due to organ accessibility and robust mechanisms for repair and regeneration. While lateral line organs in zebrafish are widely used to study hair cell physiology, zebrafish also contain hair cell organs in their inner ears that have the potential to be more comparable to mammalian counterparts. Using single-cell RNA seq in embryonic through adult stages, this study found that hair cells and supporting cells of the zebrafish inner ear are distinct from those of the lateral line. Additionally, they identified distinct cellular subtypes within the maculae and cristae of the ear.

This work provides some novel findings, including identifying distinct markers in the macula and crista hair cells and supporting cells as well as detecting a domain in the zebrafish utricle that coincides with features of the striolar region in mouse utricle. However, while much of the focus of in situ expression analysis was the spatial separation of calcium-binding proteins capb1b and capb2b in the maculae, it was not clear how Capb1 & 2 expression corresponds to striolar and extrastriolar regions in the mammalian utricle, where Capb2 is expressed throughout. In addition, the authors assumed the zebrafish utricle and saccule perform similar functions (i.e. hearing) and contain hair cells with similar frequency tuning. It would improve this study to consider the unique functions and frequency sensitivities of the utricle and the saccule, given that different expression of voltage-gated channels may give insight into the specific physiology of these two sensory organs.

Reviewer #2 (Public Review):

This is an interesting study with a primary value in generating new transcriptional data sets for zebrafish hair cells and non-sensory cells in the inner ear. The data will, no doubt, be useful for future studies of hair cell function, development, and regeneration. The data also reveal transcriptional differences between similar cell types in different structures and transcriptional similarities between fish and mammalian cell types within analogous structures. Overall the strength of evidence in support of the results is strong.

Reviewer #3 (Public Review):

The authors describe the use of single-cell RNA sequencing (scRNA-seq) of the zebrafish inner ear at various stages ranging from embryos to adults and they characterize 3 major cell types: supporting cells, progenitor cells, and hair cells. While scRNA-seq experiments have been performed on adult inner ear tissues and the lateral line previously, a detailed characterization of the cellular subtypes in the inner ear at the embryonic through adult stages has not been accomplished before at the transcriptomic and spatial levels and is an important contribution to the field.

In the manuscript, the authors describe the transcriptomic profiles of the inner ear at single-cell resolution followed by spatial validation. In agreement with previously published research, they identify 3 major cell types in the inner ear and use advanced bioinformatic analysis to identify distinct support and hair cell subtypes that reside in the hearing vs balance organs. They elucidate the transcriptomic differences between support and hair cell types that reside in the larval lateral line vs inner ear and demonstrate that these systems are different. Finally, they provide the groundwork for comparisons between zebrafish and mouse transcriptomic profiles and show conservation in the hair cell population. Most importantly, the authors validate their transcriptomic sequencing findings at the single-cell level with spatial information in the inner ear tissues using in situ hybridization assays.

The work performed takes several stages of inner ear development as well as sub-organ dissections coupled to scRNA-seq to carefully identify key cell types and map them to their matching mouse counterparts (when they exist). This work represents the groundwork for many comparative studies across species at the molecular level.

Reviewer #1 (Public Review):

The authors Rem et al., examine the mechanism of action of APP, a protein implicated in Alzheimer's disease pathology, on GABAB receptor function. It has been reported earlier that soluble APP (sAPP) binds to the Sushi domain 1 of the GABAB1a subunit. In the current manuscript, authors examine this issue in detail and report that sAPP or APP17 interacts with GABABR with nano Molar affinity. However, binding of APP to GABAB receptor does not influence any of the canonical effects such as receptor function, K+ channel currents, spontaneous release of glutamate, or EPSC in vivo. The experimental evidence provided to support the conclusions is thorough and statistically sound. The range of techniques used to address each of the aims has been carefully curated to draw meaningful conclusions.

The authors use HEK293T heterologous cell line to confirm the affinity of APP17 for the receptor, ligand displacement, and receptor activation. They also use this method to study PKA activation downstream of the GPCR. They use slice electrophysiology to measure changes in glutamatergic transmission EPSC and then in vivo 2-photon microscopy to measure functional changes in vivo.

The work is significant for the field of Alzheimer's and also GABAB receptor biology, as it has been assumed for sAPP acts via GABAB receptors to influence neurotransmission in the brain. The results presented here open up the question yet again, what is the physiological function of sAPP in the brain?

The manuscript is clearly written and easy to follow. The main criticism would be that the manuscript fails to identify the mechanism downstream of APP17 interaction with GB1a SD1.

Reviewer #2 (Public Review):

Amyloid-β precursor protein (APP) regulates synaptic activity in part through the release of secreted APP (sAPP) acting at cell-surface receptors. In 2019 two articles (Dinamarca et al, 2019; Rice et al, 2019) were published showing that sAPP binds with high affinity with GABAB receptors. These receptors regulate neuronal excitability and synaptic release. In the Rice et al. paper, it was concluded that sAPP plays a physiological role by regulating GABAB receptors by modulating synaptic transmission, consistent with the direct activation of these receptors by sAPP. This article has received major attention in the field of Alzheimer's disease and synaptic biology.

The present work was designed to fully explore the functional consequences of sAPP binding to GABAB receptors, in particular, because it was unclear how a conformational change in SD1 - the region of GABAB receptors that binds sAPP - potentially induced by sAPP could increase GBR activity.<br /> The work does confirm that the peptide APP17 which derives from sAPP binds with nanomolar affinity with GABAB receptors. The authors use a diverse range of techniques, ranging from biophysical assays in recombinantly expressed receptors to electrophysiology and live imaging in cultured neurons, slices, and in vivo neuronal activity. In none of these assays, could the authors demonstrate any functional effect of sAPP mediated by an action on GABAB receptors.

This work from a team that has exquisite knowledge of the different aspects of GABAB receptors represents an important and very convincing clarification for the field, and it would therefore be very useful if this information is rapidly available.

Reviewer #1 (Public Review):

This paper by Hubmann et al investigates the role of a large ECM protein in lymphangiogenesis using the zebrafish model to test genetic interactions and with state-of-the-art reporter readouts to evaluate expression patterns and vascular behaviors. The experiments examine the effects of svep1 deletion on facial lymphatics and produce the surprising result that svep1 and vegfc appear to be required in non-overlapping areas of this lymphatic vascular bed. They then use these tools and mutants for tie1 or tie2 to examine genetic interactions in this area and in parachordal lymphangioblast migration in the trunk, showing that tie1 but not tie2 shows epistasis with svep1. The data are overall rigorous and quantified to show the statistical significance of the stated results. The work provides important insights into a complex signaling pathway that is widely utilized in vascular development. The evidence is convincing in supporting the findings and is predicted to be of interest to vascular biologists and others interested in Tie/Tek signaling such as cancer biologists.

Reviewer #2 (Public Review):

In their current manuscript Hussmann et al., present a very detailed phenotypic analysis of the role of svep1 in lymphatic development in zebrafish. They show that svep1 is essential for the development of particular aspects of facial lymphatics (the FCLV and BLECs) in a fashion complementary to VEGF-C. Furthermore, they show that the loss of tie1 phenocopies svep1 mutants not only with respect to lymphatic defects but also in blood vessels (DLAV).

Overall, the manuscript is clearly written, the experiments are carefully executed, and the quality of data is very high and support the author's main conclusions: 1) that Svep1 and Tie1 genetically interact during lymphatic and blood vessel development and 2) that this function is independent and complementary to VEGF-C. 3) The authors confirm and extend on a previous study (Jiang et al. 2020) showing that tie2 (tek) has no overt role in vascular development in zebrafish in blood as well as in lymphatic vessels.

The strength of the paper lies in the careful combination and comparison of different mutant alleles and the use of state-of-the-art imaging. These analyses show that Svep1 and Tie1 interact at the genetic level. In vivo cell tracking experiments show that Tie1 and Svep1 regulate particular aspects of lymphatic cell migration.

An obvious remaining question concerns the epistatic relationship and the molecular mechanism of Tie1/Svep1 interaction. The authors suggest a non-autonomous requirement of Svep1 in the ECM regulating the availability of Tie1 ligands (Ang-1/-2?) in LECs. Since bona fide ligands for Tie1 have not yet been identified in zebrafish further studies will be needed to test this model.

Reviewer #3 (Public Review):

This manuscript describes a Vegfc-independent mechanism of lymphatic vessel formation that is controlled by Svep1 and an orphan endothelial receptor tyrosine kinase Tie1. Based on similarities in the phenotype of svep1 and tie1 mutant zebrafish in the head and trunk vasculature, as well as genetic interaction between the two during parachordal lymphangioblast migration, the authors propose that svep1 is a component of the tie1 signaling pathway. Specifically, svep1 and tie1 mutants show a unique phenotype with the absence of facial collecting lymphatic vessel (that forms in Vegfc mutants) while other facial vessels (that are dependent on Vegfc/Vegfr3 signaling) were only partially affected. Svep1 and tie1 mutants also show similar defects in the formation of brain LECs, the number and migration of parachordal lymphangioblasts from the horizontal myoseptum, and DLAV formation. In contrast, tie2, which is the major angiopoietin receptor in mammals, was found to be dispensable for vascular development in zebrafish.

The presented experiments are performed well and the data are conclusive. The novel findings are the identification of a role of tie1 in zebrafish lymphatic development, and svep1 as a component of the tie1 signaling pathway. The latter raises the possibility that svep1 regulates the activity of Angiopoietin or even acts as a ligand for tie1. However, the conclusion on Svep1 and Tie1 being in the same pathway is based solely on the comparison of mutant phenotypes and genetic interaction studies. Any biochemical data on how svep1 regulates tie1 signaling would greatly strengthen this conclusion.

Reviewer #1 (Public Review):

Gormley et al. conduct a comprehensive Mendelian randomisation (MR) analysis to study the causal effect of obesity and metabolic disorder on oral and oropharyngeal cancer. This work follows on from observational studies that found evidence of associations between these variables and continued on from a previous MR study that focused on the effect of circulating lipid traits on these cancer outcomes. In this study, the authors found limited evidence for an effect of obesity-related traits on either cancer type, contradicting the previous observational studies and thus implying that the latter may have been affected by confounding. Sensitivity analyses that adjusted for potential pleiotropy and outlying instrumental variants agreed with the main study findings. Risk factors including smoking, risk-taking (a proxy for sexual behavior), and alcohol consumption were also stratified for, and only evidence was found for smoking as a mediator of the observed effect.

Strengths:<br /> Strengths of this study include the thorough MR analysis conducted, with all required sensitivity analyses and checks conducted and summarised appropriately, including the use of recent MR methods such as SIMEX where required. For the disease outcomes, oral and oropharyngeal cancer, the authors used summary statistics from the single largest trans-ethnic published meta-analysis GWAS available. Known risk factors were stratified for in sensitivity analyses to follow up any results of interest. Where measures of risk factors were not directly available (e.g. sexual behaviour), genetically-correlated proxy measures were used.

Weaknesses:<br /> While this study has several strengths, there are also a number of limitations present, many of which the authors outline in the paper. In particular, the sample size of the outcome GWAS used was rather small which may hinder the power. Also, some of the cancer subtypes of interest that had previously been investigated in observational studies were not available as GWAS, and so could not be included in this study. The authors only used European or trans-ancestry GWAS, as these are the only ancestries presently available, but a future investigation into other ethnicities is warranted. They also used risk-taking (self-defined via questionnaire) as a proxy for sexual behaviour, which despite having a strong genetic correlation, may not be the best substitute.

Reviewer #2 (Public Review):

The authors used Mendelian randomisation to study the relationships between metabolic traits and oral/oropharyngeal/head and neck cancers. This study was conducted as the relationships between these traits and cancers are unclear based on observational data. Evidence for relationships between these traits and cancers is inconclusive, which is a relevant finding in the context of previous observational data.

Strengths include using large studies to develop the instrumental variables used in MR and examining multiple metabolic traits. Weaknesses include relatively low power to detect associations and a lack of discussion around any possible pleiotropy of SNPs associated with any of the metabolic traits. Based on these strengths and weaknesses, it is unclear whether the authors achieved their goal and whether the results support their conclusions.

This work is relevant to researchers interested in oral cancers and their etiology. Several issues would need to be addressed to make the evidence more reliable.

Reviewer #1 (Public Review):

This is an elegant article where the authors define valuable criteria to identify and classify high-confidence miRNA in fungi. The data supporting the conclusions are solid, and the results are essential to unify the annotation criteria in these organisms. Interestingly, the paper shows that miRNAs in fungi look and position within the genome more similarly to their animal counterpart but may act like plant miRNAs.

The conclusions of this paper are well supported by data, but some aspects need to be clarified and extended.

Reviewer #2 (Public Review):

The work systematically reassesses fungal mi/miRNA-like characteristics and annotation confidence and identifies that many of the loci fail to meet the key points of the methods developed for animal or plant miRNAs. Therefore the authors establish a set of criteria suitable for the annotation of fungal miRNAs and provide a centralized annotation of identified mi/milRNA hairpin RNAs in fungi based on their established rules.

Here are some comments and suggestions for the manuscript to be improved:<br /> 1. The title mentions "ancestral links", however, the main context of this paper does not include the evolution of fungal mi/milRNAs or show the origins of conserved mi/milRNAs in fungi. The authors are suggested to consider a more appropriate title for this work.<br /> 2. The work proposes a fungal mi/milRNAs hairpin precursor recovery pipeline with three minimal criteria to annotate fungal mi/milRNA loci, which allows nearly half of the loci to pass these rules. To highlight the innovation of this annotation, it is strongly suggested that the authors compare their established pipeline and criteria for fungi with those used in animal or plant miRNAs in detail, and emphasize the advantages of the established pipeline. A figure showing the established pipeline and detailed parameters is needed.<br /> 3. The established "standard rules" for fungal mi/milRNA annotation still require more evaluation. It would be better if there is experimental validation to improve confidence.

Reviewer #3 (Public Review):

The authors provide a centralized annotation of miRNA and miRNA-like hairpins in fungi. They aim to develop a standardized pipeline and criteria for miRNA annotation in fungi focusing only on sRNAs derived from hairpin structures, seeking to identify essential characteristics of fungal miRNA and miRNA-like.

Overall this paper will be of interest to readers trying to understand the characteristics and functions of miRNA and miRNA-like hairpins in fungi. The conclusions of this paper are mostly well supported by data, but some aspects of the methodology need to be clarified and extended. The absence of follow-up experiments somewhat limits the impact of this paper. Subsequent work should focus on searching and validating targets of miRNA in fungi. In particular, the strong mi/milRNAs candidates detected in their work.

Reviewer #1 (Public Review):

In this manuscript, the author characterizes the lattice of kinesin-decorated microtubule reconstituted from porcine tubulins in vitro and Xenopus egg extract using cryo-electron tomography and subtomogram averaging. Using the SSTA, they looked at the transition in the lattice of individual microtubules. The authors found that the lattice is not always uniform but contains transitions of different types of lattices. The finding is quite interesting and probably will lead to more investigation of the microtubule lattice inside the cells later on for this kind of lattice transition.

The manuscript is easy to read and well-organized. The supporting data is very well prepared.

Overall, it seems the conclusion of the author is justified. However, the manuscript appears to show a lack of data. Only 4 tomograms are done for the porcine microtubules. Increasing the data number would make the manuscript statistically convincing. In addition, having the same transition with the missing wedge orientation randomly from different subtomograms will allow a better average of transition without the missing wedge artifact.

Another thing that I found lacking is the mapping of the transition region/alignment in the raw data. However, it is not easy for me or the reader to understand how each segment is oriented relative to each other apart from the simplified seam diagrams in the figures, and also the orientation of the seam corresponding to the missing wedge in the average. With these improvements, I think the conclusion of the manuscript will be better justified.

Reviewer #2 (Public Review):

Differences in protofilament and subunit helical-start numbers for in vitro polymerized and cellular microtubules have previously been well characterized. In this work, Guyomar et al. analyze the fine organization of tubulin dimers within the microtubule lattice using cryo-electron tomography and subtomogram averaging. Microtubules were assembled in vitro or within Xenopus egg cytoplasmic extracts and plunge frozen after addition of a kinesin motor domain to mark the position of tubulin dimers. By generating subtomogram averages of consecutive sections of each microtubule and manually annotating their lattice geometry, the authors quantified changes in lattice arrangement in individual microtubules. They found in vitro polymerized microtubules often contained multiple seams and lattice-type changes. In contrast, microtubules polymerized in the cytoplasmic extract more frequently contained a single seam and fewer lattice-type transitions.

Overall, their segmented subtomogram averaging approach is appropriately used to identify regions of lattice-type transition and quantify their abundance. This study provides new data on how often small holes in the lattice occur and suggests that regulators of microtubule growth in cells also control lateral tubulin interactions. However, not all of the claims are well supported by their data and the presentation of their main conclusions could be improved.

Reviewer #3 (Public Review):

Protofilament number changes have been observed in in vitro assembled microtubules. This study by Guyomar and colleagues uses cryo-ET and subtomogram averaging to investigate the structural plasticity of microtubules assembled in vitro from purified porcine brain tubulin at high concentrations and from Xenopus egg extracts in which polymerization was initiated either by addition of DMSO or by adding a constitutively active Ran. They show that the microtubule lattice is plastic with frequent protofilament changes and contains multiple seams. A model is proposed for microtubule polymerization whereby these lattice discontinuities/defects are introduced due to the addition of tubulin dimers through lateral contacts between alpha and beta tubulin, thus creating gaps in the lattice and shifting the seam. The study clearly shows quantitatively the lattice changes in two separate conditions of assembling microtubules. The high frequency of defects they observe under their microtubule assembly conditions is much higher than what has been observed in vivo in intact cells. Their observations are clear and supported by the data, but it is not at all clear how generalizable they are and whether the defect frequencies they see are not a result of the assembly conditions, dilutions used and presence of kinesin with which the lattice is decorated. The study definitely has implications for mechanistic studies of microtubules in vitro and raises the question of how these defects vary for protocols from different labs and between different tubulin preparations.