Reviewer #1 (Public Review):

This group previously demonstrated that trisomy 21 causes an increase in PCNT levels, and this increase leads to pericentrosomal crowding and inhibition of ciliogenesis in fibroblasts. The authors here use trisomy and tetrasomy 21 retinal pigment epithelium cells generated by microcell-mediated chromosome transfer (MMCT) and previously generated mouse models of human trisomy 21. The well-quantified data and well-reasoned paper compellingly demonstrate that modestly increased PCNT levels can attenuate ciliogenesis and may result in trisomy 21-associated phenotypes such as cerebellar growth defects.

Reviewer #2 (Public Review):

This manuscript builds on previous work from the Pearson lab showing that one aspect of the trisomy 21 phenotype could be caused by an increase in the amount of pericentrin (PCNT), a component of the centrosome. The earlier work showed that the increase in PCNT is sufficient to reduce the frequency of ciliation in trisomy 21 cells and that the increased PCNT is often in the form of protein aggregates along microtubules proximal to the centrosome (in a preprint). Here they use several models with 3 or 4 copies of human chromosome 21 or the mouse equivalent to examine the defect in cilium formation at the level of specific proteins and the signaling function of the cilium. This is a substantial contribution that furthers the evidence for the authors' favored model of excess PCNT causing some form of pericentrosomal crowding that hinders the ability of other molecules, complexes, and/or vesicles to get to the right place at the right time. The work makes excellent use of cell lines and mice previously generated for the study of trisomy 21, making for well-controlled experiments in a situation where this is particularly important (only 1.5 x increased expression). One does wish that it were possible to do the PCNT depletion in more of the experiments than the single one shown, but that is understandable given the amount of work required and the uncertainty associated with RNAi depletion.

Reviewer #3 (Public Review):

The PCNT gene is found on human chromosome 21, and the same group previously showed that its increased expression is associated with reduced trafficking to the centrosome and reduced cilia frequency, which suggests a possible connection between cilia and ciliary trafficking, SHH signaling, and Down syndrome phenotypes. Jewett et al build upon this prior work by closely examining the trafficking phenotypes in cellular models with different HSA21 ploidy, or its mouse equivalent, thereby increasing the copy number of PCNT (3 or 4 copies of HSA21). They show that most of the trafficking defects can be reversed through the knockdown of PCNT in the context of HSA21 polyploidy. They also begin to examine the in vivo consequences of these trafficking disruptions, using a mouse model (Dp10) that partially recapitulates trisomy 21, including an increased copy number of PCNT. While I think this work advances our understanding of the trafficking defects caused by increased PCNT and has significant implications for our understanding of the cellular basis of a major hereditary human disorder, some improvements can be made to strengthen the conclusions and improve readability.

Major points:

I'm a little confused by the authors' conclusion that the increased PCNT levels in T21 and Q21 result in delayed but not attenuated ciliogenesis. The data show lower percentages of ciliated cells at all time points analyzed (Fig 1E) by quite a large margin in both T21 and Q21. Do the frequencies of cilia in the T21 or Q21 cells ever reach the same level as D21, say after 48-72 hours? If not it seems like not simply a delay. A bit more clarity about this point is needed.

The in vivo analysis of the cerebellum was interesting and important but it felt a bit incomplete given that it was a tie between the cell biology and a specific DS-associated phenotype. For example, it is interesting that the EGL of the P4 Dp10 pups is thinner. Does this translate into noticeable defects in cerebellar morphology later? Is there a reduction in proliferation that follows the reduced cilia frequency? I think it would be possible to look at the proliferation and cerebellar morphology at some additional stages without becoming an overly burdensome set of experiments. At a minimum, are there defects in cerebellar morphology at P21 or in the adult mice? The authors allude to developmental delays in these animals - maybe that complicates the analysis? But additional exploration and/or discussion on this point would help the paper.

It was a bit unclear to me why specific cell lines were used to model trisomy 21 and why this changed part way through the paper. I understand the justification for making the Dp10 mice- to enable the in vivo analysis of the cerebellum, but some additional rationale for why the RPE cell line is initially used and then the switch back to mouse cells would improve readability.

Reviewer #1 (Public Review):

The authors describe a feeding system for killifish that allows high precision control of feeding amount and schedule on a per-tank basis. The system permits automation of this task using open-source and affordable components and software. Due to this emphasis, the system appears amenable to manufacture by individual research groups and the approach appears very scalable (although more detailed build, programming and assembly instructions and videos might be useful for groups with little experience with microcontrollers and manufacturing). An exciting aspect of the system is the possibility to modify the system for different purposes. For example, it might be possible to reduce the minimum feeding amount, thereby allowing more fine grained exploration of effects related to feeding shedule. I am very enthusiastic about the open-source "maker" aspects of this work.

The authors next explore two interesting applications of the system. First, they show that precise control of food allows automated investigation of lifespan extension under calorie restriction (CR) conditions. This is an important use case for a system of this type and showing that it is fit for this application is important.

Secondly, the authors show an exciting modification of the system that involves only addition of a simple red light LED. This modification allows use of the system in a associative learning / conditioning paradigm.

Finally, they show that there is an age-dependent decline in learning as evaluated by this conditioning paradigm. I am very enthusiastic about this additional function and, again, this example demonstrates the flexibly and open nature of the technology, suggesting that others can likely modify and expand the system to suite their own questions and applications. In summary, I am enthusiastic about the technology described and about the approach by which the system was developed.

However, at the current stage, the biological applications are essentially validation experiments - e.g. showing that CR can be implemented and that the system can be used for learning and memory experiments. Neither of these aspects is pursued beyond the basic validation experiments (showing that lifespan extension can be achieved and that there is age-dependent decline in associative learning).

Reviewer #2 (Public Review):

Here, McKay, et al. describe a new automated system to feed killifish, and use it to explore dietary restriction effects on killifish lifespan and to develop an associative learning assay, two important goals in the KF/longevity field.

Fig. 1-2- The first figures focus on the design and evaluation of the feeding system. It appears that the feeding system works well and achieves what the authors set out to do.<br /> Fig. 3 explains the DR and overfeeding setup, and effects on growth and reproduction; demonstrates that the automated feeding system does achieve DR.<br /> Fig. 4 explains the DR setup and results on male and female KF, highlighting the fact that DR only extends the lifespan of males. This sex-specific effect seems somewhat surprising, and warrants further follow-up studies.<br /> Fig. 5 describes the associative learning assay, which is based on the ability of the fish to sense a red light and learn that it is associated with feeding. It is great that the authors have been able to develop a learning assay, which will no doubt become an important tool in the killifish researcher's arsenal, but additional experiments are necessary to increase the general impact of the work.

Overall, while the results seem sound, the current version of the manuscript may be pitched to a small audience (killifish researchers) who will benefit from the development of this methodology. Perhaps the paper could be re-structured to focus less on the methodology and more on the results, fleshing out the associative learning results even more (are there mutants that extend the length of associative learning? Does it require conserved genes? etc). Further exploration of the sex-specific effects of DR on lifespan (why does this only affect males) would also raise the general interest of the work, but both the DR and associative learning aspects of the paper would need to be studied quite a bit more to move this beyond a methods paper.

Reviewer #3 (Public Review):

The work presented by McKay et al. details the development of a new wireless network-enabled automated feeder system with which diet amount and schedule can be controlled across individually housed killifish. The manuscript describes the characterization of the system and demonstrates the robustness, precision, and high fidelity in feeding control achieved due to modular design.

The technique in principle can be applied to hundreds of tanks and to other species that are reared in similar tank system racks.

Strengths:

- The authors provide a convincing account of the use of automated feeder systems for implementing experiments where diet is controlled precisely. The experimental design allows the authors to clearly demonstrate feeding schedules optimal for killifish growth, reproduction, and longevity. Their characterization and results will be highly valuable for a growing community of researchers who are beginning to use killifish in laboratory settings and can choose the regimen most suited to their research goals. The system presented in this study may also allow for better husbandry practices with the potential to mimic the ephemeral natural habitats of this species more closely in the laboratory.<br /> - The authors also conducted additional experiments comparing restricted food delivery schedules. The conclusion they reach that a time and quantity restricted feeding regimen increases the lifespan of males based on this experiment is well justified from the data presented. The differences between the sexes are interesting to note as the authors observed similar results with two different cohorts, though cohorts can differ in the median and maximum lifespan.

Weaknesses:

- The authors imply the value of automated feeders is in scaling to hundreds of individual animals/tanks. I agree with the author's assessment of this need in research labs, however, it is not easy to infer exactly how many automated feeders were operating simultaneously in this study. Estimates of the costs of building, and operating (maintenance, server use, and cloud computing costs) for conducting 1 experiment (2 conditions, 24 animals per condition) running over 100 days will be valuable for other researchers interested in adapting this resource. A clearer supplementary video 1 that demonstrates the entire feeder properly, in the home tank will also be valuable for the researchers interested in adapting the system.<br /> - The proof of concept experiment showing associative learning is extremely interesting but is quite difficult to assess, based on the detail provided in the results and the method. The rationale behind key considerations for behavioral measures, whether based on previous studies or, due to technical constraints are difficult to judge. This needs a better description. In particular, results mention a "pipeline", but this is obscure, in the methods section. Clearer definitions would also be needed to evaluate if an objective scoring system for was used in measures such as the"startle" response. In principle, as all trajectories are recorded, it should be possible to describe a range of acceleration/velocity changes that quantify most parameters such as startle, unless it was manually scored. As this will be a first, clarity on how "early" and "late" sessions were categorized; exact experimental design on the number of trails that made up a session; whether all animals went through same number of trials in Figure 5, etc. will improve the description and future adaptations of the experimental design.<br /> - One more cautionary note is in the interpretation that young individuals had significantly higher learning index scores than old individuals, as the size of the effects can't be estimated from the type of data provided and the analysis used. Given the fairly small sample size for animals used in learning index calculations (< 15), and as the authors demonstrate in diet restriction experiments there can be cohort-dependent differences as well, I would caution against such an interpretation. The p values reported in Suppl. Figure 4E especially brings home the need to move away from dichotomous thinking of yes/no based on a threshold, without taking into account effect sizes. Please refer to this recent post in eNeuro on the inherent issues with such interpretations, and methods to overcome them (https://doi.org/10.1523/ENEURO.0091-21.2021). The deficiency in "old" may not be as large, and it would be important to interpret this appropriately. Other normalization issues, rather than learning could account for small differences between the young and the old. For instance, a small latency in the average velocity and/or other locomotion kinematics differences between fish categorized as old vs. young could result in the criterion of "3 seconds before the food drops" to meet the "threshold of learning" being unmet. The data available in the paper at present can't be used to evaluate such a point.

Reviewer #1 (Public Review):

Leukemic cells are known to remodel bone marrow niche to promote their expansion and to suppress normal hematopoiesis. However, molecular mechanisms remain largely unknown. In this manuscript, authors developed new experimental models in mice to address this issue, using mouse BCR-ABL-driven ALL cells marked with YFP, or DOX-inducible MLL-AF9 AML cells. After transplantation of either of these cells, authors discovered suppression of host hematopoiesis. Using these systems, authors tested their hypothesis on lymphotoxin receptor-mediated interaction of the leukemic cells and stroma cells.

The main conclusions here are: 1) lymphotoxin signaling through its receptor mediates IL7 down regulation and alters gene expression related to inflammation etc. in stroma cells, 2) IL7 down regulation leads to reduction in B lymphoid cells but not myeloid cells, 3) lymphotoxin expression in leukemic cells is induced by DNA damage response, and 4) CXCR4, which is known to be induced in B cells in response to stroma cells, collaborates with DNA damage in induction of lymphotoxin in leukemic cells. Taken together, authors suggest that a positive feedback loop of leukemic cells and stroma cells for leukemic cell proliferation and normal hematopoietic suppression, involving lymphotoxin and CXCR4 in leukemic cells and lymphotoxin receptor in stroma cells. Generally, these conclusions and the model of the positive feedback regulation are supported, to a reasonable level, by the experimental results provided in the manuscript. However, some of the results show small effects of manipulations, leaving the pathological significance of the feedback model as a future issue.

Reviewer #2 (Public Review):

In the current manuscript, Feng et al. investigate the mechanisms used by acute leukemia to get an advantage for the access to the hematopoietic niches at the expense of normal hematopoietic cells. They propose that B-ALLs hijack the niche by inducing the downmodulation of IL7 and CXCL12 by stimulating LepR+ MSCs through LTab/LTbR signaling. In order to prove the importance of LTab expression in B-ALL growth, they block LTab/LTbR signaling either through ligand/receptor inactivation or by using a LTbR-Ig decoy. They also show that CXCL12 and the DNA damage response induce LTab expression by B-ALL. They finally propose that similar mechanisms also favor the growth of acute myeloid leukemia.

Although the proposed mechanism is of particular interest, further experiments and controls are needed to strongly support the conclusions.

1/ Globally, statistics have to be revised. The authors have to include a "statistical analysis" section in the Material and Methods to explain how they proceeded and specify for each panel in the figure legend which tests they used according to the general rules of statistics.

2/ The setup of each experiment is confusing and needs to be detailed. Cell numbers are not coherent from one experiment to the other. As an example, there are discrepancies between Fig1 and Fig2. Based on the setup of the experiment in Fig.2 (Injection of B-ALL to mice followed by 2 injections of treatment every 5 days), mice have probably been sacrificed 12-14 days post leukemic cell injection. However, according to Fig.1, B cells and erythroid cells at this time point should be decreased >10 times while they are only decreased 2-4 times in Fig.2. This is also the case in Fig.4B-J or Fig.5D with even a lower decrease in B cells and erythroid cells despite a high number of leukemic cells. Please explain and give the end point for each experiment in each figure (main and supplemental).

3/ To formally prove that the observed effect is really due to LTab/LTbR signaling, the authors must perform further control experiments. LTbR signaling is better known for its positive role on lymphocyte migration. They cannot rule out by blocking LTbR signaling, that they inhibit homing of leukemic cells into the bone marrow through a systemic/peripheral effect, more than through an impaired crosstalk with BM LepR+ cells. They must confirm for inhibited/deficient LTbR signaling conditions, as compared to control, that similar B-ALL numbers home to the BM parenchyma at an early time point after injection. Furthermore, they cannot exclude that the effect on the expression of IL7 (and other genes), and consequently the effect on B cell numbers, is not simply due to the tumor burden. Indeed, B-ALL numbers/frequencies are different between control and inhibited/deficient signaling conditions at the time of analysis. The analyses should thus be performed at similar low and high tumor burden in the BM for both control and inhibited/deficient LTbR signaling conditions.

4/ LT/LTbR signaling is particularly known for its capacity to stimulate Cxcl12 expression. How do the authors explain that they see the opposite?

5/ The authors show that CXCL12 stimulates LTa expression in their cell line. They then propose that CXCR4 signaling in leukemic cells potentiates ALL lethality by showing that a CXCR4 antagonist reverses the decrease in IL7 and improves survival of the mice. This experiment is difficult to interpret. CXCL12 has been shown to be important for migration/retention of B-ALL in the BM and the decreased tumor burden is probably linked to a decreased migration more than an impaired crosstalk with LepR+ cells (see also point 3). If CXCL12 increases LTab expression, CXCR4 blockade should do the opposite. This result should be presented. The contradiction is that if B-ALLs induce a decrease in CXCL12 in the BM (in addition to IL7) and that CXCL12 regulates LTab levels, leukemic cells should be exhausted. Similarly, IL7 has been previously shown to stimulate LTab expression and B-ALL cells express the IL7R. Again, a decrease in IL7 should be unfavorable to B-ALL. How do they explain these discrepancies?

6/ In Supp 4A, RAG-/- mice are blocked at the pro-B cell stage and do not have pre-B cells. Please compare LTa and LTb expression by Artemis deficient pre-B cell to wt pre-B cells. In this experiment, the authors show that similarly to B-ALL artemis-/- pre-leukemic pre-B cells express high levels of LTab and induce IL7 downmodulation. Using mice deficient for LTbR in LepR+ cells, they show that IL7 expression is increased. However, in opposition to leukemic cells (see Figure 4F), pre-leukemic cells are increased in absence of LTab/LTbR signaling. Please explain this discrepancy. The authors use only one B-ALL model cell line for their demonstration (BCR-ABL expressing B-ALL). Another model should be used to confirm whether LTab/LTbR signaling does favor leukemic/pre-leukemic B cell growth.

7/ Pre-B cells are composed of large pre-B cells (pre-BCR+) and small pre-B cells (pre-BCR-). BCR-ABL B-ALL cells express the pre-BCR. What is the level of expression of LTa and LTb by each of these 2 subsets as compared to BCR-ABL B-ALL?

Reviewer #2 (Public Review):

This modeling paper looks at how single spikes in the cortex are able to evoke patterns of sequential neural response in the surrounding neural network, an effect observed in the visual cortex of turtles, rodents, and the middle temporal cortex of humans, and possibly generalizable across many other species and brain areas. The results are anchored by population recordings from the turtle cortex, recapitulating those data and exploring how single spikes might be able to have such an outsized effect on broad-scale neural activity. The authors aim to show which kinds of network connectivity support this kind of response.

The results reveal that sparse, but strong connections in a neural network are the necessary ingredient for the reliable triggering of network sequences by single spikes. Dense, but weaker networks can give rise to different sequences when triggered. One of the most intriguing results of the paper is the interaction of sequences triggered by different single spikes that are part of a strong, sparse sub-network. These concurrent sequences appear to be separable and potentially supported a wide repertoire of response states to very targeted and combinatorially expressive inputs.

The work is careful and well-executed and the work will be of interest to systems and computational neuroscientists. In particular, the work speaks to how to reliably trigger a wide array of broad-scale population sequence patterns. This could be important for signaling salient, complex external stimuli, especially in a dynamic environment. The work will also be of interest to the machine learning community working on recurrent neural networks and their computational capacity.

Reviewer #1 (Public Review):

There were two parts to this paper. The first was to build a network model with parameters carefully adjusted to match those seen in the turtle cortex. The second was to simulate the circuit, and show that it could produce reasonably repeatable patterns of activity in response to a single, externally added, spike.

As a model of the turtle cortex, the paper was pretty convincing. And the explanation for the repeatable patterns of activity - a small number of very strong connections and a very low background firing rate - seemed eminently reasonable. This paper should serve as a very good starting point for understanding computing in the turtle cortex.

However, average firing rates in the turtle are extremely low - 0.1 Hz, at least in these simulations. Their model is unlikely, therefore, to account for activity in the mammalian cortex, which exhibits a much higher background firing rate, and for which there's not a lot of evidence for the extremely strong connections seen in the turtle.

Reviewer #3 (Public Review):

Riquelme et al. develop a spiking neural network model based on experimental measurements from ex vivo turtle visual cortex (neuronal parameters, connectivity profiles, synaptic strength distributions). Within the constraints given, the connectivity is random. The analyses in the manuscript are based on multiple instantiations (300) of the network and multiple simulations of each. The principle finding is that, if a randomly selected excitatory neuron is induced to emit an action potential, a reliable sequence of spikes follows (in more than 90% of cases). They then examine the role of connectivity in this phenomenon, including the frequency of specific motifs in the spike cascade and the comparative role of strong and weak connections. In particular, the authors show that rare strong connections are vital for producing (long) reliable sequences. The authors then examine how the sequences can be broken down into sub-sequences that may or may not occur for a given trigger. They show that the sub-sequences are characterized by strong internal connections (compared to those between sub-sequences). Moreover, they show that the spike sequence can be routed by exciting or depressing the 'gate' neurons (i.e. those at the beginning of a particular sub-sequence) raising the intriguing possibility of context-driven routing of activity. Finally, the authors demonstrate that their model has interesting combinatorial properties, as the results of triggering two sequences at once cannot be accounted for in a linear fashion. All in all, this is a solid piece of work with well-thought-through analyses which is an interesting contribution to the fundamental question of how the brain manages reliable computation in a noisy world.

Strengths

"Ensemble approach" I appreciated the approach to generate many networks from the same distributions rather than (as is often the case) basing all their conclusions on one instantiation. In general, the statistical rigour is high.

Well-chosen analyses to tease apart the relationships between structure and dynamics.

Figures (for the most part) clearly support the conclusions of the paper.

Weaknesses

The spontaneous activity of the network is extremely low, with [0.02 0.09] spks/s considered as a high activity range. Granted, this is based on ex vivo measurements. However, if this phenomenon is to be considered computationally relevant, as the authors claim, the paper should have examined the reliability of propagation and routing with in vivo activity levels.

The above weakness is a special case of the issue that the limits of applicability/robustness of results to model assumptions have not been well established. In particular, it is not clear how strong the strongest weights must be whilst still enabling long sequences, and what is the dependence of the results on the parameters of the distance-dependent connectivity.

The figures are too densely packed and many of the elements are too small or too fine to be distinguished, especially if your eyesight is not the greatest. Although many people read online, where zooming is possible, the aim should still be that all elements of the figure can be perceived by a person over 45 who has printed the paper on regular A4 paper.

Reviewer #1 (Public Review):

Polarization in cells and organs is often dictated by opposing polarity domains. In grass subsidiary cells, several proteins (including PAN1) were previously found to polarize in a discrete patch prior to asymmetric division. Zhang et al. identify POLAR via transcriptional profiling of Bdmute, a mutant that lacks subsidiary cells The authors effectively show that Bdpolar mutants have defective subsidiary cells. A distinctive and exciting localization pattern of POLAR is demonstrated, which is opposite to PAN1. This localization pattern is further contextualized by showing that PAN1 and MUTE are both required for POLAR's distinctive localization; however, PAN1 polarization is unaffected in both polar and mute. The integration of MUTE, POLAR, and PAN1 is particularly important as it integrates how polarity proteins and fate factors interact with each other.

Bdpolar mutants have defects in subsidiary cells that lead to defects in stomatal function. The authors carefully and quantitatively compare the phenotypes of pan1 and polar and conclude distinct roles for the two proteins based on differences in phenotypes including nuclear polarization, division site specification, and repeated rounds of cell division. The discovery and localization of POLAR are very exciting, but the comparison between single alleles of pan1 and polar and the extrapolation requires scrutiny. In particular, the data on division site specification in pan1 seem inconsistent with the % defective subsidiary cells and nuclear migration defects. However, these are addressable and given the exciting nature of the localization and pathway determination, the paper's impact stands.

Reviewer #2 (Public Review):

Grasses develop morphologically unique stomata for efficient gas exchange. A key feature of stomata is the subsidiary cell (SC), which laterally flanks the guard cell (GC). Although it has been shown that the lateral SC contributes to rapid stomatal opening and closing, little is known about how the SC is generated from the subsidiary mother cell (SMC) and how the SMC acquires its intracellular polarity. The authors identified BdPOLAR as a polarity factor that forms a polarity domain in the SMC in a BdPAN1-dependent manner. They concluded that BdPAN1 and BdPOLAR exhibit mutually exclusive localization patterns within SMCs and that formative SC division requires both. Further mutant analysis showed that BdPAN1 and BdPOLAR act in SMC nuclear migration and the proper placement of the cortical division site marker BdTANGLED1, respectively. This study reveals a unique developmental process of grass stomata, where two opposing polarity factors form domains in the SMC and ensure asymmetric cell division and SC generation.

The findings of this study, if further validated, are novel and interesting. However, I feel that the data presented in the current manuscript do not fully support some crucial conclusions. The lack of dual-color images is the weakest point of this study. If it is technically impossible to add them, alternative analyses are needed to validate the main conclusions.

1. Is BdPOLAR-mVenus functional? Although the authors interpret that weak BdPOLAR-mVenus expression partially rescued the bdpolar mutant phenotype in Fig. S4D, the localization pattern visualized by BdPOLAR-mVenus may not be completely reliable with this partial rescue activity.<br /> 2. Regardless of the functionality of the tagged protein, the authors need to provide more information on their localization. For example, is there a difference in polarity pattern depending on expression level? Does overexpressed BdPOLAR-mVenus invade the BdPAN1 zone? In such cases, might the loss of BdPOLAR polarity in the bdpan1 mutant be a side effect of overexpression, not PAN1 exclusion? Does BdPOLAR expression (no tag) show a dose-dependent effect, similar to the mVenus-tagged protein?<br /> 3. A major conclusion of this study was that the polarity domains of BdPOLAR and BdPAN1 are mutually exclusive. However, not all the cells in the figures were consistent with this statement. For example, the BdPOLAR signals at the GMC/SMC interphase appear to match BdPAN1 localization (compare 0:03 s in Video 1 and 0:20 s in Video 2 [top cell]). The 3D rendered image in Fig. 2F shows that BdPOLAR is excluded near the GMC on the front side of the SMC, where BdPAN1 is not localized. Some cells did not exhibit polarization (Fig. 3A, bottom left; Fig. 3E, bottom left). The most convincing data are the dual-color images of these two proteins. Otherwise, a sophisticated image analysis is required to support this conclusion.<br /> 4. Another central conclusion was that BdPOLAR was excluded at the future SC division site, marked with BdTANGLED1. However, these data are also not very convincing, as such specific exclusion cannot be seen in some figure panels (e.g., Fig. 3A, bottom left; Fig. 3E, all three cells on the left). If dual-color imaging is not feasible, a quantitative image analysis is needed to support this conclusion.<br /> 5. I could not find detailed imaging conditions and data processing methods. Are Figs. 2B and 2E max-projection or single-plane images? If they are single-plane images, which planes of the SMC are observed? In addition, how were Figs. 2C and 2F rendered? (e.g., number of images, distance intervals, processing procedures). This information is important for data interpretations.<br /> 6. [Minor point] The authors should clearly describe where BdPAN1 is expressed and localized. Is it expressed in the GMC and localized at the GMC/SMC interface? Alternatively, is it expressed and localized in the SMC?

Reviewer #3 (Public Review):

In this manuscript, the authors characterize the localization and function of two proteins, BdPOLAR and BdPAN1 in the asymmetric cell divisions required for stomatal patterning in Brachypodium distachyon (Bd). The authors clearly demonstrate these proteins are required for normal stomatal complex formation. Most excitingly, the authors reveal that these proteins occupy two opposing polar domains during stomatal formation, particularly the localization of BdPOLAR defines a novel polar domain that is dependent on BdPAN1 for its unique accumulation. The authors clearly link the functions of these proteins to cell division orientation and division potential and show an impact on stomatal function. The data presented here are clearly described and well documented and the figures are clear and well constructed. Their results support a broadly interesting hypothesis whereby polarization of cell fate-dependent and -independent factors pattern stomata in this grass. It will be very interesting to see how/if similar or other new polarity domains function in other developmental contexts in plants where control of cell division orientation is critical for cell fate and tissue function.

The authors' careful and elegant experiments clearly demonstrate a fascinating new avenue for exploration into plant cell polarity and cell division control. Their results will be of interest to scientists interested in development and cell biology across species, as well as those broadly interested in plant biology topics. Developmental patterning of the stomata in grasses is an elegant system to address important basic biological questions about the regulation of cellular asymmetries, cell division, and cell morphology. Additionally, the function of stomata is critical to the productivity and survival of plants, including in carbon intake (for photosynthesis). Understanding the developmental framework underlying pore formation provides insights into plant patterning processes and, importantly, provides a toolbox from which plant biologists can work to engineer improved crop plant performance in a rapidly changing climate.

Reviewer #1 (Public Review):

The research investigates the genetic basis for resistance to high CO2 levels in the human pathogenic fungus Cryptococcus neoformans. Screening collections of over 5,000 gene deletion strains revealed 96 with impaired growth, including a set of genes all related to the same RAM signaling pathway. Further genetic dissection was able compellingly to place where this pathway lies relative to upstream inputs and through the isolation of suppressor mutants as potential downstream targets of the pathway. Given the high levels of CO2 encountered by fungi in the human host, this work may provide new directions for the control of disseminated fungal disease.

The research presents both strengths and weaknesses.

Strengths include:

(1) One of the largest scale analyses of genes involved in growth under high CO2 concentrations in a fungus, revealing a set of just under 100 mutants with impaired growth.<br /> (2) Elegant genetic epistasis analysis to show where different components fit within a pathway of transmission of CO2 exposure. For example, over expression of one of the kinases, Cbk1, can overcome the CO2-sensitivity of mutations in the CDC24 or CNA1 genes (but not in the reciprocal overexpression direction).<br /> (3) Isolation of suppressor mutations in the cbk1 background, now able to grow at high CO2 levels, was able to lead to the identification of two genes. Follow up characterization, which included examining in vitro phenotypes, gene expression analysis, and impact during mouse infection was able to reveal that the two suppressors restore a subset of the phenotypes impacted by mutation of CBK1. Indeed, one conclusion from this careful work is that the reduced virulence of the cbk1 mutant is not due to its sensitivity to high levels of CO2, perhaps an unexpected finding given the original goals of the study towards linking CO2 sensitivity with decreased virulence.

Weaknesses include:

(1) What is the rationale for examining gene expression using the NanoString technology of 118 genes rather than a more genome-wide approach such as RNA-sequencing?<br /> (2) Without additional species examined, some of the conclusions about differences in impact between ascomycetes and basidiomycetes might instead reflect differences between species. For example, RAM mutants in other strains of C. neoformans do not exhibit so strong a temperature sensitive phenotype. Or to extend the comparison further, one might assume given the use of CO2 for Drosophila manipulations that the RAM pathway components in an insect would not be required for surviving high CO2.<br /> (3) Given the relative ease of generate progeny of this species, it would have been informative to explore if the suppressors of cbk1 also suppressed the loss of genes like CDC24, CNA1, etc, equivalent to the experiment performed of overexpression of CBK1 in those backgrounds.

Reviewer #2 (Public Review):

In the paper by Chadwick et al., the authors identify the molecular determinants of CO2 tolerance in the human fungal pathogen Cryptococcus neoformans. The authors have screened a collection of deletion mutants to identify the genes that are sensitive at 37oC (host temperature) and elevated CO2 levels. The authors identified that the genes responsible for CO2 sensitivity are involved in the pathways responsible for thermotolerance mechanisms such as Calcineurin, Ras1-Cdc24, cell wall integrity, and the Regulator of Ace2 and Morphogenesis (RAM) pathways. Moreover, they identified that the mutants of the RAM pathway effector kinase Cbk1 were most sensitive to elevated temperature and CO2 levels. This study uncovers the previously unknown role of the RAM pathway in CO2 tolerance. Transcriptome data indicates that the deletion of CBK1 results in an alteration in the expression of CO2-related genes. To identify the potential downstream targets of Cbk1, the authors performed a suppressor screen and obtained the spontaneous suppressor mutants that rescued the sensitivity of cbk1 mutants to elevated temperature and CO2. Through this screen, the authors identified two suppressor groups that showed a modest improvement in growth at 37{degree sign}C and in presence of CO2.<br /> Interestingly, from the suppressor screen, the authors identified a previously known interactor of Cbk1 which is SSD1, and an uncharacterized gene containing a putative Poly(A)-specific ribonuclease (PARN) domain named PSC1 (Partial Suppressor of cbk1Δ) which acts downstream of Cbk1. Deletion of these two genes in cbk1 null mutants rescued the sensitivity to elevated CO2 levels and temperature but did not fully rescue the ability to cause disease in mice.

This study highlights the underappreciated role of the host CO2 tolerance and its importance in the ability of a fungal pathogen to survive and cause disease in host conditions. The authors claim to gain insight into the genetic components associated with carbon dioxide tolerance. The experimental results including the data presented, and conclusions drawn do justice to this claim. Overall, it is a well-written manuscript. However, some sections need improvement in terms of clarity and experimental design.

• One major drawback of the study is the virulence assay performed to test the ability of cbk1 mutants to cause the disease in the mouse model. The cbk1 null mutants are thermosensitive in nature. Using these mutants, establishing the virulence attributes in mice would undermine the mutants' ability to infect mice as they won't be able to survive at the host body temperature.

• The rationale for choosing the genes to test further is not clear in two instances in the study. a) From a list of 96 genes, how do the authors infer the pathways involved? Was any pathway analysis performed that helped them in shortlisting the pathways that they subsequently tested? A GO term analysis of the list of genes identified through the genetic screen would be more helpful to get an overview of the pathways involved in CO2 tolerance. b) The authors do not clearly mention why they chose only four genes to test for the CO2 sensitivity out of 16 downregulated genes identified from the nano string analysis.

• It would be more useful to the readers if the authors could also include a thorough analysis of the presence of the putative PARN domain-containing protein across various fungal species rather than mentioning that it is only observed in C. neoformans and S. pombe. Also, the authors may want to discuss the known role(s) of SSD1, if any, in pathogenic ascomycetous yeasts so that the proposed functional divergence is supported further.

Reviewer #3 (Public Review):

In this work the authors identify genes and pathways important for CO2 and thermotolerance in Cryptococcus neoformans. They additionally rule out the contribution of the bicarbonate or cAMP-dependent activation of adenylyl cyclase to this pathway, which is important for CO2 sensing in other fungi, further solidifying the need to characterize CO2 sensing in basidiomycetes. The authors establish the importance of focusing on CO2 tolerance by testing the impact of CO2 on fluconazole susceptibility with varied pH, suggesting the ability of CO2 to sensitize cryptococcal cells to fluconazole. Furthermore, the authors compared the CO2 tolerance of clinical reference strains to environmental isolates. The characterization of the RAM pathway Cbk1 kinase illustrated the integration of multiple stress signaling pathways. By using a series of CBK1OE insertions in strains with deletions in other pathways, the ability of Cbk1 over-expression to rescue several strains from CO2 sensitivity was apparent. Additionally, NanoString expression analysis comparing cbk1∆ to H99 validated the author's screen of CO2-sensitive mutants as 16/57 downregulated genes were found in their screen, further confirming the interconnected nature of these pathways. The importance of the RAM pathway in maintaining CO2 and thermotolerance was also incredibly clear.

Perhaps most interestingly, the authors identify suppressor colonies with distinctive phenotypes that allowed for the characterization of downstream effectors of the RAM pathway. These suppressor colonies were found to have mutations in SSD1 and PSC1 which somewhat restore growth at 37oC with CO2 exposure. Further confirming the importance of the RAM pathway, the cbk1∆ strain had markedly attenuated virulence during infection. Interestingly, the generated suppressor strains had varying impacts on fungal infection in vivo. While the sup1 suppressor was completely cleared from the lungs during both intranasal and IV infection, the sup2 strain, containing mutations in SSD1, maintained a high fungal load in the lungs and was able to disseminate into host tissues during IV infection but not intranasal infection.

The authors make a strong case for the exploration of thermotolerance and CO2 tolerance as contributors to virulence. Through screening and characterization of RAM pathway kinase CBK1's ability to rescue other mutants from CO2 sensitivity, the overlapping contributions of several signaling pathways and the importance of this kinase were revealed. This work is important and will be valuable to the field. However, the cbk1∆ strain does show reduced melanization, urease secretion, and higher sensitivity to cell wall stressor Congo Red in SI Appendix, Figure S4. While the authors make a strong argument that these well-established virulence factors are not perfect predictors of virulence in vivo, the cbk1∆ strain is not an example of such a case as it does have defects in these important factors in addition to thermotolerance and CO2 tolerance. Not acknowledging the changes in these virulence factors in the cbk1∆ and their potential contribution to phenotypes observed is a weakness of the manuscript. Interestingly, the sup1 and sup2 strains also rescue these virulence factors compared to cbk1∆. Additionally, the assertion that "the observation that only sup2 can survive, amplify, and persist in animals stresses the importance of CO2 tolerance in cryptococcal pathogens" due to the sup2's slightly higher CO2 tolerance compared to sup1, could be better supported by the data. These suppressors did not restore transcript abundances of the differentially expressed genes to WT levels, suggesting post-transcriptional regulation. However, there may be differences in the ability of sup2 to resist stress better than sup1 especially given the known Ssd1 repression of transcript translation in S. cerevisiae. Finally, pH appears to impact the sup1 and sup2 strain's sensitivity to CO2 in SI Appendix Figure 4. This could be better explained and interrogated in the manuscript. Finally, this work includes a variety of genes in several signaling pathways. The paper would be greatly clarified by a graphical abstract indicating how CBK1 may be integrating these pathways or by indicating which genes belong to which pathways in the Figure 1 legend to make this figure easier to follow.

Reviewer #1 (Public Review):

This paper makes an important contribution to the current debate on whether the diversity of a microbial community has a positive or negative effect on its own diversity at a later time point. In my view, the main contribution is linking the diversity-begets-diversity patterns, already observed by the same authors and others, to genomic signatures of gene loss that would be expected from the Black Queen Hypothesis, establishing an eco-evolutionary link. In addition, they test this hypothesis at a more fine-grained scale (strain-level variation and SNP) and do so in human microbiome data, which adds relevance from the biomedical standpoint. The paper is a well-written and rigorous analysis using state-of-the-art methods, and the results suggest multiple new experiments and testable hypotheses (see below), which is a very valuable contribution.

That being said, I do have some concerns that I believe should be addressed. First of all, I am wondering whether gene loss could also occur because of environmental selection that is independent of other organisms or the diversity of the community. An alternative hypothesis to the Black Queen is that there might have been a migration of new species from outside and then loss of genes could have occurred because of the nature of the abiotic environment in the new host, without relationship to the community diversity. Telling the difference between these two hypotheses is hard and would require extensive additional experiments, which I don't think is necessary. But I do think the authors should acknowledge and discuss this alternative possibility and adjust the wording of their claims accordingly.

Another issue is that gene loss is happening in some of the most abundant species in the gut. Under Black Queen though, we would expect these species to be most likely "donors" in cross-feeding interactions. Authors should also discuss the implications, limitations, and possible alternative hypotheses of this result, which I think also stimulates future work and experiments.

Regarding Figure 5B, there is a couple of questions I believe the authors should clarify. First, How is it possible that many species have close to 0 pathways? Second, besides the overall negative correlation, the data shows some very conspicuous regularities, e.g. many different "lines" of points with identical linear negative slope but different intercept. My guess is that this is due to some constraints in the pathway detection methods, but I struggle to understand it. I think the authors should discuss these patterns more in detail.

Finally, I also have some conceptual concerns regarding the genomic analysis. Namely, genes can be used for biosynthesis of e.g. building blocks, but also for consumption of nutrients. Under the Black Queen Hypothesis, we would expect the adaptive loss of biosynthetic genes, as those nutrients become provided by the community. However, for catabolic genes or pathways, I would expect the opposite pattern, i.e. the gain of catabolic genes that would allow taking advantage of a more rich environment resulting from a more diverse community (or at least, the absence of pathway loss). These two opposing forces for catabolic and biosynthetic genes/pathways might obscure the trends if all genes are pooled together for the analysis. I believe this can be easily checked with the data the authors already have, and could allow the authors to discuss more in detail the functional implications of the trends they see and possibly even make a stronger case for their claims.

Reviewer #2 (Public Review):

The authors re-analysed two previously published metagenomic datasets to test how diversity at the community level is associated with diversity at the strain level in the human gut microbiota. The overall idea was to test if the observed patterns would be in agreement with the "diversity begets diversity" (DBD) model, which states that more diversity creates more niches and thereby promotes further increase of diversity (here measured at the strain-level). The authors have previously shown evidence for DBD in microbiomes using a similar approach but focusing on 16S rRNA level diversity (which does not provide strain-level insights) and on microbiomes from diverse environments.

One of the datasets analysed here is a subset of a cross-sectional cohort from the Human Microbiome Project. The other dataset comes from a single individual sampled longitudinally over 18 months. This second dataset allowed the authors to not only assess the links between different levels of diversity at single timepoints, but test if high diversity at a given timepoint is associated with increased strain-level diversity at future timepoints.

Understanding eco-evolutionary dynamics of diversity in natural microbial communities is an important question that remains challenging to address. The paper is well-written and the detailed description of the methodological approaches and statistical analyses is exemplary. Most of the analyses carried out in this study seem to be technically sound.

The major limitation of this study comes with the fact that only correlations are presented, some of which are rather weak, contrast each other, or are based on a small number of data points. In addition, finding that diversity at a given taxonomic rank is associated with diversity within a given taxon is a pattern that can be explained by many different underlying processes, e.g. species-area relationships, nutrient (diet) diversity, stressor diversity, immigration rate, and niche creation by other microbes (i.e. DBD). Without experiments, it remains vague if DBD is the underlying process that acts in these communities based on the observed patterns.

Another limitation is that the total number of reads (5 mio for the longitudinal dataset and 20 mio for the cross-sectional dataset) is low for assessing strain-level diversity in complex communities such as the human gut microbiota. This is probably the reason why the authors only looked at one species with sufficient coverage in the longitudinal dataset.

Analyzing the effect of diversity at a given timepoint on strain-level diversity at a later timepoint adds an important new dimension to this study which was not assessed in the previous study about the DBD in microbiomes by some of the authors. However, only a single species was analysed in the longitudinal dataset and comparisons of diversity were only done between two consecutive timepoints. This dataset could be further exploited to provide more insights into the prevailing patterns of diversity.

Finally, the evidence that gene loss follows increase in diversity is weak, as very few genes were found to be lost between two consecutive timepoints, and the analysis is based on only a single species. Moreover, while positive correlation were found between overall community diversity and gene family diversity in single species, the opposite trend was observed when focusing on pathway diversity. A more detailed analysis (of e.g. the functions of the genes and pathways lost/gained) to explain these seemingly contrasting results and a more critical discussion of the limitations of this study would be desirable.

Reviewer #3 (Public Review):

This work provides a series of tests of hypothesis, which are not mutually exclusive, on how genomic diversity is structured within human microbiomes and how community diversity may influence the evolution of a focal species.

Strengths:<br /> The paper leverages on existing metagenomic data to look at many focal species at the same time to test for the importance of broad eco-evolutionary hypothesis, which is a novelty in the field.

Weaknesses:<br /> It is not very clear if the existing metagenomic data has sufficient power to test these models.<br /> It is not clear, neither in the introduction nor in the analysis what precise mechanisms are expected to lead to DBD.<br /> The conclusion that data support DBD appears to depend on which statistics to measure of community diversity are used. Also, performing a test to reject a null neutral model would have been welcome either in the results or in the discussion.

Reviewer #1 (Public Review):

This is an interesting paper that presents a novel idea for the identification of risk factors amongst highly correlated traits in a Mendelian randomization paradigm - a previous investigation (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8438050/) has considered PCA, but not sparse PCA. There are clear conceptual reasons why sparse PCA may be an improvement, as detailed in this paper. Overall, the paper does a good job in terms of motivating this work and comparing the methods. A large chunk of the motivation for the method is conceptual (rather than empirical), and it's unlikely that any method would outperform others in all circumstances, but the authors do a good job of illustrating differences and giving a clear and qualified recommendation.

Reviewer #2 (Public Review):

The main analysis performed in the paper is to determine causal associations of 118 highly correlated lipid metabolites with coronary heart disease (CHD), using summary data from two genome-wide association studies, with 148 genetic variants identified for the exposures. A standard multivariable MR analysis is problematic in this case, as the genetic variants are not simultaneously relevant for all exposures, as clearly indicated by very low values of the conditional F-statistics. In order to reduce the multicollinearity problem, the use of (sparse) principal components techniques is proposed. For the summary data used here, this entails determining the (sparse) principal components from the matrix of the estimated univariate associations of the exposures and the genetic markers. This implicitly constructs linear combinations of the exposures. In a simulation study, this approach is shown to work well for determining whether an exposure has a causal association with the outcome. A conditional F-statistic is developed to evaluate the strength of relevance of the genetic markers for the principal components. In the application, these F-statistics show that instruments are jointly relevant for the transformed exposures. For the sparse methods, the transformed exposures are loaded on VLDL, LDL, and HDL traits, hence obtaining causal estimates for intervening on biologically meaningful pathways.

The dimension reduction techniques and the results obtained are very interesting. As the analysis is performed on summary statistics, the univariate associations are treated as data, on which to perform the principal components analysis. This could be explained more and contrasted with a standard PCA when one has all the individual-level data available.

Reviewer #3 (Public Review):

To motivate the proposal, Karageorgiou et al. first identify a problem in applying current multivariable MR (MVMR) methods with many correlated exposures. I believe this problem can really be broken into two pieces. The first is that MVMR suffers from weak instrument bias. The second is that some traits may have nearly co-linear genetic associations, making it hard to disentangle which trait is causal. These problems connect in that inclusion of co-linear traits amplifies the problem of weak instrument bias - traits that are nearly co-linear with another trait in the study will have no or very few conditionally strong instruments.<br /> The authors then propose a solution: Apply a dimension reduction technique (PCA or sparse PCA) to the matrix of GWAS effect estimates for the exposures. The identified new components can then be used in MVMR in place of the directly measured exposures.

I think that the identified problem is timely and important. I also like the idea of applying dimension reduction techniques to GWAS effect estimates. However, I don't think that the manuscript in its current form achieves the goals that it has set out. Specifically, I will outline the weaknesses of the work in three categories:<br /> 1. The causal effects measured using this method are poorly defined.<br /> 2. The description of the method lacks important details.<br /> 3. Applied and simulation results are unconvincing.<br /> I will describe each of these in more detail below.

1. To me, the largest weakness of this paper is that it is not clear how to interpret the putatively causal effects being measured. The authors describe the method as measuring "the causal effect of the PC on outcome" but it is not obvious what this means.

One possible implication of this statement is that the PC is a real biological variable (say some hidden regulator) that can be directly intervened on. If this is the intention it should be discussed. However, this situation would imply that there is one correct factorization and there is no guarantee that PCs (or sparse PCs) come close to capturing that.

The counterfactual implied by estimating the effects of PCs in MVMR is that it is possible to intervene on and alter one PC while holding all other PCs constant.<br /> In the introduction, the authors note (and I agree) that one weakness of MR applied to correlated traits is that "MVMR models investigate causal effects for each individual exposure, under the assumption that it is possible to intervene and change each one whilst holding the others fixed." However, it is not obvious that altering one PC while holding the others constant is more reasonable.

2. This section combines a few items that I found unclear in the methods section. The most critical one is the lack of specification on how to select instruments.<br /> For the lipids application, the authors state that instruments were selected from the GLGC results, however, these only include instruments for LDL, HDL, and TG, so 1) it would not be possible to include variants that were independently instruments for one of the component traits alone and 2) there would be no instruments for the amino acids. There is no discussion of how instruments should be selected in general.<br /> This choice could also have a dramatic impact on the PCs estimated. The first PC is optimized to explain the largest amount of variance o of the input data which, in this case, is GWAS effect estimates. This means that the number of instruments for each trait included will drive the resulting PCs. It also means that differences in scaling across traits could influence the resulting PCs.

The other detail that is either missing or which I missed is what is used as the variant-PC association in the MVMR analysis. Specifically, is it the PC loadings or is it a different value? Based on the computation of the F-statistic I suspect the former but it is not clear. If this is the case, what is the effect of using loadings that have been shrunk via one of the sparse methods? It would be nice to see a demonstration of the bias and variance of the resulting method, though it is not clear to me what the "truth" would be.

3. In the lipids application, the fact that M.LDL.PL changes sign in MVMR analysis are offered as evidence of multicollinearity. I would generally associate multicollinearity with large variance and not bias. Perhaps the authors could offer some more insight on how multicollinearity would cause the observation.<br /> A minor point of confusion: I was unable to interpret this pair of sentences "Although the method did not identify any of the exposures as significant at Bonferroni-adjusted significance level, the estimate for M.LDL.PL is still negative but closer to zero and not statistically significant. The only trait that retains statistical significance is ApoB." The first sentence says that none of the exposures were significant while the second sentence says that Apo B is significant. The GRAPPLE results don't seem clearly bad, indeed if only Apo B is significant, wouldn't we conclude that of the 118 exposures, only Apo B is causal for heart disease? It would help to discuss more how the conclusions from the PC-based MVMR analysis compare to the conclusions from GRAPPLE.

It is a bit hard to interpret Table 4. I wasn't able to fully determine what "VLD, LDL significance in MR" means here. From the text, it seems that it means that any PC with a non-zero lodaing on VLDL or LDL traits was significant, however, this seems like a trivial criterion for the PCA method, since all PCs will be dense. This would mean this indicator only tells us whether and PCs were found to "cause" heart disease.

In simulations, I may be missing something about the definition of a true and false positive here. I think this is similar to my confusion in the previous paragraph. Wouldn't the true and false positive rates as computed using these metrics depend strongly on the sparsity of the components? It is not clear to me what ideal behavior would be here. However, it seems from the description that if the truth was as in Fig 7 and two methods each yielded one dense component that was found to be causal for Y, these two methods would get the same "score" for true positive and false positive rate regardless of the distribution of factor loadings. One method could produce a factor that loaded equally on all exposures while the other produced a factor that loaded mostly on X1 and X2 but this difference would not be captured in the results.

Reviewer #1 (Public Review):

This paper describes detailed experiments to characterize the morphology and deformability, of red blood cells (RBCs) from COVID patients as compared to healthy individuals. Deformability is characterized by the visualization of cell shapes during flow in a microfluidic channel at high strain rates. One important feature of the study is that it considers the changes in patient RBCs when placed in healthy plasma and vice versa. An important observation is that the changes to RBCs properties appear, from this report, to be reversible - diseased cells revert to normal morphology and deformability upon immersion in healthy plasma. It also reports metabolics and proteomics analyses to shed light on the connections between the biochemical environment and RBC properties. One important question with regard to the changes in COVID-RBC properties with respect to plasma composition is whether the effect is simply due to dilution - are the factors responsible for the pathological morphology just diluted away when the cells are immersed in plasma that does not contain them? The studies are performed at very low hematocrit, so the composition equilibrium established here will not correspond to physiological conditions. This issue needs further discussion.

Reviewer #2 (Public Review):

The authors addressed a timely and challenging topic, namely the role played by red blood cells (RBCs) and blood plasma in Covid-19 disease.

A remarkable feature reported here is that RBC from patients exhibits a notable morphological change, whereas when suspended in plasma control (healthy) exhibit normal shapes. Conversely, RBCs from healthy donors suspended in patients' plasma undergo similar morphological alteration as do RBCs of patients suspended in their plasma. Another important fact reported here is that RBCs affect plasma composition in a nontrivial way.

The data reported here cover a large panel of features, ranging from RBC morphological changes, plasma metabolites, and protein alteration, to collective RBC formation, in the form of clusters. They should constitute a precious enrichment of relevant information regarding the intricate response of organisms to the Covid-19 virus.

This work will be of the potential impact on the community aiming to decipher the multifactorial impacts of blood components on patients suffering Covid-19.

Reviewer #1 (Public Review):

This paper estimates the selective effects of loss-of-function mutations in each gene, ultimately providing an estimate of the overall distribution of fitness effects, and point estimates for each gene. Unlike some measures of intolerance such as pLI, the parameter the authors estimate (effectively the compound parameter hs) is interpretable in terms of evolutionary fitness. The most comparable analysis is by Weghorn et al (2019) which estimates the same parameter, but on a smaller sample and using a different approach.

The point estimates will be broadly useful for future analyses, and the overall distribution is an interesting result. The enrichment in various disease cohorts is unexpected but nice to demonstrate. Overall, I found the approach to be elegant and it has the nice property that it can be easily generalized to more complicated models. The data cleaning and filtering is quite extensive but all seems well done and appropriate. Qualitatively, the results clearly make a lot of sense (Figure 3 is an excellent figure) My only major questions are around how quantitatively robust this analysis is to the choice of parameters and hyperparameters including priors, mutation rates, and demography. I don't think that extensive work is required, but it would be helpful to see some quantification of this uncertainty.

Reviewer #2 (Public Review):

This study models the fitness costs of loss-of-function mutations in a large cohort of a human database of 55,855 individuals. The modeling indicates different values for autosomal genes, X-linked genes, and those present in the pseudo-autosomal regions of the X and Y chromosomes. The study details the frequency of de novo mutations in zygotes and examined the relationship to a few specific genetic diseases. The authors have composed a well-written manuscript, have explicitly detailed their assumptions, and have noted caveats to interpretations. The results are a valuable documentation of the effects of loss-of-function mutations in humans.

Reviewer #3 (Public Review):

This manuscript presents a new method to estimate the selective effect of heterozygous loss of function mutations. The authors offer a sequential Monte Carlo algorithm coupled with ABC estimates based on forward population genetics simulations. The method is of obvious interest to the field. The result confirms that DFE distribution for PTVs is broad with the mean and median exceeding 1% and ~20% of genes associated with more than 10% loss in fitness. The new quantitative estimates are likely an improvement over the state-of-the-art. Importantly, the authors include estimates for PTVs on the X chromosome, which are expectedly higher. The authors demonstrate that de novo PTVs leading to a substantial fitness loss are highly enriched in individuals affected by severe complex disorders including neuropsychiatric disorders. They also provide estimates of allelic ages for variants with specific selection coefficients. This work is of interest to both population and medical geneticists.

Reviewer #1 (Public Review):

The authors note contradictory clinical data on the effects of functional FAAH mutations on body weight in clinical samples. They aim to resolve this issue via animal modes and genetic approaches combined with endocrine manipulations to vary "context". Major strengths are comprehensive evaluation of FAAH variant in several models of neuroendocrine changes in body weight (CORT, leptin, gherlin) and provide some mechanistic insight at the signal transduction level. Localization of FAAH modulation to AGRP neurons is a strength. Weaknesses include lack of cellular mechanisms, i.e. how AEA release from AGRP neurons affects ongoing cellular/synaptic activity to regulate behavioral/physiological phenotypes. The work is impactful as it potentially reconciles contradictory clinical data, is comprehensive and rigorous in many ways. These data will provide insight into how FAAH activity regulates body weight in the context of distinct hormonal signals and will likely have a major impact on the field.

Reviewer #2 (Public Review):

Interestingly, prior analysis of the 385A allele indicated a post-translational mechanism that led to instability of the protein and an ~50% reduction in protein concentrations and FAAH activity. In addition, FAAH degrades AEA, one of several known endocannabinoids, suggesting that FAAH is a significant part of the endocannabinoid signaling although there are several other endocannabinoids that are not affected by FAAH.

At the basal state, normal chow and home cage conditions, wild type mice were not different from homozygous mutant FAAH mice in terms of body weight and body composition. However, the FAAH mutants had reduced food intake that was compensated for by lower energy expenditure. This finding strongly suggests that compensatory mechanisms are in play during lifelong changes in strengths of AEA signaling.

The authors go on to perform increasingly shorter durations of manipulations of glucocorticoid manipulations (down to several hours) to examine the impact of the FAAH mutation. Thus, the authors are able to conclude that the FAAH mutation leads to acute changes of feeding.

Examination of the biochemical signaling pathway showed that AMPK activity is affected by GC/FAAH experimental manipulation although the relevance of the finding should be somewhat tempered by the later studies in hypothalamic AGRP neurons and FAAH since the measures were not neuron specific.

Finally, the authors examine the role of FAAH expression in hypothalamic AGRP neurons since their measures of AEA concentrations showed changes only in the hypothalamus after CORT treatments. Virally mediated knockdown of FAAH, using a AAV CRISPR/Cas9 single vector system, indicated that knockdown of FAAH in AGRP neurons is sufficient to recapitulate the authors' findings on GC-modulated feeding.

The data are convincing and settles the issue of variability in the evidence regarding the role of FAAH genetic variants in feeding.

Reviewer #1 (Public Review):

When we tilt our heads, we do not perceive objects to be tilted or rotated. In this study, the authors investigate the underlying neural underpinnings by characterizing how neurons in monkey IT respond to objects when the entire body is tilted. They performed two experiments. In the first experiment, the authors record single neuron responses to objects rotating in the image plane, under two conditions - when the animals were tilted +20{degree sign} or -20{degree sign} relative to the gravitational vertical. Their main finding is that neural tuning curves for object orientation were highly correlated under these conditions. This high correlation is interpreted by the authors as indicative of encoding of object orientations relative to an absolute gravitational reference frame. To control for the possibility that the whole-body tilt could have induced compensatory torsional rotations of the eyes, the authors estimated the eye torsional rotation between the {plus minus}20{degree sign} whole-body tilt to be only {plus minus}6{degree sign}. In the second experiment, the authors recorded neural responses to objects rotated in the image plane with no whole-body tilt but with a visual horizon that could be tilted by the same {plus minus}20{degree sign} relative to the gravitational vertical. Here too they find many neurons whose tuning curves were correlated between the two horizon tilt conditions. Based on these results, the authors argue that IT neurons represent objects relative to the gravitational or absolute vertical.

The question of whether the visual system encodes objects relative to the gravitational vertical is an interesting and basic one, and I commend the authors for attempting this question through systematic testing of object selectivity under conditions of whole-body tilt. However, I found this manuscript extremely difficult to read, with important analyses and controls described in a very cursory fashion. I also have several major concerns about these results.

First, the high tuning correlation in the {plus minus}20{degree sign} whole-body tilt conditions could also occur if IT neurons encoded object orientation relative to other fixed contextual cues in the surrounding, such as the frame of the computer monitor. The authors ideally should have some experiment or analysis to address this potential confound, or else acknowledge that their findings can also be interpreted as the encoding of object orientation relative to contextual cues, which would dilute their overall conclusions.

Second, I do not fully understand torsional eye movements myself, but it is not clear to me whether this is a fixed or dynamic compensation. For instance, have the authors measured torsional eye rotations on every trial? Is it fixed always at {plus minus}6{degree sign} or does it change from trial to trial? If it changes, then could the high tuning correlation between the whole-body rotations be simply driven by trials in which the eyes compensated more? The authors must provide more data or analyses to address this important control.

Third, I find that when the objects were presented against a visual horizon, different object features are occluded at each orientation. This could reduce the correlation between the neural response in the retinal reference frame, thereby biasing all results away from purely retinal encoding. The authors should address this either through additional analyses or acknowledge this issue appropriately throughout.

Reviewer #2 (Public Review):

In this paper, the authors investigate the intriguing question of what orientation reference frame the visual selectivity of neurons in the IT cortex is expressed in - a world-centered gravitational one, or a retinal one? To address this, the authors physically rotate a monkey to dissociate a gravitational from a retinal reference frame. They find surprising and compelling evidence that many cells encode selectivity in a gravitational frame. The finding raises questions about whether the function of the IT cortex is solely object recognition, or whether it might play an important role in physical scene understanding.

In general, I found the paper clearly written, the analyses appropriate, and the results supportive of the conclusions. I think the work should spur new thinking about what the IT cortex is accomplishing. The notion that IT cells are receiving vestibular signals is likely to be unsettling for many who think of it as simply the endstage of a convolutional neural network.

Reviewer #3 (Public Review):

This is a very interesting study examining for the first time the influence of lateral tilt of the whole body on orientation tuning in macaque IT. They employed two types of displays: one in which the object was embedded in a scene that had a horizon and textured ground surface, and a second one with only the object. For the first type, they examined the orientation tuning with and without tilting the subject. However, the effect of tilt for the scene stimuli is difficult to interpret in terms of gravitational reference frame since varying the orientation of the object relative to the horizon leads to changes in visual features between the horizon and object. If neurons show tolerance for the global orientation of the scene (within the 50{degree sign} manipulation range) then the consistent orientation tuning across tilts may just reflect tuning for the object-horizon features (like the angle between the object and the horizon line/surface) that is tolerant for the orientation of the whole scene. Thus, the effects of tilt can be purely visually-driven in this case and may reflect feature selectivity unrelated to gravitation. The difference between retinal and gravitational effects can just reflect neurons that do not care about the scene/horizon background but only about the object and neurons that respond to the features of the object relative to the background. Thus, I feel that the data using scenes cannot be used unambiguously as evidence for a gravitational reference frame. The authors also tested neurons with an object without a scene, and these data provide evidence for a gravitational reference frame. The authors should concentrate on these data and downplay the difficult-to-interpret results using scenes. Furthermore, the analysis of the single object data should be improved and clarified.

Reviewer #2 (Public Review):

In this report, the authors evaluate the possibility that LEC neurons send direct projection onto MEC cells, thus revising the current model of LEC and MEC sending independent inputs to the DG, whose role is to eventually combine both inputs. They demonstrate that L2a SCs in the LEC that receive neocortical inputs, send collaterals to L1 MEC, thus identifying a new indirect route by which MEC neurons can integrate cortical information. Vandrey et al., show that L2a SCs in the LEC contact directly with both inhibitory and excitatory cells in the MEC, but superficial principal cells with a higher probability. Therefore, L2 LEC neurons can exert control of the MEC activity, thus shaping its inputs to the hippocampus. By controlling the firing activity in superficial MEC, this newly identified LEC-MEC connection may participate in the combination of spatial inputs with sensory and high-order signals and thus "provide a substrate for the integration of 'what' and 'where' components of episodic memories".

The manuscript is well-written and the experimental design is well-suited to answer the question. The data presented here is a thorough, well-explained, and detailed work describing a new communication route between the LEC and MEC.

Reviewer #1 (Public Review):

In this manuscript, Vandrey et al characterize axonal projections from fan cells in the lateral entorhinal cortex (LEC) to the medial entorhinal cortex (MEC). Their findings are important and the manuscript is well-written.

Reviewer #3 (Public Review):

The manuscript by Vandry et al analyzes the circuitry connecting LEC to MEC, identifying a new connection with potential significance for cortico-hippocampal coding and memory. Using a combination of viral tracing, patch-clamp electrophysiology, and optogenetics, the authors reveal a new excitatory projection from Fan cells of LEC layer 2 to superficial neurons of MEC. Specifically, Fan cells synapse on MEC L2 stellate and pyramidal neurons, as well as layer 1 and layer 2 local interneurons, which provide fast and slow local feedforward inhibition to MEC excitatory neurons. The authors observe substantial cell-to-cell heterogeneity in the excitatory-to-inhibitory ratio, which does not seem to be a result of anatomical location. This heterogeneity is conserved during theta-like stimulation. This new connection allows for a kind of unidirectional "cross-talk", in which LEC can speak to MEC prior to or during communication of both of these regions with the hippocampus.

The results are generally clear and well-contextualized by the text. The authors use multiple complementary anatomical methods to identify the LEC to MEC connection, all of which agree. This is supported by the electrophysiological measurements, which are straightforward and generally convincing. The results provide important data for understanding the previously underappreciated reciprocal circuitry between MEC and LEC, which, as the authors nicely lay out in the introduction, is likely key for understanding the operation of memory networks.

The work described in this manuscript, which is all in vitro, appears nicely conducted and solid and is well presented and analyzed appropriately. However, it is not clear how this information can be used to glean an improved understanding of how LEC and MEC interact in the intact system, which is obviously the big question. In vivo experiments of this kind are quite challenging, but without some observation or perturbation of circuit dynamics in the intact animal, or at the very least a compelling model of how hippocampal/memory information processing is influenced by this new circuit, it may be hard for readers to know what to make of the new data the authors provide.

Reviewer #1 (Public Review):

This is a brief set of experiments that tells a nice story that is relevant to a very important area of biology, namely senescence. The authors identify a role for lncRNA H19 in senescence and delve into the upstream and downstream factors that could describe the phenomenon. They identify CTCF and p53 as upstream regulators of H19 in senescing cells and propose that sponging of let-7 could be a contributing factor to H19's effects via altered regulation of EZH2.

The work is backed up by strong data in cell models. However, the work could benefit from additional mechanistic data to support the most important conclusions. For example, does H19 sponge let-7 in these cells? What are the relative levels of expression of H19 compared to let-7 in these cells? Is the let-7 binding site on H19 required for the effects of H19? And does let-7 directly regulate EZH2 in these cells? Can a direct role for H19 in affecting EZH2 be ruled out in these cells?

Reviewer #2 (Public Review):

The study aims to characterize the role of lncRNA H19 in senescence and proposes a mechanism involving CTCF and the activation of p53. The authors suggest that H19 loss induces let7b-mediated repression of EZH2, which is a critical component in the regulation of senescence-associated genes. Additionally, the authors state that H19 is required for inhibition of senescence by the mTOR inhibitor rapamycin.

The experiments appear to be performed to a high standard, and the individual observations, and conclusions about the importance of the individual players in senescence appear solid. For example, the authors convincingly show that H19 decreases in expression in aged cells/tissues and that its knockdown leads to entry into senescence. These results are consistent with recent studies in other systems (e.g., ref 38). Also, the knockdown of CTCF convincingly leads to senescence. However, these observations are largely not very surprising/novel. The premise of the manuscript is a connection between these components into a particular "axis" that regulates entry into senescence. This connection between the different regulators studied (H19, CTCF, EZH2, p53), and in particular, their specificity, which is key to the proposed "axis" remains insufficiently supported, and many of the results, unfortunately, appear to be over-interpreted.

Major comments

1. In Figure 1, the authors claim that H19 levels are reduced during aging in vitro and in vivo and that H19 levels are maintained by rapamycin treatment. To state the connection between H19 and rapamycin and its relation to aging, there is a need to show what happens in "young" cells treated with rapamycin.

Furthermore, the authors state that H19 "is essential for the inhibitory effect of rapamycin on cellular senescence". There doesn't appear to be sufficient evidence to support such a claim; additional data emphasizing the direct connection between H19 and rapamycin is needed - e.g., show that in H19-null cells rapamycin does not affect senescence.

2. CTCF is a general regulator involved in various cellular processes and supporting progression through the cell cycle; therefore, its perturbation can lead to global effects on cell health that are not necessarily related to H19. The data shown in figure 2 is insufficient to indicate a direct correlation between CTCF and H19. This will require showing that mutating specifically the CTCF binding sites near H19 affects senescence.

The same applies to the connection between H19 and let-7b shown in Figure 5. It is not very surprising that let-7b, a general antagonist of proliferation, positively regulates senescence. Here as well, the direct connection to H19 is weak. Can the authors rescue the cells that enter senescence following H19 depletion by H19 expression? If so - is this rescue capacity lost when let-7 sites are mutated? Is it possible to rescue by expressing an artificial let-7 sponge instead of H19? Otherwise, let-7b could very well be another factor related to senescence and/or regulated, but not the main mediator of the effects of H19, or part of an axis that includes H19, as proposed in the manuscript.

3. In figures 2d,3f,5i/j the authors present only representative tracks and regions from CUT&Tag-experiments, and its not clear to what extent these changes are significant when considering genome-wide data, replicates etc., and so these data are uninterpretable. This is important, as these panels are used as evidence for specific connections between members of the axis. The authors should provide a statistical test for all the regions in the genome, based on replicates, and show that these changes are significant to use these data to support their model. Otherwise, the specific connection between CTCF and H19 remains weak, and the specific change in p53 regulation of CTCF in the context of senescence is not convincing. In any case, the number of replicates and the QC of the data should be presented, and the data should be made available to the reviewers.

4. The authors state in the Discussion that the mechanism that lead to decreased H19 expression as part of the senescence program consists of two phases: an acute response driven by p53 activation and a prolonged response dictated by the loss of CTCF. There doesn't appear to be enough evidence to support this claim, as the individual experiments don't measure any such bi-phasic phenomena.

Reviewer #1 (Public Review):

In the work by Van Eyndhoven et al., the authors aim to determine if the cell state present in the cells that first produce Type I Interferon (IFN-I, an antiviral cytokine) is stochastically regulated or may be epigenetically inheritable. This work builds from previous studies demonstrating that IFN-I responses occur in two waves: a small proportion of early responding "precocious" cells which induce population-wide responses through autocrine and paracrine signaling. The authors contextualize their study well within the literature, and discuss the hypotheses of stochasticity or determinism driving early responding cell fate. Within this context, the authors set out to characterize and model the nature of these "first responder" cells during IFN-I antiviral signaling. Developing a quantitative imaging approach to measure IRF7 translocation, the authors measure the proportion of first responder cells as defined by higher ratios of nuclear/cytosolic IRF7 expression. Transfection of Poly(I:C) induces IFN-I signaling and leads to ~2% first responders, in line with previously published work. The authors then show that responder frequencies increase following treatment with a DNA methyltransferase inhibitor, suggesting a relationship between epigenetic regulation and responder potential. To test the hypothesis that the first responder cell state occurs stochastically, the authors adapted the Luria-Delbruck fluctuation test by evaluating responder frequency as a function of cell division or generation. First witnessing high variability of responder frequencies using limiting dilution clonal expansion followed by low stable frequencies after 100 divisions (similar to regular cultures), the authors suggest that the first responder state may be partially heritable and develop a mathematical model of transient heritability. Finally, to assess whether cell density and quorum sensing contribute to this transient heritability, cells plated at different densities were interrogated for responder frequencies after a fixed number of divisions; only low density seeding led to high and variable responder frequencies.

The interrogation of IFN-I early responding cells by Van Eyndhoven et al. is well executed and supports the claim that first responder events are non-stochastic. However, the use of transgenic reporter cells in vitro may limit the findings reported in the manuscript to this system, and awaits further experimentation to assess the generalizability of these findings to overall cellular decision-making during inflammatory responses. Identifying the mechanisms responsible for transient heritability and the density-dependent regulation will be of high interest.

1) Context and definitions for stochasticity and heritability: The authors provide well-referenced introductions and explanations throughout the manuscript. However, key understanding of concepts for their central hypothesis on transient heritability are not shared until well into the results sections (Lines 215-227), leaving the introduction somewhat unclear on the authors thinking and motivation. The manuscript would benefit by including clear definitions of "stochastic", "transiently heritable", and "heritable" and their relationships to "intrinsic" and "deterministic" in the introduction.

2) Generalizability of findings to other cell types, systems, and triggers: The cell line and Poly(I:C) delivery method used by the authors lacks sufficient characterization to extend the conclusions derived from its use. Notably, the NIH3T3-IRF7-CFP cell line expresses IRF7 constitutively and thus may only be a good model for cells with similar expression levels; many primary cells only express IRF7 at low levels or not at all until stimulated (PMID: 2140621). The conclusions would be greatly strengthened by demonstrating similar first responder dynamics/heritability in other cell types. The experiments measuring the efficiency of Poly(I:C) delivery by transfection lack sufficient resolution to determine if the Poly(I:C) is intracellular or membrane bound. IFN-I response kinetics, and potentially quality, would likely be distinct between cytosolic and endosomal sensing and may impact the likelihood of becoming a first responder.

3) Epigenetic regulation of transient heritability: To test the contribution of epigenetic regulation on first responder fate, the authors treat their cells with DNMTi. While treatment with this drug does increase the proportion of first responder cells, the authors don't provide evidence that the mechanism of action is mediated by inhibiting DNA methylation. This is further confounded by the reduced responder frequencies in DNMTi treated cells transduced with Poly(I:C) (Fig 4g). The authors offer an explanation for this observation, but their reported data (Fig 4h) doesn't measure whether DNMTi, leads to latent retrovirus activation, broader demethylation, or a combination of the two.

4) Temporal experimental data to validate and extend transient heritability and quorum sensing: Developing a model for cellular-decision making during early IFN-I responses, the authors formalize and test the hypothesis of transient heritability. While the data largely fit the model proposed (Fig 6D-F), the reported data points lack sufficient temporal resolution to validate the model during the earlier and more variable generations. Given that by generation 9 variability in first responder frequency has almost stabilized, there is only one data point (generation 6) to evaluate the fit of the ODE described. More densely sampled data points below generation 10 are necessary to validate the model. Moreover, a discussion of Kon calculation/observation, meaning, and validation is missing. To partially test their claim that Kon is a function of density (i.e., quorum sensing), the authors plate cells at different densities and measure the responder frequency at generation 6. This analysis lacks contextualization of other autocrine and paracrine signals potentially impacting IFN-I response. Moreover, these signals will be diverse in different cell types and could impact Kon and/or the overall model.

Reviewer #2 (Public Review):

In this manuscript, Eyndhoven and colleagues develop an experimental and analytical setup to test the role of cell-intrinsic factors in guiding fate decisions to viral infections. The study is motivated by the observations that early antiviral response mediated by type 1 interferon (IFN-1) is not fully penetrant in response to virus, and is initiated only in 1-3% of the cells. Using a combination of IFN-1 reporter system, automated image segmentation, DNMT inhibitors, and Luria-Delbrück fluctuation test in a murine cell line model, the authors state that cell intrinsic factors guide IFN-1 response in rare cells. This response (measured with IRF7 translocation) is predetermined and heritable over several generations. Lastly, the authors report cell density effects on IFN-1 response, a phenomena the authors refer to as "quorum sensing", and rationalize their observations with an ODE-based mathematical model.

Overall, this is a well-designed, well-controlled, and timely study, given the rapidly increasing reports documenting heritable cell states that can guide fate choices in single cells. The manuscript has elegant experiments and is generally clear to follow and the figures are easy to understand. While the authors largely state what they find, some of their claims and terminology are not supported by their experiments. Additionally, many figures lacked scale bars, axis, labels, and detailed captions. The authors are also encouraged to cite a wider set of seminal studies, acknowledging their contributions to transient cell states guiding fate choices.

Reviewer #3 (Public Review):

In this paper, Van Eyndhoven et al. use a quantitative and system immunology approach to dissect the factors contributing to the fate of early IFN-I responders. Overall, this manuscript is quite elegant and technically very strong. My questions/comments are limited to (1) the fraction of cells that respond in the absence of Poly(I:C), (2) the source of stimulation for the second responders in this system.

1. For the small fraction of cells that respond in the absence of Poly(I:C), are these cells just showing IRF7 translocation or are they fully responding with IFNB production? Has this been observed in other experimental systems or contexts? Do you also observe secondary responders in the unstimulated samples (as shown in the stimulated in Fig. 2G-I)?

2. Do the second responders only arise through direct IFN-I production by first responders? Is it possible that this response has any relationship with the initial transfection with Poly(I:C)?

Reviewer #1 (Public Review):

The study by Lehmann et al. reports novel structures of the human ferroportin (SLC40A1), which is responsible for iron transport in the body. Specifically, ferroportin controls the plasma concentration of iron by transporting Fe2+ out of the cell. To regulate plasma iron concentrations, the liver releases hepcidin, a peptide-based hormone that inhibits ferroportin activity. Specific inhibitors of ferroportin are being developed to treat thalassemia and sickle cell disease, which are diseases that result in reduced red blood cell function.

The present study reports the structure of human ferroportin in complex with one such inhibitor, vamifeport, which is currently in clinical trials for sickle cell disease. The authors use their structures to suggest a mechanism for vamifeport binding to ferroportin and support the structural data with in vitro binding assays to study the specific interactions made in the binding site. In addition, one of the structures obtained was a novel protein conformation, an occluded state. This is the first occluded state observed for ferroportin, enabling the authors to discuss the implications for understanding the transport mechanism. However, this appears to have resulted in a slightly confusing analysis.

Overall the study is well presented, although in several places appears overly wordy and might benefit from being edited to focus on the main points the authors wish to highlight. For example, the title focuses on the new insights gained from the vamifeport complex. Yet, the discussion section focuses almost entirely on the transport mechanism, with little additional analysis of the mechanism of vamifeport inhibition. In my view, the paper suffers from this disconnect, as the functional data support the vamifeport structure, not the transport mechanism. Yet, the discussion focuses heavily on the transport mechanism, with little reference to the results. Rather, the discussion relies on an in-depth understanding of secondary active transport literature (MFS, NRAMP, etc.).

The data is high quality, and the conclusions drawn about the orientation of the drug in the binding site are sound. This study represents an important advance in understanding iron homeostasis in the human body and current methods to modulate iron transport to treat human disease.

Reviewer #2 (Public Review):

This manuscript by Lehman et al. details the structural characterization of human Ferroportin, which builds on the previous structural characterisation of this protein. Here, through the use of synthetic nanobodies, the authors capture the protein in the outward-facing state that has been obtained previously, and a new conformation in an occluded state, information which would advance understanding of the Ferroportin transport mechanism. In addition, the authors capture Ferroportin in complex with the first clinical-stage Ferroportin inhibitor, Vamifeport, which provides insight that could be used to improve inhibitor efficacy to treat human disease. The structural data is very well supported by clear, well-executed, and informative binding and transport studies. These data reveal that the purified protein is functionally active, able to interact with the peptide-based inhibitor hepcidin in addition to Vamifeport and that hepcidin and vamifeport bind competitively. Site-directed mutagenesis and binding assays were used to convincingly validate the Vamifeport binding site.

Overall, the conclusions in this manuscript are well supported by the data, in particular those relating to inhibitor binding. However, as the authors point out, the occluded state captured here contains an unexpectedly large aqueous cavity compared to the size of the transported substrate. With this peculiar observation in mind, the requirement for the presence of Sy3 nanobody to capture this state and the positioning of the nanobody in between the 2 lobes of the transporter, raises the question of whether this conformation is physiologically relevant, or whether its formation is merely a consequence of Sy3 binding.

Reviewer #3 (Public Review):

To determine how the clinical-stage inhibitor vamifeport interacts with ferroportin, the authors used cryogenic electron microscopy (cryo-EM) to determine several structures of ferroportin in complex with newly isolated sybodies. They found that the highest resolution structure shows an occluded state of the transporter bound to sybody 3 and vamifeport. The inhibitor occupies a small portion of a large occluded cavity, interacts with both the N and the C lobe of the transporter, and overlaps with the binding site for both hepcidin and the iron ion binding site 2. The authors also use binding assays and mutagenesis to confirm that the residues in the vamifeport binding site are important for binding.

As the authors point out, the vamifeport inhibitor can readily be modeled in two orientations. The authors provide a reasonable argument that one orientation provides more specific interactions, but the case would be stronger if the structure had a high enough resolution to distinguish between the two orientations, or if the authors could provide some complementary supporting evidence. Still, the manuscript provides convincing evidence to explain how the compound inhibits ion transport and the similarities and differences between this inhibitor and the endogenous regulatory protein hepcidin.

The authors describe the occluded conformation that they resolve with bound sybody 3 and vamifeport as "on the transport pathway". However, this occluded conformation was captured in the presence of two ligands that are not on-pathway, the inhibitor and the sybody. It seems plausible (maybe even likely?) that the conformation is off-pathway and trapped by these additional ligands. The study would therefore benefit from additional evidence as to whether this conformation is indeed on-pathway.

Reviewer #1 (Public Review):

Andreyeva et al. developed a novel purification/mass spec approach to identify SuUR-associated proteins. From this biochemical tour de force, they identify a complex consisting of the insulator-associated protein Mod(Mdg4) and SuUR that they term, SUMM4. They show that this complex (at least SuUR) has ATPase activity, which is an exciting result was no known biochemical activity associated with SuUR. Given SuUR's function in the under-replication of Drosophila salivary glands, the authors show that SuUR and Mod(Mdg4) at least partially localize on polytene chromosomes and that SuUR displays at least a partial dependence on Mod(Mdg4) for localization to IH, but not PH regions. Finally, using two independent genetic reporters, they show that SuUR itself has an insulator function, which is a new function for SuUR and exciting as it is likely a diploid cell-specific function for SuUR. The authors then attempt to show the Mod(Mdg4) functions in under-replication. Unfortunately, under-replication is minimally, if at all, changed in the Mod(Mdg4) mutant. While the authors bring up several possible scenarios of why this could be, it is still uncertain whether Mod(Mdg4) has a direct effect on under-replication.

Strengths:<br /> The authors developed a very useful strategy to identify protein interactions through multiple purification steps using mass spectrometry. This approach can be applied to different systems and will be generally useful to the community. Through this approach, they provide very compelling data that SuUR and Mod(Mdg4) form a complex. Furthermore, the experiments all have been rigorously performed and the data is of high quality.

Weaknesses:<br /> The way the paper is written, its main focus is on under-replication. What the authors were not able to conclusively demonstrate is whether Mod(Mdg4) functions in under-replication.

Reviewer #2 (Public Review):

This paper from the Fyodorov lab reports the isolation of a native protein complex of SUUR, a Drosophila SNF2-related factor, in a complex with Mdg4, an established chromatin boundary protein. The discovery of this native complex, called SUMM4, was enabled by the development of a mass spec-linked proteomic analysis of fractions from an unbiased, conventional multi-step chromatographic purification of low-abundance protein complexes. The authors validate the native interactions by co-immunoprecipitation and show further with recombinant proteins that SUUR displays ATPase activity, a property not previously shown, and which is stimulated by Mdg4. From a functional perspective, authors demonstrate that both components SUUR and Mdg4 mediate activities of the Drosophila gypsy insulator that blocks enhancer-promoter interactions and acts as a heterochromatin-euchromatin barrier, and moreover, has a role in the under-replication of intercalary heterochromatin.

Overall, this work is a substantial contribution to the field in two respects. First, it provides a new approach to the identification of novel native complexes that are of low abundance and difficult to isolate and identify by conventional biochemistry and mass spectrometry. Second, the interaction between Mdg4 and SUUR is novel and offers an ATP-driven pathway to be further investigated for understanding the mechanism of insulator (gypsy) function. Together, these advances are supported by the compelling quality and quantity of data. However, the paper does not read smoothly and can benefit from rewriting for readers who are not familiar with mass-spec proteomics or Drosophila biology.

Reviewer #2 (Public Review):

Kintscher et al present a nice study on the responses of Adora2a and D1R expressing cells in the tail of the striatum/amygdala transition zone during auditory fear conditioning. Overall the conclusions are that (1) D1R cells show plasticity in activity patterns during the task, with the emergence of tone/movement co-modulated cells; (2) Adora2a cells show less of such changes; (3) gain of function of activity does little where (4) loss of function of activity in each cell class has moderate effects on the learned behavior (i.e. freezing to the CS). There is a nice section on rabies tracing which maps inputs to both cell types which then motivates an analysis of insular cortex inputs onto both cell types and reveals that (5) CS/US pairing alters insular inputs to both cell types.

Overall the paper is well done and the conclusions are believable. Furthermore, this brain area is understudied yet potentially very important.

The analysis of the fluorescence transients is heavy handed. This leads to potential for error and could obscure what appear to be large differences that could be extracted more easily. In some instances, the data are interpreted too optimistically, especially that the silencing experiments implicate plasticity of the neurons rather than the need for activity.

Reviewer #1 (Public Review):

This work identifies distinct contribution of direct (D1+) and indirect (Adora+, D2+) amygdalostriatal medium spiny cells in fear learning and plasticity. The authors combined freely moving calcium imaging with auditory fear learning assay to reveal tone, foot-shock and behavior (movement)-evoked activity of the two MSN population. While D1+ cells show plastic changes driven by fear learning and reaching their maximum tone responsiveness (PSTH) at fear retrieval, Adore+ cells activation remained constant. Furthermore, using optogenetic silencing they showed that the two MSN groups differently contribute to retrieval of fear memory. Both cells receive topographically organized insular cortical inputs which go through learning-induced long-term synaptic changes with opposite direction: postsynaptic LTP at D1 cells, while presynaptic LTD at Adora+ cells. These synaptic changes provide some level of explanation for distinct behavioral contribution of the two cell types in fear learning.

This study focuses on a so far neglected member of the 'extended' amygdalar circuitry, the amygdalostratal transition zone. The data is well-presented, the experiments are in logical order, built on each other and the paper is easy to read and follow.

However, some information regarding the connectivity (and function) of Astr have been presented in recent and earlier papers are missing from, or contradicting with, the present work. One reason to explain these is that the targeted striatal regions vary between experiments, and so, it is difficult to judge when the Astr and when the other part of the caudal (tail) striatum is examined. As these striatal regions are involved in different neuronal networks, their functional consequences could also be distinct. Without precisely clarifying and consistently targeting the aimed striatal region, it is difficult to interpret the findings of the present study (though those are relevant and important).

Reviewer #3 (Public Review):

Schneggenburger and colleagues set out to reveal roles for D1R+ and Adora+ amygdala-striatal transition zone neurons in fear learning. In the first two experiments, the authors expressed fluorescent calcium indicators in D1R+ or Adora+ neurons, measuring change in fluorescence during habituation, training and testing of tone-shock conditioning. In the next experiments, the authors expressed archeorhodopsin (or a control fluorophore) in D1R+ or Adora+ neurons and illuminated with yellow light just before and after foot shock delivery. Freezing was quantified during training and retrieval. Finally, retrograde tracing was performed to reveal direct synaptic inputs on D1R+ and Adora+ neurons.

The paper is potentially interesting. However, some important weaknesses include: the authors use of only male mice, the lack of validation of the Cre lines used in the study, and the data acquisition pipeline.

Reviewer #1 (Public Review):

This is a relatively straightforward manuscript describing an r package that attempts to address issues in color-blindness in the interpretation of multicolor overlapping plots. The demonstration of its usefulness is solid and the findings will be significant in that they should become one of the standards that the scientific community strives to achieve for greater inclusiveness.

Reviewer #2 (Public Review):

The authors present an R/Bioconductor package, scatterHatch, aimed at providing a novel framework for the creation of color-vision deficiency (CVD) accessible plots. The authors lay out that in increasingly common dimensionality reduction plots, like UMAPs and tSNEs, color tends to be the primary factor for distinguishing points of distinct groups. Although color palettes created with accessibility to CVDs in mind are often helpful, none adequately cater to all forms of CVD. Further, when too many colors are needed, even viewers with full-color vision may struggle. The authors lay out the current primary alternative to color, using point shape, which only works for sparse plotting regions, but most data points in UMAP and tSNE plots are not in sparse regions of the plot. All very true, thus demonstrating the need for a tool like scatterHatch, which can overlay hatch patterns both over regions in dense portions of a scatter plot, and also over points within automatically detected sparsely populated regions. The primary function of scatterHatch produces such plots from a given data frame and the names of columns to use for x, y, and color. The authors go on to demonstrate, with example figures, how the hatch patterns are indeed helpful in cases where color is not enough on its own. They demonstrate that the user can delineate custom hatch patterns, which gives flexibility to the user over how much to rely on hatch patterns versus color. Of particular note, the authors show how scatterHatch can be helpful for readers with monochromatic vision, a population that other visualization tools designed with CVD-accessibility in mind often still fail to aid.

Reviewer #1 (Public Review):

The layered costs and benefits of translational redundancy by Raval et al. aim to investigate the impact of gene copy number redundancy on E. coli fitness, using growth rate in different media as the primary fitness readout. Genes for most tRNAs and the three ribosomal RNAs are present in multiple copies on the E. coli chromosome. The authors ask how alterations in the gene copy number affect the growth rate of E. coli in growth media that support different rates of growth for the wild type.

While it was shown before that mutants with reduced numbers of ribosomal RNA operons grow at reduced rates in rich medium (LB), this study extends these findings and reaches some important conclusions:

1) In a poor medium (supporting slow growth rates), the mutants with fewer rRNA operons actually grow faster than the wild type, showing that redundancy comes at a cost.

2) The same is true for mutants with reduced gene copy number of certain tRNAs and correlates with slower rates of protein synthesis in these mutants.

3) That rRNA operon gene copy number is more decisive for growth rate than any tRNA gene copy number (>1).

In addition, measurements of strains with deletions of genes encoding tRNA-modification enzymes that affect tRNA specificity are included. While interesting, no unifying conclusion could be reached on the impact of these mutations on growth rate.

The well-known "growth law" relationships between growth rate and macromolecular composition (RNA/protein ratio, for example) specifically concern steady-state growth rates. It is concerning that all growth rates in this work were measured on cultures that were only back-diluted 1:100 from overnight LB precultures. That only allows 6-7 doubling times before the preculture OD is reached again. The exponential part of growth would end before that, allowing perhaps only 3-4 generations of growth in the new medium before the growth rate was measured. Thus, the cultures were not in balanced growth ("steady state") when the measurements were made, rather they were presumably in various states of adapting to altered nutrient availability.

A second concern is the use of the term "tRNA expression levels" in the text in Figure 4. I believe the YAMAT-seq method reports on the fractional contribution of a given tRNA to the total tRNA pool. Thus, since the total tRNA pool is larger in fast-growing cells than in slow-growing cells, a given tRNA may be present at a higher absolute concentration in the fast than in the slow-growing cells but will be reported as "higher in poor" in figure 4, if the given tRNA constitutes a smaller fraction of the total tRNA pool in rich than in poor medium. For this reason, the conclusions regarding the effect of growth medium quality on tRNA levels are not justified.

Reviewer #2 (Public Review):

Rava et al. by creating a series of deletion mutants of tRNAs, rRNAs, and tRNA modifying enzymes, have shown the importance of gene copy number redundancy in rich media. Moreover, they successfully showed that having too many tRNAs in poor media can be harmful (for a subset of the examined tRNAs). Below, please find my comments regarding some of the methodologies, conclusions, and controls needed to stratify this manuscript's findings.

Figure 2 presents Rrel as a relative measurement (GRmut/GRwt). Therefore, I'm confused as to how Rrel can be negative, as shown in supplemental file 3 (statistics).<br /> Does Figure 3 show the mean of 4 biological replicates or technical replicates? It should be stated clearly in the legend of figure 3.

Do all strains (datapoint on figure 3 left panel) significantly perform better than the WT in nutrient downshift? Looking at supplemental file 3 I see this is not the case. Please mark the statistically significant points. I suggest giving each set a different symbol/shape and coloring the significant ones in red.

Another issue is that in the statistics of figure 2 (in supplemental file 3), positive values reflect cases where the mutant performs poorly compared to the WT, while in figure 3 the negative values indicate this. Such discrepancy is not very clear. And again, how can Rrel be negative?

Both axes say glycerol. What about galactose?

Lines 414-419: The authors state that "all but one had a growth rate that was comparable to WT (16 strains) or higher than WT (10 strains) after transitioning from rich to poor media (i.e. during a nutrient downshift, note data distribution along the x-axis in Fig 3; Supplementary file 3). In contrast, after a nutrient upshift, 11 strains showed significantly slower growth in one or both pairs of media, and only 2 showed significantly faster growth than WT (note data distribution along the y-axis in Fig 3; Supplementary file 3)".

Looking at the Rrel values when transitioning from TB to Glycerol and vice versa suggests no direction in the effect of reducing redundancy. During downshift, four strains perform better, and three strains perform worse than the WT. During upshift, four stains perform better, and six strains perform worse. Only during downshift and upshift from TB to Gal and vice versa give a strong signal.

The authors should write it clearly in the text because the effect is specific to that transition/conditions and not of general meaning is written in the text (e.g., transition from every rich to every poor media and vice versa). I am convinced that the authors see an actual effect when downshifting or upshifting from TB to galactose and vice versa. In that case, the conclusion is that redundancy is good or bad depending on the conditions one used and not as a general theme.

Also, this is true just for some tRNAs, so I don't think the conclusion is general regarding the question of redundancy.

Figures are indicated differently along the text. Sometimes they are written "figure X", sometimes FigX. Referring to the supplemental figures are also not consistent.<br /> Line 443-444: "In fact, 10 tRNAs were significantly upregulated in the poor medium relative to the rich medium".

This result contradicts the author's hypothesis. If redundancy is bad in poor media because the cells have more tRNAs than they need, the tRNAs level will be downregulated, not upregulated. How do the authors explain this?

Line 445-447: "In contrast (and as expected), all tested tRNA deletion strains had lower expression of focal tRNA isotypes in the rich medium (Fig 4B, left panel), showing that the backup gene copies are not upregulated sufficiently to compensate for the loss of deleted tRNAs".

It is great that the authors validated the expression in their strains. However, for accuracy, please indicate that it was done in four strains to avoid the impression that they did it in all the strains.

Finally, across the manuscript, the authors reveal that deleting some tRNAs or modifying enzymes can be deleterious in rich media or advantageous in poor media. However, I think this result and the conclusions derived from it could be more convincing if the authors would show in a subset of their strains that expressing the deleted tRNAs or modifying enzymes from a plasmid can rescue the phenotype.

Reviewer #3 (Public Review):

In this manuscript, Raval et al. investigated the cost and benefit of maintaining seemingly redundant components of the translation machinery in the E. coli genome. They used systematic deletion of different components of the translation machinery including tRNA genes, tRNA modification enzymes, and ribosomal RNA genes to create a collection of mutant strains with reduced redundancy. Then they measured the effect of the reduced redundancy on cellular fitness by measuring the growth rate of each mutant strain in different growth conditions.

This manuscript beautifully shows how maintaining multiple copies of translation machinery genes such as tRNA or ribosomal RNA is beneficial in a nutrient-rich environment, while it is costly in nutrient-poor environments. Similarly, they show how maintaining parallel pathways such as non-target tRNA which directly decodes a codon versus target tRNA plus tRNA modifying enzymes which enable wobble interactions between a tRNA and a codon have a similar effect in terms of cost and benefit.

Further, the authors show the mechanisms that contribute to the increased or reduced fitness following a reduction in gene copy number by measuring tRNA abundance and translation capacity. This enables them to show how on one hand reduced copy numbers of tRNA genes result in lower tRNA abundance in rich growth media, however in nutrient-limiting media higher copy number leads to increased expression cost which does not lead to an increased translation rate.<br /> Overall, this work beautifully demonstrates the cost and benefits of the seemingly redundant translation machinery components in E. coli.

However, in my opinion, this work's conclusion should be that the seeming redundancy of the translation machinery is not redundant after all. As mentioned by the authors, it is known that tRNA gene copy number is associated with tRNA abundance (Dong et al. 1996, doi: 10.1006/jmbi.1996.0428), this effect is also nicely demonstrated by the authors in the section titled "Gene regulation cannot compensate for loss of tRNA gene copies". Moreover, this work demonstrates how the loss of the seeming redundancy is deleterious in a nutrient-rich environment. Therefore, I believe the experiments presented in this work together with previous works should lead to the conclusion that the multiple gene copies and parallel tRNA decoding pathways are not redundant but rather essential for fast growth in rich environments.

Reviewer #1 (Public Review):

In this paper from Geisberg et al., the authors examined cleavage site usage in yeast and human cells that express Pol II mutants with faster or slower elongation rates. The authors focused on two types of alternative cleavage sites, one being multiple sites clustered within a short range (called within cluster sites) and the other being multiple sites that are distant from one another (called between cluster sites). The authors identified polarity site usage of within a cluster in cells expressing mutant Pol II. Slower Pol II leads to more proximal site usage whereas faster Pol II mutant to distal site usage. In contrast, these trends were not observed with sites between clusters. The authors made four conclusions based on these observations. Overall this is a very well-written paper revealing some fundamental features associated with cleavage site choice. Most conclusions are supported by their data. I do, however, have some concerns about their between-cluster analysis.

Reviewer #2 (Public Review):

In this manuscript, Geisberg et al. present profiles of poly(A) site usage in cells with RNA Polymerase II variants transcribing at different elongation rates. It was known that transcript termination sites in cell populations occur as clusters at the 3'UTR of genes but how the choice of poly(A) site may be influenced by transcription elongation speed was not known.

The strength of their study involves using 3' READ technologies and data analyses that they have previously developed. A weakness of the study is that since the speed of elongation of Pol II is central for the data obtained and conclusions drawn, it would be important to actually measure the speed of elongation by the slow, fast, and wt Pol II used in these studies within the genes analyzed. Although the findings presented in this manuscript are not surprising, they are novel and contribute a missing piece of how the transcription machinery determines which poly(A) site to utilize at the end of genes.

Reviewer #3 (Public Review):

In this study, the authors explore an under-studied but widely observed phenomenon that polyA site selection often occurs in clusters leading to the excepted interpretation that cleavage and polyadenylation are imprecise. Here, the authors use 3READS to map polyA sites in yeast and human cells to define trends in intra-cluster polyA site usage as it relates to RNAPII speed. They observe clear trends in cleavage events that correlate with either increased or decreased RNAPII elongation rate and make a further identification that downstream GC content also correlates with these trends. The potential impact of this work is to explain the imprecise behavior of cleavage and Polyadenylation as a component of local elongation rates that are influenced by nucleotide content.

Reviewer #1 (Public Review):

Gao et al developed various genetic permutations of mouse models of kindlin-2 deficiency in the hepatocytes to explore its function. Hepatocyte-specific loss of kindlin-2 results in severe inflammatory liver injury, accelerated fibrosis/portal hypertension, and massive hepatocyte cell death by apoptosis. These effects are reversed by ablation of TNF signally or by caspase 8 deletion. AAV-mediated replacement of kindlin-2 protects the mice from chemically induced acute liver injury.

Reviewer #2 (Public Review):

As shown in this study, the focal adhesion protein, kindlin-2, plays an essential role in liver function in that its genetic inactivation leads to severe liver fibrosis and death in young mice. This lethality is attributed to activation of TNF-mediated inflammation and caspase-8-dependent cell death since effects of kindlin-2 (Fermt2 gene) knockout can be reversed by genetic inactivation of TNF or caspase signaling. Evidence is also presented that kindlin-2 overexpression can have a mildly protective effect on acute liver toxicity. Overall, this work successfully connects kindlin-2 with normal liver function and raises the possibility that modulation of kindlin-2 could have therapeutic potential for treating liver disease.

On the other hand, the underlying mechanism explaining why kindlin-2 loss stimulates TNF, caspase 8, inflammation, and fibrosis is not explored. As a major component of focal adhesions via its interaction with integrins, kindlin2 has primary functions in regulating cell-ECM signaling and mechanotransduction. However, this study does not connect these known functions with the liver fibrosis and inflammation observed. For example, only cursory analysis is provided concerning the effects of kindlin-2 loss on hepatocyte-ECM interactions, cytoskeletal structure, or focal adhesion distribution. Also, the slightly protective effects of Kindlin-2 overexpression on D-galactosamine/LPS-induced liver toxicity and death are not connected to the rest of the study. Also, one might question whether extending mouse survival by approx. 3-4 hrs with kindlin-2 overexpression is a potentially clinically relevant finding.

brief review:

Some interesting material that one doesn't hear in the broader media. I had wished there would have been more solutions in the end, but honestly simply debunking the broadly held myths by itself is a tall enough order to shift the conversation.

More of the material here should be commonly known and discussed. American can definitely do a better job than sticking to the myths of our past.

I've been reading David Graeber's Bullshit Jobs at the same time as this and the two make interesting companions.

Reviewer #1 (Public Review):

Ruesseler and colleagues combine careful paradigm design, psychophysical and EEG analyses to determine whether information leakage during decision formation is strategically adjusted to meet changing task demands. Participants made motion direction judgments that required monitoring a continuous stream of dot motion for 'response periods' characterised by a sustained period of coherent motion in a leftward or rightward direction. Coherence was modulated on a frame-to-frame basis throughout the task furnishing a parametric regressor that could be used to interrogate the longevity of sensory samples in the decision process and their influence on corresponding EEG signals. Participants completed the task under varying conditions of response period length and frequency. Psychophysical kernel analyses suggest that sensory samples had a more short-lived impact on the participants' choices when response periods were rare, suggestive of greater information leakage. When the stimulus perturbations were regressed against the EEG data, it highlighted a centro-parietal component that showed increased responsiveness to large shifts in evidence when those shifts were more rare, suggestive of a role in representing surprise. An additional triphasic component was found to correlate with the time constant of integration as estimated from the kernel analyses.

This is a very timely paper that addresses an important and difficult-to-address question in the decision-making field - the degree to which information leakage can be strategically adapted to optimise decisions in a task-dependent fashion. The authors apply a sophisticated suite of analyses that are appropriate and yield a range of very interesting observations. The paper centres on analyses of one possible model that hinges on certain assumptions about the nature of the decision process for this task which raises questions about whether leak adjustments are the only possible explanation for the current data. I think the conclusions would be greatly strengthened if they were supported by the application and/or simulation of alternative model structures.

The behavioural trends when comparing blocks with frequent versus rare response periods seem difficult to tally with a change in the leak. The greater leak should result in a reduction in the rate of false alarms yet no significant differences were observed between these two conditions. Meanwhile, false alarms did vary as a function of short/long target durations which did not show any leak effect in the psychophysical kernel analyses. Are there other models that could reproduce such effects? For example, could a model in which the drift rate varies between Rare and Frequent trials do a similar or better job of explaining the data? This ties in to a related query about the nature of the task employed by the authors. Due to the very significant volatility of the stimulus, it seems likely that the participants are not solely making judgments about the presence/absence of coherent motion but also making judgments about its duration (because strong coherent motion frequently occurs in the inter-target intervals). If that is so, then could the Rare condition equate to less evidence because there is an increased probability that an extended period of coherent motion could be an outlier generated from the noise distribution? Note that a drift rate reduction would also be expected to result in fewer hits and slower reaction times, as observed.

Some adjustment of the language used when discussing FAs seems merited. If I have understood correctly, the sensory samples encountered by the participants during the inter-response intervals can at times favour a particular alternative just as strongly (or more strongly) than that encountered during the response interval itself. In that sense, the responses are not necessarily real false alarms because the physical evidence itself does not distinguish the target from the non-target. I don't think this invalidates the authors' approach but I think it should be acknowledged and considered in light of the comment above regarding the nature of the decision process employed on this task.

The authors report that preparatory motor activity over central electrodes reached a larger decision threshold for RARE vs. FREQUENT response periods. It is not clear what identifies this signal as reflecting motor preparation. Did the authors consider using other effector-selective EEG signatures of motor preparation such as beta-band activity which has been used elsewhere to make inferences about decision bounds? Assuming that this central ERP signal does reflect the decision bounds, the observation that it has a larger amplitude at the response on Rare trials appears to directly contradict the kernel analyses which suggest no difference in the cumulative evidence required to trigger commitment.

P11, the "absolute sensory evidence" regressor elicited a triphasic potential over centroparietal electrodes. The first two phases of this component look to have an occipital focus. The third phase has a more centroparietal focus but appears markedly more posterior than the change in evidence component. This raises the question of whether it is safe to assume that they reflect the same process.

Reviewer #2 (Public Review):

In this manuscript, Ruesseler and colleagues use a continuous task to examine how neural correlates of decision-making change when subjects face conditions with different durations and frequencies of occurrence of signals embedded in noise. The authors develop a novel task where subjects must report the direction of relatively sustained (3 or 5 s) signal changes in average coherence of a random dot kinetogram that are intermittent among relatively transient noise fluctuations (<1 s) of motion coherence that is continuous. Subjects adjust their behavior to changes in the duration of signal events and the frequency of their occurrence. The authors estimate a decay time constant of leaky integration of evidence based on the average coherence leading up to decision responses. Interestingly, there is considerable inter-subject variability in decay time constants even under identical conditions. In addition, the average time constants are shorter when signal periods occur more frequently as opposed to when they are more rare. The authors use EEG to find that a component of the Centroparietal Positivity (CPP) regressed to the magnitude of changes in the noise coherence is larger in conditions when the signal periods occur less frequently. Using a control condition, the authors show that this component of the CPP is not simply based on surprise because it is smaller for changes in motion coherence in irrelevant directions with matched statistics as the changes in relevant directions. The authors also find that a different component of the CPP related to the magnitude of the motion coherence co-varies with the inter-subject variability in decay time constants estimated from behavior.

Overall, the authors use a clever experimental design and approach to tackle an important set of questions in the field of decision-making. The manuscript is easy to follow with clear writing. The analyses are well thought-out and generally appropriate for the questions at hand. From these analyses, the authors have a number of intriguing results. So, there is considerable potential and merit in this work. That said, I have a number of important questions and concerns that largely revolve around putting all the pieces together. I describe these below.

1) Quite sensibly, the authors hypothesize that "decay time constant" for past evidence and "decision threshold" would be altered between the different task conditions. They find clear and compelling evidence of behavioral alterations with the conditions. They also have a method to estimate the decay time constant. However, it is unclear to what extent the decision threshold is changing between subjects and conditions, how that might affect the empirical integration kernel, and how well these two factors can together explain the overall changes in behavior.

To be more specific, the authors state that the lower false alarm rates and slower reaction times for the LONG condition are consistent with a more cautious response threshold for LONG. The empirical integration kernels lead to the suggestion that the decay time constant is not changing between SHORT and LONG, while it is changing between FREQUENT and RARE. Does the lack of change in false alarm rate between FREQUENT and RARE imply no change in the decision threshold? Is this consistent with the behavior shown in Figure 2? I would expect that less decay in RARE would have led to more false alarms, higher detection rates, and faster RTs unless the decision threshold also increased (or there was some other additional change to the decision process). The CPP for motor preparatory activity reported in Fig. 5 is also potentially consistent with a change in the decision threshold between RARE and FREQUENT. If the decision threshold is changing, how would that affect the empirical integration kernel? These are important questions on their own and also for interpreting the EEG changes.

2) The authors find an interesting difference in the CPP for the FREQUENT vs RARE conditions where they also show differences in the decay time constant from the empirical integration kernel. As mentioned above, I'm wondering what else may be different between these conditions. Do the authors have any leverage in addressing whether the decision threshold differs? What about other factors that could be important for explaining the CPP difference between conditions? Big picture, the change in CPP becomes increasingly interesting the more tightly it can be tied to a particular change in the decision process.

I'll note that I'm also somewhat skeptical of the statements by the authors that large shifts in evidence are less frequent in the RARE compared to FREQUENT conditions (despite the names) - a central part of their interpretation of the associated CPP change. The FREQUENT condition obviously has more frequent deviations from the baseline, but this is countered to some extent by the experimental design that has reduced the standard deviation of the coherence for these response periods. I think a calculation of overall across-time standard deviation of motion coherence between the RARE and FREQUENT conditions is needed to support these statements, and I couldn't find that calculation reported. The authors could easily do this, so I encourage them to check and report it.

3) The wide range of decay time constants between subjects and the correlation of this with another component of the CPP is also interesting. However, in trying to interpret this change in CPP, I'm wondering what else might be changing in the inter-subject behavior. For instance, it looks like there could be up to 4 fold changes in false alarm rates. Are there other changes as well? Do these correlate with the CPP? Similar to my point above, the changes in CPP across subjects become increasingly interesting the more tightly it can be tied to a particular difference in subject behavior. So, I would encourage the authors to examine this in more depth.

Reviewer #3 (Public Review):

The authors are designing a novel continuous evidence accumulation task to look at neural and behavioral adaptations of continuously changing evidence. They particularly focus on centroparietal EEG potential that has been previously linked with evidence accumulation. This paper provides a novel method and analysis to investigate evidence accumulation in a continuous task set-up.

I am not familiar with either the EEG or evidence accumulation literature, therefore cannot comment on the strength of the findings related to centroparietal EEG in evidence accumulation. I have therefore commented only on the coherence and details of the method and clarity of the argumentation and results.

The main strength is in the task design which is novel and provides an interesting approach to studying continuous evidence accumulation. Because of the continuous nature of the task, the authors design new ways to look at behavioral and neural traces of evidence. The reverse-correlation method looking at the average of past coherence signals enables us to characterize the changes in signal leading to a decision bound and its neural correlate.<br /> By varying the frequency and length of the so-called response period, that the participants have to identify, the method potentially offers rich opportunities to the wider community to look at various aspects of decision-making under sensory uncertainty.

The main weaknesses that I see lie within the description and rigor of the method. The authors refer multiple times to the time constant of the exponential fit to the signal before the decision but do not provide a rigorous method for its calculation and neither a description of the goodness of the fit. The variable names seem to change throughout the text which makes the argumentation confusing to the reader. The figure captions are incomplete and lack clarity.<br /> The authors claim that the method enables continuous analysis of decision-making and evidence accumulation which is true. The analysis of the signals that come prior to the decision provides a rich opportunity to characterize decision bound in this task. The behavioral and neural analyses globally lack clarity and description and thus do not strongly support the claims of the paper. The interpretation of the figures within the figure caption and the lack of a neutral and exhaustive description of what is being shown prevent the claims to be strongly supported.

The continuous nature of the task and the computation of those evidence kernels are valuable methods to look at evidence accumulation that could be of use within the community. However, due to the lack of rigor in the analysis and description of the method, it is hard to know if the current dataset is under-exploited or whether the choice of the parameters for this set of experiment does not enable stronger claims.

Reviewer #1 (Public Review):

The authors provide a comprehensive series of experiments to show that IF promotes rapid hepatocyte proliferation driven by the dual action of systemic FGF15 (intetinally-derived) and localized WNT signaling. Hepatocyte proliferation during periods of IF maintains a steady liver-to-body-mass ratio. This study provides the first example of the dietary influence on adult hepatocyte proliferation and is highly relevant to the putative beneficial effects of IF in multiple chronic diseases. Additionally, it challenges the view that liver tissue is quiescent except in patholgical injury.

Reviewer #2 (Public Review):

The authors set out to study whether there is altered liver regeneration under physiological homeostatic conditions depending on whether an experimental model is offered continuous feeding or intermittently fasted. They report, using a series of murine models in male mice, that hepatic adjustments to fasting/refeeding occur including hyperproliferation of pericentral hepatocytes during a period of relative liver enlargement. It is interesting to note that this occurs 1 week after daily fasting/feeding cycles and appears to occur very quickly following the reintroduction of food. During fasting, they show that the liver shrinks relative to body weight then, as demonstrated by a series of lineage tracing experiments, undergoes relative hyperproliferation, particularly by pericentral hepatocytes. This was shown using an Axin2-based reporter and additionally through zonal analysis or a confetti-multicolored reporter used to trace individual clones. This response appears stable then for upto 3 months. Ideally, additional data showing the liver and body weight individually would help to give an impression of whether the predominant effect is due to changes in body weight or liver weight but it appears implicit that there is an active contraction of liver and hepatocyte size and number during fasting. This is then followed by rapid growth upon refeeding, presumably without major changes in body weight.

It is not clear whether the length of fasting is critical and what the proliferative and metabolic state of the liver is immediately prior to refeeding. It is also unclear whether the relative expansion of pericentral hepatocytes results in an expansion of the pericentral zone or whether these hepatocytes then repopulate other zonal compartments of the liver. They do provide single-cell transcriptomic data which supports the expansion of pericentral transcripts, however, whether this represents a functionally advantageous liver metabolism and how this is achieved remains will be important questions for the future. The link changes to bile acids to altered expression of Cyp7A1, which suggests a role for altered bile acid metabolism in the fasting state. It would be interesting, in the future, to explore whether a liver-to-intestinal feedback loop exists utilising the altered hepatic bile acids occurring during fasting/refeeding to signal back to the intestine for example. This would also then potentially have implications for liver disease states including cholestatic liver diseases.

Mechanistically the authors use hepatocyte-targeted FGF receptor depletion (Klb) or Wnt/b-catenin transcription factor depletion (Tbx3), through efficient adeno-associated viral vector targeting to manipulate these axes combined with hepatocellular FGF overexpression. They demonstrate that the FGF receptor Klb is expressed throughout the lobule and that its global knockout results in the loss of the pericentral proliferative response in fasting/refeeding. It is interesting to note that with the loss of Klb particularly a senescence response occurs in the areas that previously underwent proliferation in response to IF. Similarly, the loss of Klb alters the metabolic rewiring which occurred during the IF response, unlike Tbx3 depletion. Tbx depletion was separately shown to result in a polyploidisation response within the normally diploid pericentral area, consistent with the previous report from this group.

Broadly the authors achieve their aim of both describing the effects resulting from fasting upon liver regenerative biology and also shedding significant insights mechanistically into this process. Overall, these results are highly provocative and raise important questions when interpreting murine studies. These include whether the experimental effect on liver pathophysiology might be explained by or influenced by altered dietary intake as a result of animal husbandry or animal pathology. It will also be interesting in the future what effect broader dietary modifications have on the liver, and other organs, physiologically. These would include but are not limited to a high-fat diet, altered microbiome, variable fasting, and background body habitus. It also has implications for what happens in response to fasting/refeeding during development and the longer-term adaptive responses to this.

Reviewer #1 (Public Review):

This study elucidates a role of EHD2 as a tumor/metastasis promoting protein. Prior work has found varying results indicating that high expression of EHD2 is either associated with good or poor outcomes. In this work the authors find that EHD2 is expressed in both the nucleus and cytoplasm, and that high cytoplasmic to nuclear expression is associated with a poor prognosis. Using WT and either shRNA knockdown or CRISPR KO cells, they show that EHD2 promotes 3D growth, migration and invasion in vitro, and tumor growth and metastasis in vivo. Importantly, re-expression of EHD2 in KO cells rescues the loss of function phenotype. Mechanistically, the investigators show that the loss of EHD2 decreases the calveoli and that this decreases the Orai1/Stim induced calcium influx. Finally, they show that inhibitors of store operated calcium entry (SOCE) phenocopies the loss of EHD2. Together the data support a protumorigenic role for EHD2 via store-operated calcium entry and reinforce the utility of targeting calveoli and SOCE in tumors with high cytosolic EHD2. This study provides a rationale for using SOCE inhibitors in a subset of breast cancers, and a potential predictive biomarker for using SOCE inhibitors based on high expression of EHD2.

Reviewer #2 (Public Review):

The manuscript by Luan et. al. describes the role of EHD2 in promoting breast tumor growth. They showed that EHD2 cytoplasmic staining predicts poor patient outcome. Both EHD2 KO or knockdown cells showed decreased cell migration/invasion abilities and significant reduction of tumor growth and metastasis in mice. The authors further showed that the levels of EHD2 and Cav1/2 correlate with each other. EHD2 KO cells showed defects on Ca2+ trafficking. Overexpressing the SOCE factor STIM1 partially rescued SOCE defects in EHD2 KO cells. Treatment of the SOCE inhibitor SKF96365 inhibited tumor cell migration in vitro and tumor growth in vivo.

Major strengths:<br /> The authors showed that EHD2 cytoplasmic levels predict patient survival and provided strong evidence that EHD2 knockout or knockdown inhibits tumor cell migration in vitro and tumor growth in vivo. The authors also showed that SKF96365, which inhibits SOCE, suppresses tumor growth in vivo.

Major weaknesses:<br /> The connection between EHD2 and SOCE is weak.

Reviewer #1 (Public Review):

The authors convincingly show directionally tuned signals in AD and RSC. RSC is found to have a lower proportion of HD cells than AD, and RSC HD cells are more sensitive to angular velocity than AD HD cells. Importantly, HD responses are shown to be tightly correlated between the two areas. Population decoding of head direction, performed on AD neuron ensembles or RSC ensembles, revealed similar shifts following visual cue rotation and also similar HD drift in darkness, indicating that the HD representation across both areas is coordinated. The study further finds that AD-to-RSC connections are relatively frequent, while RSC-to-AD connectivity is very sparse. This asymmetry in functional connectivity is matched by viral tracing results. Together, the results lead the authors to the conclusion that this corticothalamic connection is likely not driving visual landmark updating of the global head direction system.

This is a welcome piece of work, providing the first assessment of the high degree of coherence between AD and RSC HD representations, using pairwise and population-level analysis methods, which had not been accessible before. It will be a valuable reference for researchers interested in inter-area interactions in the head direction system, leaving the question of how and where visual reference updates are fed into the HD circuit open for further investigations.

Reviewer #2 (Public Review):

Van der Goes et al recorded HD cells in the retrosplenial cortex and anterodorsal nucleus of the mouse during the rotation of a prominent visual cue. They describe the temporal coordination of the HD representation between the two structures, also in the dark condition. They provide evidence for a near-simultaneous realignment of the HD representation in the two structures (no consistent temporal offset during the cue shift). This finding is interesting and quite surprising, in light of the existing literature postulating a role of the retrosplenial cortex in a binding visual landmark and HD information. I am not sure whether the authors' conclusions are convincingly supported by the data.

Reviewer #3 (Public Review):

The work provides direct evidence for the coherent activity of head-direction (HD) cells in the anterior thalamus and retrosplenial cortex (RSC). RSC is one of two major direct cortical recipients of the subcortical HD signal, the other being the postsubiculum (POS). While it is established that POS inherits its HD tuning from ADN (Peyrache et al, 2015), it is not known whether HD cells in RSC show similar coordination with ADN. The manuscript employs technically challenging dual electrophysiological recordings from ADN and RSC to establish that the local internal representations of HD encoded in ADN and RSC are coherent during free exploration but also show coordinated realignment after cue rotation as well as coordinated drift in darkness. The work thus provides evidence that HD and RSC assemblies represent the same internal heading direction, at least in the behavioural paradigms tested and at the investigated temporal resolution. The manuscript also makes a claim that the RSC is unlikely to mediate the realignment of the HD signal following cue rotation because the HD signal realigns itself synchronously across the two brain regions. This claim is additionally supported by the sparse anatomical projection and the paucity of putative direct synaptic connections from RSC to ADN.

The manuscript convincingly demonstrates overall ADN-RSC coordination in two different paradigms. While such coordination is expected in instances when HD representations in both areas are precisely aligned with the current HD, it may not be the case in instances of sensory conflict or limited sensory information. The fact that internal HD in both ADN and RSC drifts coherently in darkness provides strong evidence of the tight functional coupling between the two areas. Additionally, while the cue rotation paradigm used in the study often failed to elicit the full realignment of the HD signal, this variability was certainly utilized to the manuscript's advantage as it makes the coupling evident even when the HD signal realigns only partially. The overall conclusions of the manuscript are largely supported by the presented data but the strength of the argument, especially with regard to the zero-lag coupling between ADN and RSC, is somewhat affected by the technical limitations.

1) The manuscript relies heavily on supervised decoding of the internal HD from population activity in RSC and ADN and in turn suffers from relatively low numbers of simultaneously recorded neurons, which is especially evident in the representative images in Figure 2C. The reported average decoding errors are much higher than those reported elsewhere (Peyrache et al, 2015; Viejo et al, 2018; Xu et al, 2019), which may occlude the effects of RSC activity on ADN that are more subtle and/or occur at shorter timescales than the bin size used in the decoding algorithm. The manuscript includes no discussion of how much these factors could contribute to the observed variability in the data.

2) RSC-HD cells recorded in the study are relatively poorly tuned to HD, which is contrary to the reports of HD cells recorded in RSC (Lozano et al, 2017; Javob et al, 2017; Keshavarzi et al, 2021). In fact, the median directional information score for RSC-HD cells is the same as that for non-HD cells in ADN (Supplementary Figure 2B). In fact, due to their relatively low HD modulation, it may be more appropriate to refer to them as 'HD-modulated' cells. While the electrode positions indicate that RSC was sampled across layers and sub-regions so missing the HD cell 'hot spots' like granular RSCb is unlikely, the apparent poor directional tuning of RSC cells could possibly be due to the nature of the recording environment (e.g. low light condition with the LED landmark being the only light source). Importantly, the manuscript lacks a control 'baseline' condition in which HD cells are recorded in a standard, well-lit open field, as well as a discussion of the discrepancy between the observed HD tuning and that reported in the literature.

3) Analysis of decoding error, which features prominently in the manuscript, is critically dependent on the quality of behavioural tracking - errors in tracking could lead to the accumulation of decoding errors and this could dominate decoding error analyses. Indeed, Figure 2A shows many gaps in the tracked HD of the mouse, which may point to the sub-optimal quality of the behavioural tracking. This is especially important for analyses like the one in Figure 2D which shows that internal HD representations in ADN and RSC are coordinated at zero lag (+/- 20ms). The observed zero-lag peak could be instead explained by errors in behavioural tracking dominating the analysis, which would affect both representations simultaneously and show spurious zero-lag positive correlations. As such, the analysis that is missing is the difference between internal HD decoded from ADN and RSC at different time lags, without reference to the HD tracked behaviourally.

4) The work often uses a number of trials as their 'n' sample size for statistical analyses and the methods state that tetrodes were regularly advanced, but there is no indication of whether multiple trials at the same tetrode position were included in the same statistical comparison (except for recordings '4 days apart' for the HD tuning and synaptic connectivity analyses). Multiple trials with a high likelihood of recording the same cell population should not be counted as separate samples when calculating statistical significance.

Reviewer #1 (Public Review):

This manuscript by the Karakas lab reports on new structures of the volume regulated LRRC8 anion channels. These ubiquitously expressed channels play key roles in cell volume regulation and in allowing efflux of organic osmolytes, neurotransmitters, and drugs. In addition to regulating cell volume LRRC8 channels might play roles in signal transduction, cell migration, apoptosis, tumor drug resistance, and stroke. Thus, elucidating their architecture and structure is of critical importance. LRRC8 channels are obligate multimers of variable stoichiometry, with the LRRC8A subunit being absolutely required for assembly of functional channels. Structures of homomeric LRRC8A and LRRC8D channels revealed a hexameric assembly with closed pores. However, the functional properties of these homomeric channels differ from those of recorded in cells, raising questions on the physiological relevance of these conformations. The authors here determine the structure of a LRRC8C-LRRC8A chimera (termed 8C-8A(IL125)) with functional properties that closely resemble those of native channels. Remarkably, the 8C-8A(IL125) chimera assembles as a heptamer with a large pore. Unexpectedly, in the structures the channel's pore is occupied by density that could correspond to lipids.

Reviewer #2 (Public Review):

Volume-regulated anion channels (VRACs), comprised of the LRRC8 family of proteins, play important roles in cell volume regulation. Physiological LRRC8 channels are heteromeric assemblies of LRRC8A and LRRC8B-LRRC8E subunits. Previous structural studies have focused on homomeric channels, which do not recapitulate functional properties of native heteromeric channels. Thus, the molecular basis of physiological VRAC assembly and function remains unknown. In this study, Takahashi and colleagues present the single-particle cryo-electron microscopy structure of a functional LRRC8 chimera, which is composed of LRRC8C and a swapped intracellular loop from LRRC8A. Surprisingly, the chimeric channel forms a heptamer, in sharp contrast to the previously reported hexamers of homomeric and heteromeric LRRC8 channels. The findings of the chimeric channel are interesting. However, the physiological implication of this chimera is unclear, and the proposal that native LRRC8 channels are heptamers is not well supported.

Reviewer #3 (Public Review):

This manuscript by Takahashi et al., reveals the structure of a chimeric VRAC channel composed by the LRRC8C and a short domain corresponding to the intracellular loop of LRRC8A. Homomeric LRRC8C channels are not functional but this chimera has been shown to "rescue" the functional and pharmacological properties of heteromeric VRAC channels. The authors obtained the Cryo-EM structure for this chimera, which provide some interesting insights about these channels. The major finding of this work is that the channel is asymmetrically formed by 7 protomers, with associated lipid-like densities that are proposed to play a role in gating. Unfortunately, critical domains of this structure could not be solved, which limit the interpretations of this new work. These missing domains include the entire LRRD, the N-terminus and the first intracellular loop containing the 25 amino acids incorporated from the LRRC8A. While this work is very interesting, the data presented are not enough to support the author claims. Particularly, the idea that the 'lipid blocked pore' is associated with gating.

Reviewer #1 (Public Review):

In 2020, Sugisawa et al. reported that Piezo1ion channels can be activated by ssRNAs, both synthetic and derived from fecal matter, suggesting that these may be the first identified natural ligands to agonize Piezo channels. Nickolls et al., provide a careful and rigorous investigation of the effect of ssRNAs and fecal extracts on Piezo channel activity in three cell lines, using both calcium imaging and electrophysiology. They find that Piezo1 is not responsive to ssRNAs nor responsible for calcium flux in response to fecal extracts in HEK293 and RIN14b cells. Overall, this study addresses the question of ssRNAs as a Piezo ligand clearly and thoroughly, with rigorous, well-controlled experiments. Overall, I am excited about this study as a necessary clarification for the field of Piezo mechanosensation.

Reviewer #2 (Public Review):

In the present study, the authors have combined calcium imaging and electrophysiology to systematically replicate the previously reported finding that the mechanical activation ion channel Piezo1 might also serve as a gut RNA sensor. The authors have employed multiple cell lines, knockout of endogenous Piezo1, and heterologous overexpression of Piezo1, Yoda1 as a positive control for chemical activation of Piezo1, and similar dosage of ssRNA used in the previous study, but clearly did not replicate the finding that ssRNA can specifically activate Piezo1. The experiments have been well designed and data quality is high. The data support the conclusion that Piezo1 is not a receptor for ssRNA in the gut.

Reviewer #3 (Public Review):

This study investigates the recently published findings by Sugisawa et al that microbial ssRNA40, a known agonist for the immune surveillance system activates the mechanically gated ion channel Piezo1. In addition to providing mechanistic insights into the study, this finding also had much broader implications as it suggested a novel role for the channel as a physiological receptor for ssRNA. Although there is nothing that prevents Piezo1 from carrying out such a role in principle, the finding caused a great deal of interest and professional skepticism among Piezo researchers and, more broadly, in the field of mechanobiology. This manuscript set out to reproduce the main findings of Sugisawa et al using the same approaches, and in addition utilized other techniques to address potential differences in experimental conditions. In summary, the authors failed to reproduce the major Piezo1-related findings reported by Sugisawa et al, while all the new experiments pointed to the absence of a functional interaction between ssRNA40 and Piezo1. The study is well-designed, with appropriate controls and statistical analyses.

Reviewer #1 (Public Review):

In this manuscript, Zhang and colleagues created a transgenic mouse strain that expresses SYT-1-tdt in all neurons. They showed that the labelled SYT-1 colocalizes with multiple synaptic markers and label synapses in different regions. More importantly, they showed that the transgenic expression does not alter synaptic function using ephys assays. This is a straightforward paper that generated a useful reagent that will be used broadly.

Reviewer #2 (Public Review):

Yang et al. produced a transgenic mouse line (Syt1-TDT) that could be used for labeling both excitatory and inhibitory synaptic sites in cultured neurons and in vivo neurons. The strength of the current study is to provide a series of thorough analyses to claim the applicability of this mouse line in the relevant neuroscience research field(s). The weakness is the potential impact/usefulness of this mouse line. To strengthen the merit of this mouse line, the authors should present evidence showing its advantage over other similar genetic approaches.

Reviewer #3 (Public Review):

Yang and colleagues provide a thorough characterization of a transgenic mouse model expressing fluorescently tagged synaptotagmin. In particular, they present key controls validating this mouse model as a tool, including co-localization of the tagged synaptotagmin with other synaptic markers as well as normalcy of synaptic transmission mediated by synaptic terminals expressing the tagged synaptotagmin. Importantly, the authors present data on the potential use of neuronal cultures obtained from these mice in synaptic co-culture assays. In these assays, synaptic cell adhesion molecules expressed on non-neuronal cell lines such as HEK-293 cells or COS cells are used to test the sufficiency of these molecules to trigger synapse assembly. This mouse model will be a useful addition to existing models expressing fluorescently-tagged synaptic vesicle proteins such as synaptophysin, synaptotagmin as well as synaptobrevin.

Reviewer #1 (Public Review):

This is an interesting and timely paper investigating the impact on participation in cancer screening programs across Italy during the COVID-19 pandemic where there was massive disruptions to health services. What is of particular interest in this analysis was the investigation of social, educational and cultural factors that might have impacted access and participation to screening.

- In the present study, the authors analyzed data collected by PASSI between 2017 and 2021, from interviews of more than 106,000 people, a representative sample of the Italian population aged 25-69 was selected but its not clear what was the representativeness by region, gender and age educational attainment? Also what is the total population (so I don't have to look it up). I am wondering if participation differed by characteristics and what approach to achieving the representative sample was made (e.g. replacement of individuals or oversampling certain strata where participation was lower).

- For figures 5-8 what is the N for the different groups not just the %?

- Table 2 to me is a key piece of information and very interesting can the authors formally test if there are signficant differences between the time periods?

Reviewer #2 (Public Review):

Giorgi Rossi et al measured in their paper the impact of COVID-19 pandemic on the main indicators used to assess the performance of national screening programs for cancers. As expected, they highlighted a significant reduction that changed during the different waves and also across geographical areas. The results of the study might be considered valid and representative as the study is relied on current data flows to assess the performance of screening programs. The paper also reports a complementary analysis on the factor associated to the access to screening that gives some more insights on the reasons behind the access. This second part of the work also relied on data collected at national level that anyway have some intrinsic limitations. Nevertheless, on the whole, the paper gives a useful contribution to the assessment of the disruption due to the pandemic that can be also used in the light of preparedness actions.

Reviewer #3 (Public Review):

This study is primarily a descriptive analysis that provides a clear and accessible account of how screening activity varied across Italy and between groups. While primarily a simple descriptive account such work is important to document what were the impacts of the pandemic on preventative health services and to understand how they differed across groups. The combination of survey responses from regional screening programmes and individuals is a useful use of two data sources. The study is very clearly written and does not over-interpret the presented data.

The methods description states that the analysis presents the "standard months" required for the programmes to recover from the service delays. The subsequent reporting of these delays in the results section did not use the same terminology and I see scope for clarification by using common language regarding this assessment throughout the paper. I see scope for further disaggregation of the regional results within the study but equally I understand why the authors might not wish to report outcomes for specific regions. I see scope for improvement in the figures within the manuscript but this is a relatively presentational matter. I would like to see some further description of the Poisson regression analysis as what is included within the manuscript appears rather brief. There is also one section of the methods that seems as if it would better belong in the introduction, but overall the manuscript was very clearly structured.

The analysis presented achieves the authors' stated aims in my view. I see a useful contribution in documenting the impact of the COVID-19 pandemic on screening in Italy. This may inform further work on assessing the eventual health impact of delays as well as work considering how best to make screening programmes more resilient to such shocks. Ultimately it will take time to observe just how significant the impacts of service interruptions were on cancer prevention. Readers should remember that many screening services may still provide good protection against cancer as long as the interruptions are limited to simply to delays in coverage rather than the longer-term loss of participation, especially for those with incomplete screening histories or of otherwise elevated risk of disease.

Further work may wish to consider how programmes prioritised capacity or what efforts have been made to restart screening. Similarly, there is scope for more detailed disaggregation assessment of who received screening as programmes restarted. Both these issues are beyond the scope of the present study however. The present submission provides a good basis for any further such exploration.

Reviewer #1 (Public Review):

The authors introduce an online tool, CausalCell, to explore causal links in single-cell datasets. The authors investigate the process through examples based on existing data, offer comparisons of different algorithms, and suggest tips about the requirements and limitations of this approach. In my opinion, the main shortcoming is that the authors do not adequately justify whether the methods included in their tool are the most suitable methods for their intended analyses. The lack of a definite "ground truth" or "gold standard" also comes in the way clearly deciding which algorithms perform the best, especially when there are considerable differences between the results of different algorithms.

Reviewer #2 (Public Review):

Wen et al. developed a useful tool for causal network inference based on scRNA-seq data. The authors comprehensively benchmarked 9 feature selection and 9 causal discovery algorithms using both synthetic data and real scRNA-seq data. Their conclusions regarding the performance of these algorithms on synthetic data are solid and valuable. I believe this tool or platform has the potential to help biologists discover novel cell type-specific signaling pathways or gene regulatory events since there is no prior knowledge (such as known pathway annotations) as inputs. However, several major concerns below need to be addressed to improve the paper.

(1) Current validation of the inferred causal networks using real scRNA-seq datasets seems quite simple and is not sufficient to support the accuracy and reliability of results. Annotations from the STRING database do not contain directions of edges among genes or proteins. However, the edge direction in the inferred network is a crucial aspect to explain the causal relationships. Besides using "spike-in" data, a systematic validation of the inferred network, especially the edge directions, should be provided.

(2) In order to illustrate the novel discovery, CausalCell should be further compared to existing gene network construction methods based on scRNA-seq data such as SCENIC (Aibar et al. Nature Methods, 2017).

(3) The authors should also claim what type of the inferred causal network represent from the biological perspective (e.g. signaling networks or gene regulatory networks?).

(4) Besides edge direction, an important feature of CausalCell is the determination of edge sign (i.e. activation or inhibition). The authors should describe its related procedures.

(5) The authors did not provide an example of constructing a causal network between cells or cell types, although they mentioned its importance in the Abstract. Such intercellular network examples can distinguish the utility of CausalCell in single-cell data analysis from bulk data analysis.

(6) If the control dataset is available, it is currently not clear whether batch effects of the query and control datasets will be removed in the data pre-processing step. Differentially expressed genes cannot be selected correctly if batch effects exist.

Reviewer #1 (Public Review):

The authors conducted a thorough analysis of the correlation between height and measures of cognitive abilities (what are essentially IQ test components) across four cohorts of children and adolescents in the UK measured between 1957 and 2018. The authors find the strength of the association between height and cognitive measures declined over this time frame--for example, among 10- and 11-year-olds born in 1958, height explained roughly 3% of the variation in verbal reasoning scores; this dropped to approximately 0.6% among those born in 2001. These associations were further attenuated after accounting for proxy measures of social class.

The authors' analyses were performed carefully and their observations regarding declining height / cognitive measure associations are likely to be robust if we interpret their results with an important caveat: these results reflect measurements aimed at assessing cognition rather than cognition itself. The importance of this distinction is evidenced by the changing correlation structure of the cognitive measures over time. For example, age 11 verbal / math scores were correlated at >= 0.75 at the first two time points but dropped to 0.33 at the most recent time point. Similar patterns are present for the other cognitive measures and time points. The authors' conclude that such changes are unlikely to impact their primary findings, but I'm less certain. For example, one interpretation of this finding is that older cognitive measures were simply worse at indexing distinct cognitive domains and instead reflected a combination of cognitive ability together with non-specific factors relating to opportunity, health, class, etc. Further, height was historically a stronger proxy for class and economic status than it is today (e.g., by capturing adequate nutritional intake, risk for childhood disease, etc.). Together, then, previously high height / cognitive measure correlations might reflect the fact that both phenotypes previously indexed socio-economic factors to a greater extent than they might today (which is still non-negligible).

Additionally, their findings add an interesting data point to a collection of recent results suggesting that the relationship between cognitive and anthropometric measures is complex and difficult to interpret. For example, studies using genetic markers to examine shared genetic bases have virtually all relied on methods assuming mating is random, which is not the case empirically. Howe et al. (doi.org/10.1038/s41588-022-01062-7) recently reported that the ostensible genetic correlation of -.32 between years of education and BMI attenuates to -.05 when using direct-effect estimates, which should theoretically be immune to the effects of non-random mating and other confounding variables. Likewise, Keller et al. (doi.org/10.1371/journal.pgen.1003451) and Border et al. (doi.org/10.1101/2022.03.21.485215) used very different approaches to arrive at the same conclusion that ~50% of the nominal genetic correlation between IQ and height could be attributed to bivariate assortative mating rather than shared causal biological factors. Given that assortative mating on both IQ measures and height involves many other traits (not just two as assumed in such bivariate models), the true extent to which height / IQ correlations reflect causal factors is plausibly even lower than these estimates suggest. For these reasons, I do not entirely agree with the authors' review of previous findings in the introduction, where they write "recent studies have suggested that links between higher cognition and taller height can be largely explained by genetic factors", though it is certainly true that this claim has been made.

Reviewer #2 (Public Review):

The authors use birth cohorts with extensive cognitive assessments and height measurements along with data on parental height and socioeconomic status. The authors estimate that the correlation between height and cognitive ability has approximately halved in the last 60 years.

Quantile regression results suggest that this is due to a stronger association between low cognitive ability and short stature in older cohorts, potentially due to environmental factors that cause both and that have been removed by improvements in the environment in the last 60 years.

While this is a plausible hypothesis, the evidence presented in the manuscript is unable to rule out alternative hypotheses, such as changes in assortative mating.

The results in the manuscript will be of interest to researchers investigating how genetics and environment lead to correlations between cognitive and physical/health traits, and to researchers interested in the relationship between social and health inequalities.

While my sense of the evidence presented is that there is fairly solid statistical evidence for a trend where the correlation between cognitive ability and height declines over time, there is no formal quantification of this trend nor measurement of the uncertainty in the trend.

Similarly, the quantile regression plots in Figure 2 appear to show a trend across the height deciles for the two oldest cohorts, but no quantification of how strong this is nor what uncertainty exists is calculated. Furthermore, if the apparent trend in the quantile regression plots is true, wouldn't this imply a non-linear association between height and cognitive ability for the older cohorts? Can this be seen in the scatterplots or in a non-linear regression?

I think the authors could have done more with their data to investigate the contribution of assortative mating to the observed trend. Looking at Figure S4, it looks like the correlation between mother's education and father's height in the 2001 cohort is substantially lower than for previous cohorts. While cognitive ability may not be available for parents, one could look at, for example, father's education and mother's height across the cohorts and see if there is a downward trend in correlation.

Reviewer #3 (Public Review):

A difficulty with the paper is the different cognitive tests used in the different cohorts; the authors address this at some length in the discussion. However, I am afraid that this matter makes the results hard or impossible to interpret along the lines of their research question. One would need to know that, if these cognitive tests were administered in a single cohort at one time, they would have the same correlation with height.

I judge that the main limitation of the method is the fact that different cognitive tests are used in the different cohorts. The tests in themselves are valid tests of cognitive functions. However, given that the focus of the study is on the change in correlations across time, then it is a worry that the tests are different; that is, the authors have the burden of proving to us that, if the environmental/social changes had NOT been operative across time, then the height-cognitive test correlations would be the same. What can the authors do to prove to us that if, say, all of these different-cohort verbal tests had been given to a single cohort on a single occasion, then they would have the same correlations with height? The same goes for the mathematics based tests. I note the tests' somewhat different distributions in Figure 1, but that is not the only thing that could lead to different correlations with, say, height. I am aware that all cognitive tests tend to correlate positively and that they all have loadings on general intelligence; however, different tests will not necessarily have the same correlations with outside variables (e.g. height). This will depend on things such as their content, their reliability/internal consistency etc.

In the Results the authors state: "Cognitive test scores were strongly-moderately positively correlated with each other, with the size of the correlation weakening across time." That's true, but perhaps, also a major concern for this study. One possible reason for the decline in verbal-maths test correlations across cohorts (old to recent) is that the nature of these tests has changed across time, either/both in terms of content (what capabilities are assessed) or something such as reliability/internal consistency/ceiling-or-floor effects (how well the capabilities are assessed). That is, given that the height-cognitive test correlations show a similarly declining pattern of correlations over cohorts, it could be that the tests' contents (of the different tests) is partly or wholly responsible. I raise that as a possibility only, and I appreciate that it might be correct, as the authors prefer, that there is an inherent lowering of intelligence-height correlations over time, but I do not think that one can rule out-with the present study's design-that it might have been due to the change in tests. For example, a reading-math correlation of 0.74 in 1946 lowered to a correlation of .32 in 2001, in the face of different tests. To show that this is not due to the different tests being used would require more information. If this is a true result, it is big news.

I have a suggestion: if the authors wish to rule out the possibility that the lowering intelligence-height correlations across cohorts are due to different cognitive tests being used, they should take all the cognitive tests used here and apply them cross-sectionally to single-year-born samples (of 11- and 16-year olds) that have also been measured for height. If the cognitive tests all correlate at the same level with height within each of these two samples (they needn't do so across the 11- and 16-year olds), then one could proceed more safely with between-cohorts (1946, 1958, 1970, 2001) comparisons of the correlations.

Reviewer #1 (Public Review):

Li et al. have designed a study that examines specific mechanisms for how different DNA sequence variants in the common cancer gene p53 (also known as TP53) influence the sensitivity of tumors to a variety of common cancer treatments. Specifically, they examine a handful of p53 variants with respect to glioblastoma and its response to platinum-based chemotherapy and to radiation therapy. The authors begin by mentioning that looking at DNA variants in cancer is useful but also incomplete: methylation, PTMs, and non-DNA sequence variants can also be critical. They then mention that they have created a model showing that nearly all cancers with p53 mutations have loss-of-function variants and that many cancers with "normal" wildtype p53 in fact have variants causing LOF. These p53 LOF tumors lead to worse patient outcomes, but the authors here show that these tumors appear to be more susceptible to radiation and platinum-based chemotherapy, which they say they have validated in glioblastoma xenografts. This potentially opens up a new avenue for precision medicine for many different sources of cancer that share common p53 LOF variants.

The authors have taken a modern approach towards cancer diagnosis and shown how this can improve targeted treatments across a large array of cancer types. They have provided a reasonably convincing proof of concept of this approach for n = 35 PDXs in one cancer type. By and large, the approach and results are reasonable, although many of the exact results concerning the genes and pathways identified that covary with the various treatments and p53 variants are unclear. For instance, the feature selection seems to be somewhat ad hoc, e.g. the method used to determine p53 LOF from p53 WT in the TCGA data was not the same method used for determining p53 LOF from p53 WT in the PDX data. The TCGA AUROCs were incredibly good - over 99% - versus more like 75% for the actual proof of concept. While any significant p-value is fine for basic research, it would be nice to know how this could be improved and bring the results in Figure 4 from ~75% to the >99% that would be necessary for use as a medical diagnostic or for treatment selection for precision medicine. However, there are significant questions regarding the specific findings uncovered: do the gene pathways identified through bioinformatic analysis fit in with the many highly-studied mechanistic roles of p53? Do the cohort selections - which vary by an order of magnitude in sample size, and come from different locations and different tissues - make statistical sense for cross-validation?

Reviewer #2 (Public Review):

The Tp53 gene is deemed as one of the most critical tumor suppressors in humans. Not surprisingly, the latter is found inactivated or mutated in the majority (if not all) of human cancers. The present study by Q. Li et al describes an attempt to predict the functional status of p53 in those tumors where no mutations on the DNA sequencing level were identified. To this end, the authors employed SVM models to train the algorithm for the detection of the 'p53 inactivation' features using normal and tumor tissues, respectively. It turned out that the 'p53 loss of function' phenotype was associated not with DNA methylation but rather with yet unknown mechanisms. Based on the fact that the p53LoF-containing tumors are similar to the p53 mutant-expressing ones with respect to platinum-based therapy, they subsequently used their SVM model on the glioblastoma samples to predict their chemosensitivity.

Reviewer #1 (Public Review):

NADPH oxidases are a family of membrane enzymes that produce reactive oxygen species (ROS). NOX2 is the most well-studied member of the NADPH oxidase family, and the proper function of NOX2 is critical for innate immunity against pathogens in mammals.

The study by Dr. Chen and colleagues used antibodies to facilitate the structural determination of the high-resolution structure of the NOX2-p22 complex, which is otherwise challenging for single-particle analysis due to its flexibility and relatively small molecular weight. The work uncovered the high-resolution information between NOX2-p22 interaction and conformational flexibility between the DH domain and the transmembrane domain of NOX2. This structural study provides valuable information for a mechanistic understanding of NOX2 activation at the molecular level.

The weakness of the paper is the lack of in-depth analyses regarding structural discoveries. In addition, a study by Noreng S et al on the structure of the NOX2-p22 complex is now available.

Reviewer #2 (Public Review):

The structure was solved in its resting (i.e. non-activated) form and was stabilized by adding an antibody that recognizes an extracellular epitope. The protein - the complex of NOX2-p22 bound to the antibody- was reconstituted from proteins expressed in human cells through baculovirus transduction. The cryoEM gridswere obtained by using nanodisc-embedded complexes. The structure clarifies the topology of the p22 subunit, showing that it comprises four transmembrane helices. Moreover, it confirms that the oxygen-reacting center is conserved among NOXs implying a similar mechanism for ROS generation. Furthermore, the 3D structure explains the effect of the many known disease-causing mutations. They mostly affect the active sites or the NOX2-p22 subunit-subunit interface. The cytosolic dehydrogenase domain is not as ordered in the cryoEM maps. Clearly, NOX2 is a highly dynamic protein where the cytosolic and membrane domains can enjoy considerable flexibility. This feature very likely underpins the mechanism of activation, which is triggered by the cytosolic subunits and remains to be understood. The manuscript suggests that the cytosolic subunits might stabilize the enzyme in the conformation that is capable of conducting electrons from the NADP-flavin site to the inner heme, thereby enabling catalysis.

Overall, this is great experimental work: the structure of NOX2 has been awaited for a long time. The data reported in this manuscript should probably be seen as the beginning of the NOX2 structural era. Indeed, a lot remains to be clarified, especially with regard to NADPH binding and the mechanism of enzyme activation. Along this line, the manuscript reads more as a preliminary report rather than a full-story manuscript. Beside this general concern, I do not have any specific comment about the presentation style: the manuscript is clearly written and nicely illustrated.

Reviewer #3 (Public Review):

This manuscript will be of interest primarily to researchers in the field of NADPH oxidases (NOXs) but also to those interested in the wider ferric reductase superfamily, also comprising members of the six-transmembrane epithelial antigen of the prostate enzymes (STEAPs). More limited interest may be expressed by investigators of ferredoxin - NADP reductases, resembling the dehydrogenase region (DH) of NOXs, expressing lesser "visibility" in the structure described in the paper. Considering the fact that NOXs are essentially electron transport machines from NADPH to dioxygen, along a multi-step redox cascade, those interested in hydride and electron transfer, at a more conceptual level, might also want to have a look at the paper. Elucidating structures of NOXs are still rare achievements, with only four published papers, so far (one coming from the group of the present main author) and, thus, any new publication profits from the aura of novelty.

Introduction<br /> This manuscript offers a detailed and in depth description of the structure of the catalytic core of the human phagocyte NADPH oxidase, NOX2, in heterodimeric association with the protein p22phox. The phagocyte NADPH oxidase is responsible for the production of reactive oxygen species (ROS), the primary molecule of which is the superoxide radical (O2.-), derived by the one-electron reduction of molecular oxygen by NADPH. NOX2 belongs to the NOX family, consisting of 7 members (NOX 1-5, and DUOX1 and DUOX2), sharing common structural characteristics but expressing a wide variety of functions. The principal but not the only function of NOX2 is as a source of ROS for the killing of pathogenic microorganisms (bacteria, fungi, protozoa) engulfed by phagocytes in the course of innate and acquired immunity.

The structures of C. stagnale NOX5, and that of murine and human DUOX1 were determined by X-ray crystallography (NOX5) and cryo-EM (DUOX1). As sources of potentially dangerous auto-toxic ROS, NOXs are subject to strict functional regulation. Whereas Nox5 and the DUOXs are regulated by Ca2+, NOXs 1, 2, and 3 are regulated by several cytosolic proteins, that associate with the Nox2-p22phox dimer forming the active O2.-generating complex. The paramount model of cytosolic regulation is Nox2 and the "dream" of structure investigators is to elucidate the structure of NOX2 in both resting and activated states.

Achievements<br /> Note: When this paper was received for review, this reviewer was not aware of any publication dealing with the structure of human Nox2. However, on October 14, 2022 a paper was published on line, dealing with the structure of Nox2 (S. Noreng et al., Structure of the core human NADPH oxidase Nox2, Nature Communications (2022)13:6079). This review will not discuss the present manuscript in relation to the paper by S. Noreng et al.

This manuscript is successful in describing the structure of the NOX2-p22phox heterodimer using cryo-EM methodology. In order to compensate for the small size of the complex, use was made of the Fab of a monoclonal anti-Nox2 antibody binding an anti-light chain tagged nanobody. In order to mimic as much as possible the milieu of NOX2-p22phox in the phagocyte membrane bilayer, the authors reconstitute the quaternary complex in a nanodisc, using soybean phosphatidylcholine (PC) and a membrane scaffold protein (MSP). To the best of my knowledge, this is the first report of studying a NOX in a nanodisc, for both function and structure. Peptidiscs were used in determining the structure of human DUOX1 by a group led by the main author of this paper, but nanodiscs offer the advantage of adding a phospholipid chosen by the investigator. The purified nanodiscs incorporating the quaternary complex led to successful structure determination of the transmembrane domain (TMD), extracellular and intracellular loops, inner and outer hemes, distances between hemes and FAD to inner heme, and a hydrophilic tunnel connecting the exterior of the cell to the oxygen-reducing center of NOX2. The structure of the dehydrogenase region (DH) was less well defined; the FAD-binding domain (FBD) was more visible than the NADPH-binding domain (NBD). The structure of p22phox and the interface between Nox2 and p22phox are well described.

The mutations in NOX2 and p22phox causative of the deficient bactericidal function in Chronic Granulomatous Disease are related in detail to the location and role of the mutated residues as revealed by the solved structure.<br /> The authors make it clear that the structure, as presented, is in the resting state. The distances between hemes are suitable for electron transfer but the distance between FAD, in the FBD, and the inner heme is too large for transfer. The poor quality of the obtained structure of the DH (especially, the NBD), even after local refinement focusing, suggests its flexibility (mobility?) relative to the TMD and that, in NOX2, the DH is "displaced" relative to the TMD, when compared to the situation in the activated (by Ca2+) DUOX1. The mobility of NBD in NOX2 also results in weak interaction with FBD, making hydride transfer from NADPH to FAD inefficient

A major achievement of the work described in this manuscript is what I believe to be the first description of the activation of recombinant NOX2-p22phox in a nanodisc, to generate O2.-, when activated by a trimeric fusion protein (trimera), consisting of the functionally important parts of the three cytosolic components, p47phox, p67phox, and Rac (see Y. Berdichevsky et al., J. Biol. Chem. 282, 22122-22139, 2007). This proves that the resting state structure of NOX2-p22phox has all that is needed to be converted to the activated state. The fact that the nature of the phospholipid in the nanodisc can be varied and that this is known to have a major effect on the affinity of the trimera for NOX2-p22phox, offers additional advantages.

Weaknesses<br /> A weakness of this, otherwise impressive work, is the difficulty for readers who are not sufficiently "structure educated" to fully understand the "displacement" of the DH of NOX2, shown in the NOX2/DUOX1 overlay (Figure 5). The meaning of "centers of mass" of FBD and FAD, in Figures 5C and 5D, respectively, is not properly explained.

Yet another weakness is the much too vague wording of the change in NOX2 conformation from the resting to the activated state by cytosolic factors as "the cytosolic factors might likely stabilize the DH of NOX2 in the "docked" conformation which is similar to that observed in the activated DUOX1 in the high-calcium state". First, the evidence from biochemical studies of NOX2 activation indicates clearly distinct targets of individual cytosolic components and not a "block" action. There is also support for the conformational change being the result of the action of a single cytosolic component (p67phox), with the other cytosolic components acting as carriers or activators of one cytosolic component by another, such as Rac-GTP acting as a carrier and inducer of a conformational change in p67phox (see J. El-Benna and P.M-C. Dang, J. Leukoc. Biol. 110, 213-215, 2021, and E. Bechor et al., J. Leukoc. Biol. 110, 219-237, 2021). Also, the concept of "docking of the DH to the TMD" seems like an oversimplification of the many locations and partners of such "docking" and ignores the possible multiple consequence of such docking. Even before the appearance of structural studies of NOXs, revealing precise distances between redox stations (NADPH-FAD; FAD-inner heme; inner heme - outer heme), as first reported for C. stagnale Nox5, by F. Magnani et al., Proc. Natl. Acad. Sci. U.S.A. 114, 6764-6769, 2017, a shortening of the distance between an electron donor and acceptor at specific locations in the redox cascade was proposed. The most popular was the NADPH - FAD hydride transfer, based on structural work by P.A. Karplus on Ferredoxin - NADP reductases, the accepted model for the DH of NOXs.

An unfair request for an unachieved task<br /> Of course, the dream of those hoping for a structure-based response to solving the molecular mechanism of NOX activation is to see the structure of the activated NOX2 in complex with three cytosolic components. The compelling finding in the present manuscript that a nanodisc-embedded recombinant NOX2-p22phox can be activated to ROS production by the use of a [p47phox-p67phox-Rac] trimera (replacing three cytosolic components) will provoke in all the readers the wish to see the structure of such a complex. The size of the trimera with a GFP tag (108 kDa) might make the use of the anti-Nox2 Fab and anti-light chain nanobody, unnecessary. Prenylation of the trimera at the Rac moiety is bound to markedly enhance its affinity for the phospholipids in the nanodisc and is likely to generate a more stable complex, most suitable for cryo-EM (see A. Mizrahi et al., J. Biol. Chem. 285, 25485-25499, 2010).

Reviewer #1 (Public Review):

Chondrosarcoma is a rare and aggressive cancer type with a poor prognosis and lacks effective treatment options. Developing an effective strategy for targeting chondrosarcoma is therefore considered an unmet clinical need. The goal of this study is to provide the molecular basis for chondrosarcoma progression and identify a potential strategy/agent for targeting chondrosarcoma. The study reveals that EZH2/hSULF1/c-Met axis is a critical signaling pathway for chondrosarcoma and provides proof of principle evidence that targeting c-MET by pharmacological approaches is an effective strategy to suppress tumor growth in chondrosarcoma mouse models. The aims to be explored for the study are novel and have been well accomplished. The conclusions from this current study are well supported by the compelling and robust datasets using diverse approaches. The study not only reveals a novel insight into how chondrosarcoma progression occurs but also offers the potential strategy for targeting chondrosarcoma, hence significantly advancing the field.

Reviewer #2 (Public Review):

In this manuscript, Lin et al. reveal a novel and fundamental discovery regarding the role of the EZH2/SULF1/cMET signaling pathway in regulating the disease progression of chondrosarcoma, a malignant cartilaginous bone tumor.

The significant strengths of the manuscript include identifying the EZH2-targeted genes in chondrosarcoma using EZH2-chromatin immunoprecipitation sequencing (ChIP-seq) and cDNA microarray profiling, deciphering the role of the EZH2/SULF1/cMET signaling pathway in regulating the progression of chondrosarcoma, verifying the therapeutic significance of this pathway using clinically used specific EZH2 and cMET pharmacological inhibitors in vitro and in vivo (in mouse tumor models), and demonstrating the clinical significance of the SULF1/cMET pathway in chondrosarcoma.

The significant weaknesses of the manuscript appear not noted. A minor drawback seems associated with the manuscript presentation.

In summary, I believe this manuscript's data well justify the authors' claims and conclusions, and this paper will significantly impact the field.

Reviewer #1 (Public Review):

It has been shown that selenium protects against the development of epilepsy, and behavioral comorbidities, as pointed out by the authors. This paper attempts to show it does if administered later after chronic seizures start. While clinically relevant, as noted by the authors, the paper seems not to be a major advance beyond the prior study. The antiseizure effect is also not very convincing because the effect size is so small and the variance so high. The data about behavior is more convincing but similar data were in the previous paper, so it is not very novel.

The data showing changes in PP2A are interesting and while logical that it contributes to the effects of sel, one would like to see proof that this is the basis of sel effects. Same for hyperphosphorylated tau, telomere length, etc. The doubt is because these are indices that change after many types of experiments and they change many aspects of brain and peripheral physiology. Regarding molecular data, how these provide insight and comparison to other data sets of this kind would be valuable.

Reviewer #2 (Public Review):

Casillas-Espinosa et al. present a well-designed study to evaluate the validity of sodium-selenate treatment in chronic epilepsy. Previous studies from the same group identified increased phospho-tau in models of seizures and epilepsy, which can be pharmacologically addressed through activation of protein phosphatase 2A with sodium-selenate. Here the authors tested the effect of delayed treatment with sodium selenate in the post-KA SE rat TLE model. Sodium selenate stopped the progression of seizures during and beyond a 4-week treatment phase compared to Levitiracetam and vehicle-treated animals. Sodium selenate further improved cognitive and sensorimotor impairments. It also persistently reduced phospho-tau and increased PP2A protein expression, and reversed TLE-associated telomere-shortening. Finally, proteome and metabolome data from the model is discussed and provides initial insights into sodium selenate treatment's molecular consequences.

This study validates the use of sodium selenate as a promising pharmacological treatment in experimental TLE that reduces seizure burden and restores cognitive deficits and pathomolecular changes. The specific strength of the study is a clinically relevant treatment paradigm, starting when recurrent seizures are fully established, and the antiepileptogenic effect with a sustainable reduction in seizure burden even after discontinuation of treatment.

The conclusions of this paper are mostly well supported by data, but some aspects of the proteome and metabolome data analysis need to be clarified and extended. The molecular data appears to be the weakest part of this study and would have benefited from adjusted sample sizes to account for interindividual variability between animals and the complex multi-dimensional nature of the data.

Reviewer #1 (Public Review):

Estimating the effects of mutations on the thermal stability of proteins is fundamentally important and also has practical importance, e.g, for engineering of stable proteins. Changes can be measured using calorimetric methods and values are reported as differences in free energy (dG) of the mutant compared to wt proteins, i.e., ddG. Values typically range between -1 kcal/mol through +7 kcal/mol. However, measurements are highly demanding. The manuscript introduces a novel deep learning approach to this end, which is similar in accuracy to ROSETTA-based estimates, but much faster, enabling proteome-wide studies. To demonstrate this the authors apply it to over 1000 human proteins.

The main strength here is the novelty of the approach and the high speed of the computation. The main weakness is that the results are not compared to existing machine learning alternatives.

Reviewer #2 (Public Review):

Summary:

This work presents a new machine-learning method, RaSP, to predict changes in protein stability due to point mutations, measured by the change in folding free energy ΔΔG.

The model consists of two coupled neural networks, a 3D self-supervised convolutional neural network that produces a reduced-dimensionality representation of the structural environment of a given residue, and a downstream supervised fully-connected neural network that, using the former network's structural representation as input, predicts the ΔΔG of any given amino-acid mutation. The first network is trained on a large dataset of protein structures, and the second network is trained using a dataset of the ΔΔG values of all mutants of 35 proteins, predicted by the biophysics-based method Rosetta.

The paper shows that RaSP gives good approximations of Rosetta ΔΔG predictions while being several orders of magnitude faster. As compared to experimental data, judging by a comparison made for a few proteins, RaSP and Rosetta predictions perform similarly. In addition, it is shown that both RaSP and Rosetta are robust to variations of input structure, so good predictions are obtained using either structures predicted by homology or structures predicted using AlphaFold2.

Finally, the usefulness of a rapid approach such as RaSP is clearly demonstrated by applying it to calculate ΔΔG values for all mutations of a large dataset of human proteins, for which this method is shown to reproduce previous findings of the overall ΔΔG distribution and the relationship between ΔΔG and the pathological consequences of mutations. The RaSP tool and the dataset of mutations of human proteins are shared.

Strengths:

The single main strength of this work is that the model developed, RaSP, is much faster than Rosetta (5 to 6 dex), and still produces ΔΔG predictions of comparable accuracy (as compared with Rosetta, and with the experiment). The usefulness of such a rapid approach is convincingly demonstrated by its application to predicting the ΔΔG of all single-point mutations of a large dataset of human proteins, for which using this new method they reproduce previous findings on the relationship between stability and disease. Such a large-scale calculation would be prohibitive with Rosetta. Importantly, other researchers will be able to take advantage of the method because the code and data are shared, and a google colab site where RaSP can be easily run has been set up. An additional bonus is that the dataset of human proteins and their RaSP ΔΔG predictions, annotated as beneficial/pathological (according to the ClinVar database) and/or by their allele frequency (from the gnomAD database) are also made available, which may be very useful for further studies.

Weaknesses:

The paper presents a solid case in support of the speed, accuracy, and usefulness of RaSP. However, it does suffer from a few weaknesses.

The main weakness is, in my opinion, that it is not clear where RaSP is positioned in the accuracy-vs-speed landscape of current ΔΔG-prediction methods. The paper does show that RaSP is much faster than Rosetta, and provides evidence that supports that its accuracy is comparable with that of Rosetta, but RaSP is not compared to any other method. For instance, FoldX has been used in large-scale studies of similar size to the one used here to exemplify RaSP. How does RaSP compare with FoldX? Is it more accurate? Is it faster? Also, as the paper mentions in the introduction, several ML methods have been developed recently; how does RaSP compare with them regarding accuracy and CPU time? How RaSP fares in comparison with other fast approaches such as FoldX and/or ML methods will strongly affect the potential usefulness and impact of the present work.

Second, this work being about presenting a new model, a notable weakness is that the model is not sufficiently described. I had to read a previous paper of 2017 on which this work builds to understand the self-supervised CNN used to model the structure, and even so, I still don't know which of 3 different 3D grids used in that original paper is used in the present work.

A third weakness is, I think, that a stronger case needs to be made for fitting RaSP to Rosetta ΔΔG predictions rather than experimental ΔΔGs. The justification put forward by the authors is that the dataset of Rosetta predictions is large and unbiased while the dataset of experimental data is smaller and biased, which may result in overfitting. While I understand that this may be a problem and that, in general, it is better to have a large unbiased dataset in place of a small biassed one, it is not so obvious to me from reading the paper how much of a problem this is, and whether trying to fix it by fitting the model to the predictions of another model rather than to empirical data does not introduce other issues.

Finally, the method is claimed to be "accurate", but it is not clear to me what this means. Accuracy is quantified by the correlation coefficient between Rosetta and RaSP predictions, R = 0.82, and by the Mean Absolute Error, MAE = 0.73 kcal/mol. Also, both RaSP and Rosetta have R ~ 0.7 with experiment for the few cases where they were tested on experimental data. This seems to be a rather modest accuracy; I wouldn't claim that a method that produces this sort of fit is "accurate". I suppose the case is that this may be as accurate as one can hope it to be, given the limitations of current experimental data, Rosetta, RaSP, and other current methods, but if this is the case, it is not clearly discussed in the paper.

Reviewer #3 (Public Review):

The authors present a machine learning method for predicting the effects of mutations on the free energy of protein stability. The method performs similarly to existing methods, but has the advantage that it is faster to run. Overall this is reasonable and a faster method will likely have some potential uses. However, not improving performance beyond the reasonable but not great performance of existing methods of course makes this a less useful advance. The authors provide predictions for a set of human proteins, but the impact of their method would be much greater if they provided predictions for all substitutions in all human proteins, for example. In places the text somewhat overstates the performance of computational methods for predicting free energy changes and is potentially misleading about when ddGs are predicted vs. experimentally measured. In addition, the comparison to existing methods is rather slim and there isn't a formal evaluation of how well RASP discriminates pathological from benign variants.

Reviewer #1 (Public Review):

This manuscript uses two OR molecues as a model to understand the mechanism behind their ligand specificity. It combines a series of targeted mutations and domain swapping followed by functional analysis in Xenopus oocyte expression system to analyse functional aspects of the modified ORs. It also models the various OR structures. The authors find that a single amino acid residue is critical for ligand specificity and that this is mediated by space constraints generated in the ligand docking region. The manuscript is generally well written and the data are clear and well represented.

Reviewer #2 (Public Review):

The antennae of insects are excellent sensors and are able to distinguish chemicals/compounds using odorant receptor proteins. Though many are promiscuous, several ORs are extremely specific and respond to only one or few related chemicals. In this study, the authors focus on two ORs from southern house mosquito, Culex quinquefasciatus namely OR10 and OR2, which respond to (show high specificity) skatole and indole respectively. Notably, these two compounds differ only by a methyl group raising the question how this is achieved. To address this question, the authors have chosen CquiOR10 (as it is more sensitive) for swapping the transmembrane domains (TMDs) from CquiOR2 and by performing heroic work, arrive at one single residue in one of the TMDs to explain the specificity in these ORs.

The major strengths of the manuscript include the careful design of the many different chimeric receptors (36 in total) and dissecting the importance of each TMD and zeroing on TMD2 as the specificity determinant, followed by zooming to a single residue in TMD2 that can change responsiveness of CquiOR10 to CquiOR2 and vice versa. This residue in TM2 is an alanine in CquiOR10, which when mutated to bulky residue becomes responsive to indole but when mutated to glycine remains specific to skatole and shows higher sensitivity. Similarly, mutating the equivalent residue in CquiOR2, Leucine 74 to a smaller residue makes this receptor now more responsive to skatole and making it more like CquiOR10.

Using RoseTTAfold and AlphaFold, the authors build models of CquiOR10 and CquiOR2, which gives them a platform to observe how ligands can bind using Rosettaligand both in native structures as well as mutants. They further ask how larger ligands or the methyl group at different position in the indole ring affects the response of the receptor, which follow a consistent trend on the key residue of Alanine 73. All these analysis allow authors to propose that the odorants or chemicals are accommodated/restricted due to the volume constraints by residues lining the cavity derived from the TMDs.

Reviewer #3 (Public Review):

Franco et al. consider two mosquito olfactory receptors that have different sensitivities to two odorants: CquiOr10 is activated by skatole while CquiOr2 is activated by indole. Starting with chimeric receptors composed of pieces from each receptor, they are ultimately able to identify a single amino acid that, when mutated, switches the specificity of the receptors. When Ala73 is mutated to a Leu in CquiOr10, the mutant receptor now preferentially binds indole, while the counterpart Leu74 to Ala substitution in CquiOr2 creates a receptor that is more sensitive to skatole. To better understand why these substitutions alter ligand-binding specificities, the authors use molecular docking to identify the likely interactions between indole or skatole and the natural or mutant CquiOr10 receptors. They conclude that the size of the amino acid at position 73 affects ligand specificity by altering the amount of space available to bind ligands.

Reviewer #1 (Public Review):

The authors investigate the relative importance of the bee host and bacterial microbiome in processing the nectar secondary metabolite amygdalin, with a focus on understanding the contributions of the different members of the microbiome, and the enzymatic basis for metabolic transformations. The manuscript clearly describes the experimental procedures, presents the results in graphically appealing figures and clear text, and puts the work into a broader context in the discussion. The conclusions are backed by sophisticated in vitro and in vivo experimental data. A particular strength of the manuscript is the combined use of genomic, gene expression, proteomic, and small metabolite analyses to pin down the mechanistic basis of the degradation of amygdalin. While at this stage the authors cannot infer the importance of their findings for bee health, their insights and methods should stimulate additional experiments into the role of microbial conversion of dietary metabolites for bee health.

Reviewer #2 (Public Review):

In this communication by Motta and colleagues, the authors address the emerging role of the gut microbiome in degrading and detoxifying plant metabolites, using bees as a study system. The experiments are elegantly controlled, spanning in vitro and in vivo work that leverages the increasing tractability of bees and their microbial symbionts. This is evident in the extensive screening of Bifidobacterium, Bombilactobacillus, Lactobacillus, and Gilliamella relative to their susceptibility to amygdalin. This provided a foundation to pinpoint which strains can degrade the cyanogenic glycoside, the potential pathways underlying that process, and the key enzymes involved. The strain Bifidobacterium wkB204 displayed elevated expression of GH3, correlating to the ability of this microbe to degrade amygdalin in vitro. Expression of the GH3 in E. coli corroborated its putative role in the transformation of amygdalin to prunasin, consistent with the single inoculation effects of Bifidobacterium wkB204 into microbiota-deprived bees. These experiments collectively point to the importance of the bee microbiota for the consistent degradation of amygdalin. The findings are nicely contextualized relative to prior work on the gut microbiome and the metabolism of the cyanogenic glycoside, including efforts on bees and rats.

Reviewer #3 (Public Review):

Motta, Erick et al. investigated the role of members of the bacterial gut microbiota of honey and bumble bees in the degradation of amygdalin, a plant cyanogenic glycoside found in almond trees and other plants. The role of the microbiota in contributing to secondary plant compounds in this system is of interest because it has been demonstrated that the genomes of these bees are depauperate in genes of detoxification enzymes relative to other insects. Using in vitro assays across a range of honey and bumble bee-derived strains of the bacterial species Bifidobacterium, Bombilactobacillus, Gilliamella, and Lactobacillus nr. melliventris the authors demonstrate strain-specific growth on amygdalin as a carbon source, clearly showing amygdalin metabolism by particular strains. The data strongly support that amygdalin degradation occurrence is not a pan-species trait, but rather strain-specific, and also that even within a bacterial species the strains metabolizing amygdalin achieve this through different pathways, with some strains producing the metabolite prunasin, but others not. Subsequent proteomics analysis suggests that a glycoside hydrolase family 3 (GH3) is likely responsible for the degradation of amygdalin. The conclusion that this GH3 is at least partially responsible for strain-specific degradation is supported by gene expression analysis of the enzyme and experiments with E. Coli transformed with the gene. Further in vivo studies demonstrate that the honey bee microbiota contributes to amygdalin metabolism, including specific strains of Bifidobacterium, but that the hosts themselves can metabolize amygdalin to prunasin in the absence of gut microbes, but not to the same degree.

The approach and evidence supporting the step-wise conclusions are comprehensive. However, further extension is required to gain a full appreciation for what the importance and relevance of the results for conclusions relating to cooperation between hosts and microbiota and particularly the consequences for host health.

Although the authors rightly do not directly interpret the attributed breakdown of amygdalin and its metabolites by specific bacterial strains as a benefit, this is alluded to in the title and parts of the discussion. Following the degradation of amygdalin through intermediates, hydrogen cyanide is produced. Hydrogen cyanide is generally considered to be detrimental. As such, it could be argued that is not appropriate to consider the production of such a compound as cooperative between host and microbiota, given that cooperation is usually to a beneficial end. Experiments exposing hosts with microbiota absent and present to amygdalin and relevant breakdown products and subsequently measuring relevant health outcomes would be an important step in aiding in the interpretation of the otherwise clear experimental outcomes. Especially given the relatively limited number of strains tested showing the ability to degrade amygdalin, it is possible that there is limited adaptive value, and/or the ability could be due to either chance or selection for the metabolism of other compounds. This is especially relevant when considering further work that may look at how health-related outcomes such as parasite resistance are affected.

This being said, the work adds to demonstrations of different functions of host gut microbiota, how they can mediate the environment encountered by hosts, and the increasing appreciation that effects derived from the microbiota can be not only dependent upon the bacterial species present but frequently the specific strains.

Reviewer #1 (Public Review):

Building upon the previous evidence of activation of auditory cortex VIP interneurons in response to non-classical stimuli like reward and punishment, Szadai et al., extended the investigation to multiple cortical regions. Use of three-dimensional acousto-optical two-photon microscopy along with the 3D chessboard scanning method allowed high-speed signal acquisition from numerous VIP interneurons in a large brain volume. Additionally, activity of VIP interneurons in deep cortical regions was obtained using fiber photometry. With the help of these two imaging methods authors were able to extract and analyze the VIP cell signal from different cortical regions. Study of VIP interneuron activity during an auditory go-no-go task revealed that more than half of recorded cortical VIP interneurons were responding to both reward and punishment with high reliability. Fiber photometry data revealed similar observations; however, the temporal dynamics of reinforcement stimuli-related response in mPFC was slower than in the auditory cortex. The authors performed detailed analysis of individual cell activity dynamics, which revealed five categories of VIP cells based on their temporal profiles. Further, animals with higher performance on the discrimination task showed stronger VIP responses to 'go trials' possibly suggesting the role of VIP interneurons in discrimination learning. Authors found that reinforcement related response of VIP interneurons in visual cortex was not correlated with their sensory tuning, unveiling an interesting idea that VIP interneurons take part in both local as well as global processing. These observations bring attention to the possible involvement of VIP interneurons in reinforcement stimuli-associated global signaling that would regulate local connectivity and information processing leading to learning.

The state-of-the-art imaging technique allowed authors to succeed in imaging VIP interneurons from several cortical regions. Advanced analyses revealed the nuances, similarities and differences in the VIP activity trend in various regions. The conclusions about reinforcement stimuli related activity of VIP interneurons made by the authors are well supported by the results obtained, however some claims and interpretations require more attention and clarification.

Reviewer #2 (Public Review):

In recent years the activity of cortical VIP+ interneurons in relation to learning and sensory processing has raised great interest and has been intensely investigated. The ability of VIP+ interneurons in the auditory cortex to respond to both reward and punishment was already reported a few years ago by some of the authors (Pi et al., 2013, Nature). However, this work importantly adds to their previous study demonstrating a largely similar and synchronous response of a large fraction of these interneurons across the neocortex to salient stimuli of different valence during the performance of an auditory discrimination task.

An additional strength of this study is the analysis and identification of the general pattern of VIP+ interneuron responses associated to specific behaviors in the different layers of the neocortex depth.

Interestingly, the authors also identified using cluster analysis 5 different classes of VIP+ interneurons, based on the dynamic of their responses, that were unequally distributed in distinct cortical areas.

This is a well performed study that took advantage of a cutting-edge imaging approach with high recording speed and good signal-to-noise ratio. Experiments are well performed and the data are properly analyzed and nicely illustrated. However, one shortcoming of this paper, in my opinion, is the "case report" structure of the data. Essentially for each neocortical area the activity of VIP+ interneurons was analyzed only in one animal. This limits the assessment of the stability of the response/recruitment of these interneurons. I appreciate the high number of recorded VIP+ interneurons per area/animal and I do understand that it would be excessively laborious to perform 3D random-access two-photon microscopy in several mice for each cortical area. On the other hand, it would be important to have some knowledge of the general variability of the responses of these neurons among animals.

In conclusion, despite the findings described in this manuscript being generally sound, additional experiments are recommended to further substantiate the conclusions.

Reviewer #3 (Public Review):

In this study Szadai et al. show reliable, relatively synchronous activation of VIP neurons across different areas of dorsal cortex in response to reward and punishment of mice performing an auditory discrimination task. The authors use both a relatively fast 2 photon imaging, as well as fiber photometry for some deeper areas. They cluster neurons according to their temporal response profiles and show that these profiles differ across areas and cortical depths. Task performance, running behavior and arousal are all related to VIP response magnitude, as has been previously shown.

Methodologically, this paper is strong: the described imaging technique allows for fairly fast sampling rates, they sample VIP cells from many different areas and the analyses are sophisticated and touch on the most relevant points. The figures are of high quality.

However, as the manuscript is now, the presentation could be clearer, the methods more complete and it is not clear whether their conclusions are entirely supported by the data.

The main issue is that reinforcement and arousal are hard to distinguish in this study. It is well known that VIP activity is correlated with arousal. And it is fairly clear that the reinforcement they use in this study - air puffs to the eye, as well as water rewards - cause arousal. It is possible that the reinforcer responses they observe in VIP neurons throughout all areas merely reflect the increases in arousal caused by these behaviorally salient events. They do discuss this caveat (albeit not fully convincingly) and in their abstract even state that the arousal state was not predictive of reinforcer responses. However their data clearly shows the tight relationship of the VIP reinforcer responses to both arousal (as measured by pupil diameter), as well as running speed of the animal. Both of these variables are well known to be tightly coupled to VIP activity.

Although barely mentioned, the authors do appear to sometimes present uncued reward (Figure S2F). If responses were noticeably different from the same events in the task context (as actual reinforcers) this could at least hint towards the reinforcement signal being distinct from mere arousal. However, this data is only mentioned in one supplementary figure in a different context (comparison with PV cells) and neither directly compared to cued reward, nor is this discussed at all. Were uncued air puffs also presented? How do the responses compare to cued air puffs/punishment?

The imaging method appears well suited for their task, however the improvements listed in table S1 make the method appear far superior to existing methods in many aspects. Published or preprinted papers with 2 photon imaging of VIP populations (eg. from Scanziani lab (Keller et al.), Carandini lab (Dipoppa et al.), deVries lab (Millman et al.), Adesnik lab (Mossing et al.), which use the much more common resonant scanning, seem to be able to image 4-7 layers at 4-8Hz with a good enough SNR and potentially bigger neuronal yield of approximately 100-200 VIP cells, depending on the field of view. While not every single cell in a volume would be captured by these studies, the only main advantage of the here-used technique appears to be the superior temporal resolution.

Even though this is not mentioned at all, it certainly appears possible, that the accousto-optical scanning emits audible noise. In this case it would be good to know the frequency range and level of this background noise, whether there are auditory responses to the scanning itself and if it interferes with the performance of the animals in the auditory task in any way. If this is not the case, this should probably simply be mentioned for non-experts.

The authors show a strong correlation between task performance (hit rate) and the response to the auditory cue on hit trials. Was there any other significant correlations of VIP cells' responses to other trial types? Was reinforcer response correlated to behavioral variables at all?

Peer review report

Title: Abnormalities in migration of neural precursor cells in familial bipolar disorder

version: 6

Referee: Shani Stern

Institution: University of Haifa

email: sstern@univ.haifa.ac.il

ORCID iD: 0000-0002-2644-7068

General assessment

I have read the study by Sukumaran et al. The authors describe that NPCs derived from bipolar disorder patients using induced pluripotent stem cells show abnormal migration and that transcriptionally there is a dysregulation of a network of genes that relate to on EGF/ERBB proteins. Overall, the study is interesting. However, some points should be addressed. See below.

The following points are important to take into consideration:

1. Some English edits are required. Also, some acronyms appear before their definition (for example MSD).

2. In the introduction, some important studies that describe transcriptomics of BD patients should be described such as Santos et al 2021 and also neuronal phenotypes such as hyperexcitability and physiological instabilities.

3. The transcriptomics is performed only for one of the control lines although there are 3 controls. More control samples should be taken for the gene expression analysis.

4. The figures are not in the correct order.

5. The damaging variants of the patients are not described. Although there are references to previous publications, since this may be central it is good to add a table describing these variants in the patients and in the controls.

6. The differentiation to which type of neurons should be briefly described (although there is a reference to previous publications, but this should be mentioned).

7. The statistical analysis is not very clear. Moreover, the images of the migration assays are not convincing enough that indeed there is a significant difference. How many times was this performed? This should be repeated with several cultures, especially since the number of patients and controls is small. It would be more convincing if this is reproducible over several repetitions.

8. It would be good to also include representative videos in the supplementary.

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):

In this work the authors study the effects of the accumulation of endogenously produced Advanced Glycation End-products (AGEs) on feeding behaviors in C. elegans. AGEs are produced during the metabolism of all organisms, and also, they are produced by the food industry through Mainard reactions. In this sense, the objectives of this study are not only to provide basic information relevant to phenomena that are likely to be conserved throughout the animal kingdom but also to provide information that could be important in human health for the understanding of disorders caused by the consumption of processed foods.

The methodology includes as read out very robust and supercharacterized assays of food intake in C. elegans, such as pharyngeal pumping and food depletion.<br /> As a general evaluation of the manuscript, I think the authors could provide more detailed mechanistic information about how MGH-1 acts on the tyraminergic pathway to potentiate food intake. While they find important players, they do not quite find how these players interact with each other, nor which cells or neural circuits are governing the processes described.

In summary, I consider the initial objective of the manuscript to be extremely significant, but I believe it falls short in the mechanistic explanation of the observations described.

Reviewer #2 (Public Review):

General description:<br /> This study elucidates how Advanced Glycation End-products (AGE), found in processed food and endogenously, drive food intake and cause some of the pathophysiological defects associated with metabolic disease. In their previous C. elegans study, the authors found that glod-4 mutants, animals that lack glyoxalase activity and thus accumulate AGEs, eat more and share some of the pathophysiological effects seen in metabolic disease. In this study, they identify a specific AGE, hydroimidazolone (MG-H1), that is sufficient to increase feeding, similar to what was previously observed in the glod-4 mutants. Gene expression studies then show expression changes in several neurotransmitter and eating genes, including the tyramine decarboxylase gene tdc-1 and its receptor. Measuring eating behaviors in animals carrying mutations in tyramine signalling genes they show that tyramine signaling system is required for the behavioral and pathophysiological effects of MG-H1. Finally, they show that the transcription factor elt-3 controls the expression of tyramine signaling components and thus is also required for the response to MG-H1.

Strength: Strengths of the paper include the elegant approach to study how toxic metabolites affect physiology and behavior in vivo. The logic behind the study is easy to follow and the paper is clearly written.

Weakness: The main weakness is that the genetic studies were generally only carried out with a single mutation that was not rescued. To corroborate the requirement of tdc-1 and elt-3 for the response to MG-H1, the results should be repeated either in a rescue strain or using a different allele. Some of the effects are subtle and there is the danger of them being caused by background mutations.

Impact: The occurrence of metabolites like AGEs in either processed food or endogenously is a topic that is not well investigated despite its general importance. In this study the authors show the functional consequences of a non-enzymatically generated metabolite and how it exerts its toxicity.

Reviewer #3 (Public Review):

This manuscript studied an interesting topic: the maillard reaction, catalyzed by glyoxalases, converts α-dicarbonyl compounds to Advanced Glycation End-products (AGEs). glod-4 is one of the glyoxalases and MG-H1 is one of AGEs which is converted from methylglyoxal (MGO). The authors discovered that both glyoxalase glod-4 KO and supplementation of MG-H1 increased pumping rates in C. elegans. MG-H1 mediated pumping rates increase is dependent on glod-4. The authors further found that tyramine synthease tdc-1 and two of the tyramine receptors ser-2 and tyra-2 are required for the increased pumping by glod-4 knockout or MG-H1. They also found the transcriptional factor elt-3 is required for pumping increase by MG-H1 and glod-4 KO, and also regulates tdc-1 transcriptional level. Lastly, they found that tdc-1 and the two tyramine receptors mutants rescue the shorter lifespan of glod-4 and neuronal loss in glod-4.

The topic is interesting, and it is a good design to show mechanistic function of neurotransmitter in regulating tasty AGEs in a model organism. Most of the results are supported by the data.

Reviewer #1 (Public Review):

This report describes an exhaustive analysis of behavior in a complex associative learning paradigm that blends aversive Pavlovian and appetitive instrumental elements. The hand-scoring technique is rigorous and documented to a greater degree than what is typically reported in papers using human raters to quantify animal behavior. Near-complete ethograms offer a novel, high-resolution look at how aversive cues exert distinct effects on appetitive and aversive behavior.

From the perspective of the rodent subject, there is quite a lot going on in the experimental chamber in this study. It's an environment in which appetitive instrumental action is set against multiple predictive cues signaling differing degrees of danger and safety. The test is fully on-baseline, occurring in the same place as training. The rich web of associations formed has a predictably complex influence on behavior. The authors contrast this complexity with much of the rest of the literature, in which freezing is reported to predominate when an aversive CS is presented. Indeed, most conventional studies of aversive associative learning train subjects on a single tone-shock association and test in a neutral context. The contrast between the common approach and the one taken by the authors suggests questions central to understanding the current report. Does being tested in an associatively complex context promote the pattern of behaviors that the authors observe? Or is it a question of learning history - would, following this kind of complex training, an off-baseline test in a neutral environment, produce the same suite of outcomes in response to the danger cue? Answers to these questions would go some distance toward nesting this paper in a wider body of knowledge about defensive reactions to aversive conditioned stimuli. Data speaking to these issues would also increase the work's impact by demonstrating the way in which a given response can be modulated by other learning.

Reviewer #2 (Public Review):

This is an important and timely characterization of a diversity of behaviors male and female rats exhibit during the acquisition of Pavlovian fear conditioning in a conditioned suppression procedure. Using hand-scored video analysis and ethogram of nine different behaviors, the authors report that auditory conditioned stimuli that predict shock with high certainty evoke not only freezing, but a variety of other behaviors including locomotion, jumping, and rearing (in addition to suppressing reward-seeking). Auditory stimuli that were followed by shock on only some trials (uncertainty condition), were less likely to evoke freezing and did not lead to a suppression of port/cup-directed behaviors (reward seeking). There were subtle sex differences in the temporal profile of freezing behavior, but not in the properties of the other behaviors under study.

Ultimately, these findings point to the importance of task variables (eg., reward seeking in a conditioned suppression procedure) and shock probability in shaping an animal's defense repertoire under threat.

An important factor that this work does not resolve is how the magnitude of the threat/shock (and presumably the state of fear that it engenders) influences an animal's defensive topography. This report used a modest/weak footshock intensity that supported very low levels of tone-elicited freezing (<20%) - a stark contrast to the extant fear conditioning literature that typically reports much higher levels of freezing behavior.

Reviewer #3 (Public Review):

The authors' goal was to explore if there were fear behaviors expressed to a conditioned fear cue other than freezing and how the timing of these behaviors may change across a discrete conditioned cue. Three separate cues representing danger (1.0 footshock probability), safety (0 footshock probability), and uncertainty (0.25 footshock probability) were used against a backdrop of operant nosepoke responding for reward in male and female Long Evans rats. All behaviors were recorded with a frame capture of 5 frames per second and manually scored afterwards blindly for one of ten behaviors.

Analyzing the repertoire of possible behaviors, beyond freezing, across a 10s conditioned cue that may be perceived as dangerous, uncertain, or safe is a strength of the study. Displaying the possible behaviors stacked across the 10s, second by second, instead of a bulk 10s average of each type of behavior highlights the dynamic nature of the defensive behaviors expressed across time. It is unclear though why the 2s post-cue were not included since the footshock was not administered until 2s after cue offset. Given their argument of defensive behaviors being adjusted as the threat becomes more imminent, this 2s period would appear to be a valuable interval for their analyses and argument.

The authors emphasize the ethology of their findings but they also acknowledge that their findings do not agree with the majority of rodent fear conditioning papers reporting upwards of 80% freezing across a cue. Since these differences could be due to a myriad of experimental differences such as cue length, cue modality, number and strength of shocks, etc, it is difficult to extrapolate and apply the reported findings to potentially broader conditions; e.g. cues that are not 10s, non-rodent species, food-restricted vs not food-restricted, an environment that is not a small, enclosed box, etc. In the end, while additional defensive behaviors were reported in response to a danger cue, the predominant behavior still appeared to be freezing, although there were interesting differences noted between males and females in that females appeared to display most of their freezing early in the cue while males express a more sustained freezing response across the cue.

This work could certainly inspire other labs to approach their video analyses in a similar fashion and, although not discussed in the paper, could potentially be interesting to also look at individual differences across ethograms, instead of the grouped data presented across the 12 males versus 12 females as shown here. These could then be used before or after a manipulation and used to try to predict how an animal may respond to a certain event or manipulation.

Reviewer #1 (Public Review):

The paper, fundamentally, is a description of the accuracy of individual model and ensemble model short-term forecasts of COVID-19. This has been done before in both weather and infectious disease. So what are the contributions of this manuscript? I see the following:

1. The authors show that ensemble prediction (a straight average) generally outperforms individual component models. This is not new and has been shown, as the authors cite, for weather, climate, and infectious disease.<br /> 2. Use of the median estimate across models, rather than the mean, buffers against outliers. This is a well-recognized workaround for right-skewed distributions, though the specific finding in this study is of some importance, as this hasn't always been the case (noted by the authors in their discussion).<br /> 3. Deaths are better forecasted than cases. This is not new, either, as the authors note, as deaths are a lagged function of cases/infections.<br /> 4. It presents the archive of European COVID-19 forecasts.

Although I don't see a lot of novelty in these findings, this COVID-19 forecasting work is important and represents a considerable effort on part of the individual modelers. The paper is well written, but it doesn't show much that is novel methodologically. For instance, it doesn't propose and validate an approach for improving forecasting or projection accuracy. Are there new ways to handle or predict behavioral, vaccination uptake, or viral changes? Are there novel post-processing approaches, other than 'ensembling' that could improve forecast accuracy?

Reviewer #2 (Public Review):

This paper by Sherratt et al. evaluated the performance of real-time predictions for COVID-19 submitted to the European COVID-19 Forecast Hub between March 8 2021 and March 7 2022. This large-scale multi-team multi-county collaboration collected short-term forecasts for COVID-19 from 26 teams generated for 32 countries in Europe, making this dataset one of the largest archives of real-time COVID-19 forecasts. The results indicate that ensemble models combining forecasts from individual models generally performs better than each individual model, and ensemble methods based on medians outperform the ones based on means. The comparison also shows that incident death forecasts are more reliable than incident case forecasts beyond two weeks into the future. The paper further included detailed discussions on several practical considerations in the operational use of forecasting models. These findings provide practical guides for generating real-time forecasts for infectious diseases and novel insights into coordinating international forecasting efforts during a public health emergency.

The conclusions of this paper are well supported by the data and analyses. A few aspects could be further discussed in the manuscript.

1. A parallel effort of real-time COVID-19 forecasting in the US (i.e., the US COVID-19 Forecast Hub) reported similar findings on the use of ensemble models. This study from Europe provides independent validation that shows the robustness of these findings. While both studies followed similar guidelines and used the same evaluation metrics (coverage and WIS), I believe there should be unique challenges associated with forecasting for multiple countries (as opposed to forecasting in a single country). As a result, it might be worthwhile to discuss those challenges and potential solutions to inform similar efforts in the future.

2. WIS is a strictly proper score for evaluating forecast performance; however, it must rely on a reference forecast model. This may create difficulties in interpreting forecast accuracy for the general public who may not understand the concept of WIS. For instance, what is a WIS score good enough to trust? The authors may want to include a simple metric (e.g., mean absolute error) as a supplement even though these metrics have some caveats. I presume the performance should be highly correlated using different evaluation metrics.

3. It might be helpful to elaborate more on the assumptions for near-term predictions in participating models (e.g., status quo, reactive change of transmission, etc.). Essentially all real-time predictions were generated based on assumptions, although sometimes those assumptions were not stated explicitly. For behavior-induced changing points (peaks or troughs), it might be challenging to predict using the status quo without considering a change in model states.

4. Data in the tables and figures were used to compare forecasts. It would be great to have a formal statistical test for comparing model performance, if possible.

Reviewer #1 (Public Review):

Earlier this year Skolnick and colleagues managed to tweak AlphaFold to predict protein complexes (reference 23 in the current manuscript). They also added a score that allows the detection of true protein-protein interactions among arbitrary protein pairs. Thus, their methodology allows reliable prediction of homo- and hetero-meric protein-protein interactions, and predicting the structure of the corresponding protein complexes. Leveraging this methodology, the current manuscript describes a very interesting application to a set of about 1,500 E. coli proteins of the outer membrane, the periplasm and the inner membrane of this Gram negative bacteria. They explore protein-protein interactions among this protein set, which they refer to as 'envelome'. Their results reproduce known protein complexes, such as the translocon, and suggest many yet unknown interactions that make biological sense.

A main strength here is the generation of ample hypotheses to be tested in experiment, i.e., all protein-protein interactions of high predicted accuracy. Another strength is that the methodology is readily applicable to other systems. However, a few outstanding issues need to be clarified.

1. Even though the methodology was already introduced, it should be described in some detail. Most importantly, AlphAfold's measures of accuracy have been part of the loss function during training/testing. What about the measure of protein-protein interaction accuracy? Was it also in the loss function?<br /> 2. Figure 1a (upper panel, PpiD) includes quite a few promising hits but only the first, third, and 12th were considered. How were these chosen? For example, why not consider the second? The lower panel (YfgM) also shows many promising hits but only the first was chosen. Why not more?<br /> 3. Likewise, only two of the top hits in Figure 4 are considered. What about the rest? For example, why taking into account the second best hit while skipping the first?<br /> 4. Authors argue that the unstructured part of OmpA, which wraps around SurA, is to be trusted, which may be the case. But a more likely explanation is that it is an artefact, in agreement with the very low confidence assigned by AlphaFold.<br /> 5. Figure 5. How is this predicted structure compare with the known structure of the complex? In particular, how similar are the predicted and known structures of the individual subunits, and how similar are the predicted docking poses to the known ones?<br /> 6. Authors should make the results easily accessible to all. Maybe as Cytoscape and CyToStruct sessions for easy visualization.<br /> 7. Finally, AlphaFold was trained and tested mostly with water-soluble protein. Thus, application to outer membrane proteins is a bit risky. Maybe authors can comment on this.

Reviewer #2 (Public Review):

It is known that bacterial outer membrane proteins must interact with a variety of cellular factors to reach their final destination safely. There is considerable biochemical evidence in the literature (primarily from crosslinking studies) that these factors interact to promote the movement of client proteins and to prevent their aggregation or misfolding, but the details of the interactions are unknown. The authors showed that they could use a novel virtual screening method together with known crystal structures of individual factors to predict the three-dimensional structures of several pairs or groups of interacting factors (supercomplexes). The predicted supercomplex structures are both fascinating and compelling because they are consistent with the published results and they help to explain the mechanism by which the cellular factors promote outer membrane protein biogenesis. I think that this study will be of interest to a wide audience because it serves as a proof-of-concept that although Alpha Fold is incredibly useful for predicting the structures of protein monomers, more sophisticated applications can be used to successfully predict the structures of protein complexes which are often the workhorses of the cell. I have only two significant concerns. First, the authors focused on high confidence supercomplexes that have known biological significance. Their method also identified other high confidence supercomplexes, but they need to explain how they can distinguish predicted supercomplexes that have potential biological significance from those that are simply "false positives". Second, one of the proposed functional models does not seem to be consistent with the results of a previous study.

Reviewer #3 (Public Review):

In this paper, the authors apply AlphaFold2 to predict the structure of membrane protein complexes in E.Coli. They scan ~1500 membrane proteins starting with one protein to predict the interactions. They present the results for four proteins and analyse them carefully to propose novel models for complexes.

The main problem with the manuscript is that the authors first claim that the method is highly specific but then cherry-pick a subset of interactions that they believe are correct (most likely they are). But the authors do not discuss the other high-scoring predictions. Are these false positives (in which case the method has very limited value) or novel interactions (which would be really interesting but needs further examination)?

Reviewer #1 (Public Review):

The authors present a study of figure-ground segregation in different species. Figure-ground segregation is an important mechanism for the establishment of an accurate 3D model of the environment. The authors examine whether figure-ground segregation occurs in mice in a similar manner to that reported in primates and compare results to two other species (Tree shrews and mouse lemurs). They use both behavioral measures and electrophysiology/two-photon imaging to show that mice and tree shrews do not use opponent motion signals to segregate the visual scene into objects and background whereas mouse lemurs and macaque monkeys do. This information is of great importance for understanding to what extent the rodent visual system is a good model for primate vision and the use of multiple species is highly revealing for understanding the development of figure-ground segregation through evolution.

The behavioral data is of high quality. I would add one caveat: it seems unfair to report that the tree shrews could not generalize the opponent motion stimulus as it seems they struggled to learn it in the first place. Their performance was below 60% on the training data and they weren't trained for many sessions in comparison to the mice. Perhaps with more training the tree-shrews might have attained higher performance on the textures and this would allow a more sensitive test of generalization. The authors should qualify their statements about the tree-shrews to reflect this issue.

Reviewer #2 (Public Review):

Luongo et al. investigated the behavioural ability of 4 different species (macaque, mouse lemur, tree shrew and mouse) to segment figures defined by opponent motion, as well as different visual features from the background. With carefully designed experiments they convincingly make the point that figures that are not defined by textural elements (orientation or phase offsets, thus visible in a still frame) but purely by motion contrast, could not be detected by non-primate species. Interestingly it appears to be particularly motion contrast, since pure motion - figures moving on a static background - could be discriminated better, at least by mice.

This is highly interesting and surprising -- especially for a tree shrew, a diurnal, arboreal mammal, very closely related to primates and with a highly evolved visual system. It is also an important difference to take into account considering the multitude of studies on the mouse visual system in recent years.

The authors additionally present neuronal activity in mice, from three different visual cortical areas recorded with both electrophysiology and imaging. Their conclusions are mostly supported by the data, but some aspects of the recordings and data analysis need to be clarified and extended.

The main issues are outlined below roughly in order of importance:

1. The most worrying aspect is that, if I interpret their figures correctly, their recordings seem not very stable and this may account for many of the differences across the visual conditions. The authors do not report in which order the different stimuli were shown, their supplemental movie, however, makes it seem as though they were not recorded fully interleaved, but potentially in a block design with all cross1 positions recorded first, before switching to cross2 positions and then on to iso... If I interpret Figure 6a correctly, each line is the same neuron and the gray scale shows the average response rate for each condition. Many of these neurons, however, show a large change in activity between the cross1 and the cross2 block. Much larger than the variability within each block that should be due to figure location and orientation tuning. If this interpretation is correct, this would mean that either there were significant brain state changes (they do have the mice on a ball but don't report whether and how much the animals were moving) between the blocks or their recordings could be unstable in time. It would be good to know whether similar dramatic changes in overall activity level occur between the blocks also in their imaging data.

The same might be true for differences in the maps between conditions in figure 4. If indeed the recordings were in blocks and some cells stopped responding, this could explain the low map similarities. For example Cell 1 for the cross stimuli seems to be a simple ON cell, almost like their idealized cell in 3d. However, even though the exact texture in the RF and large parts of the surround for a large part of the locations is exactly identical for Cross1 and Iso2, as well as Cross2 and Iso1, the cells responses for both iso conditions appear to only be noise, or at least extremely noise dominated. Why would the cell not respond in a phase or luminance dependent manner here?

This could either be due to very high surround suppression in the iso condition (which cannot be judged within condition normalization) or because the cell simply responded much weaker due to recording instability or brain state changes. Without any evidence of significant visual responses, enough spikes in each condition and a stable recording across all blocks, this data is not really interpretable. Instability or generally lower firing rates could easily also explain differences in their decoding accuracy.

Similarly, it is very hard to judge the quality of their imaging data. They show no example field of views or calcium response traces and never directly compare this data to their electrophysiology data. It is mentioned that the imaging data is noisy and qualitatively similar, but some quantification could help convince the reader. Even if noisy, it is puzzling that the decoding accuracy should be so much worse with the imaging data: Even with ten times more included neurons, accuracy still does not even reach 30% of that of the ephys data. This could point to very poor data quality.

2. There is no information on the recorded units given. Were they spike sorted? Did they try to distinguish fast spiking and regular spiking units? What layers were they recorded from? It is well known that there are large laminar differences in the strength of figure ground modulation, as well as orientation tuned surround suppression. If most of their data would be from layer 5, perhaps a lack of clear figure modulation might not be that surprising. This could perhaps also be seen when comparing their electrophysiology data to the imaging data which is reportedly from layer 2/3, where most neurons show larger figure modulation/tuned surround suppression effects. There is, however, no report or discussion of differences in modulation between recording modalities.

3. There is an apparent discrepancy between Figure 5d and i. How can their modulation index be around -0.1 for cross (Figure 5d) - which would correspond to on average ~20% weaker responses to a figure than to background, when their PSTH (5i) shows an almost 50% increase of figure over ground. This positive figure modulation has also been widely reported in the literature (Schnabel, Kirchberger, Keller). Are there different populations of cells going into these analyses?

4. In a similar vein, it is not immediately clear why the average map correlation would be bigger for random cell pairs (~0.2, Fig 3g) than for the different conditions of the same cell (~0, Fig 5b). Could this be due to differences in recording modality (imaging in 3g and ephys in 5b)?

5. The maps in Figure 4 should show the location of the RF, because they cannot be interpreted without knowledge of the RF center and size. For example cell 4 in the iso 1 condition could be a border cell, or could respond to the center of the figure. It is impossible to deduce without knowledge of the location of the RF.

6. It could help the reader to discuss the interpretation of the map correlations in Fig 5 a and b in more detail. My guess is that negatively correlated maps (within cross or iso condition) could come from highly orientation tuned neurons, whereas higher correlation values point to more generally figure/contextually modulated cells (within this condition). While the distribution is far from bimodal, this does not rule out a population of nicely figured modulated cells at the high end of the distribution. It might not be necessary at the level of V1 that the figure modulation be consistent across all textures. It would not be surprising, if orientation contrast-defined, phase contrast-defined and motion contrast-defined figures could be signalled to higher areas by discrete populations of V1 or even LM cells.

7. Some of the behavioural results warrant a little more explanation or discussion, as well. In Figure 2h, the mice seem significantly better on the static version of the iso task, than on the moving one. If statistically significant, this should be discussed. Is this because the static frame was maximally phase offset? Then the figure would indeed be better visible better (bigger phase contrast in more frames) than in the moving condition.

Figure 2 and extended Figure 1c: why is the mouse lemur performing so poorly on average? It also appears to have biggest problems with the cross stimulus early on in training.

Tree shrews seem not to be able to memorize the textures as well as the mice do. Is this because of less deprivation/motivation? Or because of the bigger set of textures in training? This would make memorization harder and could thus lower their overall performance. The comparative aspects are very interesting but the absolute differences in performance could be discussed in more detail or explained better.

8. In Figure 7b, why wouldn't the explanation for the linear decodability in cross also hold for iso? There are phase offsets at the borders that simple cells should readily be able to resolve, just as in the case of orientation discontinuities. Could they make a surround phase model, similar to their surround orientation model, that could more readily capture the iso discontinuities?

Reviewer #1 (Public Review):

This study used a multidimensional stimulus-response mapping task to determine how monkeys learn and update complex rules. The subjects had to use either the color or shape of a compound stimulus as the discriminative dimension that instructed them to select a target in different spatial locations on the task screen. Learning occurred across cued block shifts when an old mapping became irrelevant and a new rule had to be discovered. Because potential target locations associated with each rule were grouped into two sets that alternated, and only a subset of possible mapping between stimulus dimensions and response sets were used, the monkeys could discover information about the task structure to guide their block-by-block learning. By comparing behavioral models that assume incremental learning, quantified by Q-learning, Bayesian inference, or a combination, the authors show evidence for a hybrid strategy in which animals use inference to change among response sets (axes), and incremental learning to acquire new mappings within these sets.

Overall, I think the study is thorough and compelling. The task is cleverly designed, the modeling is rigorous, and the manuscript is clear and well-written. Importantly there are large enough distinctions in the behavior generated by different models to make the authors' conclusions convincing. They make a strong case that animals can adopt mixed inference/updating strategies to solve a rule-based task. My only minor question is about the degree to which this result generalizes beyond the particulars of this task.

Reviewer #2 (Public Review):

The authors trained two monkeys to perform a task that involved sequential (blocked) but unsignalled rules for discriminating the colour and shape of visual stimulus, by responding with a saccade to one of four locations. In rules 1 and 3, the monkeys made shape (rule 1) or colour (rule 3) discriminations using the same response targets (upper left / lower right). In rule 2, the monkeys made colour judgments using a unique response axis (lower left/upper right). The authors report behaviour, with a focus on time to relearn the rules after an (unsignalled) switch for each rule, discrimination sensitivity for partially ambiguous stimuli, and the effect of congruency. They compare the ability of models based on Q-learning, Bayesian inference, and a hybrid to capture the results.

The two major behavioural observations are (1) that monkeys re-learn faster following a switch to rule 2 (which occurs on 50% of blocks and involves a unique response axis), and (2) that monkeys are more sensitive to partially ambiguous stimuli when the response axis is unique, even for a matched feature (colour). These data are presented clearly and convincingly and, as far as I can tell, they are analysed appropriately. The former finding is not very surprising as rule 2 occurs most frequently and follows each instance of rule 1 or 3 (which is why the ideal observer model successfully predicts that the monkeys will switch by default to rule 2 following an error on rules 1 or 3) but it is nevertheless reassuring that this behaviour is observed in the animals. It additionally clearly confirms that monkeys track the latent state that denotes an uncued rule.

The latter finding is more interesting and seems to have two potential explanations: (i) sensitivity is enhanced on rule 2 because it is occurs more frequently; (ii) sensitivity is enhanced on rule 2 because it has a unique response axis (and thus involves less resource sharing/conflict in the output pathway).

The authors do not directly distinguish between these hypotheses per se but their modelling exercise shows that both results (and some additional constraints) can be captured by a hybrid model that combines Bayesian inference and Q learning, but not by models based on either principle alone. A Q-learning model fails to capture the latent state inference and/or the rule 2 advantage. The Bayesian inference model captures the rapid switches to rule 2 (which are more probable following errors on rule 1 and rule 3) but predicts matched discrimination performance for partially ambiguous stimuli on colour rules 2 and 3. This is because although knowing the most likely rule increases the probability of a correct response overall it does not increase discriminability and thus boosts the more ambiguous stimuli. I wondered whether it might be possible to explain this result with the addition of an attention-like mechanism that depends on the top-down inference about the rule. For example, greater certainty about the rule might increase the gain of discrimination (psychometric slope) in a more general way.

The authors propose a hybrid model in which there is an implicit assumption that the response axis defines the rule. The model infers the latent state like an ideal observer but learns the stimulus-response mappings by trial and error. This means that the monkeys are obliged to constantly re-learn the response mappings along the shared response axis (for rules 1/3) but they remain fixed for rule 2 because it has a unique response axis. This model can capture the two major effects, and for free captures the relative performance on congruent and incongruent trials (those trials where the required action is the same, or different, for given stimuli across rules) on different blocks.

I found the author's account to be plausible but it seemed like there might be other possible explanations for the findings. In particular, having read the paper I remained unclear as to whether it was the sharing of response axis per se that drove the cost on rule 3 relative to 2, or whether it was only because of the assumption that response axis = rule that was built into the authors' hybrid model. It would have been interesting to know, for example, whether a similar advantage for ambiguous stimuli on rule 2 occurred under circumstances where the rule blocks occured randomly and with equal frequency (i.e. where there was response axis sharing but no higher probability); or even whether, if the rule was explicitly signalled from trial to trial, the rule 2 advantage would persist in the absence of any latent state inference at all (this seems plausible; one pointer for theories of resource sharing is this recent review: https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(21)00148-0?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS1364661321001480%3Fshowall%3Dtrue). No doubt these questions are beyond the scope of the current project but nevertheless it felt to me that the authors' model remained a bit tentative for the moment.

Reviewer #1 (Public Review):

In the submitted manuscript, the authors observed that Glycine treatment could phenocopy deficiency of NINJ1, a recently discovered cell surface molecule critical for plasma membrane rupture, and also inhibit the aggregation of NINJ1. However, whether Glycine directly inhibits NINJ1 was not examined, and thus, the manuscript falls short of having a significant impact in the field.

Strengths of the manuscript:

1. Timely. There is great interest in understanding the mechanism of plasma membrane rupture.<br /> 2. The data provided using several mouse and human cell culture systems overall support the conclusion that Glycine targets NINJ1-mediated plasma membrane rupture (as the title says).

However, most of the presented data is predictable from previous publications. Direct evidence of the mechanism by which NINJ1 is inhibited by Glycine, or in other words, NINJ1 as the direct target of Glycine, was not provided in this manuscript. It is therefore still possible that Glycine acts indirectly upstream of NINJ1. This possible indirect mechanism can be inferred from previous reports where other amino acids such as Serine also could inhibit cell lysis (reviewed in PMID: 27066896).

Reviewer #2 (Public Review):

In this manuscript, the authors investigated the mechanism by which glycine prevents cell membrane rupture. They found that deficiency of NINJ1 (the key executioner of plasma membrane rupture by forming oligomers) phenocopies the cytoprotection of glycine during lytic cell death, and glycine treatment inhibits the oligomerization of NINJ1. Based on these observations, they claimed that glycine executes its inhibitory effect on cell lysis by targeting and inactivating NINJ1. This study addresses an important subject, because how glycine prevents cell membrane rupture is not understood and the literature is full of the implausible conclusion that it works as an osmoprotectant and that pyroptotic cell rupture is secondary to osmotic changes in cells undergoing pyroptosis, even though the gasdermin pore is very large and should allow the free passage of ions and many small proteins.

Reviewer #3 (Public Review):

This manuscript investigates the basis for the cytoprotective effect of exogenous glycine, which has been known to limit cell lysis in response to various stimuli. The authors propose Ninjurin 1 (NINJ1) as a possible regulator or target of glycine-induced blockade of cell lysis, which is an attractive model, given the recently-described role of NINJ1 in inducing membrane rupture downstream of gasdermin cleavage in response to apoptotic and pyroptotic stimuli. The data that support the conclusion are that the authors report that glycine treatment phenocopies NINJ1 deficiency. They go on to conclude, using both native gel western electrophoresis and fluorescence microscopy to assay NINJ1 aggregation, that glycine treatment prevents higher order NINJ1 oligomerization. The authors test these observations in primary human and mouse cells as well as in human and murine macrophage cell lines. The analysis of the role of glycine in both human and murine cells is a strength of the work. This topic is of broad importance, as the mechanism and manner by which cells die impacts host defense against infection, cancer, and autoinflammatory disease. The mechanisms of terminal cell lysis remain surprisingly unclear as recent studies have found that gasdermin cleavage and oligomerization are not sufficient to mediate cell lysis and that cells can survive in the presence of functional gasdermin D pores. Previous studies have reported that glycine treatment limits the release of some cytoplasmic contents during the activation of pyroptosis, but does not affect the secretion of IL-1 cytokines. This property of glycine phenocopies NINJ1 deficiency, suggesting a possible link between the two. This work, therefore, has the potential to shed further light on the regulation of cell lysis, if the studies can be made more definitive with better quantification and more robust controls, which are currently missing for a large portion of the data.

Overall, the area and topic being investigated are of broad interest. While the manuscript attempts to make inroads into how glycine functions as a cytoprotectant, in its current form, the manuscript does not provide definitive evidence that glycine functions through NINJ1, and the data that are currently provided require substantial development, including the addition of key controls and better quantification of microscopy in order for the authors to robustly make the conclusions that they would like to make.

Reviewer #2 (Public Review):

This manuscript provides additional data about how smell is encoded by insects. The study includes both new experimental measurements and simulations. At present, there are questions about whether simulations are appropriately performed to support experimental measurements.

The main experimental finding reported here is that the same olfactory receptor neurons (ORN) can respond with different temporal dynamics to different odorants. This finding is of interest. However, it is very important to discuss whether the differences in temporal dynamics can be explained by differences in how this odorant is carried by air, as has been described here: https://pubmed.ncbi.nlm.nih.gov/23575828/.

There are several questions that need to be addressed regarding the simulations part of the manuscript.

1) There is a mismatch between the number of ORNs used in the model and in the insect system studied.

2) The demonstration in Figure 5 that motif switching improves odor classification includes motif switching for a given odorant, which is not observed experimentally.

3) The methodology for estimating neural temporal dynamics needs to be corrected to apply to the natural stimuli used here.

Reviewer #1 (Public Review):

In order to study odor response dynamics in the olfactory peripheral organ, Kim et al. employs extracellular sensillum recording from the locust antenna to a set of 4 odors at different concentrations. Using spike sorting to assign odor responses to single olfactory sensory neurons (OSNs), the authors demonstrate that OSNs exhibit four distinct response motifs comprising two types of excitation, namely fast and delayed excitatory responses, as well as inhibitory responses in form of offset responses and inhibition. Notably, OSNs can switch between these four motifs depending on the odor applied. This finding is highly interesting and facilitates odor classification as demonstrated by computational modeling in this study. Furthermore, the authors demonstrate that each response motifs follows different adaptation profiles which further results in an increased coding space. The authors conclude and provide evidence with their model that the experimentally observed response dynamics also facilitate determining the distance to the odor source. The obtained results are novel and demonstrate a new dimension of odor response properties at the peripheral level. However, given that the authors used a very limited set of chemically similar odors and considering that the broad tuning and wiring of OSNs in the locust is special and follows different rules compared to the olfactory circuitry of OSNs in other insects (i.e. locust OSNs do not converge onto a single glomerulus but target multiple glomeruli), I wonder whether the observed distinct response motifs are a general phenomenon or a rather special case. I therefore recommend that the authors discuss their findings in the light of these key issues before general conclusions with regard to odor coding rules is being drawn. Do these response motifs also occur for highly ecologically relevant odors, such as PAN, where a rather specialized olfactory circuit would be assumed? Hence, the MS would benefit if those questions would be addressed as well. In addition, the computational modeling approach is written in specialized terms and is therefore difficult to grasp for readers lacking modeling expertise.

Reviewer #3 (Public Review):

In this contribution, the authors align an extensive analysis of in vivo recordings of olfactory receptor neuron (ORN) responses to odors in the locust with a data-driven mathematical model of ORN population coding. Their results provide novel insights into the temporal dynamics of peripheral encoding of time-varying and naturalistic olfactory input.

The manuscript presents three central experimental results: 1) ORNs odor responses can be grouped into 4 distinct response motifs (response profiles). This has partly been known with respect to the typical excitatory phasic-tonic motif and odor offset responses. The exhaustive data set here is however unprecedented. 2) Individual ORNs can switch their response motif, e.g. from excitatory to inhibitory responses. To my knowledge, this is entirely new, highly interesting, and has strong implications. For one it implies an increased coding space and odor separability, which is supported by the authors' model study. It also bears implications for our understanding of processing in the antennal lobe where projection neurons were shown to exhibit property but this has largely been attributed to network processing within the AL. The authors discuss ephaptic interactions as a possible underlying mechanism. 3) ORNs not only show classical within and across pulse adaptation where the response amplitude reduces, but also the novel result that the offset response can increase across repeated pulses with short inter-stimulus intervals. The data-driven model reproduces the experimental observations and a population model that confirms the assumed increase in coding space. In the temporal domain, the authors then perform simulations that mimic realistic stimulus statistics with stochastic arrival of odor packets of variably short duration. The model with a trained linear filter and a non-linear transfer function faithfully predicts the experimental firing rates.

These results, based on an exhaustive set of experimental data, provide a novel view of peripheral odor coding in insects and they will have a particularly strong impact on biologically realistic computational (spiking) circuit models of sensory processing and sensory-to-motor transformations during odor source navigation in naturalistic simulated odor environments where conclusive data and analysis of ORN signaling has thus far been lacking.

Oppenheimer interacts with his zettelkasten almost daily. He reviews them more often than he originally thought he would.

Reviewer #1 (Public Review):

The current manuscript by Schwager and colleagues describes a mechanism by which poorly migratory MDA-MB-231 cells can be metastatic. This study follows a recent paper from the same group (published in January) demonstrating that these poorly migratory cells are more metastatic than their highly migratory counterparts, and that this is due at least in part to E-Cadherin expression and the ability to form circulating tumour cell (CTC) clusters. In the current study, the authors show that the low migratory cells secrete unique EVs that can activate fibroblasts, concomitant with metastatic progression, and that this function is dependent on the presence of Tg-2. The novelty of this work is in the phenotypic heterogeneity of tumour cells, even within cell lines, and the importance the microenvironment in mediating metastasis associated with this diversity. While interesting, this work uses only one model, which was very recently published. The study, I think, would require repetition within additional models, as well as the inclusion of mechanistic studies designed to determine why the EV cargo differs between the highly and poorly migratory subclones.

Reviewer #2 (Public Review):

This fascinating study describes a possible effect of cancer-generated microvesicles on fibroblasts. Microvesicles from a particularly metastatic line promote more contractile and proliferative fibroblasts, and there is a key role for at least one microvesicle factor - the crosslinking enzyme Transglutaminase-2. A wide range of studies help identify and elucidate these effects, but a few aspects remain unclear.

1. MV- has more crosslinking TGM2 but also less MMP14 degradation, and so ECM is more stable either way. The authors should describe any other factors that would give a similar effect as these. The authors should address: do other genes change with TGM2 knockdown; does MMP14 change? If the latter changes, does it have a more important role than TGM2?

2. Perhaps the cleanest and important study of MV effects is in Fig.6j,k, but it shows in vivo differences that are barely significant or not significant, and compares to 'SF' serum free media as a control. Are serum components detected in Mass Spec? If so, wouldn't this suggest a serum supplemented media is a better control? The serum is usually from another species, which is a further (xenogenic) concern that motivates care and discussion about dose -- especially given the high frequency of injection. Also, is there a survival difference for the mice?

Reviewer #1 (Public Review):

It has previously been shown that deletion of the GluA3 subunit in mice leads to alterations in auditory behavior in adult mice that are older than a couple of months of age. The GluA3 subunit is expressed at several synapses along the auditory pathway (cochlea and brainstem), and in ko mice changes in brainstem synapses have been observed. These previously documented changes may account for some of the deficits in hearing in adult ko mice.

In the current study, the authors investigate an earlier stage of development (at 5 wks) when the auditory brainstem responses (ABRs) are normal, and they ask how transmission persists at inner hair cell (ihc) ribbon synapses in GluA3 ko mice. They discovered that deletion of GluR3A significantly changed 1) the relative expression of Glu A2 (dramatically downregulated) and A4 subunits at SGN afferents, and 2) caused morphological changes in ihc ribbons (modiolar side) and synaptic vesicle size (pillar).

The changes documented in the 5 wk old GluA3ko mice were not necessarily predicted because in general the mechanisms involved in shuffling GluA receptors at this synapse (or other sensory synapses) are not completely understood; furthermore, much less is known about the role of differentiation of ihc-sgn synapses along a modiolar-pillar axis. With that said, the only shortcoming of the study is a lack of explanation for the observed changes in the synaptic structure; but this is not specific to this study.

Given the quality of the data and the clarity of presentation of results, this is a very valuable study that will aid and motivate researchers to further explore how auditory circuitry develops, and becomes differentiated, at the level of ihc-sgn synapses.

Reviewer #2 (Public Review):

The goal of the study by Rutherford and colleagues was to characterize functional, structural, and molecular changes at the highly specialized cochlear inner hair cell (IHC) - spiral ganglion neuron (SGN) ribbon synapse in GluA3 AMPA receptor subunit knockout mice (GluA3KO). Previous work by the authors demonstrated that 2-month-old GluA3KO mice experienced impaired auditory processing and changes in synaptic ultrastructure at the SGN - bushy cell synapse, the next synapse in the auditory pathway.

In the present study, the authors investigated whether GluA3 is required for ribbon synapse formation and physiology in 5-week-old mice using a series of functional and light- and electron microscopy imaging approaches. While deletion of GluA3 AMPAR subunit did not affect hearing sensitivity at this age, the authors reported that cochlear ribbon synapses exhibited changes in the molecular composition of AMPARs and pre- and postsynaptic ultrastructural alterations. Specifically, the authors demonstrated that GluA3KO ribbon synapses exhibit i) a global reduction in postsynaptic AMPARs, which is also reflected by smaller AMPAR arrays, ii) a reduction in GluA2 and an increase in GluA4 protein expression at individual postsynaptic sites, and iii) changes in the dimensions and morphology of the presynaptic specialization ("ribbon") and in the size of synaptic vesicles. These reported structural changes are linked to the side of innervation with respect to the IHC modiolar-pillar axis.

The results presented by the authors are conceptually very interesting as the data support the notion that potentially detrimental changes in the molecular composition of a sensory synapse can be compensated to sustain synaptic function to a certain extent during development. The conclusions of the study are mostly well supported by the data, but some experimental details or control experiments are missing or need to be clarified to allow a full assessment.

1. The authors tested which GluA isoforms are expressed in SGNs of GluA3KO mice and reported that only GluA2 and GluA4, and not GluA1, receptor subunits are present in the cochlear. It is, however, a bit difficult to understand why immunolabelling for GluA1 was only performed on brainstem sections (Fig. 1B right) and not in the cochlear to probe for postsynaptic localization at ribbon synapses as it was done for the other isoforms (Fig. 2 and 6) given that GluA3KO IHCs exhibited a larger number of ribbons that lacked GluA2 and 3 (lone or 'orphaned' ribbons; Fig. 6B). It is also not clear why immunolabelling for GluA2 and 4 was performed to probe for expression of these receptor subunits on SGN cell bodies in the cochlear spiral ganglion. Which neurons are expected to synapse onto these somata?

2. The authors state in the text that GluA3 expression is completely abolished in GluA3KO IHCs, however, there appears to still be a faint punctate immunofluorescence signal visible when an antibody directed against GluA3 was used (Fig. 2C). Providing additional information on the specificity of this (and the other) antibodies used in the study would be helpful.

3. The authors reported changes in the volume of the presynaptic ribbon and postsynaptic density surface area in GluA3KO KO animals. The EM data as presented are however not sufficiently convincing.<br /> i) There appears to be a mismatch between the EM data shown in Fig. 3 and 4 and the information in the text with respect to the number of data points in the plots and the reported number of reconstructed synapses. This raises several questions with respect to the analysis. For instance, it is unclear whether certain synapses were reconstructed but excluded from the analysis. If so, what were the exclusion criteria?<br /> ii) The authors compare PSD surface areas in reconstructions from 3D serial sections, but for some of the shown reconstructions (i.e. Fig. 3A' and B' and 4B'), it appears as if PSDs were only incompletely reconstructed.

4. The immunolabelling experiments shown in Fig. 2 and 6 are of very high quality and the quantitative analysis of the light microscopy data (Fig. 6-9) is clearly very detailed, but slightly difficult to interpret the way it is presented. Specifically, it is unclear how the number of synapses per IHC (Fig. 6B) and the separation into modiolar and pillar side (Fig. 8) was achieved based on the shown images without the outlines of individual cells being visible.

5. Adding more detailed information about important parameters (mean, N/n, SD/SEM) and the statistical tests used for the individual comparisons presented in the Figures would help strengthen the confidence in the presented data.

6. In general, the authors report a series of molecular and structural changes in IHCs and reach the conclusion that GluA3 subunits may have a role in "trans-synaptically" determining or organizing the architecture of both the pre- and post-synapse. However, some of the arguments are very speculative and many of the claims are not supported by experimental data presented in the paper. The authors should consider to also compare their findings to studies that investigated ultrastructural changes of AMPAR subunit knockouts in other synapse types, and discuss alternative interpretations (e.g. homeostatic changes).

Reviewer #1 (Public Review):

In this study, Lefebvre et al. investigate the interplay between tissue geometry and the expression patterns of Runt and Tartan in establishing anisotropic myosin localization during germband extension in the Drosophila embryo. Using live and fixed light sheet imaging, computational analysis, and modeling, the authors establish a global time-resolved map of Runt expression and myosin localization during germband extension. They show that a posterior Runt stripe increasingly deviates from the dorsoventral (DV) axis during elongation, while myosin anisotropy in this region transiently deviates from the DV axis and then realigns with this axis after a delay. The authors attribute this delay to the timescale of myosin turnover and the realignment to an unidentified geometric cue. The authors develop a model that can largely account for myosin localization in wild-type, eve mutant, and twist mutant embryos using a myosin lifetime parameter representing myosin turnover. These results provide evidence for a static signal that aligns myosin anisotropy with the DV axis during elongation.

The strengths of this paper are the combination of modeling and quantitative measurements. Powerful in toto measurements show that myosin anisotropy becomes increasingly misaligned with Runt, an essential regulator of myosin planar polarity, at later stages of elongation in posterior regions of the embryo. In addition, the authors present a simple model in which changes in one parameter representing the myosin lifetime can recapitulate the relationship between myosin and edge orientation in wild-type, eve mutant, and twist mutant embryos.

The main weakness of the paper is that the authors do not directly test if their model correctly predicts the myosin lifetime in eve mutants, twist mutants, or in Fat2-RNAi embryos with altered geometry. As myosin turnover is the key parameter in their model, measuring myosin dynamics in these backgrounds would provide an important first test of their model. In addition, the authors should attempt to relate their measurements of myosin dynamics in wild-type embryos to the myosin lifetime value predicted by their model, and they should consider alternative explanations that could account for their observations in wild-type and mutant embryos.

Reviewer #2 (Public Review):

The manuscript by Lefebvre et al. investigates how the tissue-scale spatial organization of protein evolves during germ band extension. The key question is whether changes in the localization of important features such as pair-rule gene (PRG) stripes and apical myosin orientation can be explained purely via passive advection without the need for additional regulatory mechanisms. In the case of the PRG, as well as TLRs, their data strongly suggests the answer is yes: the authors show that the deformation of the characteristic stripe pattern closely matches that predicted by advecting the initial pattern in a velocity field extracted from the observed tissue flow. By contrast, the authors find that anisotropic myosin orientation cannot be explained purely in terms of the local velocity field, in particular the fact that myosin remains robustly oriented with the DV axis. This leads the authors to postulate that myosin orientation is continually re-established via a static source aligned with said axis, which dominates over re-orientation due to advection. A simple model of myosin reorientation is developed from this hypothesis, which produces qualitatively similar relationships between orientation and local vorticity to that seen both in WT and in several mutants.

The strongest feature of this paper is illustrated by the results in Figure 2. The result it presents, which the authors summarize as "PRGs flow with tissue while myosin does not," is a very nice application of recent advances in using toto microscopy for embryonic systems to extract and quantify whole embryo expression patterns and flow fields, which are needed information for this kind of result. Tissue flow is a complicated, active process, and identifying which parts of the dynamics can be sufficiently explained by passive transport can tremendously simplify the conceptual challenges of germ band extension and related tissue movements found during neurulation or organogenesis. The resistance the authors found that myosin exhibits to re-orientation is likewise very interesting because it implies that information about global geometry (the direction of the DV axis) is somehow maintained at the cellular level throughout the convergent extension.

The principle weakness in this manuscript is the vagueness of the proposed static source mechanism and the lack of direct evidence for it in experiments. The FRAP experiments performed here suggest that binding/unbinding happens on the right timescale to play a role in anisotropy maintenance, but if the principle question is 'how does myosin remain oriented along the DV axis' then the static source hypothesis just kicks the can down the road to ask 'how does the static source remain oriented along the DV axis'? The minimal model the authors employ has the benefit that it lets them relate angular deviation to vorticity, at the cost that it is agnostic to the form and nature of the source term, so it cannot be used to extract useful constraints. This said the evidence provided regarding the connection between vorticity and binding rates to myosin deflection is sufficient indirect evidence of the hypothesized mechanism that I suspect it will be of interest to a good number of people interested in epithelial morphogenesis.

Reviewer #3 (Public Review):

Lefevbre et al combine in toto imaging with "tissue cartography" to investigate the respective roles of pair-rule (PR) and toll-like receptor (TLR) gene expression, and embryo geometry, in shaping anisotropic distributions of myosin II during germband elongation (GBE) in Drosophila embryos. The authors find that the simple dependence of Myosin II on PR and TLR expression gradients cannot explain observed global patterns of myosin II. PR and TLR expression patterns evolve continuously as expressing cells are advected by tissue flow during GBE, while myosin II anisotropies remain roughly stationary even as myosin-rich junctions are advected and reoriented by tissue flows. The authors show that the observed spatiotemporal evolution of myosin II anisotropies in wild-type and certain mutant embryos can instead be explained by a simple model in which a geometric cue promotes myosin II accumulation of vertically oriented junctions, flows advect myosin-rich junctions, and myosin II turns over on a ~5-minute timescale.

The core findings are well-supported by rigorous quantitative analysis and modeling; they provide a fresh perspective on the role of geometry in the dynamic control of myosin II anisotropies. Thus they are likely to stimulate further experimental work to identify and characterize the underlying basis for this geometric control.

Key strengths

A key strength is the use of in toto light sheet imaging and tissue cartography, plus the high stereotypy of early Drosophila development, which allows the authors to assimilate data across multiple embryos to extract robust quantitative signatures of gene expression, protein localization, and tissue flows that allows robust analysis of relationships between these different factors in the wild type and across different mutants.

A second strength is the introduction of a very simple model for the evolution of myosin II anisotropy driven by local tissue rotation and myosin turnover which allows decomposing of their respective contributions.

Weaknesses

The power of the model is tested only by its sufficiency to reproduce observed features of myosin II anisotropy over time. There is no direct test/verification of a core model assumption - that the local binding of myosin II is biased with respect to a static geometric signal. Similarly, the inference from the model fits that myosin binding times are reduced in eve mutants has not been confirmed (e.g. by FRAP experiments).

There are a number of (clearly fixable) issues with the clarity of presentation - especially if the authors wish to make their work accessible to a broad audience. The comparison of model predictions and experimental observations is presented in a somewhat confusing way. Ditto for the analysis of mutant phenotypes and the conclusions drawn from this analysis. Some key information about the choices made to justify a very simple model (i.e. why alternative hypotheses and/or additional complexity in the junctional dynamics can be ignored) is presented only in the Supplementary text and should be summarized in the main text.

Reviewer #1 (Public Review):

In this manuscript, Winter and colleagues define the sensitivity of cancer cells lacking the mitochondrial AAA+ ATAD1 to proteasome inhibition. They show that ATAD1 is often co-deleted with PTEN¬ in many different types of cancer. Using two complementary CRISPR screens in two distinct cell models, they identified the mitochondrial E3 ubiquitin ligase MARCH5 as a gene whose deletion is synthetically lethal with ATAD1. Since MARCH5 was previously reported to function to attenuate apoptotic signaling through mechanisms including promoting degradation of pro-apoptotic factors including BIM1, they sought to define the specific role of ATAD1 in regulating pro-apoptotic factor. They present evidence that ATAD1 extracts the pro-apoptotic protein BIMEL from mitochondria to facilitate its inactivation by mechanisms including degradation and inhibitory phosphorylation - a mechanism that appears enhanced during proteasome inhibition. This suggested that ATAD1-deficient cells could be preferentially sensitive to proteasome inhibitors. Consistent with this, expression of ATAD1 in ATAD1-deficient cells decreases sensitivity to proteasome inhibition. Similarly, depletion of ATAD1 in PC3 cells increased sensitivity to proteasome inhibition in xenografts, although somewhat curiously a corresponding increase in BIM was not readily observed (NOXA levels did increase). Finally, the authors show that prostate cancer patients with combined PTEN1/ATAD1 deletion show improved survival as compared to tumors where PTEN1 was deleted alone. Ultimately, these results support a model whereby ATAD1 promotes tumor cell survival and highlights that ATAD1 deletion may represent a vulnerability that can be exploited to treat tumors through the use of proteasome inhibitors.

Overall, this is an interesting and generally well-performed study that defines the mechanistic and functional implications of a genetic 'hitchhiker' in the context of cancer cell survival. The synthetic lethality for ATAD1 and MARCH5 observed using two different genetic approaches (deletion/overexpression) in two different cell models underscores a strong link between these two genes. Further, the data showing an important role for ATAD1 in regulating BIM mitochondrial localization/cytosolic phosphorylation are interesting. The evidence demonstrating relationships between ATAD1 and proteasome sensitivity is also convincing. However, there are some weaknesses. For example, the direct relationship between ATAD1-dependent prosurvival activities and BIM is not clearly defined. This is evident as BIM1 depletion did not influence ATAD1-deficient PC3 cells' sensitivity to bortezomib and BIM was not significantly impacted in the xenograft models. BIM deletion did partially rescue synthetic lethality in Jurkat cells deficient in both MARCH5 and ATAD1, indicating a potential role in those cells. While the authors do address this, these results do create a disconnect within the studies that complicates the overall interpretation, as the specific importance of BIM regulation by ATAD1 in different models is not consistent or always clear. Regardless, this study does reveal new insights into the genetic relationship between ATAD1 deficiency and proteasome inhibition that could have direct therapeutic potential to improve the treatment of patients. Further, considering that the anti-apoptotic roles for ATAD1 appear to extend beyond BIM regulation, this will open new avenues for investigation of the underlying molecular mechanisms whereby ATAD1 contributes to regulating apoptotic signaling in cancer and other models. With that being said, tempering the writing to better highlight that BIM regulation does not explain the ATAD1 protection observed across cancer cell models (it is the case in some, but not all) would be helpful. While there is value in the new mechanistic insight provided into the potential mechanism of ATAD1-dependent apoptotic regulation, more focus on the specific relationship between ATAD1 deficiency and proteasome inhibitor sensitivity would better suit the current work.

Reviewer #2 (Public Review):

This manuscript by Winter et al represents an analysis of the function of the ATAD1 gene in cancer. At present, the manuscript makes a number of interesting observations, with strong experimental support. First, the authors show that tumors with PTEN deletions frequently have additional mutations in ATAD1, and that prostate tumors with both mutations are associated with a shorter period of survival. Second, tumors lacking ATAD1 are more sensitive to proteotoxic stress, based in part on an increased tendency to apoptosis. Third, the ATAD1 protein interacts with BIM, and interactions with BIM contribute in part to an increased tendency to apoptosis. Fourth, ATAD1 and MARCH5 have at least moderate synthetic sick/lethal interactions; together with other data, this suggests they control the release of BIM from the OMM, contributing to its degradation. Overall, the data suggest that tumors with ATAD1 deletions may be particularly vulnerable to drugs that induce proteotoxic stress, suggesting new potential therapeutic regimens, which would be a valuable contribution to the field. The level of data presented here is already substantial; however, some additional experiments to support the authors' contentions would strengthen the work. Some claims about the mechanism are overstated given the current body of data and should be qualified.

Reviewer #1 (Public Review):

Pathogen effectors promote parasitism either in the apoplast or cytoplasm. Unexpectedly, the work described here suggests that FolSpv1 first interacts with SlPR1 in the apoplast and then translocates SlPR1 into the nucleus of tomato plant cells. The authors suggested that the FolSpv1-mediated translocation of SlPR1 into the nucleus prevented the generation of CAPE1, leading to compromised immunity in tomato plants. The study additionally showed that acetylation of FolSpv1 K167 protects the protein from ubiquitination and proteasome-mediated degradation in both the fungal cell and plant cell. Overexpression of SlPR1 or exogenous application of CAPE1 enhanced resistance to F. oxysporum, indicating that CAPE1 contributes to disease resistance to the pathogen in tomato plants. This is consistent with prior reports that CAPE1 positively regulates plant immunity. Y2H screen followed by BiFC and co-IP supported SlPR1 as a target of FolSpv1. Most importantly, incubation of the SlPR1 recombinant protein with FolSvp1 led to uptake of both FolSvp1 and SlPR1 by tomato root protoplasts and nuclear localization of both proteins. Consistent with their model, NLS sequence is required for FolSpv1 virulence function and re-localization of SlPR1 in the nucleus. Furthermore, disease resistance conferred by SlPR1 overexpression in tomato plants could be reversed by overexpression of FolSpv1 in the fungus. Overall, the work represents a potentially significant advance in effector biology of phytopathogens. However, it is too early to exclude the possibility that the nucleus-dependent virulence function of FolSpv1 is independent of CAPE1. It is a bit strange why nuclear localization of SlPR1 is required for preventing CAPE1 generation. The following concerns need to be addressed.

1. Fig 6E shows that CAPE1 is released only upon Fol infection. This appears to contradict with the notion that FolSpv1 prevents CAPE1 release. However, Fol strain overexpressing FolSpv1 prevented the release of CAPE1. It is necessary to compare WT and the mutant strain in which the FolSvp1 gene is deleted. One would expect that the mutant strain induces significantly more CAPE1 release. Similarly, mutant strain complemented with the nls1 construct needs to be tested to see whether nuclear localization is required for preventing CAPE1 release.<br /> 2. SlPR1 is localized in the apoplast in a manner dependent on the signal peptide (Fig 5-figure supplement 1). Overexpression of SlPR1 with added NLS but lacking the signal peptide failed to enhance disease resistance to Fol infection (Fig 7G). What about overexpression of SlPR1 lacking the signal peptide without the added NLS? Does retention of SlPR1 in the cytoplasm sufficient to abolish its function? It is not even discussed why SlPR1 has to be in the nucleus to prevent CAPE1 release.<br /> 3. FolSvp1 carrying the PR1 signal peptide interacted with SlPR1 in the apoplast (Fig 6D and Fig 6-figure supplement 2). Why weren't these proteins translocated into the nucleus? These seem to contradict the in vitro uptake data. It seems that either no or only a very small proportion of SlPR1 transiently expressed in tobacco cells is located in the nucleus. Fig 7C shows that infection of the WT strain, but not the nls1 mutant strain, allowed detection of SlPR1 in the nucleus of tomato cells. However, it is not clear how much of SlPR1 remain in the apoplast or cytoplasm. Is the FolSpv1 protein secreted by Fol sufficient to translocate a significant portion of SlPR1 into the nucleus? The authors are suggested to examine apoplastic and cytoplasmic protein fractions for the relative amounts of SlPR1 after Fol infection.<br /> 4. Fig 7J and 7K, a better experiment would be to pretreat WT tomato plants with CAPE1 prior to inoculation with WT and FolSpv1 OE strains. The pretreatment should eliminate the virulence function of FolSpv1 OE if the virulence is solely dependent on the prevention of CAPE1 release.

Reviewer #2 (Public Review):

In this work, the authors were trying to prove the model that the fungal pathogen Fusarium oxysporum f. sp. lycopersici (Fol) utilizes the acetyltransferase FolArd1 to induce the acetylation of the K167 residue of the effector protein FolSvp1. This acetylation prevents the K152, K258 and K284 ubiquitination-mediated degradation of FolSvp1 in Fol, and meanwhile inhibits the K167 ubiquitination-mediated degradation of FolSvp1 in tomato plants. In the host plants, FolSvp1 interacts specifically with the apoplastic defense protein SlPR1 and translocates it to the nucleus, which suppresses the SlPR1-derived CAPE1 peptide-induced fungal resistance. Overall, the experiments were well designed and the large amount of data justified most of their conclusions. The work sheds novel insight into the virulence mechanisms of fungal effectors by showing that acetylation modification can stabilize a fungal effector, which is able to mis-localize a key defense protein to dampen the host immunity.

There are two issues that need to be addressed.

1. As far as I know, the apoplastic PR1 proteins may have a fungicide activity. When the authors tested the interaction between FolSvp1 and SlPR1 in Nicotiana benthamiana by BiFC, both apoplastic and nuclear interactions could be detected. Therefore, the authors should discuss the possibilities whether the binding of FolSvp1 to SlPR1 remained in the apoplast can inhibit (i) its anti-Fol activity and (ii) the cleavage of SlPR1 to produce the CAPE1 peptide. In other words, although translocating SlPR1 to the nucleus by FolSvp1 is effective for suppressing CAPE1 production, this may not be the only way.

2. The FolSvp1 produced in N. benthamiana was using the SlPR1 signal peptide and lacked the acetylation modification. It is possible that the acetylation of FolSvp1 can affect the interaction affinity or localization between FolSvp1 and SlPR1. The K167Q mutation of FolSvp1 might not be able to faithfully mimic the K167 acetylation.

Reviewer #1 (Public Review):

This report describes evidence that the main driving force for stimulation of glycolysis in DGC neurons by electrical activity comes from influx of Na+ including Na+ exchanging into the cell for Ca2+. The findings are presented very clearly and the authors' interpretations seem reasonable. This is important and impactful because it identifies the major energy demand in excited neurons that stimulates glycolysis to supply more ATP.

Strengths are the highly rigorous use of fluorescent probes to directly monitor the concentrations of NADH/NAD, Ca2+ and Na+. The strategies directly test the roles of Na+ and Ca2+.

Reviewer #2 (Public Review):

This study seeks to determine how neuronal glycolysis is coupled to electrical activity. Previous studies had found that glycolytic enzymes cluster within nerve terminals (in C. elegans) during activity. Furthermore, the glucose transporter GLUT4 is recruited to synaptic surface during activity. The authors previously showed that Ca2+ does not stimulate glycolysis in active neurons. Here, the authors show that the cytosolic Na+, not Ca2+, and the activity of the Na/K pump drive glycolysis. However, it is important to note that in this study, glycolysis was examined in the soma, not nerve terminals, where some of the previous studies were conducted. A few other caveats in the interpretation of the findings are listed below:

1. The NADH/NAD ratio is used throughout as the only measurement reflecting glycolytic flux.<br /> 2. It has been hypothesized that the close association of glycolytic enzymes with ion transporters (such as the Na+/K+ pump) is meant to provide localized ATP to power these pumps. How does bulk glycolysis (monitored with NADH/NAD ratio) relate to localized/compartmentalized glycolysis?<br /> 3. Related to point 2, most of the peredox measurements in the paper have been made at baseline, in the absence of electrical activity. Therefore, it is not clear how the findings relate to activity-driven glycolysis.<br /> 4. The finding that inhibition of SERCA during stimulation actually elevates cytosolic NADH level argues against Na+ being the only ion that regulates glycolysis.<br /> 5. The finding that "SBFI ΔF/F transients were longer in duration than the RCaMP LT transient" does not necessarily mean that Na+ elevation lasts longer than Ca2+ in the cell. This could be an artefact of the SBFI on/off rate relative to RCaMP. In fact, prolonged elevation of cytosolic Na+ would make neurons refractive to depolarization in AP trains.

Reviewer #3 (Public Review):

Meyer et al have studied the mechanisms of glycolysis activation in the hippocampus during neuronal activity. The study is logically laid out, uses sophisticated fluorescence lifetime imaging technology and smart experimental designs. The support for intracellular [Na+] vs [Ca2+] rise driving glycolysis is strong. The evidence for the direct involvement of the Na+/K+ pump is based only on pharmacology using ouabain but the Na+/K+ pump is admittedly not an easy subject for specific perturbations. I still think that the Authors should strengthen the support for the pathway.

Also, there is a long list of publications on the connection between the Na+/K+ pump and glycolysis. It might be useful to highlight the role of the NCX- Na+/K+ pump coupling in the activation of glycolysis in the title.

HLTH Code on ProvenExpert

Reviewer #1 (Public Review):

The 'ForensOMICS' approach is an exciting new area that clearly needs further attention. Despite the current paper being a proof-of-concept, the authors have taken due care and diligence to present the findings of the work in a transparent manner, being careful not to draw hard conclusions based on preliminary experimentation.

Despite being one of the most critical aspects of forensic investigations involving human remains, the estimation of PMI still presents significant challenges. This issue forms the premise of the current work, and this is clearly addressed in both the results and the thorough discussion. The selection of bone tissue as the target matrix is also quite unique and valuable, particularly in scenarios where other more common matrices (like soft tissues) are depleted, as is explained in the work. It is clear that, given further studies and validation, this approach could have a profound impact on the operational world of forensics.

Reviewer #2 (Public Review):

The authors applied an innovative and very interesting approach based on different -omics platforms to study the biological post-mortem transformations of human bone. Despite the study being a proof-of-concept, because of the small number of collected samples and the lack of external validation, the methodology is promising. The study will have a strong impact on the field and present the state-of-art of -omics sciences.

Reviewer #3 (Public Review):

In this paper, for the first time, metabolomics, proteomics, and lipidomics are combined to multi-dimensionally obtain more objective and scientific clues about early and advanced PMI, compared to the traditional methods of PMI estimation that relies on the subjective judgment of morphology. The "ForensOMICS" pipeline establishes a multi-omics analysis pipeline based on the LC-MS platform, which will bring influence and inspiration to the related research of PMI estimation based on molecular biological markers in the foreseeable future. However, due to the limitation of the availability of bone samples and metadata (which might contain covariates with latent influences on the PMI estimation), the current research is still a proof-of-concept study which is incomplete for the "ForensOMICS" approach to be applied in court.

Strengths:

Combing multiple omics and bioinformatics, as claimed by the authors, the "ForensOMICS" approach is more accurate and precise than the conventional morphological methods and molecular biological methods using single omics. Moreover, the research does not stop at developing time-dependent models using several omics biomarkers but carries on the enrichment analysis of relevant markers to further explore the pathophysiology mechanism behind the great changes in the internal environment after death, so as to provide meaningful reference data for the basic forensic research of death.

Data Integration Analysis for Biomarker discovery using Latent variable approaches for Omics studies (DIABLO) method and multiple features selecting tools are used in the bioinformatic process to analyze multiple omics data, and PMI classification model constructed based on PLS-DA, with parameters optimized by 3-fold/100 repeats cross-validation. The overall analysis process is relatively complete, and the data and classification model provided have scientific values for reference.

The "ForensOMICS" workflow in principle is compatible across metabolomics, proteomics, and lipidomics data obtained in different domains of proof-of-concept studies focusing on forensic-related time estimation (e.g. post-mortem submersion interval and time since deposit), for offering relatively complete analysis process.

Weaknesses:

Although the paper does have strengths in principle, the limitation of the availability of bone samples and metadata leads to the major weaknesses of the paper. Therein, age bias samples with single bone type and lack of analysis for environmental factors are the major weaknesses that argue against the key claims in the manuscript by the data presented.

The mean age of body donors is 74 years with {plus minus}11.6 years of standard deviation, while there was only one type of bone tissue (left anterior midshaft tibia). Different structures and locations of the sampled bone tissue as well as metabolic changes and bone degeneration caused by aging may lead to significant discrepancies in different multi-omics data. Moreover, most of the dead found at crime scenes are in the prime of life, and in addition to the tibia, other skeletal remains found at the scenes are commonly skull, ribs, upper limb bones, and teeth. Therefore, the relevant conclusions obtained from the research based on the limited bone samples cannot meet the actual needs for estimating the PMI of skeletal remains. As mentioned by the authors in the discussion, due to the difficulty in acquiring human remain samples with definite post-mortem intervals, this study is still proof-of-concept. If possible, the authors can focus on a larger sample set of different bone remains in younger age groups in future studies.

It is suggested that metadata which may be influence factors of PMI such as temperature, humidity, UV-exposure, and deposition context (which is already recorded) should be recorded and statistically analyzed, so as to further optimize the "ForensOMICS" classification model by considering these possible environmental covariates. In addition, according to the No Free Lunch theorem, PLS-DA is very likely not to be the optimal solution for all the above-mentioned PMI classification tasks based on multi-omics data under different environmental conditions. It is recommended to develop and compare more different classification models for improving the generalization performance of the "ForensOMICS" approach.

Due to the limitation of sample size and the discrete-time gradients, the omics data obtained in the paper could only be applied to build a classification model rather than the regression model. Since such a model does not give a specific predicted PMI with MSE and RMSE indicating its performance, and the current "ForensOMICS" approach failed to distinguish different samples of late PMI (219-834 days), there is still a distance for "ForensOMICS" approach to apply in the actual forensic practice.

Reviewer #1 (Public Review):

The authors of this manuscript report that human DUX4 and mouse Dux4 interact with STAT1 and inhibit interferon-stimulated gene transcription (ISG). The different functional domains of DUX4 were investigated to evaluate which ones are necessary for ISG. DUX4 transcriptional activity was found not to be necessary for ISG, rather the DUX4 C-terminal domain (CTD) was necessary and sufficient to suppress ISG. Employing liquid chromatography-mass spectroscopy (LC-MS), the DUX4 CTD was found to interact with several polypeptides present in human myoblasts. Two key regulators of innate immune signaling, STAT1, and DDX3X, ranked at the top of the list of candidate DUX4-CTD interactors. Immunoprecipitation confirmed DUX4-CTD interaction with STAT1, DDX3X, and several other polypeptides identified by LC-MS. Two regions of DUX4 were found to mediate interaction with STAT1. Amino acids 271-372 were necessary for co-IP of STAT1, and amino acids 372-424, containing (L)LxxL(L) motifs, could enhance binding to phosphorylated STAT1. IFN-gamma treatment enhanced DUX4-CTD binding to wild-type STAT1 and of the STAT1-S727A mutant. In contrast, IFN-gamma did not enhance the binding of DUX4 to the STAT1-Y701A mutant, indicating that DUX4-CTD and STAT1 interaction is promoted by STAT1 -Y701 phosphorylation. A mechanistic investigation of DUX4-STAT1 interaction was conducted by chromatin immunoprecipitation which revealed reduced IFN-gamma-induced STAT1 binding and Pol-II recruitment at promoters of several ISGs. Treatment with IFN-gamma of myoblasts derived from patients affected by facioscapulohumeral dystrophy (FSHD) showed that myoblasts expressing endogenous DUX4 failed to express the IDO1 gene which was, on the other hand, expressed in FSHD myoblasts not expressing DUX4. The majority of Ewing fusion-negative small blue round cell sarcomas have a genetic rearrangement between the CIC and DUX4 genes creating a fusion protein containing the C-terminal (L)LxxL(L) motif of DUX4. The Kitra-SRS sarcoma cell line expresses CIC-DUX4. IFN-gamma treatment of the Kitra-SRS cells showed very low induction of ISGs. Knock-down of the CIC-DUX4 fusion RNA resulted in a substantially increased IFN-gamma induction of ISGs whereas a corresponding knock-down in human myoblasts, which do not express CIC-DUX4, did not alter ISG induction.

This is an important and compelling study that sheds light on a molecular mechanism by which DUX4 inhibits IFN-mediated immune response with potential translational relevance for the treatment of DUX4-expressing cancers. The experiments are rigorously executed and controlled for, and the conclusions are well supported by the presented data.

Reviewer #2 (Public Review):

The goal of this study was to understand the molecular mechanism of how transcription factor DUX4, which has a role in cancer, inhibits the induction of genes stimulated by interferon-gamma. The authors achieved this goal, and their results mostly support their conclusions. They found that DUX4, in their experimental model, interacts with STAT1, thereby decreasing STAT1 and Pol-II recruitment to sites of gene transcription.

The present study has many strengths: The topic is of broad interest, the findings are novel and intriguing, the experiments are well-designed and controlled, the data, with one exception, is carefully interpreted, and the manuscript is very well-written.

Two major weaknesses were identified. One is that all experiments, except Figure 6, rely on one experimental setup, which is a human skeletal muscle cell line with an integrated doxycycline-inducible transgene. The concern is that both the treatment of cells with the drug doxycycline and the fact that signaling pathways could be disrupted in this (immortalized?) cell line could lead to artifacts that skew results. Indeed, results in Figure 4C indicate that total STAT1 is completely localized in the nucleus even prior to interferon stimulation when it should be in the cytoplasm. The other weakness is the use of the DUX4-C-terminal-domain (DUX4-CTD) mutant for the majority of the mechanistic experiments. The concern here is that although the phenotype of ISG repression is observed in this truncated mutant, important regulatory domains could be missing that modulate the interaction with STAT1 or other proteins. Is the NLS added after the flag tag identical to the endogenous NLS? Related, I disagree with the interpretation of Figure 4C that "this interaction happens within the nuclei of DUX4-CTC expressing cells". The interaction could happen prior to STAT1 shuttling to the nucleus.

Reviewer #1 (Public Review):

The authors sought to identify the relationship between social touch experiences and the endogenous release of oxytocin and cortisol. Female participants who received a touch from their romantic partner before a stranger exhibited a blunted hormonal response compared to when the stranger was the first toucher, suggesting that social touch history and context influence subsequent touch experiences. Concurrent fMRI recordings identified key brain networks whose activity corresponded to hormonal changes and self-report.

The strengths of the manuscript are in the power achieved by collecting multi-faceted metrics: plasma hormones across time, BOLD signal, and self-report. The experiment was cleverly designed and nicely counterbalanced. Data analysis was thorough and statistically sophisticated, making the findings and conclusions convincing.

This work sheds new light on potential mechanisms underlying how humans place social experiences in context, demonstrating how oxytocin and cortisol might interact to modulate higher-level processing and contextualizing of familiar vs. stranger encounters.

Reviewer #2 (Public Review):

To test how oxytocin impacts the brain and the psychological, neural, and hormonal response to touch, the authors tested human females during two counterbalanced fMRI sessions wherein females were stroked on the arm or the palm, by a real-world romantic partner or a stranger, while blood levels of oxytocin and cortisol were collected at multiple time points.

This combination of measures, and the number of hypotheses that could be tested with them, is remarkable - virtually unheard of. This impressive, difficult, and more ecological design than is typical for the field is a major strength of the study, which allowed the authors to test many important hypotheses concurrently and to show contextual effects that could not otherwise be observed. The only potential drawback perhaps is that with such a large design, including many measures, the authors produced so many significant interactions and results that it could be hard for the casual reader to appreciate the importance of each.

The authors supported their hypothesis that oxytocin effects are context-sensitive, as they found a key interaction wherein experiencing the partner first increased oxytocin for the partner relative to when they came first the OT levels were low but then increased if they were preceded by the partner (excepting one timepoint). Cortisol responses (which reflect hormonal stress) were also higher when the stranger came first than when he was preceded by the partner). In addition, touch was experienced more positively on the arm than on the palm, supporting the role of c-fibers in conveying specifically felt responses to warm, tender touch.

These data indicate significant context sensitivity with real-world implications. For example, experiencing warm touch on the arm can make us more receptive to other people in subsequent encounters. Conversely, when strangers try to approach and get close to us "out of the blue" people experience this as stressful, which reduces the pleasantness of the interaction and may reduce trust in the moment...perhaps even subsequently.

This research is critical to the basic science of neurohormonal modulation, given that most of this research occurs in rodents or in simplified studies in humans, usually through intranasal oxytocin administration with unclear impacts on circulating levels in the brain and blood. Oxytocin in particular has suffered from oversimplification as the "love drug" - wherein people assume that it always renders people more loving and trusting. The reality is more complex, as they showed, and these demonstrations are needed to clarify for the field and the public that neurohormones adaptively shift with the context, location, and identity of the social partner in an adaptive way. These results also help us understand the many null effects of oxytocin on trusting strangers in human neuroeconomic studies. In a modern world that is characterized by significant loneliness, interactions with strangers and outsiders, and touch-free digital interactions, our ability to understand the human need for genuine social contact and how it impacts our response to outsiders (welcomed in versus a source of stress) is critical to human health and the wellbeing of individuals and society.

Reviewer #3 (Public Review):

In an ambitious, multimodal effort, Handlin, Novembre et al. investigated how the endogenous release of oxytocin and cortisol as well as functional brain activity are modulated by social touch under different contextual circumstances (e.g. palm vs. arm touch, stranger vs. partner touch) in neurotypical female participants.

Using serial sampling of plasma hormone levels in blood during concurrent functional MRI neuroimaging, the authors show that the familiarity of the interactant during social touch not only impacts current hormonal levels but also subsequent hormonal responses in a successive touch interaction. Specifically, endogenous oxytocin levels are significantly heightened (and cortisol levels dampened) during touch from a romantic partner compared to touch from an unfamiliar stranger, at least during the first touch interaction. During the second touch interaction, however, oxytocin levels plummeted when being touched by a stranger following partner touch (although a recovery was made), whereas the normally elevated oxytocin responses to partner touch were dampened when following stranger touch. These results are paralleled by similar familiarity- and order-related effects in neural regions involving the hypothalamus, dorsal raphe, and precuneus.

However, an important distinction to be made is that, although a significant main effect of familiarity was encountered in several brain regions when taking peak plasma oxytocin levels into account, subsequent t-tests showed no activation differences in the BOLD response between partner and stranger touch within the same subjects. Significant interaction maps seem thus mainly driven by between-subject effects at the different time points, which is arguably due to differences between subjects in their initial calibration of neural/hormonal responses, and not session-to-session changes within the same subjects.<br /> A similar comment can be made for the reported covariance between (changes in) maximal oxytocin levels and (changes in) BOLD activity for the hypothalamus.

In an effort to delineate the complex cascade of responses induced by afferent tactile stimulation, the authors report an exploratory regression analysis to identify BOLD activation that precedes the pattern of serial plasma changes in oxytocin levels (looking backwards; i.e. implying changes in brain activation drive changes in hormonal plasma levels). Although the authors are appropriately modest about the significance of the encountered effects, additional control analyses could bring further clarifications about the temporal (e.g., can similar covariations also be found when looking forward) and hormonal specificity (e.g. can similar findings be found for cortisol-variations) of the encountered results. Nevertheless, despite the 'dynamically' covarying relationships between BOLD and max plasma oxytocin levels (i.e. dynamic as in the sense across conditions, not across timepoints), claims about the directionality of this effect (i.e. 'hormonal neuromodulation' vs. 'neural modulation of hormonal levels') remain speculative.

A particular strength of this study is the employment of a "female-first" strategy since experimental data concerning endogenous oxytocin levels in women are sparse. Adequate control analyses are reported to take potential variability due to differences in contraception and phase in the hormonal cycle into account.

Reviewer #1 (Public Review):

Overview:

In this work, the authors set to study the effects of topographic connectivity in a hierarchical model of neural networks. They hypothesize that the topographic connectivity, often observed in cortical networks, is essential for signal propagation and allows faithful transmission of signals.

To study the effects of topographic connectivity on the dynamics, the authors consider a network composed of several layers. Each layer is a recurrent neural network with excitatory and inhibitory subpopulations. The excitatory neurons in each layer enervate a subpopulation of the following layer. The receiving excitatory subpopulation targets a specific group in the next layer and so on. This procedure leads to separate channels that carry the inputs through the network. The authors study how the degree of specificity in each targeted projection, called 'modularity,' affects signal propagation through the network.

The authors find that the network reduces noise above a critical level of network modularity: the deep layers show a clear separation of an active channel and inactive channels, despite the noisy input signal. They study how different dynamical and structural properties affect the signal propagation through the network layers and suggest that the dynamics can implement a winner-takes-all computation.

Strengths and novelty:

- Topographic projections, in which subpopulations of neurons target specific cells in efferent populations, are common in the central nervous system. The dynamic and computation benefits of this organization are not fully understood. With their simple model, the authors were able to quantify the amount of topographic structure and selectivity in the network and study its impact on the network's steady-state. In particular, a bifurcation point suggests a qualitative difference between networks with and without sufficient topographic modularity.<br /> - The theoretical analysis in the paper is rigorous, and the mean-field study shows good agreement with computer simulations of the model.<br /> - The authors describe simulation results of networks with different dynamical properties, including rate-based networks, integrate-and-fire neurons, and more realistic conductance-based spiking neurons. All simulations exhibit similar qualitative behavior, supporting the conclusion that the behavior due to structural modularity will carry to more complex and biologically relevant neural dynamics.<br /> - Overall, the authors convince that the topographic structure of the network can lead to noise reduction, given that the input to the network is provided as distinct channels.

Weaknesses:

The authors support their hypothesis and show a relation between topographic connection and noise reduction in their model. However, I find the study limited and struggle to see the impact it will have on the field. The paper is purely theoretical; it does not provide any physiological evidence that supports the conclusion. On the other hand, and this is the key issue, I do not find real theoretical insights in this work. In the following, I elaborate on why I hold this opinion.

- The hypothesis is that topographic projections in cortical areas allow faithful signal propagation. However, as the authors point out, reliable transmission can be achieved in other ways, such as by direct routing of information (lines 17-19). Furthermore, denoising can be accomplished by a simple feedforward network (e.g., ref 38) without E/I balance and with plasticity rules that do not require topographic connectivity. Thus, I find the computational model not well motivated.<br /> - The task studied here is a simple classification of static inputs: the efferent readout needs to identify the active channel. Again, this could be achieved by a single layer of simple binary neurons [Babadi and Sompolinsky 2014]. The recurrent connectivity and E/I balance suggest that dynamics should play an essential part in the model. However, the task is not well suited for understanding the role of dynamics.<br /> - The authors perform a mean-field study to explain how modularity affects signal propagation. At the heart of their argument is that the E/I network exhibit bistability. However, bistability can be achieved by an excitatory population with a threshold [Renart et al., 2013]. The role of the inhibitory population does not seem crucial for the task and questions the motivations for this analysis.<br /> - Active and inactive channels are decided by the two stable states of the network: the high and the low activity regimes. However, noise fluctuations and their propagation through the network may have a prominent role in the overall dynamics. I find that noise fluctuation analysis is bluntly missing in this work.<br /> - The main finding is a critical level of modularity, m=~0.83, above which the network shows denoising properties of silencing inactive channels and increasing the mean activity of active ones. However, the critical modularity is numerically demonstrated and is not derived theoretically. For a theoretical insight into this transition between denoising and mixing properties of the network, I would have liked to see a more rigorous discussion on the critical value. What does the critical point depend on? The authors show that the single-neuron dynamics do not affect the critical value, but what about other structural elements such as the relative efficacies of the E/I and the feedforward connectivity matrices? Do the authors suggest that m=0.83 is a universal number? I expect a more detailed analysis and discussion of this core issue in a theoretical paper.

To conclude my main criticism, I believe that a theoretical paper should offer a more in-depth analysis and discussion of the core ideas presented and not rely mainly on simulations. For example, to provide theoretical insight, the authors should address central questions such as the origin of the critical modularity, the role of the recurrent balance connectivity, and how the network can facilitate computations other than winner-takes-all among channels. Alternatively, if the authors aim to describe a neural dynamics model without deep theoretical insights, I would expect to see physiological evidence supporting the suggested dynamics.

Conclusions:

The model studied by the authors is novel and provides a valuable way of exploring the effects of modularity and topographic connectivity on signal propagation through hierarchical recurrent neural networks. However, the study lacks theoretical insights into cortical circuit functions in its current version. I believe that for this work to impact the field, it needs to show further analysis and not rely on a numerical study of the model with limited theoretical derivations.

Reviewer #1 (Public Review):

This paper tests the hypothesis that 1/f exponent of LFP power spectrum reflects E-I balance in a rodent model and Parkinson's patients. The authors suggest that their findings fit with this hypothesis, but there are concerns about confirmation bias (elaborated on below) and potential methodological issues, despite the strength of incorporating data from both animal model and neurological patients.

First, the frequency band used to fit the 1/f exponent varies between experiments and analyses, inviting concerns about potentially cherry-picking the data to fit with the prior hypothesis. The frequency band used for fitting the exponent was 30-100 Hz in Experiment 1 (rodent model), 40-90 Hz in Experiment 2 (PD, levodopa), and 10-50 Hz in Experiment 3 (PD, DBS). Ad-hoc reasons were given to justify these choices, such as " to avoid a spectral plateau starting > 50 Hz" in Experiment 3. However, at least in Experiment 3 (Fig. 3), if the frequency range was shifted to 1-10 Hz, the authors would have uncovered the opposite effect, where the exponent is smaller for DBS-on condition.

Second, there are important, fine-grained features in the spectra that are ignored in the analyses, which confounds the interpretation.

One salient example of this is Fig. 2, where based on the plots in B, one would expect that the power of beta-band oscillations to be higher in the Med-On condition, as the oscillatory peaks rise higher above the 1/f floor and reach the same amplitude level as the Med-OFF condition (in other words, similar total power is subtracted by a smaller 1/f power in the Med-ON condition). But this impression is opposite to the model-fitting results in C, where beta power is lower in the Med-ON condition.

Another example is Fig. 1C, where the spectra for high and low STN spiking epochs are identical between 10 and 20 Hz, and the difference in higher frequency range could be well-explained by an overall increase of broadband gamma power (e.g. as observed in Manning et al., J Neurosci 2012, Ray & Maunsell PLoS Biol 2011). This increase of broadband gamma power is trivially expected, as broadband gamma power is tightly coupled with population spiking rate, which was used to define the two conditions.

The above consideration also speaks to a major weakness of the general approach of considering the 1/f spectrum a monolithic spectrum that can be captured by a single exponent. As the authors' Fig. 1C shows, there are distinct frequency regions within the 1/f spectrum that have different slopes. Indeed, this tripartite shape of the 1/f spectrum, including a "knee" feature around 40-70 Hz which is well visible here, was described in multiple previous papers (Miller et al., PLoS Comput Biol 2009; He et al., Neuron 2010), and have been successfully modeled with a neural network model using biologically plausible mechanisms (Chaudhuri et al., Cereb Cortex, 2017). The neglect of these fine-grained features confounds the authors' model fitting, because an overall increase in the broadband gamma power - which can be explained straightforwardly by the change in population firing rates - can result in the exponent, fit over a larger spectral frequency region, to decrease. However, this is not due to the exponent actually changing, but the overall increase of power in a specific sub-frequency-region of the broadband 1/f activity.

Reviewer #2 (Public Review):

In their manuscript, Wiest and colleagues focus on testing two primary hypotheses. The first is that the aperiodic exponent from the intracranial EEG / LFP reflects to population EI balance, and the second is that Parkinson's disease is specifically associated with reduced inhibition-concomitant excessive excitation-in the STN.

To accomplish this, they make use of data from 24 patients with Parkinson's disease who have undergone surgery to implant a deep brain stimulator as part of the treatment of their disease. These patients provide a rare opportunity to record high signal-to-noise EEG/LFP data directly from the human brain. These data are complemented by an additional dataset collected from eight 6-OHDA-lesioned rats, which provide a model of Parkinson's disease. The rat data includes both single-unit spiking activity, which allows Wiest and colleagues to examine periods of relatively high- or low-firing as a proxy for excitatory tone, as well as LFP data which allows them to bridge to the human data and more directly test their first hypothesis that the aperiodic exponent reflects EI balance.

Overall this is a very strong paper. The cross-species approach is especially convincing, and the methods are well-implemented and sound. The authors use appropriate analysis tools and statistical methods, and their inferences are clear, but measured. Their results are convincing, and the potential for aperiodic activity to serve as a potential physiologically interpretable index of Parkinsonian state.

Reviewer #3 (Public Review):

In their manuscript Christoph Wiest and colleagues tested the recently established excitation/inhibition (E/I) hypothesis in data from both patients suffering from Parkinson's disease (PD) and a PD rodent model. In particular, they study activity from the basal ganglia, primarily the subthalamic nucleus (STN). It is a thoughtful work which uses sound methods and is well-written and well-structured. The figures are strikingly good.

The authors demonstrate that the aperiodic exponents and power at 30-100 Hz in such data reflect changes in basal ganglia network activity.

Strengths:<br /> - The clear aim and the rare and valuable rodent and patient data under study.<br /> - The cross-species approach.<br /> - Clear perspective towards adaptive deep brain stimulation application.<br /> - Excellent integration in the existing body of knowledge.

Weaknesses:<br /> - No clear link between findings and symptom severity.<br /> - Relatively low number of animals/patients.<br /> - Limited consistency of results across individual data set.<br /> - In parts weak correlations.

All in all, the present manuscript provides initial evidence that the E/I hypothesis is also valid for neurophysiological data from the STN in PD patients and corresponding rodent models.<br /> This is an important finding which will strengthen the idea of the E/I hypothesis in general and also further substantiates our knowledge about neurophysiological activity of the STN.

Reviewer #1 (Public Review):

The manuscript by Vitet et al. reveals the role of the motor adaptor protein Huntingtin in regulating the pool of synaptic vesicles via its phosphorylation and binding to Kinesin-3 motor protein on one end and synaptic vesicle precursors on the other. The authors use both genetic models of mice harboring mutations in the HTT gene that either mimic constitutive phosphorylation of Huntingtin protein or a phospho-dead version of it. Despite previous reports suggesting no functional outcome for these mutations, using modified motor tests, the authors identified that constitutive phosphorylation of huntingtin impairs the motor skill learning of mice. Next, in a set of elegant and multidisciplinary methods, including electrophysiological recordings in acute slices, TEM imaging, knock-out rescue assay, and biochemical and in-vitro approaches, the authors suggest the mechanism for this dysfunction is through the accumulation of synaptic vesicles in the constitutive phosphorylation mode of huntingtin which increases the release probability and the corticostriatal network. The authors show that this accumulation is mediated by enhanced interaction between vesicular and phosphorylated huntingtin with Kinesin-3 motor proteins which drives the anterograde transport of synaptic vesicle precursors towards the axons and synaptic terminals.

Altogether, this reviewer finds this manuscript well written, well performed, comprehensive and convincing. The new findings in this work are a fundamental addition to the understanding of both basic mechanisms of neuronal function, as well as their dysfunction in neurodegenerative diseases, in this case, Huntington's disease.

Reviewer #2 (Public Review):

This work presents valuable evidence of the connection between Huntingtin's (HTT) phosphorylation state and the recruitment of Kif1A in the axonal anterograde trafficking of synaptic vesicles precursors (SVPs). In brief, the authors describe how phosphorylation of HTT in Serine 421 determines the recruitment of the anterograde molecular motor Kif1A to SVPs, increasing their rate of transport along the axons to the synapse. This conclusion is substantiated by the measured impact of HTT phosphorylation on motor skills learning ability.

The study presents a variety of investigative angles, combining both ex vitro and in vivo approaches. The use of custom microfluidics chamber to recreate neuronal circuits is a point of strength as it allows for in depth analysis of the transport phenotype. This tool could be a very useful tool for the community to explore for a variety of similar studies. The use of mouse models also adds credibility to the physiological importance of the findings.<br /> The evidence presented supports the claims, though more emphasis could be added to the explanation and mechanisms behind how an increased transport dynamic of SVPs due to HTT phosphorylation, results in a detrimental effect on motor skill learning. This finding is perhaps the most critical as it reiterates the importance of balance in SVPs transport and highlights how the system is finely regulated and sensitive to both down and upregulation. This fine tuning might ensure the presence of the proper quantity of SVs at synapses to guarantee an effective synaptic function.

This works adds an important angle to role of HTT phosphorylation, which could open new avenues of treatment for HTT disease based on the manipulation of HTT phosphorylation state.

Reviewer #1 (Public Review):

In the current manuscript, scRNA data of the early "ventral nerve cord" and optic system of the adult brain are compared. The authors generated scRNAseq data for the embryo and integrated existing data sets from other labs and extracted repo-positive glial sets to present a description of the transcriptional landscape of glial cells. The main message of the paper is that morphological diversity among glial cells in a given class is not a strong predictor of transcriptional identity.

However, the data on embryonic "ventral nerve cord" glia are generated from whole embryos, and even provided that the ventral nerve cord harbors 75% of all glia and thus the majority is ventral nerve cord, the data should not be called vnc-specific. The vnc-specific data set (adult CNS) that is already published (Allen et al., 2020) is strangely not even mentioned in the current manuscript. The idea of having a comprehensive description of glial transcriptional profiles is great - but I was missing the integration of the midline glial cells, which can be considered as ensheathing glial cells that - as the cortex glia - also express wrapper (Stork et al., 2009).

Unfortunately, I found most of what is reported in this work not to be entirely new. The classification of glial diversity in the adult brain was presented by the Meinerzhagen and Gaul labs (Edwards and Meinertzhagen, 2010; Edwards et al., 2012; Kremer et al., 2017). The description of two astrocyte-like cell types is a reduction of data that defined three morphologically distinct astrocyte-like cells (Peco et al., 2016), which is not discussed. Some other aspects were ignored, too. Two other morphological distinct types of ensheathing glia exist, ensheathing glia and ensheathing/wrapping or track-associated glia were described but this is not discussed (Kremer et al., 2017; Peco et al., 2016).

Reviewer #2 (Public Review):

Ines Lago-Baldaia et al. investigate the connection between transcriptional and morphological diversity of glial cells. This is an important question to answer in the glial biology field and has been amplified by recent advances in single-cell sequencing. It remains unclear if transcriptional diversity that is often reported in scRNA analysis equates to morphologically distinct glia. To explore the correlation between transcriptome and morphology, the authors utilize the strength of the Drosophila model system to demonstrate that although morphotypes of glia can be identified in the nervous system, the morphotypes do not correlate with a distinct transcriptional profile. Overall, the paper is well written and the conclusion matches the results that are presented. This work will be an important contribution to the glial biology field.

Reviewer #3 (Public Review):

Our brain is comprised of both electrically active neurons that transmit information and an equal number of a set of cells called glial cells, which are actually comprised of many different cell types with a variety of functions. Compared to neurons, we know much less about glia and therefore need model systems in which they can be studied.

This study reports the generation of an atlas of glial cells in the Drosophila fly model. Drosophila glia have many similarities with those of vertebrates and are a useful model system for the interrogation of glia due to their simplicity and ease of genetic manipulation to better understand glial cell biology. This study catalogued the morphology of various types of glia in different areas of both the developing and adult fly, building a repository that will be of immense value to researchers. The study also aimed to determine how the shape of different glia, or even within glia of the same type related to the genes that they expressed, and their molecular state. The study found that while cell morphology was tightly linked to gene expression state in some cases, in others it was not, meaning that cells with very different shapes had very similar gene expression profiles/ molecular states. This latter finding suggests that at least some glial cells' shapes are more likely controlled by their interactions with the environment or molecular events that are independent of gene expression per se. The study is very impressive in its depth of characterisation and will come to represent a very useful resource for the community of biologists who employ Drosophila to understand glial cell biology.

Reviewer #1 (Public Review):

The authors show that metformin reduced the elevated intraocular pressure in mice with steroid-induced ocular hypertension and attenuated damage to the cytoskeleton of the ocular trabecular meshwork. In human trabecular meshwork cells, the authors showed that the protective effects of metformin against oxidative injury were exerted by regulating cytoskeleton remodeling through integrin/ROCK signals.

Strengths of the paper include the rigorous methodology and support of the data for the conclusions. The work has the potential to advance glaucoma research but also the use of metformin for reversing other states of oxidative injury, such as fundamental aging mechanisms, in multiple tissues.

Reviewer #2 (Public Review):

What the authors were trying to explore is very interesting with translational potential toward glaucoma treatment. They used a topical dexamethasone (dex) induced mouse model showing ocular hypertension and a culture model using human TM cells treated with tBHP to induce TM oxidative stress. Their results suggested that metformin protected TM cells from cytoskeletal destruction by enhancing the integrin/ROCK pathway and alleviated elevated IOP in the mouse model. However, the provided simulative results were vague and the research needs extra experimental data to support its conclusion.

Reviewer #1 (Public Review):

Iyer et al. address the problem of how cells exposed to a graded but noisy morphogen concentration are able to infer their position reliably, in other words how the positional information of a realistic morphogen gradient is decoded through cell-autonomous ligand processing. The authors introduce a model of a ligand processing network involving multiple "branches" (receptor types) and "tiers" (compartments where ligand-bound receptors can be located). Receptor levels are allowed to vary with distance from the source independently of the morphogen concentration. All rates, except for the ligand binding and unbinding rates, are potentially under feedback control. The authors assume that the cells can infer their position from the output of the signalling network in an optimal way. The resulting parameter space is then explored to identify optimal "network architectures" and parameters, i.e. those that maximise the fidelity of the positional inference. The analysis shows how the presence of both specific and non-specific receptors, graded receptor expression and feedback loops can contribute to improving positional inference. These results are compared with known features of the Wnt signalling system in Drosophila wing imaginal disc.

The authors are doing an interesting study of how feedback control of the signalling network reading a morphogen gradient can influence the precision of the read-out. The main strength of this work is the attention to the development of the mathematical framework. While the family of network architectures introduced here is not completely generic, there is enough flexibility to explore various features of realistic signalling systems. It is exciting to find that some network topologies are particularly efficient at reducing the noise in the morphogen gradient. The comparison with the Wnt system in Drosophila is also promising.

Major comments:

- The authors assume that the cell estimates its position through the maximum a posteriori estimate, Eq.(5), which is a well-defined mathematical object; it seems to us however that whether the cell is actually capable of performing this measurement is uncertain (it is an optimal measurement in some sense, but there is no guarantee that the cell is optimal in that respect). Notably, this entails evaluating p(theta), which is a probability distribution over the entire tissue, so this estimate can not be done with purely local measurements. Can the authors comment on this and how the conclusions would change if a different position measurement was performed?

- One of the features of the signalling networks studied in the manuscript is the ability of the system to form a complex (termed a conjugated state, Q) made of two ligands L, one receptor and one non-signalling receptor. While there are clear examples of a single ligand binding to two signalling receptors (e.g. Bmps), are there also known situations where such a complex with two ligands, one receptor, and one non-signalling receptor can form? In the Wnt example (Fig. 10a), it is not clear what this complex would be? In general, it would be great to have a more extended discussion of how the model hypothesis for the signalling networks could relate to real systems.

- The authors consider feedback on reaction rates - it would seem natural to also consider feedback on the total number of receptors; notably, since there are known examples of receptors transcriptionally down-regulated by their ligands (e.g. Dpp/Tkv)? Also it is not clear in insets such as in Fig. 7b, if the concentration plotted corresponds to the concentration of receptors bound to ligands?

- The authors are clear about the fact that they consider the morphogen gradient to be fixed independently of the reaction network; however, that seems like a very strong assumption; in the Dpp morphogen gradient for instance over expression of the Tkv receptor leads to gradient shortening. Can the authors comment on this?

- Fig. 10f is showing an exciting result on the change in endocytic gradient CV in the WT and in DN mutant of Garz. Can the authors check that the Wg morphogen gradient is not changing in these two conditions? And can they also show the original gradient, and not only its CV?

Reviewer #2 (Public Review):

The work of Iyer et al. uses a computational approach to investigate how cells using multiple tiers of processing and multiple parallel receptor types allow more accurate reading of position from a noisy signal. Authors find that combining signaling and non-signaling types of receptors together with additional feedback increases the accuracy of positional readout against extrinsic noise that is conveyed in the morphogen signal. Further, extending the number of layers of signal processing counteracts the intrinsic stochasticity of the signal reading and processing steps. The mathematical formulation of the model is general but comprehensive in the way it handles the difference between branches and tiers for the processing of channels with feedbacks. The results of the model are presented from simple one-branch and one-tier architecture to two-branch and two-tier architecture with feedbacks. Interestingly authors find that adding more tiers results in only very small improvements in the accuracy of positional readout. The model is tested against a perturbation experiment that impairs one of the signaling branches in the Drosophila wing disc, but the comparison is only qualitative as further experiment-oriented work is planned in a separate paper.

Strengths

There is a clear statement of objectives, model, and how the model is evaluated. In particular, the objective is to find what number of receptor types and their concentrations for a given number of tiers and feedback types is resulting in the most accurate positional readout. The employed optimization procedure is capable to find signalling architectures that result in one cell diameter positional precision for most of the tissue with 3-4 cells at the tissue end that is most distant to the morphogen source. This demonstrates that employing additional complexity in signal processing results in a very accurate positional readout, which is comparable with estimates of positional precision obtained in other developmental systems (Petkova et al., Cell 2019, Zagorski et al., Science 2017).

The optimal signalling architectures indicate that both signalling (specific) and non-signalling (non-specific) receptors affect the precision of positional readout, but the contributions of each type of these receptors are qualitatively different. Even slight perturbation of signalling receptors drives the system out of optimum, resulting in a decrease in positional precision. In contrast, the non-signalling receptors could accommodate much larger perturbations. This observation could provide a biophysical explanation for how cross-talk between different morphogen species could be realized in a way that positional precision is kept at the optimum when morphogen signaling undergoes extrinsic and intrinsic perturbations.

Last, the model formulation allows to specifically address perturbations of signalling and feedbacks, that could be explored to validate model predictions experimentally in Drosophila wing disc, but also in other developmental tissues. The authors present a proof-of-concept by obtaining consistent results of variation of output profiles in two-tier two-branch architectures with non-signaling branch removed and intensity profiles of Wg in wing disc where the CLIC/GEEC endocytic pathway was perturbed.

Weaknesses

The list of model parameters is long including more than 20 entries for two-tier two-branch architectures. This is expected, as the aim of the model is to describe the sophisticated signalling architecture mimicking the biological system. However, this also makes it very challenging or impossible to provide guiding principles or understanding of the system behaviour for the complete space of signalling architectures that optimize positional readout. Although, the employed optimization procedure finds solutions that exhibit very high positional accuracy, there is only very limited notion how these solutions depend on variation of different parameters. The authors do not address the following question, whether these solutions correspond to broad global optima in the space of all solutions, or were rather fine-tuned by the optimization procedure and are quite rare.

It is unclear how contributions from the intrinsic noise affect the system behaviour compared to contributions from extrinsic noise. In principle, the two-branch one-tier architecture results in an already very accurate positional readout across the tissue. The adding of another tier seems to provide only a very weak improvement over a one-tier solution. It is possible that contributions from intrinsic noise for the investigated signalling architectures are only mildly affecting the system compared with contributions from extrinsic noise. Hence, it is difficult to assess whether the claim of reducing intrinsic noise by adding another tier is supported by the presented data, as the contributions from intrinsic noise could overall very weakly affect the positional readout.

Reviewer #1 (Public Review):

Hyperactivation of WNT/b-catenin signaling has been implicated in cancer. How b-catenin enters the nucleus is not completely understood. Using a heterologous model system of budding yeast, authors find that nuclear translocation of b-catenin is mediated by Kap104, the orthologue of TPO1/2. Authors further showed that a PY like motif in the C-terminus of b-catenin binds TPO1 and serves as a nuclear localization signal (NLS). Mutation of the PY like motif or inhibition of TPO1/2 inhibits b-catenin mediated transcription. Overall, this is an interesting study. The evidence that the PY like motif can serve as a NLS in yeast is convincing. However, how much this motif contributes to nuclear localization of full-length b-catenin in mammalian cells is not clear. Authors have relied on transcription readout of b-catenin, which has many caveats. Direct measurement of the level of b-catenin in the nucleus is important.

Reviewer #2 (Public Review):

Hwang et al take an unconventional approach to address a longstanding problem in the field of Wnt signaling and cancer: the mechanism of beta-catenin nuclear import. The authors introduce expression of Xenopus beta-catenin in budding yeast, a heterologous model system that does not harbor any known Wnt signaling components but carries highly conserved nuclear transport machinery. They find that GFP-tagged beta-catenin is actively transported to the yeast nucleus in a Ran-GEF-dependent process, indicating NTR-dependent transport. An elegant rapamycin treatment-dependent Anchor-Away method is applied to systematically inhibit 10 budding yeast NTRs, for which orthologues exist in human cells. Significant and specific inhibition of beta-catenin nuclear import is identified when Kap104 (orthologue of Kapbeta2/Transportin-1 (TNPO1) was anchored to the plasma membrane. Furthermore, nuclear import depends on a PY-like NLS sequence in the beta-catenin C-terminus, which was shown to mediate a direct interaction with TNPO1. A role of the vertebrate paralogs tnpo1/2 and the PY-like NLS was confirmed in Xenopus, using double axis formation assays, and in mouse and human cell lines, combining tnpo1/2 depletion with nuclear localization and reporters for beta-catenin dependent transcription. Finally, the authors provide proof that responses of MEF cells to Wnt3a or human beta-catenin overexpression can be inhibited by treatment with a TNPO1/2 blocking peptide (M9M).

Overall, the results of this study provide a valuable addition to the longstanding and ongoing discussions on the mechanisms of beta-catenin nuclear import. The conclusions are based on a well-focused and solid set of experiments and are confirmed across species in a diverse set of model systems, and findings are discussed against the state of the field. Although the identified TNPO1/2-dependent beta-catenin nuclear import pathway was shown to be a target for peptide-based inhibitory strategies, these findings remain to be confirmed in relevant (colorectal) cancer model systems in which levels of beta-catenin are inappropriately enhanced and inhibition of its nuclear entry is most warranted.

Reviewer #2 (Public Review):

This manuscript puts forward a new idea that topography in neural networks helps to remove noise from inputs. The neural network consists of multiple stages. At each stage, the network is structured to be balanced in terms of the strength of inhibitory and excitatory signals. Because of topography, the networks become "dis-balanced" and receive more recurrent excitatory signals locally for those regions that receive strong initial inputs. This leads to error correction. The main weakness in the manuscript is that the approach will only work for inputs that are constant-in-time. It is important to acknowledge this limitation in both the title and throughout the manuscript.

Reviewer #1 (Public Review):

This manuscript reports the function of FIO1, a mammalian METTL16 homolog, in Arabidopsis. The authors found FIO1 affects early flower phenotype through regulating splicing via U6 m6A modification. This paper confirmed FIO1-mediated m6A methylation on U6 RNA, consistent with two recently published reports. The manuscript contains quite a thorough splicing analysis on how splicing is affected in the fio1 mutant where U6 m6A is absent, and a detailed explanation of how m6A could affect base pairing and secondary structure involving U6 at different temperatures.

1. FLC mRNA can be m6A methylated. The authors appear to suggest the effect is secondary. More analysis and explanation are required. For instance the authors could measure m6A level on FLC in fio1 mutant, mta mutant, and compare it with that of wt.

2. The authors used nanopore m6A sequencing to map m6A in mRNA from wt and fio1 mutant strains. I would suggest either RIP-seq or mass spectrometry measurement to confirm the loss of fio1 leads to limited mRNA m6A changes.

Reviewer #2 (Public Review):

Overall, I think that the screen or mutants in the Arabidopsis flowering pathway and its outcome are biologically interesting and important. The authors show that FIO1 methylates U6 snRNA and not (or rarely) mRNA. However, subsequent to this, the results are entirely from bioinformatics of RNAseq data from the derived mutants; there are no further experiments performed, either to confirm or test newly-derived hypotheses. Furthermore, the main hypothesis, that 5'SS pos.+4 identity is critical for sensitivity to U6 N6-methylation, was already described in yeast S. pombe, based on data from mutants in the pombe ortholog Mtl16. Minimally, the conclusions based on bioinformatics should be confirmed with experimental data. In addition, there are examples throughout the manuscript where the authors state results or conclusions without providing any data; this is not acceptable and data supporting these assertions must be included.

Reviewer #1 (Public Review):

In the article "Neuroendocrinology of the lung revealed by single cell RNA sequencing", Kuo et. al. described various aspects of pulmonary neuroendocrine cells (PNECs) including the scRNA-seq profile of one human lung carcinoid sample. Overall, although this manuscript does not have any specific storyline, it is informative and would be an asset for researchers exploring various new roles of PNECs.

Major comments:<br /> The major concern about the work is most results are preliminary, and at a descriptive level, conclusions or sub-conclusions are derived from scRNA-seq analysis only, lacking in-depth functional analysis and validation in other methods or systems. There are many open-end results that have been predicted by the authors based on their scRNA-seq data analysis without functional validation. In order to give them a constructive roadmap, it would be better to investigate literature and put them in a potential or probable hypothesis by citing the available literature. This should be done in each section of the result part.<br /> The paper lacks a main theme or specific biology question to address. In addition, the description about the human lung carcinoid by scRNA-seq is somehow disconnected from the main study line. Also, these results are derived from the study on only one single patient, lacking statistical power.

Reviewer #2 (Public Review):

Pulmonary neuroendocrine cells (PNECs) are known to monitor oxygen levels in the airway and can serve as stem cells that repair the lung epithelium after injury. Due to their rarity, however, their functions are still poorly understood. To identify potential sensory functions of PNECs, the authors have used single-cell RNA-sequencing (scRNA-seq) to profile hundreds of mouse and human PNECs. They report that PNECs express over 40 distinct peptidergic genes, and over 150 distinct combinations of these genes can be detected. Receptors for these neuropeptides and peptide hormones are expressed in a wide range of lung cell types, suggesting that PNECs may have mechanical, thermal, acid, and oxygen sensory roles, among others. However, since some of these cognate receptors are not expressed in the lung, PNECs may also have systemic endocrine functions. Although these data are largely descriptive, the results represent a significant resource for understanding the potential roles of PNECs in normal biology as well as in pulmonary diseases and cancer and are likely to be relevant for understanding neuroendocrine cells in other tissue contexts.

However, there are several aspects of the data analysis that are unclear and require clarification, most notably the definition of a neuroendocrine cell (points #1 and #2 below).

1. Figure S1 shows the sorting strategy used for isolation of putative PNECs from Ascl1CreER/+; Rosa26ZsGreen/+ mice, and distinguishes neuroendocrine cells defined as ZsGreen+ EpCAM+ and "neural" cells defined as ZsGreen+ EpCAM-; the figure legend also refers to the ZsGreen+ EpCAM- cells as "control" cells. However, the table shown in panel D indicates that the NE population combines 112 ZsGreen+ EpCAM+ cells together with 64 ZsGreen+ EpCAM- cells to generate the 176 cells used for subsequent analyses. Why are these ZsGreen+ EpCAM- cells initially labeled as neural or control, but are then defined as neuroendocrine? If these do not express an epithelial marker, can they be rigorously considered as neuroendocrine?

2. Similarly, in the human scRNA-seq analysis, how were PNECs defined? The methods description states that these cells were identified by their expression of CALCA and ASCL1, but does not indicate whether they also expressed epithelial markers.

3. The presentation of sensitivity and specificity in Figure 1 is confusing and potentially misleading. According to Figure 1B, Psck1 and Nov are two of the top-ranked differentially expressed genes in PNECs with respect to both sensitivity and specificity. However, the specificity of these two genes appears to be lower than that of Scg5, Chgb, and several other genes, as suggested in Figure 1C and Figure S1E. In contrast, Chgb appears to have higher specificity and sensitivity than Psck1 in Figures 1C and E but is not shown in the list of markers in Figure 1B.

4. The expression of serotonin biosynthetic genes in mouse versus human PNECs deserves some comment. The authors fail to detect the expression of Tph1 and Tph2 in any of the mouse PNECs analyzed, but TPH1 is expressed in 76% of the human PNECs (Table S8). Is it possible that Tph1 and Tph2 are not detected in the mouse scRNA-seq data due to gene drop-out? If serotonin signaling by mouse PNECs is due to protein reuptake, as implied on p. 5, is there a discrepancy between serotonin expression as detected by smFISH versus immunostaining?

5. The smFISH and immunostaining analyses are often presented without any indication of the number of independent replicate samples analyzed (e.g., Figure 2B, Figure 3F, G).

6. It would be helpful to provide a statistical analysis of the similarities and differences shown in the graphs in Figures 1E and G.

Reviewer #3 (Public Review):

The authors present a comprehensive profile of signals and sensors expressed in mouse and human PNECs by single-cell RNA sequencing. Analyses revealed a myriad combination of neuropeptide, neurotransmitter, receptor, and channel genes in PNECs. A diverse transcript combination is further enriched by alternative posttranscriptional and posttranslational processing. The authors also surveyed cognate receptors expressed in epithelial cells, endothelial cells, stromal cells, immune cells, and pulmonary sensory neurons, identifying potential local targets for the PNECs signals. The scRNA-seq profile from lung carcinoid tumors suggests that selected PNECs are susceptible to carcinoid transformation. Together, these data indicate that PNECs serve as sentinels to perceive multiple airway stimuli and express a variety of signals that either act locally or potentially through circulation to regulate homeostasis.

Reviewer #1 (Public Review):

In this manuscript, Dhurandhar, Cecchi and Meyer present a model that aims to predict the discrimination performance of human subjects in an odor mixture discrimination task using low-dimensional features, which include intensity, pleasantness and a set of 19 semantic descriptors. Specifically, the authors aim to find a metric of odor mixture similarity in feature space that accurately captures similarity (or discriminability) as judged by human subjects. The semantic descriptors are obtained from a chemoinformatic model previously developed by the authors. A mixture's feature vector is defined as the average of the features of the individual components. A Mahalanobis distance is defined between two mixtures, whose parameters are fit using experimental data from Bushdid et al, Science, 2016 and applied to three other independent datasets. They show that the RMSE in prediction outperforms a previously published model in two of the datasets.

Strengths:

The idea to relate the embedding vector of individual odor components to the embedding of a mixture so as to predict mixture discrimination performance is novel and interesting.

Weaknesses:

1) The authors claims are not supported by the data presented in the Figures. A trivial model which predicts a constant can potentially achieve better predictive performance:

It is difficult to gauge the performance of the model solely from the RMSE as the data and predictions are not plotted (except in a pooled format in Figure 4b, which is however masked by the density plot). The RMSE should at the minimum be compared to the standard deviation of the dataset and plotted as the fraction of variance unexplained. Without knowledge of the standard deviation of the experimental data, it is not possible to judge the quality of the prediction.

An examination of the inset in Figure 2a and Figures 4 shows that the data spans from ~0.54 to ~0.75. Since this was quite comparable to the RMSE of ~0.17 obtained by the author's prediction, I examined the data from the four datasets provided as a supplement by the authors. It turns out that the standard deviations of the discrimination performance (the output variable) are: Bushdid 0.176, Ravia 0.144, Snitz1 0.124, Snitz2 0.119. As these numbers indicate, simply using the constant mean as a prediction will lead to an RMSE of 0.176 for the Bushdid dataset.

This appears to contradict the Middle inset in Figure 2a, which seemingly looks like a good fit. Closer examination of the two plots shows that the experimental data in the two are not the same (note for example the two datapoints with y < 0.45 in the left plot which are absent in the right). Since the authors have not clarified in the caption whether this is an illustration or if it is actual data, it is unclear how to interpret this plot.

2) The data transformations performed to obtain the mixture embedding vector seem arbitrary. For a mixture of 30 components (or even 10), this involves taking an average of 30 feature vectors, which will very likely average out. The authors should explain the rationale for taking the average and not for instance the most common descriptors that appears in the mixture components.

3) Other comments - i) the authors use linear regression to model a classification task. The justification for this choice is not explained. ii) Although this is not primary data from the authors, the authors should perhaps comment on why the minimal performance is not chance level (33%) but instead around 50 percent, even when the percent overlap between the mixtures is close to 100%. Iii) The authors do not define the Direct model. How is the RMSE of the Direct model on the Ravia dataset (0.45) much larger than the standard deviation of the dataset (0.144)?

Reviewer #2 (Public Review):

The authors introduce a model based on textual data for predicting odor properties of a mixture of chemicals. Modelling approach is relevant to olfactory scientists and experimental neuro-scientists.

Work is relevant because it unifies and studies multiple mixture odor datasets, achieving satisfactory results. Work is novel because modelling for mixture datasets is scarce, this work introduces a grounded approach for modeling such data. Model is directly interpretable since it relies on a linear model (lasso) to build mapping between features (metric learning).

The authors's evidence supports most of the conclusions of the work with some room for improvement.

This work can be of the many in the future trying to further modelling approaches for mixture data.

Reviewer #3 (Public Review):

It has been difficult to predict perceptual quality of odor mixtures. In this study, Dhurandhar and colleagues developed a computational method to predict perceptual discriminability of odor mixtures. The authors previously developed a method to predict natural language descriptors from chemical structures of monomolecular odorants (Gutierrez et al., Nat. Commun. 2018). In the new model developed in the present study, the authors used these predicted natural language descriptors to predict the discriminability of odor mixtures. This was done by first averaging the values of natural language descriptors across component odorants in a mixture. The authors then used a Lasso regression to predict the fraction of subjects that correctly discriminated these odors from the Mahalanobis distance between the average descriptors of two odorants. The performance of the model was compared against a "Direct model" in which chemical structures were used directly to compute the vector angles based on the cosine similarity metric.

The authors address an important question and the model that the authors propose is potentially interesting to the community. The method is relatively simple and the manuscript was written relatively clearly. However, I have some concerns on the approach or methods used.

Major concerns

1. The authors compare the new model against the Direct model. The performance was compared based on the root mean squared errors (RMSE). While the result indicates statistically significant improvement, the models differ in multiple ways, and it is unclear what components in the new model contributed to the improvement. The authors should compare a model in which discrimination performance was predicted based on chemical structures using a Lasso regression. Comparison to this model would be necessary to demonstrate that transforming to the natural language descriptors was critical for the improvement, and not due to just the use of Lasso.

2. The authors should compare their model against other classes of model proposed before.

Reviewer #1 (Public Review):

The sequencing of a genome is the first step in identifying the functional regions of that genome. The identification of the regions that encode sequences that will become proteins (protein coding genes) is made complicated by the transcription of the DNA into multiple versions of RNA (isoforms) from the same genome locus. Often these RNA isoforms have different start and stop positions in the genome and also have different sequences (exons) that are used for the protein coding process. Taking advantage of considerable improvements in a recently developed computer algorithm that predicts the most stable three-dimensional (3D) folding of protein sequences (AlphaFold2) Sommer, et al describe a strategy to use this information to evaluate among the multiple isoforms generated by each gene. This approach provides additional information along with sequence conservation, synteny and other genes that are co-regulated that can potentially rank order among isoforms to aid in annotating the protein coding human transcriptome. This capability is needed in determining the boundaries, exon sequences, evolutionary relationships of genes to their ancestral homologues, gene function and the structural regions responsible for disease.

A troubling issue of using this approach is pointed out by the authors themselves, namely, the fact that many functional genes express isoforms that make proteins with poor Local Distance Difference Test (pLDDT) scores. Thus, the 3D structures of a proteins arising from two different isoforms cannot be the only criteria used to identify the gene structure encoded in a locus. However, an isoform encoding a protein with a high pLDDT (estimated to be >80/100) is likely to help define at least a conservative set of boundaries and structures for the annotation for a gene. It would have been useful to have some overall estimate as to the false positive and negative rates of using this strategy. Without this information this approach while useful, could be considered an incremental improvement in the annotation process.

Reviewer #2 (Public Review):

The study by Sommer et al. applies alphafold to the CHESS selection of transcripts with the goal of generating predicted 3D protein structures and a quality measure of folding, the pLDDT score. From these data, the authors build up a database for result exploration. In addition, they provide examples to underline this approach. Examples include proteins, where the authors propose the pLDDT score as a measure of presumed superior biological functionality over other isoforms. The authors also use the generated data to propose novel functionally relevant isoforms, e.g. in the mouse.

The study is based on the elegant idea to aid genome annotation through 3D structure prediction. This is a very powerful approach that allows large-scale data generation for functional interpretation. This approach appears technically sound and well executed (although I may miss details not being a protein expert). However, in my opinion, the authors could make more use of the potential of their approach. From the big-data start, they seem to directly restrict themselves to interesting examples. I am missing a global analysis that shows the bigger picture of their results. Given that they have generated structures from 90,415 isoforms, each associated with a pLDDT score, conservation scores, length, expression levels and other quantifiable data listed on page 18. I would wish for a comprehensive analysis of these data and their potential before applying the focus on a few (admittedly very nice) examples.

Furthermore, one of the weak spots of such an analysis is the relationship between foldability and functional relevance. Disordered regions would imply reduced relevance due to poor pLDDT scores, which may be a misleading conclusion. While this may be a problem difficult to solve with their approach, I think this still needs to be addressed and discussed throughout the paper and particularly as part of the global analysis, not just in the context of examples.

As a minor point, I would like to motivate the authors to be more explicit with some quantifications. For example, when focusing on proteins < 500 aa long, what does this mean in relation to what they are not representing in their analysis? How many isoforms will they miss? Is there going to be a bias (e.g. against scaffolding proteins, kinases like ATM, etc.)?

Overall, I consider the idea of the paper very elegant and well executed, yet focusing too much on trees, while I, as a reader, would like to know more about the forest.

Reviewer #1 (Public Review):

The majority of polygenic scores have been developed in individuals of European descent and the analysis of the generalisability and applicability of these PRSes in diverse populations has hitherto been limited. In this study, the authors make an important contribution to addressing this gap by evaluating utility of common PRS, curated in the Polygenic Score (PGS) Catalog, in predicting the risk of the commonly diagnosed cancers with high genetic predisposition (breast, prostate, colorectal, and lung) in a prospective cohort comprising 21,694 participants of East Asian descent in Singapore.

Two major strengths in this paper are that this is one of the largest prospective Asian cohorts with long term follow up data, and the authors have completed the evaluation of a large number of PRSes (although it should be pointed out that not all of which are independent of each other).

However, the authors have only described the results of the best performing PRS and attempted to describe PRSes across 4 major common cancers as a group. In so doing, there is a missed opportunity to describe what lessons we might learn in the applicability of PRSes discovered in one population in another diverse population. In addition, it is not clear what benefits may be gleaned from the analysis of the PRSes as a group, rather than individually.

Reviewer #2 (Public Review):

In this work, Li, Dorajoo, and colleagues use national Singaporean data to demonstrate the associations of previously published polygenic risk scores (PRS) for 4 cancers (breast, prostate, colorectal, and lung) with incident cases over 20 years of follow-up. Using available PRS for the four cancers from the Polygenic Score Catalog, they used recommended metrics to evaluate the distribution, discrimination, risk association, and calibration of the PRS. Although the PRS were derived from predominantly European populations, the authors confirmed all PRS-disease associations in this ethnic Chinese population, with per-standard deviation effect sizes ranging from hazard ratio 1.17 for lung cancer to 1.73 for prostate cancer.

The strengths of this work include the use of an apparently unbiased national population with 20 years of follow-up and near-complete outcomes ascertainment. The authors use state-of-the-art methods for genotyping, imputation, and PRS construction, and they use recently published PRS reporting standards to evaluate the PRS and organize the presentation of their work. Although the authors used an unbiased approach to their initial selection of PRS to evaluate (all 1,706 entries with <10,000 predictors in the PGS Catalog at the time), a significant weakness is the lack of detail in how the final 110 cancer PRS were selected for evaluation. Notable absences from these 110 are the PRS from the largest prostate cancer GWAS to date (PGS000662) and a Chinese-specific lung cancer GWAS (PGS000070). The latter absence is particularly notable as the authors report poorest performance of the lung cancer PRS they did evaluate.

Nonetheless, this work confirms prior observations of imperfect portability of PRS derived in one population to another, particularly of different genetic ancestry. The practical consequences of this performance differential will depend on the proposed use of the PRS. One important distinction the authors rightly point out is whether a PRS is intended for individual- or population-level application. The authors do not quantify the potential consequences of applying these PRS to the Singaporean population in different use cases (e.g., screening programs based on PRS), but interested readers will be able to use these findings to make such projections on their own.

Reviewer #1 (Public Review):

This paper has significant strengths in taking a rich, quantitative, neurally-grounded approach to the development of human walking. It provides a rich empirical dataset of EMG and kinematic data at this challenging age, as well as sophisticated analyses of these data in terms of motor primitives, which are a concept that has recently been usefully applied to understanding human walking and its development.

STRENGTHS

It builds on emerging literature in this field and adds data at the key age of infancy-toddlerhood.

It takes a longitudinal approach, sampling children at the ages of newborn, 3 months, and newly walking. This is still reasonably rare in developmental research and allows for a powerful, robust interpretation of data: the authors should be commended for taking this approach.

WEAKNESSES

Some aspects of the work could have been more clearly introduced. This includes neural aspects: the location of the CNS control centres at the spinal level, and which higher centres control them (e.g. brainstem); the justification for understanding primitives as modular (no cross-talk or feedback). It also includes developmental aspects: introducing the stepping reflex, and behavioural aspects of infant motor variability (e.g. Adolph, Hoch & Cole, TICS, 2018).

The patterns relate to walking in a stereotypical manner, yet children's walking is full of skips, jumps, and climbs - both in relation to external obstacles and on even ground. Indeed, it is a challenge to get children to 'walk normally' in a lab. Thus, variability is in fact greater than is discussed here and this should be acknowledged.

The analyses are based on a limited sample of the data. (1) I am not clear on what basis the coders selected cycles, and why 5 cycles were selected. (2) It is not clear why certain movement parameters (cycle duration and flexion/extension proportions) and not others (e.g. step length, double support time) were selected. In particular, it is not clear why the authors focus on temporal, rather than spatial, variability. (3) Some data are based on stepping, and some on kicking. Because it's not clear that these are really equivalent, and because there are small samples of each (n<10), it's not clear that there is enough data to allow us to come to strong conclusions. The sample size should be justified - on the basis of power analyses and/or previous work in this area (e.g. Dominici, Science, n=40). From the results, where p values often hover around p=0.06, the paper seems underpowered to detect a decrease in variability with age for stepping kinematics and primitives.

There are some points of interpretation that could have been clearer, for example highlighting how one might distinguish between variability as incidental (motor noise) or purposeful (for exploration); and how studying the time around walking onset can contribute to the broader literature on this topic.

Reviewer #2 (Public Review):

In this work, the authors attempt to resolve an apparent paradox in human locomotor development. Previous works have reported that neonates exhibit highly variable movement, which is believed to be important for driving exploration-based motor skill learning. Yet, other recent studies have also demonstrated that locomotor behaviors of newborn babies are generated by a very small number of invariant motor primitives that may underpin stereotypical innate motor behaviors. Indeed, as infants acquire the ability to walk independently, the number of motor primitives tends to increase while the overall motor variability decreases. Hinnekens et al. propose that this apparent paradox can be explained by following the variability of the activations of the motor primitives (or motor modules) as the locomotor behaviors of infants mature. The authors collected bilateral EMGs from infants longitudinally at 3 time points (from ~4 days old to walking onset) and used a well-known machine learning algorithm (non-negative matrix factorization) to extract both spatial and temporal motor modules, along with their activations, from the EMGs. They found that at birth, the cycle-to-cycle activations of the small number of modules were highly variable. But as the infants developed into toddlers, while the number of motor modules increased, their activations across cycles also became less variable. The authors conclude that early motor exploration is driven by the variable activation of a small number of motor modules, which would later fractionate into more modules that are more stably recruited across step cycles.

STRENGTHS:

Overall, this work is a valuable addition to the growing literature on the development of motor modules. It not only emphasizes how motor variability is a hallmark of typical motor development, but also suggests the relatively new concept that development-related motor variability originates from the variable activations of early motor modules. Indeed, recent works have proposed that in human adults, the motor variability that drives early motor skill learning may likewise originate from the variable recruitment of motor modules. With this work, it may become possible to conceptually unite the provenance of motor variability that drives both early development and adult learning under the modularity framework. The authors are also commended for their huge effort in collecting this very valuable data from newborn infants and following them with multiple recording sessions till their walking onset. The demonstration of the same longitudinal trend in variability and modules in two different motor behaviors (stepping and kicking) is also highly appreciated.

WEAKNESSES:

The analysis of EMGs relies on a model of motor modules that assumes that multi-muscle activities across step cycles are generated by the variable activations of fixed spatial modules and fixed temporal modules (line 511); thus, by design, after the identification of the spatial (w_j in equation 511) and temporal (w_i(t) in equation 511) modules, the only variable that is adjustable for explaining motor variability is the modules' activation coefficient (a_ijs in equation 511). But it is possible that the observed EMG or kinematic variability may be equally, if not better, accounted for by the cycle-to-cycle variation of the spatial and/or the temporal modules themselves. In fact, the variances of any combination of w_j, w_i(t), and a_ijs may all contribute to EMG variability, even though with the present model, the variance of w_j and w_i(t) are not considered. Therefore, the conclusion that motor variability is generated by variable activations of fixed modules can only be argued based on how well a single model (i.e., line 511) describes the data, rather than by excluding other alternatives (but equally legitimate a priori) models with perhaps less explanatory power. Notably, recent works (e.g., Cheung et al., 2020, IEEE-OJEMB; Berger, d'Avella et al., 2022, JNP) have shown or implied that the variability of the spatial/temporal modules themselves, in addition to their activation coefficients, may be a source of learning-related motor variability.

Reviewer #3 (Public Review):

Hinnekens et al. examined the development of humans' leg movements as they learn to step, kick, and independently walk during infancy. An established theory argues that motor movements can be composed of a finite set of building blocks ("motor primitives"), just like any word can be composed of a finite set of letters. In their paper, Hinnekens et al. follow up this theory by longitudinally recording muscle activations of infants using EMG (at three time points: a few days after birth, at 3 months, and shortly after they learned to walk independently). The authors examined two modules that underlie the infants' stepping and two modules that underlie toddler walking, all based on previous literature. The authors also examined different modules that underlie infants' upright stepping and supine kicking. The authors used supervised machine learning (an advanced version of factor analysis) to identify the modules and to track their change at the different developmental time points. The authors found that trial-to-trial variability in the structure of primitives reduces from newborns to toddlers, even though the number of primitives increased. The authors relate these findings to motor exploration by arguing that newborns generate high variability with a low number of primitives.

The paper has one clear strength - its longitudinal recordings. Unlike most papers in this area of research, the authors follow the same individuals from birth until they learn to walk and the comparison between the use of primitives is done on the same infants. This is certainly novel.

That said, the contribution of the paper to the literature is unclear and it suffers from some critical weaknesses that challenge the current conclusions in the paper, based on the existing data.

1. Although the data is based on longitudinal recordings, and this is certainly desirable, the paper is based only on 10 infants. Moreover, only seven infants contributed supine data at the first time points and only six infants contributed upright data at the different time points. The paper would benefit from a more reliable dataset that includes more infants and time points to compare. To conclude the authors' conclusions, much richer data is required.

2. Relatedly, although the strength of longitudinal data is compared between individuals and has significant insights into individual differences in development, this was not clearly (sometimes not at all) discussed in the paper. The work would benefit from more focus on individual differences and a clear explanation of its contribution to the field from that aspect. The key arguments in the paper focus on the ratio between the number of primitives and the variability in each time point, but none of this from the lens of individual differences. This is challenging to do because there are not many individuals who contribute to the dataset but otherwise, it is not clear what the paper contributes to previous work and more critically.

3. The motivation for the paper is unclear. Why did the authors do what they did? Why is this important to do it the way they did? In the current manuscript, it is not clear why they used this design to get those conclusions.

4. The data selection process is also not clear. At each time point and from each infant, the authors examined 5 cycles from the same leg. The definition of a cycle was hip-flexion onset to another hip-flexion onset on one side of hip extension. It is not clear what variability (measured by % of the cycle in flexion and extension) means in this case because infants hold their legs in one position for a long time. What are those 5 cycles? Why five? A lot of information is missing there about the arbitrary selection of analytic parameters. In addition, the authors argue they performed the same analyses with different parameters and that they got similar results. However, those results are not given in detail and it is hard to compare them with the authors' report.

5. The recording times are not common across individuals. One newborn was recorded after 1 day and the other after 21 days. Not sure this is comparable, especially if the main contribution of the paper is the longitudinal data. Moreover, the second recording was conducted between 74 days to 122 days. This range is too broad. Same for the third time point - one walk onset is not reported, some infants were recorded at <380 days and some >500 days. This difference challenges the reliability of the data.

6. Conceptually, I'm not sure I understand why the authors selected leg alternation (and not other types of movements) as their modules. I was not convinced that leg alternations reflect their real-life locomotor experience (e.g., short bouts in all directions), and therefore the variability measured in this work does not reflect the variability of infants' natural locomotor behaviour.

7. There is not enough rationale for why the specific measurements (IEV, VAF, IRV, etc.) were used and why those are the appropriate ones for the address the questions in the paper. What is the justification for using those measurements?

8. Some of the conclusions, especially those that relate to motor exploration, are not based on sufficient data. Motor exploration was not explicitly measured in this study, and how motor exploration is reflected by the current data and analyses is not clear.