eLife assessment

This paper reports a valuable new set of new results. The main claim is that the projection from adult-born granule cells in the dentate gyrus to the hippocampal subfield CA2 is necessary for the retrieval of social memories formed during development. However, the reviewers agreed that evidence for this major claim is currently incomplete.

eLife assessment

This important study illustrates the value of museum samples for understanding past genetic variability in the genomes of populations and species, including those that no longer exist. The authors present genomic sequencing data for the extinct Xerces Blue butterfly and report convincing evidence of declining population sizes and increases in inbreeding beginning 75,000 years ago, which strongly contrasts to the patterns observed in similar data from its closest relative, the extant Silvery Blue butterfly. Such long-term population health indicators may be used to highlight still extant but especially vulnerable-to-extinction insect species – irrespective of their current census population size abundance.

eLife assessment

This study presents valuable new insights from the protist Tetrahymena regarding radial spokes, conserved protein complexes that are relevant for cilia motility. The work employs interdisciplinary approaches to provide convincing support for radial spoke composition with some experiments, but there are weaker areas with partially incomplete support, such as relying on knockouts alone rather than including localization studies of tagged proteins.

eLife assessment

The study presents a tool for searching molecular dynamics simulation data, making such data sets accessible for open science. The authors provide convincing evidence that it is possible to identify useful molecular dynamics simulation data sets and their analysis can produce valuable information.

eLife assessment

The demonstration that the PARG dePARylation enzyme is required in S phase to remove polyADP-ribose (PAR) protein adducts that are generated in response to the presence of unligated Okazaki fragments is potentially valuable, but the evidence is incomplete, and identification of relevant PARylated PARG substrates in S-phase is needed to understand the role of PARylation and dePARylation in S-phase progression. Their observation that human ovarian cancer cells with low levels of PARG are more sensitive to a PARG inhibitor, presumably due to the accumulation of high levels of protein PARylation, suggests that low PARG protein levels could serve as a criterion to select ovarian cancer patients for treatment with a PARG inhibitor drug.

eLife assessment:

This valuable study provides molecular-level insights into the functional mechanism of bacterial ice-nucleating proteins, detailing electrostatic interactions in the domain architecture of multimeric assemblies. The evidence supporting the claims of the authors is solid, with results from protein engineering experiments, functional assays, and cryo-electron tomography, while the proposed structural model of protein self-assembly remains hypothetical. The work is of broad interest to researchers in the fields of protein structural biology, biochemistry, and biophysics, with implications in microbial ecology and atmospheric glaciation.

eLife assessment

This is an important study because it provides evidence that specific neuronal firing patterns in deep cerebellar nuclei map onto specific behavioral movement disorder phenotypes. The optogenetic manipulations and resulting neuronal and behavioral outcomes are highly compelling, but the development of the classifier tool was incomplete. This study contributes to the fields of cerebellar physiology and movement disorders because it puts forth a map of relationships between neuronal firing patterns and multiple distinct movement phenomena, providing a comprehensive view that goes beyond most studies which typically examine one phenomenon in isolation.

eLife assessment

This study provides valuable insight into the role of miR-199a/b-5p in cartilage formation. The evidence supporting the significance of the identified miRNA and its target mRNA transcripts is convincing, however further experiments and a broader contextual analysis are warranted to draw a more robust conclusion. This paper will likely primarily benefit scientists focused on diseases related to this biological process, such as osteoarthritis. Furthermore, researchers interested in miRNAs as a broader subject may find the computational model development methodology particularly helpful.

eLife assessment

This fundamental work quantifies the stochastic dynamics of neural population activity in the lateral intraparietal area (LIP) of the macaque monkey brain during single perceptual decisions. These single-trial dynamics have been subject to intense debate in neuroscience, and they have implications for modelling decision-making in various fields including neuroscience and psychology. Through a combination of state-of-the-art recordings from many LIP neurons and theory-driven data analyses, the authors provide solid evidence for the notion that single-trial neural population dynamics in LIP encode the decision variable postulated by the drift-diffusion model of decision-making.

eLife assessment

This study offers a compelling molecular model for the organization of rootlets, a critical organelle that links cilia to the basal body, ensuring proper anchoring. While previous research has explored rootlet structure and organization, this study delivers an unprecedented level of resolution, important to the centrosome and cilia field. The model proposed by the authors will serve as a reference for future studies.

eLife assessment

The authors developed a tool to improve our understanding of tissue-specific activation of Free Fatty Acid receptor 2 (FFA2). Convincing in vitro and in vivo validation of the tool is presented via the development of new antibody reagents that constitute an important advance in the field. Some of the technical details could be presented more clearly.

eLife assessment

This manuscript describes an important web resource for kinases connected to cytokines. The compelling information will be highly used by researchers across a number of fields including analysts, modelers, and wet lab experimentalists – and clinician-researchers – who are looking to improve our understanding of pathologies and means to correct them through modulating the immune response.

eLife assessment

This valuable work by Rivera et al. probes to understand how the regulation of oligodendrocyte progenitor cell (OPC) remyelination and function contributes to the treatment of multiple sclerosis. The authors provide incomplete evidence for the platelets to mediate OPC differentiation and remyelination. Both reviewers have raised significant questions. This work will be of broad interest to biologists in general.

eLife assessment

The authors used ribosome profiling in conjunction with standard biochemical approaches to investigate the role of eIF2A in translation initiation in yeast under optimal growth conditions or stress. The convincing data demonstrate that eIF2A does not play a substantial role in translation initiation in yeast. These important findings challenge the current view that eIF2A substitutes for eIF2 under stress and are thus anticipated to spur future investigation on the role(s) of eIF2A. Considering the broad scope of cellular functions attributed to eIF2A, this study should be of interest to a wide spectrum of biomedical researchers ranging from those studying mechanisms of translation regulation to virologists and cancer biologists.

eLife assessment

This important study proposes a new method for tracking neurons recorded with Neuropixel electrodes across days. The methods and the strength of the evidence are convincing, but the authors do not adequately address whether their approach can be generalized to other brain areas, species, behaviors, or tools. Overall, this method will be potentially of interest to many neuroscientists who want to study long-term activity changes of individual neurons in the brain.

eLife assessment

This important study analyses the role of post-translational modifications of tubulin regulate the function of the microtubule cytoskeleton in vivo? The authors generate a large panel of tubulin mutants designed to lack specific modifications and describe their effects using endogenous editing and touch receptor neurons in C. elegans as an in vivo model. While the work presents an impressive amount of data, it is in part incomplete, since the presence and absence of specific tubulin modifications and their effects on microtubules are not demonstrated in all cases.

Reviewer #2 (Public Review):

The tubulin subunits that make up microtubules can be posttranslationally modified and these PTMs are proposed to regulate microtubule dynamics and the proteins that can interact with microtubules in many contexts. However, most studies investigating the roles of tubulin PTMs have been conducted in vitro either with purified components or in cultured cells. Lu et al. use CRISPR/Cas9 genome editing to mutate tubulin genes in C. elegans, testing the role of specific tubulin residues on neuronal development. This study is a real tour de force, tackling multiple proposed tubulin modifications and following the resulting phenotypes with respect to neurite outgrowth in vivo. There is a ton of data that experts in the field will likely reference for years to come as this is one of the most comprehensive in vivo analyses of tubulin PTMs in vivo.

This paper will be very important to the field, however would be strengthened if: 1) the authors demonstrated that the mutations they introduced had the intended consequences on microtubule PTMs, 2) the authors explored how the various tubulin mutations directly affect microtubules, and 3) the findings are made generally more accessible to non C. elegans neurobiologists.

(1) The authors introduce several mutations to perturb tubulin PTMs, However, it is unclear to what extent the engineered mutations affect tubulin in the intended way i.e. are the authors sure that the PTMs they want to perturb are actually present in C. elegans. Many of the antibodies used did not appear to be specific and antibody staining was not always impacted in the mutant cases as expected. For example, is there any evidence that S172 is phosphorylated in C. elegans, e.g. from available phosphor-proteomic data? Given the significant amount of staining left in the S172A mutant, the antibody seems non-specific in this context and therefore not a reliable readout of whether MTs are actually phosphorylated at this residue. As another example, there is no evidence presented that K252 is acetylated in C. elegans. At the very least, the authors should consider demonstrating the conservation of these residues and the surrounding residues with other organisms where studies have demonstrated PTMs exist.

(2) Given that the authors have the mutants in hand, it would be incredibly valuable to assess the impact of these mutations on microtubules directly in all cases. MT phenotypes are inferred from neurite outgrowth phenotypes in several cases, the authors should look directly at microtubules and/or microtubule dynamics via EBP-2 when possible OR show evidence that the only way to derive the neurite phenotypes shown is through the inferred microtubule phenotypes. For example, the effect of the acetylation or detyrosination mutants on MTs was not assessed.

(3) There is a ton of data here that will be important for experts working in this field to dig into, however, for the more general cell biologist, some of the data are quite inaccessible. More cartoons and better labeling will be helpful as will consistent comparisons to control worms in each experiment.

(4) In addition, I am left unconvinced of the negative data demonstrating that MBK does not phosphorylate tubulin. First, the data described in lines 207-211 does not appear to be presented anywhere. Second, RNAi is notoriously finicky in neurons, thus necessitating tissue-specific degradation using either the ZF/ZIF-1 or AID/TIR1 systems which both work extremely well in C. elegans. Third, there appears to be increasing S172 phosphorylation in Figure 3 Supplement 2 with added MBK-2, but there is no anti-tubulin blot to show equal loading, so this experiment is hard to interpret.

Author Response

We thank the reviewers for their positive comments and constructive feedback following their thorough reading of the manuscript. In this provisional reply we will briefly address the reviewer’s comments and suggestions point by point. In the forthcoming revised manuscript, we will more thoroughly address the reviewer’s comments and provide additional supporting data.

(1) The expression 'randomly clustered networks' needs to be explained in more detail given that in its current form risks to indicate that the network might be randomly organized (i.e., not organized). In particular, a clustered network with future functionality based on its current clustering is not random but rather pre-configured into those clusters. What the authors likely meant to say, while using the said expression in the title and text, is that clustering is not induced by an experience in the environment, which will only be later mapped using those clusters. While this organization might indeed appear as randomly clustered when referenced to a future novel experience, it might be non-random when referenced to the prior (unaccounted) activity of the network. Related to this, network organization based on similar yet distinct experiences (e.g., on parallel linear tracks as in Liu, Sibille, Dragoi, Neuron 2021) could explain/configure, in part, the hippocampal CA1 network organization that would appear otherwise 'randomly clustered' when referenced to a future novel experience.

As suggested by the reviewer, we will revise the text to clarify that the random clustering is random with respect to any future, novel environment. The cause of clustering could be prior experiences (e.g. Bourjaily M & Miller P, Front. Comput. Neurosci. 5:37, 2011) or developmental programming (e.g. Perin R, Berger TK, & Markram H, Proc. Natl. Acad. Sci. USA 108:5419, 2011).

(2) The authors should elaborate more on how the said 'randomly clustered networks' generate beyond chance-level preplay. Specifically, why was there preplay stronger than the time-bin shuffle? There are at least two potential explanations:

(2.1) When the activation of clusters lasts for several decoding time bins, temporal shuffle breaks the continuity of one cluster's activation, thus leading to less sequential decoding results. In that case, the preplay might mainly outperform the shuffle when there are fewer clusters activating in a PBE. For example, activation of two clusters must be sequential (either A to B or B to A), while time bin shuffle could lead to non-sequential activations such as a-b-a-b-a-b where a and b are components of A and B;

(2.2) There is a preferred connection between clusters based on the size of overlap across clusters. For example, if pair A-B and B-C have stronger overlap than A-C, then cluster sequences A-B-C and C-B-A are more likely to occur than others (such as A-C-B) across brain states. In that case, authors should present the distribution of overlap across clusters, and whether the sequences during run and sleep match the magnitude of overlap. During run simulation in the model, as clusters randomly receive a weak location cue bias, the activation sequence might not exactly match the overlap of clusters due to the external drive. In that case, the strength of location cue bias (4% in the current setup) could change the balance between the internal drive and external drive of the representation. How does that parameter influence the preplay incidence or quality?

Based on our finding that preplay occurs only in networks that sustain cluster activity over multiple decoding time bins (Figure 5d-e), our understanding of the model’s function is consistent with the reviewers first explanation. We will provide additional analysis in the forthcoming revised manuscript in order to directly test the first explanation and will also test the intriguing possibility that the reviewer’s second suggestion contributes to above-chance preplay.

(3) The manuscript is focused on presenting that a randomly clustered network can generate preplay and place maps with properties similar to experimental observations. An equally interesting question is how preplay supports spatial coding. If preplay is an intrinsic dynamic feature of this network, then it would be good to study whether this network outperforms other networks (randomly connected or ring lattice) in terms of spatial coding (encoding speed, encoding capacity, tuning stability, tuning quality, etc.)

We agree that this is an interesting future direction, but we see it as outside the scope of the current work. There are two interesting avenues of future work: 1) Our current model does not include any plasticity mechanisms, but a future model could study the effects of synaptic plasticity during preplay on long-term network dynamics, and 2) Our current model does not include alternative approaches to constructing the recurrent network, but future studies could systematically compare the spatial coding properties of alternative types of recurrent networks.

(4) The manuscript mentions the small-world connectivity several times, but the concept still appears too abstract and how the small-world index (SWI) contributes to place fields or preplay is not sufficiently discussed.

For a more general audience in the field of neuroscience, it would be helpful to include example graphs with high and low SWI. For example, you can show a ring lattice graph and indicate that there are long paths between points at opposite sides of the ring; show randomly connected graphs indicating there are no local clustered structures, and show clustered graphs with several hubs establishing long-range connections to reduce pair-wise distance.

How this SWI contributes to preplay is also not clear. Figure 6 showed preplay is correlated with SWI, but maybe the correlation is caused by both of them being correlated with cluster participation. The balance between cluster overlap and cluster isolation is well discussed. In the Discussion, the authors mention "...Such a balance in cluster overlap produces networks with small-world characteristics (Watts and Strogatz, 1998) as quantified by a small-world index..." (Lines 560-561). I believe the statement is not entirely appropriate, a network similar to ring lattice can still have the balance of cluster isolation and cluster overlap, while it will have small SWI due to a long path across some node pairs. Both cluster structure and long-range connection could contribute to SWI. The authors only discuss the necessity of cluster structure, but why is the long-range connection important should also be discussed. I guess long-range connection could make the network more flexible (clusters are closer to each other) and thus increase the potential repertoire.

We agree that the manuscript would benefit from a more concrete explanation of the small-world index. We will revise the text and add illustrative figures.

We note that while our most successful clustered networks are indeed those with small-world characteristics, there are other ways of producing small-world networks which may not show good place fields or preplay. We will test another type of small-world network if time permits.

Our discussion of “cluster overlap” is specific to our type of small-world network in which there is no pre-determined spatial dimension (unlike the ring network of Watts and Strogatz). Therefore, because clusters map randomly to location once a particular spatial context is imposed, the random overlap between clusters produces long-range connections in that context (and any other context) so one can think of the amount of overlap between clusters as representing the number of long-range connections in a Watts-Strogatz model, except, we wish to iterate, such models involve a spatial topology within the network, which we do not include.

(5) What drives PBE during sleep? Seems like the main difference between sleep and run states is the magnitude of excitatory and inhibitory inputs controlled by scaling factors. If there are bursts (PBE) in sleep, do you also observe those during run? Does the network automatically generate PBE in a regime of strong excitation and weak inhibition (neural bifurcation)?

During sleep simulations, the PBEs are spontaneously generated by the recurrent connections in the network. The constant-rate Poisson inputs drive low-rate stochastic spiking in the recurrent network, which then randomly generates population events when there is sufficient internal activity to transiently drive additional spiking within the network.

During run simulations, the spatially-tuned inputs drive greater activity in a subset of the cells at a given point on the track, which in turn suppress the other excitatory cells through the feedback inhibition.

(6) Is the concept of 'cluster' similar to 'assemblies', as in Peyrache et al, 2010; Farooq et al, 2019? Does a classic assembly analysis during run reveal cluster structures?

Yes, we are highly confident that the clusters in our network would correspond to the functional assemblies that have been studied through assembly analysis and will present the relevant data in a revision.

(7) Can the capacity of the clustered network to express preplay for multiple distinct future experiences be estimated in relation to current network activity, as in Dragoi and Tonegawa, PNAS 2013?

We agree this is an interesting opportunity to compare the results of our model to what has been previously found experimentally and will test this if time permits.

Reviewer # 2

Weaknesses:

My main critiques of the paper relate to the form of the input to the network.

First, because the input is the same across trials (i.e. all traversals are the same duration/velocity), there is no ability to distinguish a representation of space from a representation of time elapsed since the beginning of the trial. The authors should test what happens e.g. with traversals in which the animal travels at different speeds, and in which the animal's speed is not constant across the entire track, and then confirm that the resulting tuning curves are a better representation of position or duration.

We agree that this is an important question, and we plan to run further simulations where we test the effects of varying the simulated speed. We will present results in the resubmission.

Second, it's unclear how much the results depend on the choice of a one-dimensional environment with ramping input. While this is an elegant idealization that allows the authors to explore the representation and replay properties of their model, it is a strong and highly non-physiological constraint. The authors should verify that their results do not depend on this idealization. Specifically, I would suggest the authors also test the spatial coding properties of their network in 2-dimensional environments, and with different kinds of input that have a range of degrees of spatial tuning and physiological plausibility. A method for systematically producing input with varying degrees of spatial tuning in both 1D and 2D environments has been previously used in (Fang et al 2023, eLife, see Figures 4 and 5), which could be readily adapted for the current study; and behaviorally plausible trajectories in 2D can be produced using the RatInABox package (George et al 2022, bioRxiv), which can also generate e.g. grid cell-like activity that could be used as physiologically plausible input to the network.

We agree that testing the robustness of our results to different models of feedforward input is important and we plan to do this in our revised manuscript for the linear track and W-track.

Testing the model in a 2D environment is an interesting future direction, but we see it as outside the scope of the current work. To our knowledge there are no experimental findings of preplay in 2D environments, but this presents an interesting opportunity for future modeling studies.

Finally, I was left wondering how the cells' spatial tuning relates to their cluster membership, and how the capacity of the network (number of different environments/locations that can be represented) relates to the number of clusters. It seems that if clusters of cells tend to code for nearby locations in the environment (as predicted by the results of Figure 5), then the number of encodable locations would be limited (by the number of clusters). Further, there should be a strong tendency for cells in the same cluster to encode overlapping locations in different environments, which is not seen in experimental data.

Thank you for making this important point and giving us the opportunity to clarify. We do find that subsets of cells with identical cluster membership have correlated place fields, but as we show in Figure 7b the network place map as a whole shows low remapping correlations across environments, which is consistent with experimental data (Hampson RE et al, Hippocampus 6:281, 1996; Pavlides C, et al, Neurobiol Learn Mem 161:122, 2019). Our model includes a relatively small number of cells and clusters compared to CA3, and with a more realistic number of clusters, the level of correlation across network place maps should reduce even further in our model network. The reason for a low level of correlation is because cluster membership is combinatorial, whereby cells that share membership in one cluster can also belong to separate/distinct other clusters, rendering their activity less correlated than might be anticipated. In our revised manuscript we will address this point more carefully and cite the relevant experimental support.

Reviewer # 3

Weaknesses:

To generate place cell-like activity during a simulated traversal of a linear environment, the authors drive the network with a combination of linearly increasing/decreasing synaptic inputs, mimicking border cell-like inputs. These inputs presumably stem from the entorhinal cortex (though this is not discussed). The authors do not explore how the model would behave when these inputs are replaced by or combined with grid cell inputs which would be more physiologically realistic.

We chose the linearly varying spatial inputs as the minimal model of providing spatial input to the network so that we could focus on the dynamics of the recurrent connections. We agree our results will be strengthened by testing alternative types of border-like input so will present such additional results in our revised version. However, given that a sub-goal of our model was to show that place fields could arise in locations at which no neurons receive a peak in external input, whereas combining input from multiple grid cells produces peaked place-field like input, adding grid cell input (and the many other types of potential hippocampal input) is beyond the scope of the paper.

Even though the authors claim that no spatially-tuned information is needed for the model to generate place cells, there is a small location-cue bias added to the cells, depending on the cluster(s) they belong to. Even though this input is relatively weak, it could potentially be driving the sequential activation of clusters and therefore the preplays and place cells. In that case, the claim for non-spatially tuned inputs seems weak. This detail is hidden in the Methods section and not discussed further. How does the model behave without this added bias input?

First, we apologize for a lack of clarity if we have caused confusion about the type of inputs (linear and cluster-dependent as we had attempted to portray prominently in Figure 1, where it is described in the caption, l. 156-157, and Results, l. 189-190 & l. 497-499, as well as in the Methods, l. 671-683) and if we implied an absence of spatially-tuned information in the network. In the revision we will clarify that for reliable place fields to appear, the network must receive spatial information and that one point of our paper is that the information need not arrive as peaks of external input already resembling place cells or grid cells. We chose linearly ramping boundary inputs as the minimally place-field like stimulus (that still contains spatial information) but in our revision we will include alternatives. We should note that during sleep, when “preplay” occurs, there is no such spatial bias (which is why preplay can equally correlate with place field sequences in any context). In the revision, we will update Figure 1 to show more clearly the cluster-dependent linearly ramping input received by some specific cells with both similar and different place fields.

Unlike excitation, inhibition is modeled in a very uniform way (uniform connection probability with all E cells, no I-I connections, no border-cell inputs). This goes against a long literature on the precise coordination of multiple inhibitory subnetworks, with different interneuron subtypes playing different roles (e.g. output-suppressing perisomatic inhibition vs input-gating dendritic inhibition). Even though no model is meant to capture every detail of a real neuronal circuit, expanding on the role of inhibition in this clustered architecture would greatly strengthen this work.

This is an interesting future direction, but we see it as outside the scope of our current work. While inhibitory microcircuits are certainly important physiologically, we focus here on a minimal model that produces the desired place cell activity and preplay, as measured in excitatory cells.

For the modeling insights to be physiologically plausible, it is important to show that CA3 connectivity (which the model mimics) shares the proposed small-world architecture. The authors discuss the existence of this architecture in various brain regions but not in CA3, which is traditionally thought of and modeled as a random or fully connected recurrent excitatory network. A thorough discussion of CA3 connectivity would strengthen this work.

We agree this is an important point that is missing, and we will revise the text to specifically address CA3 connectivity (Guzman et al., Science 353 (6304), 1117-1123 2016) and the small-world structure therein due to the presence of “assemblies”.

eLife assessment

This study presents a valuable finding on the spontaneous emergence of structured activity in artificial neural networks endowed with specific connectivity profiles. The evidence supporting the claims of the authors is potentially solid but still incomplete at this stage, as the authors would ideally demonstrate that similar properties are observed with more diverse inputs and in more complex environments. The work will be of interest to systems and computational neuroscientists.

Reviewer #2 (Public Review):

Summary:

The authors show that a spiking network model with clustered neurons produces intrinsic spike sequences when driven with a ramping input, which are recapitulated in the absence of input. This behavior is only seen for some network parameters (neuron cluster participation and number of clusters in the network), which correspond to those that produce a small world network. By changing the strength of ramping input to each network cluster, the network can show different sequences.

Strengths:

A strength of the paper is the direct comparison between the properties of the model and neural data.

Weaknesses:

My main critiques of the paper relate to the form of the input to the network.

First, because the input is the same across trials (i.e. all traversals are the same duration/velocity), there is no ability to distinguish a representation of space from a representation of time elapsed since the beginning of the trial. The authors should test what happens e.g. with traversals in which the animal travels at different speeds, and in which the animal's speed is not constant across the entire track, and then confirm that the resulting tuning curves are a better representation of position or duration.

Second, it's unclear how much the results depend on the choice of a one-dimensional environment with ramping input. While this is an elegant idealization that allows the authors to explore the representation and replay properties of their model, it is a strong and highly non-physiological constraint. The authors should verify that their results do not depend on this idealization. Specifically, I would suggest the authors also test the spatial coding properties of their network in 2-dimensional environments, and with different kinds of input that have a range of degrees of spatial tuning and physiological plausibility. A method for systematically producing input with varying degrees of spatial tuning in both 1D and 2D environments has been previously used in (Fang et al 2023, eLife, see Figures 4 and 5), which could be readily adapted for the current study; and behaviorally plausible trajectories in 2D can be produced using the RatInABox package (George et al 2022, bioRxiv), which can also generate e.g. grid cell-like activity that could be used as physiologically plausible input to the network.

Finally, I was left wondering how the cells' spatial tuning relates to their cluster membership, and how the capacity of the network (number of different environments/locations that can be represented) relates to the number of clusters. It seems that if clusters of cells tend to code for nearby locations in the environment (as predicted by the results of Figure 5), then the number of encodable locations would be limited (by the number of clusters). Further, there should be a strong tendency for cells in the same cluster to encode overlapping locations in different environments, which is not seen in experimental data.

Reviewer #3 (Public Review):

Summary:

This work offers a novel perspective on the question of how hippocampal networks can adaptively generate different spatial maps and replays/preplays of the corresponding place cells, without any such maps pre-existing in the network architecture or its inputs. Unlike previous modeling attempts, the authors do not pre-tune their model neurons to any particular place fields. Instead, they build a random, moderately-clustered network of excitatory (and some inhibitory) cells, similar to CA3 architecture. By simulating spatial exploration through border-cell-like synaptic inputs, the model generates place cells for different "environments" without the need to reconfigure its synaptic connectivity or introduce plasticity. By simulating sleep-like random synaptic inputs, the model generates sequential activations of cells, mimicking preplays. These "preplays" require small-world connectivity, so that weakly connected cell clusters are activated in sequence. Using a set of electrophysiological recordings from CA1, the authors confirm that the modeled place cells and replays share many features with real ones. In summary, the model demonstrates that spontaneous activity within a small-world structured network can generate place cells and replays without the need for pre-configured maps.

Strengths:

This work addresses an important question in hippocampal dynamics. Namely, how can hippocampal networks quickly generate new place cells when a novel environment is introduced? And how can these place cells preplay their sequences even before the environment is experienced? Previous models required pre-existing spatial representations to be artificially introduced, limiting their adaptability to new environments. Other models depended on synaptic plasticity rules which made remapping slower than what is seen in recordings. This modeling work proposes that quickly-adaptive intrinsic spiking sequences (preplays) and spatially tuned spiking (place cells) can be generated in a network through randomly clustered recurrent connectivity and border-cell inputs, avoiding the need for pre-set spatial maps or plasticity rules. The proposal that small-world architecture is key for place cells and preplays to adapt to new spatial environments is novel and of potential interest to the computational and experimental community.

The authors do a good job of thoroughly examining some of the features of their model, with a strong focus on excitatory cell connectivity. Perhaps the most valuable conclusion is that replays require the successive activation of different cell clusters. Small-world architecture is the optimal regime for such a controlled succession of activated clusters.

The use of pre-existing electrophysiological data adds particular value to the model. The authors convincingly show that the simulated place cells and preplay events share many important features with those recorded in CA1 (though CA3 ones are similar).

Weaknesses:

To generate place cell-like activity during a simulated traversal of a linear environment, the authors drive the network with a combination of linearly increasing/decreasing synaptic inputs, mimicking border cell-like inputs. These inputs presumably stem from the entorhinal cortex (though this is not discussed). The authors do not explore how the model would behave when these inputs are replaced by or combined with grid cell inputs which would be more physiologically realistic.

Even though the authors claim that no spatially-tuned information is needed for the model to generate place cells, there is a small location-cue bias added to the cells, depending on the cluster(s) they belong to. Even though this input is relatively weak, it could potentially be driving the sequential activation of clusters and therefore the preplays and place cells. In that case, the claim for non-spatially tuned inputs seems weak. This detail is hidden in the Methods section and not discussed further. How does the model behave without this added bias input?

Unlike excitation, inhibition is modeled in a very uniform way (uniform connection probability with all E cells, no I-I connections, no border-cell inputs). This goes against a long literature on the precise coordination of multiple inhibitory subnetworks, with different interneuron subtypes playing different roles (e.g. output-suppressing perisomatic inhibition vs input-gating dendritic inhibition). Even though no model is meant to capture every detail of a real neuronal circuit, expanding on the role of inhibition in this clustered architecture would greatly strengthen this work.

For the modeling insights to be physiologically plausible, it is important to show that CA3 connectivity (which the model mimics) shares the proposed small-world architecture. The authors discuss the existence of this architecture in various brain regions but not in CA3, which is traditionally thought of and modeled as a random or fully connected recurrent excitatory network. A thorough discussion of CA3 connectivity would strengthen this work.

eLife assessment

The study presents an important ecosystem designed to support literature mining in biomedical research, showcasing a methodological framework that includes tools like Pubget for article collection and labelbuddy for text annotation. The solid evidence presented for these tools suggests they could streamline the analysis and annotation of scientific literature, potentially benefiting research across a range of biomedical disciplines. While the primary focus is on neuroimaging literature, the applicability of these methods and tools might extend further, offering an advance in the practices of meta-research and literature mining.

eLife assessment

This study provides valuable insights into how chromatin-bound PfMORC controls gene expression in the asexual blood stage of Plasmodium falciparum. By interacting with key nuclear proteins, PfMORC is predicted to affect expression of genes relating to host invasion and variable subtelomeric gene families. Correlating transcriptomic data with in vivo chromatin analysis, the study provides convincing evidence for the role of PfMORC in epigenetic transcriptional regulation.

Author response:

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

eLife assessment

This study provides valuable insights into how chromatin-bound PfMORC controls gene expression in the asexual blood stage of Plasmodium falciparum. By interacting with key nuclear proteins, PfMORC appears to affect expression of genes relating to host invasion and subtelomeric var genes. Correlating transcriptomic data with in vivo chromatin insights, the study provides solid evidence for the central role of PfMORC in epigenetic transcriptional regulation through modulation of chromatin compaction.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

The study provides valuable insights into the role of PfMORC in Plasmodium's epigenetic regulation, backed by a comprehensive methodological approach. The overarching goal was to understand the role of PfMORC in epigenetic regulation during asexual blood stage development, particularly its interactions with ApiAP2 TFs and its potential involvement in the regulation of genes vital for Plasmodium virulence. To achieve this, they conducted various analyses. These include a proteomic analysis to identify nuclear proteins interacting with PfMORC, a study to determine the genome-wide localization of PfMORC at multiple developmental stages, and a transcriptomic analysis in PfMORCHA-glmS knockdown parasites. Taken together, this study suggests that PfMORC is involved in chromatin assemblies that contribute to the epigenetic modulation of transcription during the asexual blood stage development.

Strengths:

The study employed a multi-faceted approach, combining proteomic, genomic, and transcriptomic analyses, providing a holistic view of PfMORC's role. The proteomic analysis successfully identified several nuclear proteins that may interact with PfMORC. The genome-wide localization offered valuable insights into PfMORC's function, especially its predominant recruitment to subtelomeric regions. The results align with previous findings on PfMORC's interaction with ApiAP2 TFs. Notably, the authors meticulously contextualized their findings with prior research, including pre-prints, adding credibility to their work.

Weaknesses:

While the study identifies potential interacting partners and loci of binding, direct functional outcomes of these interactions remain an inference. The authors heavily rely on past research for some of their claims. While it strengthens some assertions, it might indicate a lack of direct evidence in the current study for particular aspects. The declaration that PfMORC may serve as an attractive drug target is substantial. While the data suggests its involvement in essential processes, further studies are required to validate its feasibility as a drug target.

Reviewer #2 (Public Review):

Summary:

This is a paper entitled "Plasmodium falciparum MORC protein modulates gene expression through interaction with heterochromatin" describes the role of PfMORC during the intra-erythrocytic cycle of Plasmodium falciparum. Garcia et al. investigated the PfMORC-interacting proteins and PfMORC genomic distribution in trophozoites and schizonts. They also examined the transcriptome of the parasites after partial knockdown of the transcript.

Strengths:

This study is a significant advance in the knowledge of the role of PfMORC in heterochromatin assembly. It provides an in-depth analysis of the PfMORC genomic localization and its correlation with other chromatin marks and ApiAP2 transcription factor binding.

Weaknesses:

However, most of the conclusions are based on the function of interacting proteins and the genomic localization of the protein. The authors did not investigate the direct effects of PfMORC depletion on heterochromatin marks. Furthermore, the results of the transcriptomic analysis are puzzling as 50% of the transcripts are downregulated, a phenotype not expected for a heterochromatin marker.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Suggestions for improved or additional experiments, data, or analyses.

• Figure 1A and Table 1: the authors should incorporate a volcano plot in their proteomic results presentation. This graphical representation can provide a more intuitive grasp of the most relevant proteins associated with PfMORC in terms of both their abundance and significance. It will aid in swiftly pinpointing proteins with the most notable differential associations. This will complement the comprehensive overview provided by the authors, referencing past research where PfMORC was detailed.

We thank the reviewer for the suggestion. We agree with the reviewer that the volcano plot we now provide does indeed bring comprehensive information on associations between PfMORC and other cellular proteins. The volcano plot presented in the revised manuscript as Figure 1A, was generated using the normalized MS/MS counts from the anti-GFP and 3D7 (control) proteomics datasets (n=3). The potential PfMORC interacting proteins were determined using the fold changes and p-values between the two datasets, as provided in Table 1.

Several protein interactors were strongly supported by statistical analysis (p-value), while others showed weaker p-value due to variability between replicates. Indeed, the total number of proteins identified in the three replicates, shown in the Venn diagram (Supplemental Figure 1D), exhibits a good overlap between the replicates but a lower number of identified proteins in the GFP-E1 sample. This variability was observed also in the statistical analysis. Indeed, by analyzing the GFP/3D7 ratios, some proteins have a significant difference in abundance (fold change greater than 1.5x) in one of the groups but do not meet the statistical threshold. For more clarity, we have included the -log p-value for the proteins listed in Table 1.

Overall, these results demonstrate that many ApiAP2 proteins and several chromatin-associated factors interact with PfMORC.

• Given the plethora of proteins detected in the PfMORC eluate, it raises the question of how many are genuine MORC interactors versus those that are merely nearby molecules acting adjacently. These might incidentally end up in the immunoprecipitate due to unintended interactions with DNA or chromatin. While the M&M section mentions that the beads were thoroughly washed, there is no specification about the washing buffer or its stringency (i.e., salinity level). At higher salinities, one could isolate core complexes of interactors associated with DNA or even RNA carryover.

We apologize for this omission and have now added the buffer composition used to wash the beads. This section now reads "To perform the co-immunoprecipitation we followed the manufacturer protocol (ChromoTek, gta-20). Samples were lysed in modified RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Nonidet P-40, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 10 µg/ml, 1 mM phenylmethylsulfonyl fluoride, benzamidine) for 30 min on ice. The lysate was precleared with 50 µl of protein A/G-Agarose beads at 4°C for 1 h and clarified by centrifugation at 10,000 × g for 10 min. The precleared lysate was incubated overnight with an anti-GFP antibody using anti-GFP-Trap-A beads (ChromoTek, gta-20). The magnetic beads were then pelleted using a magnet (Invitrogen) and washed 3 times with wash buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.05 % Nonidet™ P40 Substitute, 0.5 mM EDTA)."

We used the same salt concentration for immunoprecipitation as was used in the lysis buffer to minimize the binding of non-specific proteins. The wash buffer composition is updated in the revised manuscript. The immunoprecipitations were done in biological triplicates to ensure reproducibility and statistical support. A number of proteins are common across all three replicates. We also used wild-type parasites (non-GFP) as a negative control to eliminate non-specific hits, and we used a log2-fold change ≥1.5 relative to wild type parasites as our cutoff between the comparison groups.

We believe that these conditions provide the stringency required to identify high confidence PfMORC interacting proteins, although this still leaves a possibility for additional lower affinity interactions. Future studies will certainly follow up candidate interaction partners to better define this complex. However, the complexity of the complex resembles that reported previously in Toxoplasma gondii (Farhat et al. 2020, Nat Microbiol) as well another report on the PfMORC complexes: https://elifesciences.org/reviewed-prepri nts/92499

• The authors demonstrate that PfMORC creates distinct peaks in and around HP1-bound areas (Figure 2F), hinting at a specific role for PfMORC in heterochromatin compaction, boundary definition, and gene silencing. This pattern is clearly depicted in an example in Figure 2F. It would be beneficial to know if this enrichment profile is replicated elsewhere and, if so, it would be worthwhile to quantify it.

This is an excellent point. Yes, this pattern is seen across the entire genome, where PfMORC is apposed to PfHP1-bound heterochromatic regions. As indicated in the manuscript, we have quantified this effect genome-wide; however, since we already display compiled data for Chromosome 2 (at both chromosome ends) pertaining to the position of PfMORC relative to PfHP1 we do not feel it is essential to provide such a figure for the entire genome as it does not alter the central message of our manuscript. Figure 2F is representative of the genome-wide distribution of PfMORC relative to PfHP1. The raw genome-wide data are available in Supplementary Information for further inspection of specific loci on other chromosomes.

Recommendations for improving the writing and presentation.

MAIN TEXT

Panel e, referenced both in the main text and legend, is missing from Figure 4. This missing panel represents a significant finding of the study, highlighting according to the authors a low correlation between ChIP-seq gene targets and RNA-seq DEGs. This observation implies that PfMORC's global occupancy is more aligned with shaping chromatin architecture than directly regulating specific gene targets. In light of this, the authors should rephrase parts of their manuscript (including abstract and title) to avoid suggesting that PfMORC acts primarily (directly) as a gene regulator, emphasizing instead its role in influencing the topological structure of chromosomes.

We have modified the title as suggested by the reviewer to more accurately reflect that PfMORC modulates chromatin architecture rather than acting as a direct regulator of specific genes. Our new title is: A Plasmodium falciparum MORC protein complex modulates epigenetic control of gene expression through interaction with heterochromatin

We apologize for the omission of Figure 4e, which is now included in the revised manuscript. We found PfMORC occupancy on all chromosomes at subtelomeric regions, which are known to harbor genes related to immune evasion and antigenic variation (including most of the var genes). This study is also in agreement with Bryant et al. (PMID 32816370) which reported PfMORC occupancy along with PfISW1 at var gene promoters. PfMORC has also been identified in complexes with various ApiAP2 proteins in a proteome-wide study (Hillier et al. Cell Rep, PMID 31390575), as well as in immunoprecipitations of PfAP2-G2 (Singh et al., Mol Micro, PMID 33368818) and PfAP2-P (Subudhi et al., Nat Microbiol, PMID 37884813). The recent study by Subudhi et al. reports that PfAP2-P is involved in the regulation of var gene expression, antigenic variation, trophozoite development and parasite egress. It is therefore possible that PfMORC may have different effects on transcriptional regulation through interactions with different ApiAP2 transcription factors. Our comparison of PfMORC with known ApiAP2 protein occupancy reveals a high level of overlap, indicating that PfMORC may affect gene expression in various ways throughout the asexual cycle. Additionally, Hillier et al. show that PfMORC interaction is not limited to ApiAP2 but also implicates several other chromatin remodellers, which is consistent with our own results. We do not imply direct regulation of transcription via PfMORC in our manuscript. To the contrary, we suggest that it interacts with heterochromatin and thereby plays a role in the epigenetic control of asexual blood stage transcriptional regulation which is also clarified in the revised abstract.

Another limitation of differential gene expression was use of the glmS ribozyme system, which resulted in only 50% depletion of the PfMORC transcript. There may still be enough PfMORC to rescue the gene expression we could not detect correctly. Therefore, it is challenging to interpret the function of PfMORC in only chromatin architecture but not in gene expression.

If we believe that PfMORC in Plasmodium isn't mainly adjusting gene expression, the authors' suggestion that MORC is targeted by some AP2s becomes puzzling. How do we make sense of these different ideas? The authors need to clarify this to maintain consistency in their findings.

Based on our data, we hypothesize that PfMORC acts as an accessory protein for ApiAP2 transcription factors. In a number of studies, including ours and the concurrent publication in eLife (https://elifesciences.org/reviewed-preprints/92499), PfMORC co-IPed with several ApiAP2 proteins, suggest it has multiple functions. In our previous study we showed that PfMORC expression is highest in mid and late asexual stages. A comparison of the PfMORC occupancy with 6 ApiAP2 (having different expression profile) suggest plasticity in PfMORC function. We have revised our discussion to make this hypothesis more transparent for the readers.

The authors should cite Farhat et al. 2020 (Extended Data Fig. 1a), as it similarly identified 3 different ELM2-containing proteins in Toxoplasma MORC-associated complexes. This previous work provides context and supports the observations made with PfMORC in this study.

Thank you for the suggestion and pointing out this omission. We have indeed cited the work of the Farhat group in the original manuscript and have now included this additional reference to corroborate the text and provide further support to our conclusions.

Minor corrections to the text and figures.

• Panel e is missing from Figure 4.

As mentioned above Panel e is now included in Figure 4.

• The captions are very minimally detailed. An effort must be made to better describe the panels as well as which statistical tests were used. As it stands, this is not really up to standard.

We have elaborated the captions with more detailed descriptions, and we now provide additional information where further clarification was necessary.

Reviewer #2 (Recommendations For The Authors):

• The study lacks a direct correlation between the inferred function of PfMORC and the heterochromatin state of the genome after its depletion. It would be interesting to perform chip-seq on known heterochromatin markers such as H3K9me3, HP1 or H3K36me2/3 to measure the consequences of PfMORC depletion on global heterochromatin and its boundaries.

While the proposed experiments are certainly interesting, they are beyond the scope of this study. The current manuscript is focused on PfMORC occupancy, its interacting partners, and its impact on differential gene regulation after PfMORC depletion in asexual parasites. Nonetheless, we did in fact compared the PfMORC occupancy with that of various heterochromatin markers (H2A.Z, H3K9ac, H3K4me3, H3K27ac, H3K18ac, H3K9me3, H3K36me2/3, H4K20me3, and H3K4me1) at 30hpi and 4hpi time points. These data are presented in Supplemental Figure 9. We did not find any significant colocalization, but documented the presence of PMORC in H3K36me2 depleted regions.

• The PfMORC depletion was performed using a glms-based genetic system and the reviewer did not find any quantification of the depletion level at 24h or 36h. This is particularly important as the authors present RNA-seq data at these time points.

We would like to clarify that RNA-seq was performed on 32hpi parasites after approximately 48 h treatment with 2.5 mM GlcN. At the trophozoite and schizont stage, PfMORC expression is high, which is why we selected these time points for RNA-seq (32hpi) and ChIP-seq (30hpi and 40hpi). PfMORC protein expression after GlcN treatment is analyzed in our previous paper (Singh et al., Sci Rep, PMID 33479315), where treatment with 2.5 mM GlcN leads to 50% reduction in PfMORC transcript at 32hpi. This is referenced in the Results section; we decided not to repeat the same experiment in the current manuscript.

• The authors performed a thorough analysis of the correlations between ApiAP2 binding, histone modification and genomic localization of PfMORC (their chip-seq data). However, they found an inverse relationship between H3K36me2, a known histone repressive mark, and PfMORC genomic localization. This is particularly surprising when PfMORC itself is presented as a heterochromatin marker. The wording of this data is confusing in the results section (lines 257-258) and never discussed further. This important data should at least be discussed to make sense of this apparent contradiction.

H3K36me2 indeed acts as a global repressive mark in P. falciparum. However, our hypothesis implies that PfMORC not only overlaps with H3K36me2 depleted region, but also interacts with other epigenetic regulators. Therefore, we propose that PfMORC is part of chromatin remodeling complexes involved in heterochromatin dynamics. Moreover, we did not see any overlap between several other heterochromatin markers, suggesting it has a unique binding preference not shared with other heterochromatin markers. Based on this study and parallel work submitted by Chahine et al. (https://elifesciences.org/reviewed-preprints/92499#abstract), it is evident that PfMORC is crucial for gene regulation and chromatin structure maintenance as shown in other organisms. Currently, we do not know what the apparent mutual exclusion between H3K36me2 and PfMORC implies mechanistically or how PfMORC interaction with heterochromatin aids in chromatin integrity. In Arabidopsis thaliana, MORC binding leads to chromatin compaction and reduces DNA accessibility to transcription factors, thereby repressing gene expression. In P. falciparum, overlap in the binding region of PfMORC with different transcription factors suggests several possibilities that require further investigation. Since there is only one gene encoding a PfMORC protein in P. falciparum, it is possible that PfMORC function is not limited to chromatin integrity, but it may also function to modulate gene expression at different stages. To fully explore the function of PfMORC will require investigating the functional role of the other interacting partners we and others have identified.

We have modified the result section per the reviewer's suggestion, and we now also discuss this finding in more detail in the discussion section.

• The ChIP-seq data are central to this manuscript. However, the presentation of this data in Figure 2A suggests that it is very noisy (particularly for Chr1). It would be of interest to present the called peaks together with the normalized data so that the reader can assess the quality of the ChIP-seq data.

Our results clearly demonstrate the enrichment of PfMORC in sub-telomeric regions and internal heterochromatic islands. These results are consistent across all of our replicates taken at two independent time points of parasite asexual blood stage development and correlate well with the results of Le Roch: https://elifesciences.org/reviewed-preprints/92499. The raw data files have been provided and can be re-analyzed by any user.

• The RNA-seq data showed that only a few genes are affected after 24 h of PfMORC depletion. Furthermore, there is an equal number of up- and down-regulated genes. It is not clear why depletion of a heterochromatin marker would induce down-regulation of genes. How these data relate to the partial depletion of PfMORC is not discussed.

We would like to clarify that RNA-seq experiment was performed at 32hpi after GlcN following knockdown as previously described (Singh et al., Sci Rep, PMID 33479315). Briefly, synchronous, early trophozoites stage (24hpi) PfMORCglmS-HA parasites were treated with 2.5 mM GlcN until they reached the trophozoite stage (32 hpi) in the next cycle. These parasites were then collected for analysis by RNA-seq. We did not detect a substantial log-fold change at this point because only 50% of the transcripts were depleted in the glmS-based PfMORC knockdown system. However, we have seen a distinctive pattern of up (60) and down (103) regulated DEGs that are comprised of egress-related genes or surface antigens. We believe that PfMORC interacts with different ApiAP2 proteins, as shown in Figure 3A, and consequently exhibits multiple functions. This finding has now been corroborated in several other recent studies (See response to Reviewer 1 above).

eLife assessment

This important study addresses the question of how wing morphology and kinematics changed as insect species miniaturized. The authors found no significant correlation between body size and wing kinematics across eight hoverfly species, and instead argue that evolutionary changes in wing size and shape enabled flight in smaller species. However, if the integrative approach to animal biomechanics is strong, the evidence supporting the general conclusion that changes in wing morphology, rather than kinematics, correlate with miniaturization is incomplete and would benefit from more detailed biomechanical analysis and improved methods for phylogenetic comparison.

eLife assessment

Bos and colleagues address the question of how two major inhibitory interneuron classes in the neocortex differentially affect cortical dynamics. They perform stability and gain analysis of simplified models with nonlinear transfer functions to show how, under specific conditions, inhibitory modulation can counter-intuitively increase both response gain and circuit stability. This effect depends on the connection strengths within the circuit model, providing valuable guidance as to when and why it arises. Support for the main conclusions is generally solid, but could be strengthened by additional analyses

Reviewer #1 (Public Review):

Summary:

In this manuscript, Shao et al. investigate the contribution of different cortical areas to working memory maintenance and control processes, an important topic involving different ideas about how the human brain represents and uses information when it is no longer available to sensory systems. In two fMRI experiments, they demonstrate that the human frontal cortex (area sPCS) represents stimulus (orientation) information both during typical maintenance, but even more so when a categorical response demand is present. That is, when participants have to apply an added level of decision control to the WM stimulus, sPCS areas encode stimulus information more than conditions without this added demand. These effects are then expanded upon using multi-area neural network models, recapitulating the empirical gradient of memory vs control effects from visual to parietal and frontal cortices. In general, the experiments and analyses provide solid support for the authors' conclusions, and control experiments and analyses are provided to help interpret and isolate the frontal cortex effect of interest. However, I suggest some alternative explanations and important additional analyses that would help ensure an even stronger level of support for these results and interpretations.

Strengths:

- The authors use an interesting and clever task design across two fMRI experiments that is able to parse out contributions of WM maintenance alone along with categorical, rule-based decisions. Importantly, the second experiment only uses one fixed rule, providing both an internal replication of Experiment 1's effects and extending them to a different situation when rule-switching effects are not involved across mini-blocks.

- The reported analyses using both inverted encoding models (IEM) and decoders (SVM) demonstrate the stimulus reconstruction effects across different methods, which may be sensitive to different aspects of the relationship between patterns of brain activity and the experimental stimuli.

- Linking the multivariate activity patterns to memory behavior is critical in thinking about the potential differential roles of cortical areas in sub-serving successful working memory. Figure 3 nicely shows a similar interaction to that of Figure 2 in the role of sPCS in the categorization vs. maintenance tasks.

- The cross-decoding analysis in Figure 4 is a clever and interesting way to parse out how stimulus and rule/category information may be intertwined, which would have been one of the foremost potential questions or analyses requested by careful readers. However, I think more additional text in the Methods and Results to lay out the exact logic of this abstract category metric will help readers better interpret the potential importance of this analysis and result.

Weaknesses:

- Selection and presentation of regions of interest: I appreciate the authors' care in separating the sPCS region as "frontal cortex", which is not necessarily part of the prefrontal cortex, on which many ideas of working memory maintenance activity are based. However, to help myself and readers interpret these findings, at a minimum the boundaries of each ROI should be provided as part of the main text or extended data figures. Relatedly, the authors use a probabilistic visual atlas to define ROIs in the visual, parietal, and frontal cortices. But other regions of both lateral frontal and parietal cortices show retinotopic responses (Mackey and Curtis, eLife, 2017: https://elifesciences.org/articles/22974) and are perhaps worth considering. Do the inferior PCS regions or inferior frontal sulcus show a similar pattern of effects across tasks? And what about the middle frontal gyrus areas of the prefrontal cortex, which are most analogous to the findings in NHP studies that the authors mention in their discussion, but do not show retinotopic responses? Reporting the effects (or lack thereof) in other areas of the frontal cortex will be critical for readers to interpret the role of the frontal cortex in guiding WM behavior and supporting the strongly worded conclusions of broad frontal cortex functioning in the paper. For example, to what extent can sPCS results be explained by visual retinotopic responses? (Mackey and Curtis, eLife, 2017: https://elifesciences.org/articles/22974).

- When looking at the time course of effects in Figure 2, for example, the sPCS maintenance vs categorization effects occur very late into the WM delay period. More information is needed to help separate this potential effect from that of the response period and potential premotor/motor-related influences. For example, are the timecourses shifted to account for hemodynamic lag, and if so, by how much? Do the sPCS effects blend into the response period? This is critical, too, for a task that does not use a jittered delay period, and potential response timing and planning can be conducted by participants near the end of the WM delay. Regardless, parsing out the timing and relationship to response planning is important, and an ROI for M1 or premotor cortex could also help as a control comparison point, as in reference (24).

- Interpreting effect sizes of IEM and decoding analysis in different ROIs. Here, the authors are interested in the interaction effects across maintenance and categorization tasks (bar plots in Figure 2), but the effect sizes in even the categorization task (y-axes) are always larger in EVC and IPS than in the sPCS region... To what extent do the authors think this representational fidelity result can or cannot be compared across regions? For example, a reader may wonder how much the sPCS representation matters for the task, perhaps, if memory access is always there in EVC and IPS? Or perhaps late sPCS representations are borrowing/accessing these earlier representations? Giving the reader some more intuition for the effect sizes of representational fidelity will be important. Even in Figure 3 for the behavior, all effects are also seen in IPS as well. More detail or context at minimum is needed about the representational fidelity metric, which is cited in ref (35) but not given in detail. These considerations are important given the claims of the frontal cortex serving such an important for flexible control, here.

eLife assessment

This work presents valuable findings that the human frontal cortex is involved in a flexible, dual role in both maintaining information in short-term memory, and controlling this memory content to guide adaptive behavior and decisions. The evidence supporting the conclusions is convincing, with a well-designed task, best-practice decoding methods, and careful control analyses. The work will be of broad interest to cognitive neuroscience researchers working on working memory and cognitive control.

Reviewer #2 (Public Review):

Summary:

The authors provide evidence that helps resolve long-standing questions about the differential involvement of the frontal and posterior cortex in working memory. They show that whereas the early visual cortex shows stronger decoding of memory content in a memorization task vs a more complex categorization task, the frontal cortex shows stronger decoding during categorization tasks than memorization tasks. They find that task-optimized RNNs trained to reproduce the memorized orientations show some similarities in neural decoding to people. Together, this paper presents interesting evidence for differential responsibilities of brain areas in working memory.

Strengths:

This paper was strong overall. It had a well-designed task, best-practice decoding methods, and careful control analyses. The neural network modelling adds additional insight into the potential computational roles of different regions.

Weaknesses:

While the RNN model matches some of the properties of the task and decoding, its ability to reproduce the detailed findings of the paper was limited. Overall, the RRN model was not as well-motivated as the fMRI analyses.

Author response:

(1) Reviewer 1 suggested that we repeat the analyses in additional ROIs in the prefrontal cortex (PFC). We appreciate this suggestion and believe it will contribute to a comprehensive understanding of the current findings. These results will be included in the revision.

(2) Reviewer 1 suggested that we also examine results in motor-related ROIs to rule out influences from response planning. We would like to note that our experimental design makes it unlikely that response planning would have influenced our results, as participants were unable to plan their motor responses in advance due to randomized response mapping on a trial-by-trial basis. Nevertheless, we agree with the reviewer that showing results from motor-related ROIs is important, and will include these results in the revision.

(3) Reviewer 1 raised a question about the effect size of the results across different ROIs. In our manuscript, we tried to avoid direct comparisons of representational strength across ROIs, by focusing on the differences in representational strength between conditions within the same ROI. Nevertheless, we agree that clarifying this issue is important, which we will address in the revision.

(4) Reviewer 2 raised a concern about the similarity between the RNN and fMRI results. We acknowledge that the complexity of our results makes it challenging to replicate all fMRI findings within a single RNN (e.g., simulating three brain regions in a single network with distinct result patterns). Nonetheless, the current RNNs effectively captured our key fMRI findings, including increased stimulus representation in frontal cortex as well as the tradeoff in category representation with varying levels of flexible control. Reviewer 2 also made several suggestions in tweaking the RNN structure and in choosing alternative analysis methods. We are happy to carry out these points as we think they could potentially increase the alignment between the two modalities.

eLife assessment

This work presents some valuable information regarding the molecular mechanisms controlling the regeneration of pancreatic beta cells following induced cell ablation in zebrafish. Specifically, the data suggest that Calcineurin is a regulator of beta cell regeneration. However, the study lacks the critical lineage tracing results to support the conclusion about the origin of the regenerated beta cells and thus is deemed incomplete.

Reviewer #1 (Public review):

Induction of beta cell regeneration is a promising approach for the treatment of diabetes. In this study, Massoz et.al., identified calcineurin (CaN) as a new potential modulator of beta cell regeneration by using zebrafish as model. They also showed that calcineurin (CaN) works together with Notch signaling calcineurin (CaN) to promote the beta cell regeneration. Overall, the paper is well organized, and technically sound. However, some evidences seem weak to get the conclusion.

Reviewer #2 (Public review):

This work started with transcriptomic profiling of ductal cells to identify the upregulation of calcineurin in the zebrafish after beta-cell ablation. By suppressing calcineurin with its chemical inhibitor cyclosporin A and expressing a constitutively active form of calcineurin ubiquitously or specifically in ductal cells, the authors found that inhibited calcineurin activity promoted beta-cell regeneration transiently while ectopic calcineurin activity hindered beta-cell regeneration in the pancreatic tail. They also showed similar effects in the basal state but only when it was within a particular permissive window of Notch activity. To further investigate the roles of calcineurin in the ductal cells, the authors demonstrated that calcineurin inhibition additionally induced the proliferation of the ductal cells in the regenerative context or under a limited level of Notch activity. Interestingly, the enhanced proliferation was followed by a depletion of ductal cells, suggesting that calcineurin inhibition would exhaust the ductal cells. Based on the data, the authors proposed a very attractive and intriguing model of the role of calcineurin in maintaining the balance of the progenitor proliferation and the endocrine differentiation. However, the conclusions of this paper are only partially supported by the data as some evidence of the lineage between ductal cells and beta cells remains suggestive.

Author response:

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

Thank you for all your recommendations to improve the manuscript. We took them into account and tried to integrate them as much as possible in the paper. I understand that the main issue is the lack of genetic lineage tracing. Unfortunately, I am no longer in a position to perform experiments and as a consequence, we cannot bring these data. However, we previously performed several experiments that attest the ductal origin of the beta cells. As a reminder, we used experiment setting where beta cell regeneration occur from the ducts in the pancreatic tail; we used a genetic approach to over-express CaN specifically in the ducts at the level of the pancreas ; and we investigate the function of CaN under Notch repression, known to trigger beta cell formation from the ducts. Altogether, our data underline the contribution of the ductal cells. In addition, as recommended by the editors, we showed that while the proportion of ductal cells EdU+ increase Figure 5 C-D, the number of ductal cells remain constant  Figure 5A supplemental. We integrate a paragraph in the discussion to remind all these points in the manuscript.  

We thank you greatly for your time and consideration for this work.

eLife assessment

This valuable study investigates the effect of noncaloric monosaccharides, sugar substitutes that are commonly used by diabetic patients, on angiogenesis in the zebrafish embryo. The authors show that noncaloric monosaccharides and glucose similarly induce excessive blood vessel formation due to the increased formation of tip cells by endothelial cells through the foxo1a-marcksl1a pathway. This solid study is of interest for the medical community in charge of the prevention and of the treatment of diabetes and other metabolic diseases.

Reviewer #1 (Public Review):

Dong Liu et al. successfully established a short-term zebrafish model by treating the embryos with high concentrations of monosaccharides, resembling the hyperangiogenic characteristics observed in proliferative diabetic retinopathy. The authors found that excessive angiogenesis induced by glucose and noncaloric monosaccharides can be achieved by activating the quiescent endothelial cells into proliferating tip cells. Importantly, the authors further confirmed the effects of monosaccharides on inducing excessive angiogenesis were mediated by the foxo1a-marcksl1a pathway. These results demonstrate the potentially detrimental effects of the noncaloric monosaccharides on blood vessel function and provided novel insights into the underlying mechanisms.

Reviewer #2 (Public Review):

In the manuscript Liu et al. observed that glucose and noncaloric monosaccharides can prompt an excessive formation of blood vessels, particularly intersegmental vessels (ISVs). They propose that these branched vessels arise from the ectopic activation of quiescent endothelial cells (ECs) into tip cells. Moreover, through single-cell transcriptome sequencing analysis of embryonic endothelial cells exposed to glucose, they noted an increased proportion of arterial and capillary endothelial cells, proliferative endothelial cells, along with a series upregulated genes in categories of blood vessel morphogenesis, development, and pro-angiogenesis. The authors provide evidence suggesting that caloric and noncaloric monosaccharides (NMS) induce excessive angiogenesis via the Foxo1a-Marcksl1a pathway.

Reviewer #3 (Public Review):

The authors have investigated the effect of noncaloric monosaccharides on angiogenesis in the zebra fish embryo. These compounds are used as substitutes of sugars to sweeten beverages and they are commonly used by diabetic patients. The authors show that noncaloric monosaccharides and glucose similarly induce excessive blood vessels formation due to increased formation of tip cells by endothelial cells. The authors show that this excessive angiogenesis involved the foxo1a-marcksl1a pathway.

A limitation of the study is that the mechanism of angiogenesis in the retinal circulation and in peripheral vasculature is certainly different.

This result suggests that these noncaloric monosaccharides share common side effects with glucose. Consequently, more caution should be taken as regard to the use of these artificial sweeteners. This work is of interest for a better management of diabetes.

Author response:

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

Reviewer #1:

(1) The authors claimed that they examined the arterial and venous identity of the hyperbranched vessels via live imaging analysis of the high glucose-treated Tg(flt1:YFP::kdrl:ras-mCherry) line, and revealed that the hyperbranched ectopic vessels comprised arteries and veins. That's good, of course. However, there are no relevant results in Figure 2. Please revise it.

Thank you very much for the suggestion. We’ve added this part of the results in Figure 2i and j.

(2) In Figures 3f and 3g, some of the ECs protruded long and intricate sprouts, and nearly all the ECs within an ISV underwent the outgrowth of filopodia in some extreme cases (Figure 3g), suggesting that the high glucose treatment induced the endothelial differentiation into tip cell-like cells. The findings are surprising and interesting. In order to further confirm the author's conclusion, in situ hybridization experiments are more appropriate to show the expression changes of tip cell-like cell marker genes in the high glucose-treated embryos.

Thank you very much for your constructive suggestions. We have performed the analysis of single-cell RNA-seq data, and the results showed that the tip cell marker genes such as esm1, apln, and cxcr4a were significantly up-regulated in arterial and capillary ECs after high glucose treatment. The results were integrated into Figure 3 of the revised manuscript.

(3) Embryos treated with AS1842856 or injected with foxo1a-MO exhibited excessive angiogenesis (Figure 5g-i), suggesting the transcription activity of foxo1 is required to maintain the quiescent state of endothelial cells. Did the downregulation of foxo1a lead to the differentiation of endothelial cells into tip-cell-like cells?

Thank you very much for the question. We examined our results carefully and marked these tip cell-like cells with arrow heads in Figure 5h of the revised manuscript.

(4) Foxo1a was significantly downregulated in arterial and capillary ECs after high glucose treatment (Figure 5c-e). More importantly, whether overexpression of foxo1a in the high glucose-treated embryos could eliminate the hyperangiogenic characteristics?

Thank you for the great questions. We performed rescue experiments, and the results suggested that the overexpression of foxo1a partially mitigated the excessive angiogenesis induced by high glucose treatment. These results were integrated into Figure 6 of the revised manuscript.

(5) The authors' results found that foxo1a was enriched in both the predicted binding sites of marcksl1a by ChIP-PCR experiments (Figure 7d). This result is reliable. However, whether these two sites are important for marcksl1a gene transcription needs to be confirmed by relevant experiments, such as luciferase reporter assays.

We’ve performed the luciferase reporter assays and added these data to Figure 8f and g.

Reviewer #2:

Suggested major experiments:

(1) A previous study (Jorgens et al., Diabetes 64, 2015) reported that high tissue glucose levels increased reactive dicarbonyl methylglyoxal (MG) concentrations in zebrafish embryos and triggered the formation of hyperbranched ISVs. Additionally, they illustrated that MG induced the vascular hyperbranching phenotype via enhancing phosphorylated VEGFR and pAKT signaling cascade. The authors must examine whether both pVEGFR and pAKT are increased in noncaloric monosaccharide (NMS)-treated embryos. The authors need also to test the crosstalks between VEGFR/AKT signaling and foxo1a-Marcksl1a pathway in glucose or NMS-treated embryos.

Thank you very much for your suggestion. We treated the embryos with AS1842856 (foxo1 inhibitor) and Lenvatinib (VEGFR inhibitor), and the results showed that Lenvatinib treatment attenuated the excessive angiogenesis induced by foxo1 inhibition. We also examined the expression level of vegfaa after AS1842856 treatment; the results suggested that foxo1 inhibition did not affect the expression of vegfaa.

Author response image 1.

(2) In this manuscript, the authors performed single endothelial cell sequencing in glucose-treated embryos, and found reduced foxo1a expression and upregulated marcksl1a . Based on these data, the authors demonstrated that glucose and NMS-induced excessive angiogenesis through the foxo1a-marcksl1a pathway. The authors must conduct endothelial scRNA-seq in NMS-treated embryos, and analyze and compare the datasets with scRNA-seq datasets from glucose-treated endothelial cells, considering the focus of the paper. In addition, ASBs have been suggested as healthy alternatives to sugar-sweetened beverages. The authors also need to examine carefully whether metabolic gene programs are altered in glucose-treated endothelial cells, which was mentioned in Jorgens et al paper.

Thank you very much for your constructive suggestions. We have performed the whole embryo transcriptome sequencing after high D-glucose and L-glucose treatment. We analyzed and compared the differentially expressed genes of control, high D-glucose-treated, and high L-glucose-treated embryos. The results revealed that 1259 and 1074 genes were up-regulated significantly in high D-glucose and high L-glucose treated embryos, respectively, compared with control.

We also analyzed some metabolic-related genes and found that some genes involved in gluconeogenesis, glycolysis, and oxidative phosphorylation were significantly changed. The results were integrated into supplementary Figure12 and 13 of the revised manuscript.

(3) Glucose or NMS treatments induce the hyperbranched endothelial vessels from the dorsal aorta and ISVs but not cardinal veins. In Figure 4i, the arterial and capillary cell population is increased in glucose-treated embryos, but the venous cell population seems to be reduced. The authors need to check whether arterial/venous differentiation and proliferation are affected in glucose- and NMS-treated embryos.

Thank you for your suggestions. We examined arterial/venous differentiation based on Tg(flt1BAC:YFP::kdrl:ras-mCherry) zebrafish line, in which the YFP is mainly expressed in arterial Endothelial cells. We found the endothelial cells of excessively formed blood vessels induced by high glucose treatment are mainly arterial (Figure 2j). This might explain why the arterial and capillary cell population was increased in glucose-treated embryos.

(4) The manuscript proposes that excessively branched vessels within ISVs arise from the ectopic activation of quiescent endothelial cells (ECs) into tip cells. To confirm this process, the authors need to detect some specific tip cell markers to demonstrate their ectopic activation.

Thank you for your constructive suggestions. We have performed the analysis of single-cell RNA-seq data, and the results showed that the tip cell marker genes such as esm1, apln, and cxcr4a were significantly up-regulated in arterial and capillary ECs after high glucose treatment. The results were integrated into Figure 3 of the revised manuscript.

(5) Disaccharides such as lactose, maltose, and sucrose did not exhibit a notable induction of excessive angiogenic phenotype. However, the specific treatment concentrations utilized in the study were not delineated. Therefore, further investigation is warranted to determine whether increased disaccharide concentrations can cause vascular hyperbranching phenotype.

Thank you very much for the suggestions. We’ve described the concentrations of monosaccharides and disaccharides in the materials and methods section of the revised manuscript. Following the suggestion, we treated zebrafish embryos with a higher concentration of the disaccharide. The results showed that higher concentrations of disaccharide treatment also caused excessive angiogenesis in zebrafish embryos. These results were integrated into supplementary Figure 8 of the revised manuscript.

(6) The authors claim that glucose and NMS (such as L-glucose) induce excessive angiogenesis through the foxo1a-marcksl1a pathway. Following exposure to elevated glucose levels, a substantial down-regulation of foxo1a was observed in arterial and capillary endothelial cells. This down-regulation led to the release of foxo1a inhibition on marccksl1a, subsequently resulting in an augmented expression of marccksl1a and the manifestation of a vascular phenotype. Consequently, it is imperative to investigate whether the foxo1a overexpression can attenuate marccksl1a expression and mitigate the vascular phenotype induced by monosaccharides. Sufficient data support is needed for the conclusion that monosaccharides induce angiogenesis via the foxo1a-marcksl1a pathway.

Thank you very much for your constructive suggestions.

We confirmed the expression of marcksl1a in foxo1a-overexpressed embryos. The results indicated that foxo1a overexpression significantly attenuated marcksl1a expression. The results were integrated into Figure 8c. We also performed the rescue experiments, which indicated that overexpression of foxo1a partially mitigated the excessive angiogenesis induced by high glucose treatment. These results were integrated into Figure 6 of the revised manuscript.

Minor corrections:

(1) Figure 2i, j has no corresponding graphs.

We’ve made the change in Figure 2.

(2) Figure 2h has no vertical coordinates.

We’ve made the change in Figure 2.

(3) All Figures should be referenced within the manuscript.

We’ve checked our manuscript carefully and made the corrections.

(4) The concentrations of monosaccharides and disaccharides employed in this study must be distinctly elucidated within the manuscript and annotated using the internationally recognized unit notation.

We’ve checked our manuscript carefully and described the concentrations of monosaccharides and disaccharides in the revised materials and methods section.

Reviewer #3:

(1) A possible limitation of the study is that the mechanism leading to angiogenesis in the retinal circulation and in peripheral vasculature is certainly different as diabetes is associated with excessive angiogenesis in the retina and a defect in angiogenesis in the peripheral circulation as shown by a reduced post-ischemic revascularization (see Silvestre et al.: DOI: 10.1152/physrev.00006.2013).

Thank you very much for your suggestions. As you said, the peripheral blood vessel model in this study does not fully represent individuals with diabetic retinopathy, which is a limitation. However, from a specific view, the phenotype and mechanism of excessive angiogenesis of peripheral blood vessels in the high glucose model may provide a reference for excessive angiogenesis in the retina; they might have similar etiology and regulation mechanisms in excessive angiogenesis.

(2) Another limitation is that angiogenesis in the embryo is not fully representative of the excessive angiogenesis observed in the diabetic retinal circulation. It would be of interest to analyse the retinal vascular tree in adult fish submitted to high glucose and to ASB.

In our future study, we will try to observe the angiogenesis phenotype in the diabetic retina and improve the disease model.

(3) Line 52: "Endothelial cell dysfunction (ECD)" instead of "Endothelial dysfunction (ECD)".

We’ve made the correction in the revised manuscript.

(4) The authors should elaborate more on the observation showing that L-glucose, D-mannose, D-ribose, and L-arabinose, which could not be digested by animals, also induce excessive angiogenesis. Is the effect indirect?

In the current manuscript, we conducted an in vivo live imaging analysis to show the phenotype of excessive angiogenesis caused by those noncaloric monosaccharides. However, we did not find differences in the phenotypes of embryos treated with noncaloric and caloric monosaccharides. Therefore, we supposed that the mechanisms underlying the phenotypes were similar. The effect might be indirect.

eLife assessment

In this important study, Baniulyte and Wade provide solid evidence that translation of a short ORF denoted toiL positioned upstream of the topAI-yjhQP operon is responsive to different ribosome-targeting antibiotics, consequently controlling translation of the TopAI toxin as well as Rho-dependent transcription termination. Strengths of the study include combining a genetic screen to identify 23S rRNA mutations that affect topA1 expression and a creative approach to map the different locations of ribosome stalling within toiL induced by different antibiotics, with ribosome profiling and RNA structure probing by SHAPE to examine consequences of different antibiotics on toiL-mediated regulation. The work could be improved by examining the physiological consequences of topAI-yjhQP activation on antibiotic exposure and by resolving discrepancies between the SHAPE data and the translation rate of toiL.

Reviewer #1 (Public Review):

Summary:

The manuscript reports that expression of the E. coli operon topAI/yjhQ/yjhP is controlled by the translation status of a small open reading frame, that authors have discovered and named toiL, located in the leader region of the operon. The authors propose the following model for topAI activation: Under normal conditions, toiL is translated but topAI is not expressed because of Rho-dependent transcription termination within the topAI ORF and because its ribosome binding site and start codon are trapped in an mRNA hairpin. Ribosome stalling at various codons of the toiL ORF, caused by the presence of some ribosome-targeting antibiotics, triggers an mRNA conformational switch which allows translation of topAI and, in addition, activation of the operon's transcription because the presence of translating ribosomes at the topAI ORF blocks Rho from terminating transcription. Even though the model is appealing and several of the experimental data support some aspects of it, several inconsistencies remain to be solved. In addition, even though TopAI was shown to be an inhibitor of topoisomerase I (Yamaguchi & Inouye, 2015, NAR 43:10387), the authors suggest, without offering any experimental support, that, because ribosome-targeting antibiotics act as inducers, expression of the topAI/yjhQ/yjhP operon may confer resistance to these drugs.

Strengths:

- There is good experimental support of the transcriptional repression/activation switch aspect of the model, derived from well-designed transcriptional reporters and ChIP-qPCR approaches.

- There is a clever use of the topAI-lacZ reporter to find the 23S rRNA mutants where expression topAI was upregulated. This eventually led the authors to identify that translation events occurring at toiL are important to regulate the topAI/yjhQ/yjhP operon. This section can be strengthened if the authors suggest an explanation for how mutant ribosomes translating toiL increased topAI expression. Is there any published evidence that ribosomes with the identified mutations translate slowly (decreased fidelity does not necessarily mean slow translation, does it?)?

- Authors incorporate relevant links to the antibiotic-mediated expression regulation of bacterial resistance genes. Authors can also mention the tryptophan-mediated ribosome stalling at the tnaC leader ORF that activates the expression of tryptophan metabolism genes through blockage of Rho-mediated transcriptional attenuation.

Weaknesses:

The main weaknesses of the work are related to several experimental results that are not consistent with the model, or related to a lack of data that needs to be included to support the model.

The following are a few examples:

- It is surprising that authors do not mention that several published Ribo-seq data from E. coli cells show active translation of toiL (for example Li et al., 2014, Cell 157: 624). Therefore, it is hard to reconcile with the model that starts codon/Shine-Dalgarno mutations in the toiL-lux reporter have no effect on luciferase expression (Figure 2C, bar graphs of the no antibiotic control samples).

- The SHAPE reactivity data shown in Figure 5A are not consistent with the toiL ORF being translated. In addition, it is difficult to visualize the effect of tetracycline on mRNA conformation with the representation used in Figure 5B. It would be better to show SHAPE reactivity without/with Tet (as shown in panel A of the figure).

- The "increased coverage" of topAI/yjhP/yjhQ in the presence of tetracycline from the Ribo-seq data shown in Figure 6A can be due to activation of translation, transcription, or both. For readers to know which of these possibilities apply, authors need to provide RNA-seq data and show the profiles of the topAI/yjhQ/yjhP genes in control/Tet-treated cells.

- Similarly, to support the data of increased ribosomal footprints at the toiL start codon in the presence of Tet (Figure 6B), authors should show the profile of the toiL gene from control and Tet-treated cells.

- Representation of the mRNA structures in the model shown in Figure 5, does not help with visualizing 1) how ribosomes translate toiL since the ORF is trapped in double-stranded mRNA, and 2) how ribosome stalling on toiL would lead to the release of the initiation region of topAI to achieve expression activation.

- The authors speculate that, because ribosome-targeting antibiotics act as expression inducers [by the way, authors should mention and comment that, more than a decade ago, it had been reported that kanamycin (PMID: 12736533) and gentamycin (PMID: 19013277) are inducers of topAI and yjhQ], the genes of the topAI/yjhQ/yjhP operon may confer resistance to these antibiotics. Such a suggestion can be experimentally checked by simply testing whether strains lacking these genes have increased sensitivity to the antibiotic inducers.

Reviewer #2 (Public Review):

Summary:

In this important study, Baniulyte and Wade describe how the translation of an 8-codon uORF denoted toiL upstream of the topAI-yjhQP operon is responsive to different ribosome-targeting antibiotics, consequently controlling translation of the TopAI toxin as well as Rho-dependent termination with the gene.

Strengths:

I appreciate that the authors used multiple different approaches such as a genetic screen to identify factors such as 23S rRNA mutations that affect topA1 expression and ribosome profiling to examine the consequences of various antibiotics on toiL-mediated regulation. The results are convincing and clearly described.

Weaknesses:

I have relatively minor suggestions for improving the manuscript. These mainly relate to the figures.

Reviewer #3 (Public Review):

Summary:

The authors nicely show that the translation and ribosome stalling within the ToiL uORF upstream of the co-transcribed topAI-yjhQ toxin-antitoxin genes unmask the topAI translational initiation site, thereby allowing ribosome loading and preventing premature Rho-dependent transcription termination in the topAI region. Although similar translational/transcriptional attenuation has been reported in other systems, the base pairing between the leader sequence and the repressed region by the long RNA looping is somehow unique in toiL-topAI-yjhQP. The experiments are solidly executed, and the manuscript is clear in most parts with areas that could be improved or better explained. The real impact of such a study is not easy to appreciate due to a lack of investigation on the physiological consequences of topAI-yjhQP activation upon antibiotic exposure (see details below).

Strengths:

>Conclusion/model is supported by the integrated approaches consisting of genetics, in vivo SHAPE-seq and Ribo-Seq.

>Provide an elegant example of cis-acting regulatory peptides to a growing list of functional small proteins in bacterial proteomes.

Author Response:

The reviewers suggested that we determine whether the functions of TopAI, YjhQ, and/or YjhP are connected to antibiotic susceptibility. 

We fully agree with the reviewers that the function of TopAI/YjhQ/YjhP is an important topic. Our preliminary studies (not included in the paper) failed to identify a function connected to antibiotic susceptibility, although these studies were far from exhaustive. There are many environmental stressors that can stall ribosomes, making it challenging to find the functionally relevant stressor(s). We feel that further work on this topic is outside the scope of this manuscript.

The reviewers suggested that the SHAPE data are inconsistent with our conclusions about translation of toiL.

We believe the SHAPE data are consistent with our model, although we acknowledge that interpretation of base reactivity is somewhat subjective. We will address the reviewers’ comments on this topic in more detail in our full response.

The reviewers suggested that published Ribo-Seq data are inconsistent with our data showing that toiL start codon/Shine-Dalgarno mutations have no effect on expression of luciferase reporters in the absence of antibiotics. 

Our assays with these mutations looked at expression of topAI, not toiL. Our model predicts that mutations that prevent toiL translation will not induce expression of the downstream genes. We did not look at the effect of these mutations on expression of toiL itself.

The reviewers suggested we use RNA-seq to complement the Ribo-seq data for cells grown +/- tetracycline (Figure 5).

In principle, RNA-seq data would allow us to determine whether tetracycline specifically induces translation of topAI, as opposed to only increasing the RNA level. We did not generate RNA-seq data because prior work from other groups suggests that topAI is too weakly expressed to accurately measure translation efficiency in non-inducing conditions. However, the major conclusion from Figure 5 is that tetracycline stalls ribosomes at start codons, including the start codon of toiL.

eLife assessment

This meta-analysis presents valuable findings that reexamine the function of butterfly eyespots in predator avoidance and report for conspicuousness over mimicry. The analysis is robust, but the evidence supporting the importance of conspicuousness is incomplete due to the limitations of the literature, and this debate would benefit from additional experiments that would strengthen these claims. This paper is of interest to evolutionary biologists and ecologists working on the evolution of morphology and predator-prey interactions.

Reviewer #1 (Public Review):

The question of whether eyespots mimic eyes has certainly been around for a very long time and led to a good deal of debate and contention. This isn't purely an issue of how eyespots work either, but more widely an example of the potential pitfalls of adopting 'just-so-stories' in biology before conducting the appropriate experiments. Recent years have seen a range of studies testing eye mimicry, often purporting to find evidence for or against it, and not always entirely objectively. Thus, the current study is very welcome, rigorously analysing the findings across a suite of papers based on evidence/effect sizes in a meta-analysis.

The work is very well conducted, robust, objective, and makes a range of valuable contributions and conclusions, with an extensive use of literature for the research. I have no issues with the analysis undertaken. The results and conclusions are compelling. It's probably fair to say that the topic needs more experiments to really reach firm conclusions but the authors do a good job of acknowledging this and highlighting where that future work would be best placed.

Author response:

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

Public Reviews: 

Reviewer #1 (Public Review):

Summary:

The question of whether eyespots mimic eyes has certainly been around for a very long time and led to a good deal of debate and contention. This isn't purely an issue of how eyespots work either, but more widely an example of the potential pitfalls of adopting 'just-so-stories' in biology before conducting the appropriate experiments. Recent years have seen a range of studies testing eye mimicry, often purporting to find evidence for or against it, and not always entirely objectively. Thus, the current study is very welcome, rigorously analysing the findings across a suite of papers based on evidence/effect sizes in a meta-analysis.

Strengths:

The work is very well conducted, robust, objective, and makes a range of valuable contributions and conclusions, with an extensive use of literature for the research. I have no issues with the analysis undertaken, just some minor comments on the manuscript. The results and conclusions are compelling. It's probably fair to say that the topic needs more experiments to really reach firm conclusions but the authors do a good job of acknowledging this and highlighting where that future work would be best placed.

Weaknesses:

There are few weaknesses in this work, just some minor amendments to the text for clarity and information.

We greatly appreciate Reviewer 1’s positive comments on our manuscript. We also revised our manuscript text and a figure in accordance with Reviewer 1’s recommendations.

Reviewer #2 (Public Review):

Many prey animals have eyespot-like markings (called eyespots) which have been shown in experiments to hinder predation. However, why eyespots are effective against predation has been debated. The authors attempt to use a meta-analytical approach to address the issue of whether eye-mimicry or conspicuousness makes eyespots effective against predation. They state that their results support the importance of conspicuousness. However, I am not convinced by this.

There have been many experimental studies that have weighed in on the debate. Experiments have included manipulating target eyespot properties to make them more or less conspicuous, or to make them more or less similar to eyes. Each study has used its own set of protocols. Experiments have been done indoors with a single predator species, and outdoors where, presumably, a large number of predator species predated upon targets. The targets (i.e, prey with eyespot-like markings) have varied from simple triangular paper pieces with circles printed on them to real lepidopteran wings. Some studies have suggested that conspicuousness is important and eye-mimicry is ineffective, while other studies have suggested that more eye-like targets are better protected. Therefore, there is no consensus across experiments on the eye-mimicry versus conspicuousness debate.

The authors enter the picture with their meta-analysis. The manuscript is well-written and easy to follow. The meta-analysis appears well-carried out, statistically. Their results suggest that conspicuousness is effective, while eye-mimicry is not. I am not convinced that their meta-analysis provides strong enough evidence for this conclusion. The studies that are part of the meta-analysis are varied in terms of protocols, and no single protocol is necessarily better than another. Support for conspicuousness has come primarily from one research group (as acknowledged by the authors), based on a particular set of protocols.

Furthermore, although conspicuousness is amenable to being quantified, for e.g., using contrast or size of stimuli, assessment of 'similarity to eyes' is inherently subjective. Therefore, manipulation of 'similarity to eyes' in some studies may have been subtle enough that there was no effect.

There are a few experiments that have indeed supported eye-mimicry. The results from experiments so far suggest that both eye-mimicry and conspicuousness are effective, possibly depending on the predator(s). Importantly, conspicuousness can benefit from eye-mimicry, while eye-mimicry can benefit from conspicuousness.

Therefore, I argue that generalizing based on a meta-analysis of a small number of studies that conspicuousness is more important than eye-mimicry is not justified. To summarize, I am not convinced that the current study rules out the importance of eye-mimicry in the evolution of eyespots, although I agree with the authors that conspicuousness is important.

We understand Reviewer 2’s concerns and have addressed them by adding some sentences in the discussion part (L506- 508, L538-L540). In addition, our findings, which were guided by current knowledge, support the conspicuousness hypothesis, but we acknowledge the two hypotheses are not mutually exclusive (L110-112). We also do not reject the eye mimicry hypothesis. As we have demonstrated, there are still several gaps in the current literature and our understanding (L501-553). Our aim is for this research to stimulate further studies on this intriguing topic and to foster more fruitful discussions.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Minor comments

Lines 59/60: "it is possible that eyespots do not involve mimicry of eyes..."

The sentence was revised (L59). To enhance readability, we have integrated Reviewer 1's suggestions by simplifying the relevant section instead of using the suggested sentence.

Line 61: not necessarily aposematism. They might work simply through neophobia, unfamiliarity, etc even without unprofitability

We changed the text in line with the comment from Reviewer 1 (L61-63).

Lines 62/63 - this is a little hard to follow because I think you really mean both studies of real lepidopterans as well as artificial targets. Need to explain a bit more clearly.

We provided an additional explanation of our included primary study type (L64-65).

Lines 93/94 - not quite that they have nothing to do with predator avoidance, but more that any subjective resemblance to eyes is coincidental, or simply as a result of those marking properties being more effective through conspicuousness in their own right.

Line 94 - similarly, not just aposematism. You explain the possible reasons above on l92 as also being neophobia, etc.

We agreed with Reviewer 1’s comments and added more explanations about the conspicuousness hypothesis (L96-97). We have also rewritten the sentences that could be misleading to readers (L428).

Line 96 - this is perhaps a bit misleading as it seems to conflate mechanism and function. The eye mimicry vs conspicuousness debate is largely about how the so-called 'intimidation' function of eyespots works. That is, how eyespots prevent predators from attacking. The deflection hypothesis is a second function of eyespots, which might also work via consciousness or eye mimicry (e.g. if predators try to peck at 'eyes') but has been less central to the mimicry debate.

The explanations and suggestions from Reviewer 1 are very helpful. We revised this part of our manuscript (L103-108) and Figure 1 and its legend to make it clearer that the eyespot hypothesis and the conspicuousness hypothesis explain anti-predator functions from a different perspective than the deflection hypothesis.

There is a third function of eyespots too, that being as mate selection traits. Note that Figure 1 should also be altered to reflect these points.

We wanted to focus on explaining why eyespot patterns can contribute to prey survival. Therefore, we did not state that eyespot patterns function as mate selection traits in this paragraph. Alternatively, we have already mentioned this in the Discussion part (L455-L465) and rewrote it more clearly (L456).

Were there enough studies on non-avian predators to analyse in any way? 

We found a few studies on non-avian predators (e.g. fish, invertebrates, or reptiles), but not enough to conduct a meta-analysis.

Line 171/72 - why? Can you explain, please.

The reason we excluded studies that used bright or contrasting patterns as control stimuli in our meta-analysis is to ensure comparability across primary studies. We added an explanation in the text (L180-181).

Line 177 - can you clarify this?

Without control stimuli, it is challenging to accurately assess the effect of eyespots or other conspicuous patterns on predation avoidance. Control stimuli allow for a comparison of the effect of eyespots or patterns. We added a more detailed explanation to clarify here (L186-188).

Line 309 - presumably you mean 33 papers, each of which may have multiple experiments? I might have missed it, but how many individual experiments in total? 

There were 164 individual experiments. We have now added that information in the manuscript (L320).

Line 320 - paper shaped in a triangle mostly?

We cannot say that most artificial prey were triangular. After excluding the caterpillar type, 57.4% were triangular, while the remaining 43.6% were rectangular (Figure 2b).

Line 406: Stevens.

We fixed this name, thank you (L417).

Discussion - nice, balanced and thorough. Much of the work done has been in Northern Europe where eyespot species are less common. Perhaps things may differ in areas where eyespots are more prevalent.

We appreciate Reviewer 1’s kind words and comments. We agree with your comments and reflected them in our manuscript (L542-545).

Line 477 - True, and predators often have forward-facing eyes making it likely both would often be seen, but a pair of eyes may not be absolutely crucial to avoidance since sometimes a prey animal may only see one eye of a predator (e.g. if the other is occluded, or only one side of the head is visible).

We were grateful for Reviewer 1's comment. We added a sentence noting that the eyespots do not necessarily have to be in pairs to resemble eyes (L490-L492).

eLife assessment

Bonnifet et al. present data on the expression and interacting partners of the transposable element L1 in the mammalian brain. The work includes important findings addressing the potential role of L1 in aging and neurodegenerative disease. However, several aspects of experimental evidence presented are preliminary and the study remains incomplete in its current form.

Reviewer #1 (Public Review):

Summary:

In this study, Bonnifet et al. profile the presence of L1 ORF1p in the mouse and human brain. They claim that ORF1p is expressed in the human and mouse brain at a steady state and that there is an age-dependent increase in expression. This is a timely report as two recent papers have extensively documented the presence of full-length L1 transcripts in the mouse and human brain (PMID: 38773348 & PMID: 37910626). Thus, the finding that L1 ORF1p is consistently expressed in the brain is not surprising, but important to document.

Strengths:

Several parts of this manuscript appear to be well done and include the necessary controls. In particular, the evidence for steady-state expression of ORF1p in the mouse brain appears robust.

Weaknesses:

Several parts of the manuscript appear to be more preliminary and need further experiments to validate their claims. In particular, the data suggesting expression of L1 ORF1p in the human brain and the data suggesting increased expression in the aged brain need further validation. Detailed comments:

(1) The expression of ORF1p in the human brain shown in Figure 1j is not convincing. Why are there two strong bands in the WB? How can the authors be sure that this signal represents ORF1p expression and not non-specific labelling? Additional validations and controls are needed to verify the specificity of this signal.

(2) The data shown in Figure 2g are not convincing. How can the authors be sure that this signal represents ORF1p expression and not non-specific labelling? Extensive additional validations and controls are needed to verify the specificity of this signal.

(3) The data showing a reduction in ORF1p expression in the aged mouse brain is confusing and maybe even misleading. Although there is an increase in the intensity of the ORF1p signal in ORF1p+ cells, the data clearly shows that fewer cells express ORF1p in the aged brain. If these changes indicate an overall loss or gain of ORF1p, expression in the aged brain is not resolved. Thus, conclusions should be more carefully phrased in this section. It is important to show the quantification of NeuN+ and NeuN- cells in young vs aged (not only the proportions as shown in Figure 3b) to determine if the difference in the number of ORF1p+ cells is due to loss of neurons or perhaps a sampling issue. More so, it would be essential to perform WB and/or proteomics experiments to complement the IHC data for the aged mouse samples.

(4) The transcriptomic data presented in Figure 4 and Figure 5 are not convincing. Quantification of transposon expression on short read sequencing has important limitations. Longer reads and complementary approaches are needed to study the expression of evolutionarily young L1s (see PMID: 38773348 & PMID: 37910626 for examples of the current state of the art). Given the read length and the unstranded sequencing approach, I would at least ask the authors to add genome browser tracks of the upregulated loci so that we can properly assess the clarity of the results. I would also suggest adding the mappability profile of the elements in question. In addition, since this manuscript focuses on ORF1p, it would be essential to document changes in protein levels (and not just transcripts) in the ageing human brain.

(5) More information is needed on RNAseq of microdissections of dopaminergic neurons from 'healthy' post-mortem samples of different ages. No further information on these samples is provided. I would suggest adding a table with the clinical information of these samples (especially age, sex, and cause of death). The authors should also discuss whether this experiment has sufficient power. The human ageing cohort seems very small to me.

(6) The findings in this manuscript apply to both human and mouse brains. However, the landscape of the evolutionarily young L1 subfamilies between these two species is very different and should be part of the discussion. For example, the regulatory sequences that drive L1 expression are quite different in human and mouse L1s. This should be discussed.

(7) On page 3 the authors write: "generally accepted that TE activation can be both, a cause and consequence of aging". This statement does not reflect the current state of the field. On the contrary, this is still an area of extensive investigation and many of the findings supporting this hypothesis need to be confirmed in independent studies. This statement should be revised to reflect this reality.

Reviewer #2 (Public Review):

Summary:

Bonnifet et al. sought to characterize the expression pattern of L1 ORF1p expression across the entire mouse brain, in young and aged animals, and to corroborate their characterization with Western blotting for L1 ORF1p and L1 RNA expression data from human samples. They also queried L1 ORF1p interacting partners in the mouse brain by IP-MS.

Strengths:

A major strength of the study is the use of two approaches: a deep-learning detection method to distinguish neuronal vs. non-neuronal cells and ORF1p+ cells vs. ORF1p- cells across large-scale images encompassing multiple brain regions mapped by comparison to the Allen Brain Atlas, and confocal imaging to give higher resolution on specific brain regions. These results are also corroborated by Western blotting on six mouse brain regions. Extension of their analysis to post-mortem human samples, to the extent possible, is another strength of the paper. The identification of novel ORF1p interactors in the brain is also a strength in that it provides a novel dataset for future studies.

Weaknesses:

The main weakness of the study is that cell type specificity of ORF1p expression was not examined beyond neuron (NeuN+) vs non-neuron (NeuN-). Indeed, a recent study (Bodea et al. 2024, Nature Neuroscience) found that ORF1p expression is characteristic of parvalbumin-positive interneurons, and it would be very interesting to query whether other neuronal subtypes in different brain regions are distinguished by ORF1p expression. The data suggesting that ORF1p expression is increased in aged mouse brains is intriguing, although it seems to be based upon modestly (up to 27%, dependent on brain region) higher intensity of ORF1p staining rather than a higher proportion of ORF1+ neurons. Indeed, the proportion of NeuN+/Orf1p+ cells actually decreased in aged animals. It is difficult to interpret the significance and validity of the increase in intensity, as Hoechst staining of DNA, rather than immunostaining for a protein known to be stably expressed in young and aged neurons, was used as a control for staining intensity. The main weakness of the IP-MS portion of the study is that none of the interactors were individually validated or subjected to follow-up analyses. The list of interactors was compared to previously published datasets, but not to ORF1p interactors in any other mouse tissue.

The authors achieved the goals of broadly characterizing ORF1p expression across different regions of the mouse brain, and identifying putative ORF1p interactors in the mouse brain. However, findings from both parts of the study are somewhat superficial in depth.

This provides a useful dataset to the field, which likely will be used to justify and support numerous future studies into L1 activity in the aging mammalian brain and in neurodegenerative disease. Similarly, the list of ORF1p interacting proteins in the brain will likely be taken up and studied in greater depth.

Reviewer #3 (Public Review):

The question about whether L1 exhibits normal/homeostatic expression in the brain (and in general) is interesting and important. L1 is thought to be repressed in most somatic cells (with the exception of some stem/progenitor compartments). However, to our knowledge, this has not been authoritatively / systematically examined and the literature is still developing with respect to this topic. The full gamut of biological and pathobiological roles of L1 remains to be shown and elucidated and this area has garnered rapidly increasing interest, year-by-year. With respect to the brain, L1 (and repeat sequences in general) have been linked with neurodegeneration, and this is thought to be an aging-related consequence or contributor (or both) of inflammation. This study provides an impressive and apparently comprehensive imaging analysis of differential L1 ORF1p expression in mouse brain (with some supporting analysis of the human brain), compatible with a narrative of non-pathological expression of retrotransposition-competent L1 sequences. We believe this will encourage and support further research into the functional roles of L1 in normal brain function and how this may give way to pathological consequences in concert with aging. However, we have concerns with conclusions drawn, in some cases regardless of the lack of statistical support from the data. We note a lack of clarity about how the 3rd party pre-trained machine learning models perform on the authors' imaging data (validation/monitoring tests are not reported), as well as issues (among others) with the particular implementation of co-immunoprecipitation (ORF1p is not among the highly enriched proteins and apparently does not reach statistical significance for the comparison) - neither of which may be sufficiently rigorous.

Author Response:

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this study, Bonnifet et al. profile the presence of L1 ORF1p in the mouse and human brain. They claim that ORF1p is expressed in the human and mouse brain at a steady state and that there is an age-dependent increase in expression. This is a timely report as two recent papers have extensively documented the presence of full-length L1 transcripts in the mouse and human brain (PMID: 38773348 & PMID: 37910626). Thus, the finding that L1 ORF1p is consistently expressed in the brain is not surprising, but important to document.  

Thank you for recognizing the importance of this study. The two cited papers have indeed reported the presence of full-length transcripts in the mouse and human brain. However, the first (PMID: 38773348) report has shown evidence of flL1 RNA and ORF1 protein expression in the mouse hippocampus (but not elsewhere) and the second (PMID: 37910626) shows full-length LINE-1 RNA expression and H3K4me3-ChIP data in the frontal and temporal lobe of the human brain, but not protein expression.  

Strengths:

Several parts of this manuscript appear to be well done and include the necessary controls. In particular, the evidence for steady-state expression of ORF1p in the mouse brain appears robust. 

Weaknesses: 

Several parts of the manuscript appear to be more preliminary and need further experiments to validate their claims. In particular, the data suggesting expression of L1 ORF1p in the human brain and the data suggesting increased expression in the aged brain need further validation. Detailed comments: 

(1) The expression of ORF1p in the human brain shown in Figure 1j is not convincing. Why are there two strong bands in the WB? How can the authors be sure that this signal represents ORF1p expression and not nonspecific labelling? Additional validations and controls are needed to verify the specificity of this signal. 

We have validated the antibody (Abcam 245249 - https://www.abcam.com/en-us/products/primary-antibodies/line-1-orf1p-antibody-epr22227-6-ab245249), which we use for Western blotting experiments like in Fig1j), by several means. We have done immunoprecipitations (IPs) and co-immunoprecipitations (co-IPs) followed by quantitative mass spectrometry (LC-MS/MS). We efficiently detect ORF1p in IPs (Western blot) and by quantitative mass spectrometry (5 independent samples per IP-ORF1p and IP-IgG: ORF1p/IgG ratio: 40.86; adj p-value 8.7e-07; human neurons in culture). We also did co-IPs followed by Western blot using two different antibodies, the Millipore or the Abcam antibody to immunoprecipitate and the Abcam antibody for Western blotting (the Millipore AB does not work well on WB in our hands) which consistently showed a double band indicating that both bands are ORF1p-derived. We can provide this data to the revised manuscript, although some of it (the MS data) is subject of another study in preparation. Abcam also reports a double band, and they suspect that the lower band is a truncated form (see the link to their website above). ORF1p Western blots done by other labs with different antibodies have detected a second band in human samples

(1) Sato, S. et al. LINE-1 ORF1p as a candidate biomarker in high grade serous ovarian carcinoma. Sci Rep 13, 1537 (2023) in Figure 1D

(2) McKerrow, W. et al. LINE-1 expression in cancer correlates with p53 mutation, copy number alteration, and S phase checkpoint. Proc. Natl. Acad. Sci. U.S.A. 119, e2115999119 (2022)) showing a Western blot of an inducible LINE-1 (ORFeus) detected by the MABC1152 ORF1p antibody from Millipore Sigma in Figure 7 3) in a publication in eLife (Walter et al. eLife 2016;5:e11418. DOI: 10.7554/eLife.11418) in mouse ES cells with an antibody made in-house from another lab (gift) – Figure 2B

The lower band might thus be a truncated form of ORF1p or a degradation product which appears to be shared by mouse and human ORF1p. We will mention this in the revised version of the paper. In addition, we have used the very well characterized antibody from Millipore (https://www.merckmillipore.com/CH/en/product/Anti-LINE-1-ORF1p-Antibody-clone-4H1,MM_NF-MABC1152?ReferrerURL=https%3A%2F%2Fwww.google.com%2F) for immunostainings and detect ORF1p staining in human neurons in the very same brain regions (Fig 2H) including the cerebellum (selectively in Purkinje cells as in mice in Fig1B panel 10: human images not shown). 

Altogether, based on our experimental validations and evidence from the literature, we are very confident that it is ORF1p that we detect on the blots. 

(2) The data shown in Figure 2g are not convincing. How can the authors be sure that this signal represents ORF1p expression and not non-specific labelling? Extensive additional validations and  controls are needed to verify the specificity of this signal.

Figure 2g shows a Western blot using an extensively used and well characterized ORF1p antibody from abcam (mouse ORF1p - (https://www.abcam.com/en-us/products/primary-antibodies/line-1-orf1p-antibody-epr21844108-ab216324; cited in at least 11 publications) after FACS-sorting of neurons (NeuN+) of the mouse brain. We have validated this ORF1p antibody ourselves in IPs (see Fig 6A) and co-IP followed by mass spectrometry (LC/MS-MS; see Fig 6, where we detect ORF1p exclusively in the 5 independent ORF1p-IP samples and not at all in 5 independent IgG-IP control samples, see Suppl Table 2). This together makes us very confident that we are looking at a specific ORF1p signal. Please note that in the IP of ORF1p shown in Fig6A, there is a double band as well, strongly suggesting that the lower band might be a truncated or processed form of ORF1p. As stated above, this double band has been detected in other studies (Walter et al. eLife 2016;5:e11418. DOI: 10.7554/eLife.11418) in mouse ES cells using an in-house generated antibody against mouse ORF1p. Thus, with either commercial or in-house generated antibodies in some mouse and human samples, there is a double band corresponding to full-length ORF1p and a truncated or processed version of it.

We noticed that we have not added the references of the primary antibodies used in Western blot experiments in the manuscript, which will be corrected in the revised version.  

(3) The data showing a reduction in ORF1p expression in the aged mouse brain is confusing and maybe even misleading. Although there is an increase in the intensity of the ORF1p signal in ORF1p+ cells, the data clearly shows that fewer cells express ORF1p in the aged brain. If these changes indicate an overall loss or gain of ORF1p, expression in the aged brain is not resolved. Thus, conclusions should be more carefully phrased in this section. It is important to show the quantification of NeuN+ and NeuN- cells in young vs aged (not only the proportions as shown in Figure 3b) to determine if the difference in the number of ORF1p+ cells is due to loss of neurons or perhaps a sampling issue. More so, it would be essential to perform WB and/or proteomics experiments to complement the IHC data for the aged mouse samples. 

The data presented in Fig3 C-I show a modest but widespread and reproducible increase in expression of ORF1p per cell. What decreases is the proportion of ORF1p+/NeuN+ cells (Fig3A, B), indicating that fewer cells might express ORF1p in the brain. However, the proportion or number/mm2 of ORF1p+ cells overall does not decrease significantly, neither does the proportion or number/mm2 of NeuN+ cells (data will be added to the revision). We show data of the % of NeuN+ and NeuN- cells in the ventral midbrain (Suppl Fig3C, quantified on confocal images)) which indeed indicates that in this region, there are less neurons in the aged mouse brain compared to the young. There might thus be a very regional decrease in neurons with age in the midbrain motor region. We will, however, as suggested, plot the number of NeuN+ and NeuN- cells per mm2 for the whole brain as well as the different regions in young vs aged to compare actual cell numbers per volume. While it is true that we cannot say that there is an overall loss or gain of ORF1p expression in the aged mouse brain, we believe that this is not of the highest importance as what most likely matters biologically in the context of aging is the quantity of ORF1p per cell (and possibly full-length LINE-1 RNA and ORF2p) and not “per brain”. 

We also plan on doing Western blots on mouse brain tissues from young and aged individuals, however, we might run into limits regarding tissue availability of aged mice. 

(4) The transcriptomic data presented in Figure 4 and Figure 5 are not convincing. Quantification of transposon expression on short read sequencing has important limitations. Longer reads and complementary approaches are needed to study the expression of evolutionarily young L1s (see PMID: 38773348 & PMID: 37910626 for examples of the current state of the art). Given the read length and the unstranded sequencing approach, I would at least ask the authors to add genome browser tracks of the upregulated loci so that we can properly assess the clarity of the results. I would also suggest adding the mappability profile of the elements in question. In addition, since this manuscript focuses on ORF1p, it would be essential to document changes in protein levels (and not just transcripts) in the ageing human brain. 

We agree that there are limitations to the analysis of TEs with short read sequencing and we will add more text on this aspect in a revised version. The approaches shown in PMID: 38773348 & PMID: 37910626 or even a combination of them, would be ideal of course. However, here we reanalyzed a unique existing dataset (Dong et al, Nature Neuroscience, 2018; http://dx.doi.org/10.1038/s41593-018-0223-0), which contains RNA-seq data of human post-mortem dopaminergic neurons in a relatively high number of brain-healthy individuals of a wide age range including some “young” individuals which is rare in post-mortem studies. Such data is unfortunately not available with long read sequencing or any other more appropriate approach yet. Limitations are evident, but all limitations will apply equally to both groups of individuals that we compare. We will add genome browser tracks of the differentially expressed elements. The general mappability profile of the full-length LINE-1 “UIDs” is shown in Suppl Fig 6A. We will color-highlight the specific elements in this graph and will add genome browser data for these elements in a revised version. 

We will not be able to document changes in protein levels in aged human dopaminergic neurons as we do not have access to this material. We have tried to obtain human substantia nigra tissues but were not able to get sufficient amounts to do laser-capture microdissection or FACS analyses, especially of young individuals. There are still important limitations to tissue availability, especially of regions of interest like the substantia nigra pars compacta affected in Parkinson’s disease.

(5) More information is needed on RNAseq of microdissections of dopaminergic neurons from 'healthy' postmortem samples of different ages. No further information on these samples is provided. I would suggest adding a table with the clinical information of these samples (especially age, sex, and cause of death). The authors should also discuss whether this experiment has sufficient power. The human ageing cohort seems very small to me. 

This is a re-analysis of a published dataset (Dong et al, Nat Neurosci, 2018; doi:10.1038/s41593-018-0223-0), available through dbgap (phs001556.v1.p1). In this original article, the criteria for inclusion as a brain-healthy control were as follows:

“…Subjects… were without clinicopathological diagnosis of a neurodegenerative disease meeting the following stringent inclusion and exclusion criteria. Inclusion criteria: (i) absence of clinical or neuropathological diagnosis of a neurodegenerative disease, for example, PD according to the UKPDBB criteria47, Alzheimer’s disease according to NIA-Reagan criteria48, or dementia with Lewy bodies by revised consensus criteria49; for the purpose of this analysis incidental Lewy body cases (not meeting clinicopathological diagnostic criteria for PD or other neurodegenerative disease) were accepted for inclusion; (ii) PMI ≤ 48 h; (iii) RIN50 ≥ 6.0 by Agilent Bioanalyzer (good RNA integrity); and (iv) visible ribosomal peaks on the electropherogram. Exclusion criteria were: (i) a primary intracerebral event as the cause of death; (2) brain tumor (except incidental meningiomas); (3) systemic disorders likely to cause chronic brain damage.”

We do not have access to the cause of death, but we will add available metadata to the manuscript.

We will perform a post-hoc power analysis and add the result to the revision. 

(6) The findings in this manuscript apply to both human and mouse brains. However, the landscape of the evolutionarily young L1 subfamilies between these two species is very different and should be part of the discussion. For example, the regulatory sequences that drive L1 expression are quite different in human and mouse L1s. This should be discussed. 

Indeed, they are very different. We will add this to the discussion.  

(7) On page 3 the authors write: "generally accepted that TE activation can be both, a cause and consequence of aging". This statement does not reflect the current state of the field. On the contrary, this is still an area of extensive investigation and many of the findings supporting this hypothesis need to be confirmed in independent studies. This statement should be revised to reflect this reality. 

We agree, this is overstated, we will change this sentence accordingly.  

Reviewer #2 (Public Review):

Summary: 

Bonnifet et al. sought to characterize the expression pattern of L1 ORF1p expression across the entire mouse brain, in young and aged animals, and to corroborate their characterization with Western blotting for L1 ORF1p and L1 RNA expression data from human samples. They also queried L1 ORF1p interacting partners in the mouse brain by IP-MS. 

Strengths: 

A major strength of the study is the use of two approaches: a deep-learning detection method to distinguish neuronal vs. non-neuronal cells and ORF1p+ cells vs. ORF1p- cells across large-scale images encompassing multiple brain regions mapped by comparison to the Allen Brain Atlas, and confocal imaging to give higher resolution on specific brain regions. These results are also corroborated by Western blotting on six mouse brain regions. Extension of their analysis to post-mortem human samples, to the extent possible, is another strength of the paper. The identification of novel ORF1p interactors in the brain is also a strength in that it provides a novel dataset for future studies. 

Thank you for highlighting the strength of our study. 

Weaknesses: 

The main weakness of the study is that cell type specificity of ORF1p expression was not examined beyond neuron (NeuN+) vs non-neuron (NeuN-). Indeed, a recent study (Bodea et al. 2024, Nature Neuroscience) found that ORF1p expression is characteristic of parvalbumin-positive interneurons, and it would be very interesting to query whether other neuronal subtypes in different brain regions are distinguished by ORF1p expression. 

We agree that this point is important to address. We do provide indications for this in the manuscript. For instance, we detect staining in mouse and human Purkinje cells of the cerebellum in accordance with data from Takahashi et al, Neuron, 2022; DOI: 10.1016/j.neuron.2022.08.011. We also know from previous work, that in the mouse ventral midbrain, dopaminergic neurons (TH+/NeuN+) express ORF1p and that these neurons express higher levels of ORF1p than adjacent non-dopaminergic neurons (TH-/NeuN+; Blaudin de Thé et al, EMBO J, 2018). Others have shown evidence of full-length L1 RNA expression in both excitatory and inhibitory neurons but much less expression in non-neuronal cells (Garza et al, SciAdv, 2023). In sum, although this has not been investigated systematically brain-wide, it does not seem as if ORF1p expression is restricted to PV cells overall. We will deepen the discussion of this aspect in the revised manuscript. To address this question experimentally, we will try to perform ORF1p stainings on different brain regions together with PV stainings and add this data to a revised version, if possible.  

The data suggesting that ORF1p expression is increased in aged mouse brains is intriguing, although it seems to be based upon modestly (up to 27%, dependent on brain region) higher intensity of ORF1p staining rather than a higher proportion of ORF1+ neurons. Indeed, the proportion of NeuN+/Orf1p+ cells actually decreased in aged animals. It is difficult to interpret the significance and validity of the increase in intensity, as Hoechst staining of DNA, rather than immunostaining for a protein known to be stably expressed in young and aged neurons, was used as a control for staining intensity. 

It would have been indeed interesting to have another marker than DNA as a control. However, this requires a protein that is indeed stably expressed throughout the brain and throughout age. We are not aware of a protein for which this has been established. DNA staining with Hoechst does control for technical artefacts. We have whole-brain imaging data for the protein Rbfox3 (NeuN) which we used as a marker for cell identity. If this protein turns out to be stable, we could add this data to a revised version. 

The main weakness of the IP-MS portion of the study is that none of the interactors were individually validated or subjected to follow-up analyses. The list of interactors was compared to previously published datasets, but not to ORF1p interactors in any other mouse tissue. 

As stated in the manuscript, the list of previously published datasets does include a mouse dataset with ORF1p interacting proteins in mouse spermatocytes (please see line 434-435: “ORF1p interactors found in mouse spermatocytes were also present in our analysis including CNOT10, CNOT11, PRKRA and FXR2 among others (Suppl_Table4).”) -> De Luca, C., Gupta, A. & Bortvin, A. Retrotransposon LINE-1 bodies in the cytoplasm of piRNA-deficient mouse spermatocytes: Ribonucleoproteins overcoming the integrated stress response. PLoS Genet 19, e1010797 (2023)). We indeed did not validate any interactors for several reasons (economic reasons and time constraints (post-doc leaving)). However, we feel that the significant overlap with previously published interactors highlights the validity of our data and we anticipate that this list of ORF1p protein interactors in the mouse brain will be of further use for the community.  

The authors achieved the goals of broadly characterizing ORF1p expression across different regions of the mouse brain, and identifying putative ORF1p interactors in the mouse brain. However, findings from both parts of the study are somewhat superficial in depth. 

This provides a useful dataset to the field, which likely will be used to justify and support numerous future studies into L1 activity in the aging mammalian brain and in neurodegenerative disease. Similarly, the list of ORF1p interacting proteins in the brain will likely be taken up and studied in greater depth. 

Reviewer #3 (Public Review):

The question about whether L1 exhibits normal/homeostatic expression in the brain (and in general) is interesting and important. L1 is thought to be repressed in most somatic cells (with the exception of some stem/progenitor compartments). However, to our knowledge, this has not been authoritatively / systematically examined and the literature is still developing with respect to this topic. The full gamut of biological and pathobiological roles of L1 remains to be shown and elucidated and this area has garnered rapidly increasing interest, year-by-year. With respect to the brain, L1 (and repeat sequences in general) have been linked with neurodegeneration, and this is thought to be an aging-related consequence or contributor (or both) of inflammation. This study provides an impressive and apparently comprehensive imaging analysis of differential L1 ORF1p expression in mouse brain (with some supporting analysis of the human brain), compatible with a narrative of non-pathological expression of retrotransposition-competent L1 sequences. We believe this will encourage and support further research into the functional roles of L1 in normal brain function and how this may give way to pathological consequences in concert with aging. However, we have concerns with conclusions drawn, in some cases regardless of the lack of statistical support from the data. We note a lack of clarity about how the 3rd party pre-trained machine learning models perform on the authors' imaging data (validation/monitoring tests are not reported), as well as issues (among others) with the particular implementation of co-immunoprecipitation (ORF1p is not among the highly enriched proteins and apparently does not reach statistical significance for the comparison) - neither of which may be sufficiently rigorous.  

Thank you for your comments on our manuscript. 

In a revised version and a more in-depth response, we will address the concerns about the machine learning paradigm. Concerning the co-IP-MS, we can confirm that ORF1p is among the highly enriched proteins as it was not found in the IgG control (in 5 independent samples), only in the ORF1p-IP (in 5 out of 5 independent samples). This is what the infinite sign in Suppl Table 2 indicates and this is why there is no p-value assigned as infinite/0 doesn’t allow to calculate a p-value. We will make this clearer in a revised version of the manuscript.

eLife assessment

This study provides important computational insights into the dynamics of PROTAC-induced degradation complexes. The findings are solid and hold significant implications for advancing cancer treatments, particularly for breast and prostate cancers. However, the major conclusions of the work could be strengthened with a more thorough analysis. This work will be of broad interest to both biochemists and biophysicists.

Reviewer #1 (Public review):

This study by Wu et al. provides valuable computational insights into PROTAC-related protein complexes, focusing on linker roles, protein-protein interaction stability, and lysine residue accessibility. The findings are significant for PROTAC development in cancer treatment, particularly breast and prostate cancers.

The authors' claims about the role of PROTAC linkers and protein-protein interaction stability are generally supported by their computational data. However, the conclusions regarding lysine accessibility could be strengthened with more in-depth analysis. The use of the term "protein functional dynamics" is not fully justified by the presented work, which focuses primarily on structural dynamics rather than functional aspects.

Strengths:

(1) Comprehensive computational analysis of PROTAC-related protein complexes.

(2) Focus on critical aspects: linker role, protein-protein interaction stability, and lysine accessibility.

Weaknesses:

(1) Limited examination of lysine accessibility despite its stated importance.

(2) Use of RMSD as the primary metric for conformational assessment, which may overlook important local structural changes.

Reviewer #2 (Public review):

Summary:

The manuscript reports the computational study of the dynamics of PROTAC-induced degradation complexes. The research investigates how different linkers within PROTACs affect the formation and stability of ternary complexes between the target protein BRD4BD1 and Cereblon E3 ligase, and the degradation machinery. Using computational modeling, docking, and molecular dynamics simulations, the study demonstrates that although all PROTACs form ternary complexes, the linkers significantly influence the dynamics and efficacy of protein degradation. The findings highlight that the flexibility and positioning of Lys residues are crucial for successful ubiquitination. The results also discussed the correlated motions between the PROTAC linker and the complex.

Strengths:

The field of PROTAC discovery and design, characterized by its limited research, distinguishes itself from traditional binary ligand-protein interactions by forming a ternary complex involving two proteins. The current understanding of how the structure of PROTAC influences its degradation efficacy remains insufficient. This study investigated the atomic-level dynamics of the degradation complex, offering potentially valuable insights for future research into PROTAC degradability.

Reviewer #3 (Public review):

The authors offer an interesting computational study on the dynamics of PROTAC-driven protein degradation. They employed a combination of protein-protein docking, structural alignment, atomistic MD simulations, and post-analysis to model a series of CRBN-dBET-BRD4 ternary complexes, as well as the entire degradation machinery complex. These degraders, with different linker properties, were all capable of forming stable ternary complexes but had been shown experimentally to exhibit different degradation capabilities. While in the initial models of the degradation machinery complex, no surface Lys residue(s) of BRD4 were exposed sufficiently for the crucial ubiquitination step, MD simulations illustrated protein functional dynamics of the entire complex and local side-chain arrangements to bring Lys residue(s) to the catalytic pocket of E2/Ub for reactions. Using these simulations, the authors were able to present a hypothesis as to how linker property affects degradation potency. They were able to roughly correlate the distance of Lys residues to the catalytic pocket of E2/Ub with observed DC50/5h values. This is an interesting and timely study that presents interesting tools that could be used to guide future PROTAC design or optimization.

Author response:

Thank you for the reviewers’ thoughtful comments and suggestions! We greatly appreciate the feedback and are committed to address all the points raised by the reviewers to strengthen our manuscript.

We plan to conduct additional local structural analyses to better demonstrate our observations of PROTAC-induced LYS-GLY interactions and lysine associability. Specifically, we will add more in-depth analysis such as computing dihedral entropies and Root Mean Square Fluctuation (RMSF) of nearby side chains and integrating various structural alignments to provide better visualization and understanding of the local structural arrangements. We plan to extend and add simulations when needed. We will review the latest available crystal and cryo-EM structures. If new structures are available, we will incorporate them into our revised analysis and discussion.

In the revision, additional figures will be included to offer a more comprehensive assessment of local conformational changes. We will also ensure that explanations of technical terminology are clear to non-expert readers and will address the grammatical and terminology errors highlighted by the reviewers. We will refine our language to more accurately describe the focus on structural dynamics in our study.

eLife assessment

This study presents a valuable finding on the role of the Inferior Colliculus in sensory prediction, cognitive decision-making, and reward prediction. The evidence supporting the claims of the authors is solid. The work will be of interest to neurobiologists working on auditory processing.

Reviewer #1 (Public review):

Summary:

This work made a lot of efforts to explore the multifaceted roles of the inferior colliculus (IC) in auditory processing, extending beyond traditional sensory encoding. The authors recorded neuronal activitity from the IC at single unit level when monkeys were passively exposed or actively engaged in behavioral task. They concluded that 1）IC neurons showed sustained firing patterns related to sound duration, indicating their roles in temporal perception, 2) IC neuronal firing rates increased as sound sequences progress, reflecting modulation by behavioral context rather than reward anticipation, 3) IC neurons encode reward prediction error and their capability of adjusting responses based on reward predictability, 4) IC neural activity correlates with decision-making. In summary, this study tried to provide a new perspective on IC functions by exploring its roles in sensory prediction and reward processing, which are not traditionally associated with this structure.

Strengths:

The major strength of this work is that the authors performed electrophysiological recordings from the IC of behaving monkeys. Compared with the auditory cortex and thalamus, the IC in monkeys has not been adequately explored.

Weaknesses:

(1) The authors cited several papers focusing on dopaminergic inputs in the IC to suggest the involvement of this brain region in cognitive functions. However, all those cited work were done in rodents. Whether monkey's IC shares similar inputs is not clear.<br /> (2) The authors confused the two terms, novelty and deviation. According to their behavioral paradigm, deviation rather than novelty should be used in the paper because all the stimuli have been presented to the monkeys during training. Therefore, there is actually no novel stimuli but only deviant stimuli. This reflects that the author has misunderstood the basic concept.<br /> (3) Most of the conclusions were made based on correlational analysis or speculation without providing causal evidences.<br /> (4) Results are presented in a very "straightforward" manner with too many detailed descriptions of phenomena but lack of summary and information synthesis. For example, the first section of Results is very long but did not convey clear information.<br /> (5) The logic between different sections of Results is not clear.<br /> (6) In the Discussion, there is excessive repetition of results, and further comparison with and discussion of potentially related work are very insufficient. For example, Metzger, R.R., et al. (J Neurosc, 2006) have shown similar firing patterns of IC neurons and correlated their findings with reward.

Reviewer #2 (Public review):

Summary:

The inferior colliculus (IC) has been explored for its possible functions in behavioral tasks and has been suggested to play more important roles rather than simple sensory transmission. The authors revealed the climbing effect of neurons in IC during decision-making tasks, and tried to explore the reward effect in this condition.

Strengths:

Complex cognitive behaviors can be regarded as simple ideals of generating output based on information input, which depends on all kinds of input from sensory systems. The auditory system has hierarchic structures no less complex than those areas in charge of complex functions. Meanwhile, IC receives projections from higher areas, such as auditory cortex, which implies IC is involved in complex behaviors. Experiments in behavioral monkeys are always time-consuming works with hardship, and this will offer more approximate knowledge of how the human brain works.

Weaknesses:

These findings are more about correlation but not causality of IC function in behaviors. And I have a few major concerns.

Comparing neurons' spike activities in different tests, a 'climbing effect' was found in the oddball paradigm. The effect is clearly related to training and learning process, but it still requires more exploration to rule out a few explanations. First, repeated white noise bursts with fixed inter-stimulus-interval of 0.6 seconds was presented, so that monkeys might remember the sounds by rhymes, which is some sort of learned auditory response. It is interesting to know monkeys' responses and neurons' activities if the inter-stimuli-interval is variable. Second, the task only asked monkeys to press one button and the reward ratio (the ratio of correct response trials) was around 78% (based on the number from Line 302). so that, in the sessions with reward, monkeys had highly expected reward chances, does this expectation cause the climbing effect?

"Reward effect" on IC neurons' responses were showed in Fig. 4. Is this auditory response caused by physical reward action or not? In reward sessions, IC neurons have obvious response related to the onset of water reward. The electromagnetic valve is often used in water-rewarding system and will give out a loud click sound every time when the reward is triggered. IC neurons' responses may be simply caused by the click sound if the electromagnetic valve is used. It is important to find a way to rule out this simple possibility.

Reviewer #3 (Public review):

Summary:

The authors aimed to investigate the multifaceted roles of the Inferior Colliculus (IC) in auditory and cognitive processes in monkeys. Through extracellular recordings during a sound duration-based novelty detection task, the authors observed a "climbing effect" in neuronal firing rates, suggesting an enhanced response during sensory prediction. Observations of reward prediction errors within the IC further highlight its complex integration in both auditory and reward processing. Additionally, the study indicated IC neuronal activities could be involved in decision-making processes.

Strengths:

This study has the potential to significantly impact the field by challenging the traditional view of the IC as merely an auditory relay station and proposing a more integrative role in cognitive processing. The results provide valuable insights into the complex roles of the IC, particularly in sensory and cognitive integration, and could inspire further research into the cognitive functions of the IC.

Weaknesses:

Major Comments:

(1) Structural Clarity and Logic Flow:<br /> The manuscript investigates three intriguing functions of IC neurons: sensory prediction, reward prediction, and cognitive decision-making, each of which is a compelling topic. However, the logical flow of the manuscript is not clearly presented and needs to be well recognized. For instance, Figure 3 should be merged into Figure 2 to present population responses to the order of sounds, thereby focusing on sensory prediction. Given the current arrangement of results and figures, the title could be more aptly phrased as "Beyond Auditory Relay: Dissecting the Inferior Colliculus's Role in Sensory Prediction, Reward Prediction, and Cognitive Decision-Making."

(2) Clarification of Data Analysis:<br /> Key information regarding data analysis is dispersed throughout the results section, which can lead to confusion. Providing a more detailed and cohesive explanation of the experimental design would significantly enhance the interpretation of the findings. For instance, including a detailed timeline and reward information for the behavioral paradigms shown in Figures 1C and D would offer crucial context for the study. More importantly, clearly presenting the analysis temporal windows and providing comprehensive statistical analysis details would greatly improve reader comprehension.

(3) Reward Prediction Analysis:<br /> The conclusion regarding the IC's role in reward prediction is underdeveloped. While the manuscript presents evidence that IC neurons can encode reward prediction, this is only demonstrated with two example neurons in Figure 6. A more comprehensive analysis of the relationship between IC neuronal activity and reward prediction is necessary. Providing population-level data would significantly strengthen the findings concerning the IC's complex functionalities. Additionally, the discussion of reward prediction in lines 437-445, which describes IC neuron responses in control experiments, does not sufficiently demonstrate that IC neurons can encode reward expectations. It would be valuable to include the responses of IC neurons during trials with incorrect key presses or no key presses to better illustrate this point.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

This work made a lot of efforts to explore the multifaceted roles of the inferior colliculus (IC) in auditory processing, extending beyond traditional sensory encoding. The authors recorded neuronal activitity from the IC at single unit level when monkeys were passively exposed or actively engaged in behavioral task. They concluded that 1）IC neurons showed sustained firing patterns related to sound duration, indicating their roles in temporal perception, 2) IC neuronal firing rates increased as sound sequences progress, reflecting modulation by behavioral context rather than reward anticipation, 3) IC neurons encode reward prediction error and their capability of adjusting responses based on reward predictability, 4) IC neural activity correlates with decision-making. In summary, this study tried to provide a new perspective on IC functions by exploring its roles in sensory prediction and reward processing, which are not traditionally associated with this structure.

Strengths:

The major strength of this work is that the authors performed electrophysiological recordings from the IC of behaving monkeys. Compared with the auditory cortex and thalamus, the IC in monkeys has not been adequately explored.

We appreciate the reviewer’s acknowledgment of the efforts and strengths of our study. Indeed, our goal was to provide a comprehensive exploration of the multifaceted roles of the inferior colliculus (IC) in auditory processing and beyond, particularly in sensory prediction and reward processing. The use of electrophysiological recordings in behaving monkeys was central to our approach, as we sought to uncover the underexplored aspects of IC function in these complex cognitive domains. We are pleased that the reviewer recognizes the value of investigating the IC, a structure that has not been adequately explored in primates compared to other auditory regions like the cortex and thalamus. This feedback reinforces our belief that our work contributes significantly to advancing the understanding of the IC's roles in cognitive processing.

We look forward to addressing any further points the reviewers may have and refining our manuscript accordingly. Thank you for your constructive feedback and for recognizing the strengths of our research approach.

Weaknesses:

(1) The authors cited several papers focusing on dopaminergic inputs in the IC to suggest the involvement of this brain region in cognitive functions. However, all those cited work were done in rodents. Whether monkey's IC shares similar inputs is not clear.

We appreciate the reviewer's insightful comment on the limitations of extrapolating findings from rodent models to monkeys, particularly concerning dopaminergic inputs to the Inferior Colliculus (IC). While it is true that most studies on dopaminergic inputs to the IC have been conducted in rodents, to our knowledge, no studies have been conducted specifically in primates. To address the reviewer's concern, we have added a statement in both the introduction and discussion sections of our manuscript:

- Introduction: " However, these studies were conducted in rodents, and the existence and role of dopaminergic inputs in the primate IC remain underexplored."

- Discussion: " However, the exact mechanisms and functions of dopamine modulation in the inferior colliculus are still not fully understood, particularly in primates. "

(2) The authors confused the two terms, novelty and deviation. According to their behavioral paradigm, deviation rather than novelty should be used in the paper because all the stimuli have been presented to the monkeys during training. Therefore, there is actually no novel stimuli but only deviant stimuli. This reflects that the author has misunderstood the basic concept.

We appreciate the reviewer's clarification regarding the distinction between "novelty" and "deviation" in the context of our behavioral paradigm. We agree that, given the nature of our experimental design where all stimuli were familiar to the monkeys during training, the term "deviation" more accurately describes the stimuli used in our study rather than "novelty."

To address this, we have revised the manuscript to replace the term "novelty" with "deviation" wherever applicable. This change has been made to ensure accurate terminology is used throughout the paper, thereby eliminating any potential misunderstanding of the concepts involved in our study.

We thank the reviewer for pointing out this important distinction, which has improved the clarity and precision of our manuscript.

(3) Most of the conclusions were made based on correlational analysis or speculation without providing causal evidences.

We appreciate the reviewer’s concern regarding the reliance on correlational analyses in our study. Indeed, we acknowledge that the conclusions drawn primarily reflect correlations between neuronal activity and behavioral outcomes, rather than direct causal evidence. This limitation is inherent to many electrophysiological studies, particularly those conducted in behaving primates, where direct manipulation of specific neural circuits to establish causality is often challenging.

This limitation becomes even more complex when considering the IC’s role as a key lower-level relay station in the auditory pathway. Manipulating IC activity could potentially affect auditory responses in downstream pathways, which, in turn, may influence sensory prediction and decision-making processes. Moreover, we hypothesize that the sensory prediction and reward signals observed in the IC may not have direct causal effects but may instead be driven by top-down projections from higher cognitive regions. However, it is important to emphasize that our study provides novel evidence that the IC may exhibit multiple facets of cognitive signaling, which could inspire future research into the underlying mechanisms and broader functional implications of these signals.

To address this, we have taken the following steps in our revised manuscript:

(1) Clarified the Scope of Conclusions: We have revised the language in the Results and Discussion sections to explicitly state that our findings represent correlational relationships rather than causal mechanisms. For example, we now refer to the associations observed between IC activity and behavioral outcomes as "correlational" and have refrained from making definitive causal claims without supporting experimental evidence.

(2) Proposed Future Directions: In the Discussion section, we have included suggestions for future studies to directly test the causality of the observed relationships. We acknowledge the need for further investigation to substantiate the causal links between IC activity and cognitive functions such as sensory prediction, decision-making, and reward processing.

We believe these revisions provide a more balanced interpretation of our findings while emphasizing the importance of future research to build on our results and establish causal relationships. Thank you for raising this critical point, which has led to a more rigorous and transparent presentation of our study.

(4) Results are presented in a very "straightforward" manner with too many detailed descriptions of phenomena but lack of summary and information synthesis. For example, the first section of Results is very long but did not convey clear information.

We appreciate the reviewer’s feedback regarding the presentation of our results. We understand that the detailed descriptions of phenomena may have made it difficult to discern the key findings and overarching themes in the study. We recognize the importance of balancing detailed reporting with clear summaries and synthesis to effectively communicate our findings.

To address this concern, we have made the following revisions to the manuscript:

(1) Condensed and Synthesized Key Findings: We have streamlined the presentation of the Results section by condensing overly detailed descriptions and focusing on the most critical aspects of the data. Key findings are now summarized at the end of each subsection to ensure that the main points are clearly conveyed.

(2) Enhanced Section Summaries: We have added summary statements at the end of each major results section to synthesize the findings and highlight their significance. This should help guide the reader through the narrative and emphasize the key takeaways from each part of the study.

(3) Improved Flow and Clarity: We have revised the structure and organization of the Results section to improve the flow of information. By rearranging certain paragraphs and refining the language, we aim to present the results in a more cohesive and coherent manner.

We believe these changes will make the Results section more accessible and informative, allowing readers to more easily grasp the significance of our findings. Thank you for your valuable suggestion, which has significantly improved the clarity and impact of our manuscript.

(5) The logic between different sections of Results is not clear.

We appreciate the reviewer’s observation regarding the lack of clear logical connections between different sections of the Results. We acknowledge that a coherent flow is essential for effectively communicating the progression of findings and their implications.

To address this concern, we have made the following revisions:

(1) Enhanced Transitions Between Sections: We have introduced clearer transitional statements between sections of the Results. These transitions explicitly state how each new section builds upon or relates to the previous findings, creating a more cohesive narrative.

(2) Integration of Findings: In several places within the Results, we have added brief synthesis paragraphs that integrate findings across sections. These integrative summaries help to tie together the different aspects of our study, demonstrating how they collectively contribute to our understanding of the Inferior Colliculus’s (IC) role in sensory prediction, decision-making, and reward processing.

(3) Clarified Rationale: At the beginning of each major section, we have clarified the rationale behind why certain experiments were conducted, connecting them more clearly to the overarching goals of the study. This should help the reader understand the purpose of each set of results in the context of the broader research objectives.

We believe these changes improve the overall coherence and readability of the Results section, allowing readers to better follow the logical progression of our study. We are grateful for this constructive feedback and believe it has significantly enhanced the manuscript.

(6) In the Discussion, there is excessive repetition of results, and further comparison with and discussion of potentially related work are very insufficient. For example, Metzger, R.R., et al. (J Neurosc, 2006) have shown similar firing patterns of IC neurons and correlated their findings with reward.

We appreciate the reviewer's insightful critique regarding the excessive repetition in the Discussion and the lack of sufficient comparison with related work. We acknowledge that a well-balanced Discussion should not only interpret findings but also place them in the context of existing literature to highlight the novelty and significance of the study.

To address these concerns, we have made the following revisions:

(1) Reduction of Repetition: We have carefully revised the Discussion to minimize redundant repetition of the Results. Instead of restating the findings, we now focus more on their implications, limitations, and how they advance the current understanding of the Inferior Colliculus (IC) and its broader cognitive roles.

(2) Incorporation of Related Work: We have expanded the Discussion to include a more comprehensive comparison with existing literature, specifically highlighting studies that have reported similar findings. For example, we now discuss the work by Metzger et al. (2006), which demonstrated similar firing patterns of IC neurons and correlated these with reward-related processes. This comparison helps contextualize our results and emphasizes the novel contributions our study makes to the field.

We believe these revisions have significantly improved the quality of the Discussion by reducing unnecessary repetition and providing a more thorough engagement with the relevant literature. We are grateful for the reviewer's valuable feedback, which has helped us refine and strengthen the manuscript.

Reviewer #2 (Public review):

Summary:

The inferior colliculus (IC) has been explored for its possible functions in behavioral tasks and has been suggested to play more important roles rather than simple sensory transmission. The authors revealed the climbing effect of neurons in IC during decision-making tasks, and tried to explore the reward effect in this condition.

Strengths:

Complex cognitive behaviors can be regarded as simple ideals of generating output based on information input, which depends on all kinds of input from sensory systems. The auditory system has hierarchic structures no less complex than those areas in charge of complex functions. Meanwhile, IC receives projections from higher areas, such as auditory cortex, which implies IC is involved in complex behaviors. Experiments in behavioral monkeys are always time-consuming works with hardship, and this will offer more approximate knowledge of how the human brain works.

We greatly appreciate the reviewer's positive summary of our work and recognition of the effort involved in conducting experiments on behaving monkeys. We agree with the reviewer that the inferior colliculus (IC) plays a significant role beyond mere sensory transmission, particularly in integrating sensory inputs with higher cognitive functions. Our study aims to shed light on these complex functions by revealing the climbing effect of IC neurons during decision-making tasks and exploring how reward influences this dynamic.

We are encouraged that the reviewer acknowledges the importance of investigating the IC's role within the broader framework of complex cognitive behaviors and appreciates the hierarchical nature of the auditory system. The reviewer's comments reinforce the value of our research in contributing to a more nuanced understanding of how the IC might contribute to sensory-cognitive integration.

We thank the reviewer for highlighting the significance of using behavioral monkey models to approximate human brain function. We are hopeful that our findings will serve as a stepping stone for further research exploring the multifaceted roles of the IC in cognition and behavior.

We will now proceed to address the specific concerns and suggestions provided by the reviewer in the following sections.

Weaknesses:

These findings are more about correlation but not causality of IC function in behaviors. And I have a few major concerns.

We appreciate the reviewer’s concern regarding the reliance on correlational analyses in our study. We acknowledge the importance of distinguishing between correlation and causality. As detailed in our response to Question 3 from Reviewer #1, we recognize the limitations of relying on correlational data and the challenges of establishing direct causal links in electrophysiological studies involving behaving primates.

We have taken steps to clarify this distinction throughout our manuscript. Specifically, we have revised the Results and Discussion sections to ensure that the findings are presented as correlational, not causal, and we have proposed future studies utilizing more direct manipulation techniques to assess causality. We hope these revisions adequately address your concerns.

Comparing neurons' spike activities in different tests, a 'climbing effect' was found in the oddball paradigm. The effect is clearly related to training and learning process, but it still requires more exploration to rule out a few explanations. First, repeated white noise bursts with fixed inter-stimulus-interval of 0.6 seconds was presented, so that monkeys might remember the sounds by rhymes, which is some sort of learned auditory response. It is interesting to know monkeys' responses and neurons' activities if the inter-stimuli-interval is variable. Second, the task only asked monkeys to press one button and the reward ratio (the ratio of correct response trials) was around 78% (based on the number from Line 302). so that, in the sessions with reward, monkeys had highly expected reward chances, does this expectation cause the climbing effect?

We thank the reviewer for raising these insightful points regarding the 'climbing effect' observed in the oddball paradigm and its potential relationship with training, learning processes, and reward expectation. Below, we address each of the reviewer's specific concerns:

(1) Inter-Stimulus Interval (ISI) and Rhythmic Auditory Response:

The reviewer suggests that the fixed inter-stimulus interval (ISI) of 0.6 seconds might lead to a rhythmic auditory response, where monkeys could anticipate the sounds. We appreciate this perspective. However, we believe that rhythm is unlikely to play a significant role in the 'climbing effect' for the following reason: The 'climbing effect' starts from the second sound in the block (Fig.2D and Fig.3B), before any rhythm or pattern could be fully established, as a rhythm generally requires at least three repetitions to form. Unfortunately, we did not explore variable ISIs in the current study, so we cannot directly address this concern with the data at hand.

(2) Reward Expectation and Climbing Effect:

The reviewer raises an important concern about whether the 'climbing effect' could be influenced by the monkeys' high reward expectation, especially given the high reward ratio (~78%) in the sessions. While it is plausible that reward expectation could contribute to the observed increase in neuronal firing rates, we believe the results from our reward experiment (Fig. 4) suggest otherwise. In this experiment, even though reward expectation was likely formed due to the consistent pairing of sounds with rewards (100%), we did not observe a climbing effect in the auditory response. The presence of reward prediction error (Fig. 4D) further suggests that while the monkeys may form reward expectations, these expectations do not directly drive the climbing effect.

To clarify this point, we have added sentences in the revised manuscript to explicitly discuss the relationship between reward expectation and the climbing effect, emphasizing that our findings indicate the climbing effect is not primarily due to reward expectation.

We believe these revisions provide a clearer understanding of the factors contributing to the climbing effect and address the reviewer's concerns effectively. Thank you for these valuable suggestions.

"Reward effect" on IC neurons' responses were showed in Fig. 4. Is this auditory response caused by physical reward action or not? In reward sessions, IC neurons have obvious response related to the onset of water reward. The electromagnetic valve is often used in water-rewarding system and will give out a loud click sound every time when the reward is triggered. IC neurons' responses may be simply caused by the click sound if the electromagnetic valve is used. It is important to find a way to rule out this simple possibility.

We appreciate the reviewer’s concern regarding the potential confounding factor introduced by the electromagnetic valve’s click sound during water reward delivery, which could be misinterpreted as an auditory response rather than a response to the reward itself. Anticipating this possibility, we took measures to eliminate it by placing the electromagnetic valve outside the soundproof room where the neuronal recordings were performed.

To address your concern more explicitly, we have added sentences in the Methods section of the revised manuscript detailing this setup, ensuring that readers are aware of the steps we took to eliminate this potential confound. By doing so, we believe that the observed reward-related neural activity in the IC is attributable to the reward processing itself rather than an auditory response to the valve click. We appreciate you bringing this important aspect to our attention, and we hope our clarification strengthens the interpretation of our findings.

Reviewer #3 (Public review):

Summary:

The authors aimed to investigate the multifaceted roles of the Inferior Colliculus (IC) in auditory and cognitive processes in monkeys. Through extracellular recordings during a sound duration-based novelty detection task, the authors observed a "climbing effect" in neuronal firing rates, suggesting an enhanced response during sensory prediction. Observations of reward prediction errors within the IC further highlight its complex integration in both auditory and reward processing. Additionally, the study indicated IC neuronal activities could be involved in decision-making processes.

Strengths:

This study has the potential to significantly impact the field by challenging the traditional view of the IC as merely an auditory relay station and proposing a more integrative role in cognitive processing. The results provide valuable insights into the complex roles of the IC, particularly in sensory and cognitive integration, and could inspire further research into the cognitive functions of the IC.

We appreciate the reviewer’s positive summary of our work and recognition of its potential impact on the field. We are pleased that the reviewer acknowledges the significance of our findings in challenging the traditional view of the Inferior Colliculus (IC) as merely an auditory relay station and in proposing its integrative role in cognitive processing.

Our study indeed aims to provide new insights into the multifaceted roles of the IC, particularly in the context of sensory and cognitive integration. We believe that this research could pave the way for future studies that further explore the cognitive functions of the IC and its involvement in complex behavioral processes.

We are encouraged by the reviewer’s positive assessment and are committed to continuing to refine our work in response to the constructive feedback provided. We hope that our findings will contribute to advancing the understanding of the IC’s role in the broader context of neuroscience.

We will now proceed to address the specific concerns and suggestions provided by the reviewer in the following sections.

Weaknesses:

Major Comments:

(1) Structural Clarity and Logic Flow:

The manuscript investigates three intriguing functions of IC neurons: sensory prediction, reward prediction, and cognitive decision-making, each of which is a compelling topic. However, the logical flow of the manuscript is not clearly presented and needs to be well recognized. For instance, Figure 3 should be merged into Figure 2 to present population responses to the order of sounds, thereby focusing on sensory prediction. Given the current arrangement of results and figures, the title could be more aptly phrased as "Beyond Auditory Relay: Dissecting the Inferior Colliculus's Role in Sensory Prediction, Reward Prediction, and Cognitive Decision-Making."

We appreciate the reviewer’s detailed feedback on the structural clarity and logical flow of the manuscript. We understand the importance of presenting our findings in a clear and cohesive manner, especially when addressing multiple complex topics such as sensory prediction, reward prediction, and cognitive decision-making.

To address the reviewer's concerns, we have made the following revisions:

(1) Reorganization of Figures and Results:

We agree with the suggestion to merge Figure 3 into Figure 2. By doing so, we can present the population responses to the order of sounds more effectively, thereby streamlining the focus on sensory prediction. This will allow readers to more easily follow the progression of the results related to this key function of the IC.

We have reorganized the Results section to ensure a smoother transition between the different aspects of IC function that we are investigating. The new structure will better guide the reader through the narrative, aligning with the themes of sensory prediction, reward prediction, and cognitive decision-making.

(2) Revised Title:

In line with the reviewer's suggestion, we have revised the title to "Beyond Auditory Relay: Dissecting the Inferior Colliculus's Role in Sensory Prediction, Reward Prediction, and Cognitive Decision-Making." We believe this title more accurately reflects the scope and focus of our study, as it highlights the three core functions of the IC that we are investigating.

(3) Improved Logic Flow:

We have added introductory statements at the beginning of each section within the Results to clarify the rationale behind the experiments and the logical connections between them. This should help to improve the overall flow of the manuscript and make the progression of our findings more intuitive for readers.

We believe these changes significantly enhance the clarity and logical structure of the manuscript, making it easier for readers to understand the sequence and importance of our findings. Thank you for your valuable suggestion, which has led to a more coherent and focused presentation of our work.

(2) Clarification of Data Analysis:

Key information regarding data analysis is dispersed throughout the results section, which can lead to confusion. Providing a more detailed and cohesive explanation of the experimental design would significantly enhance the interpretation of the findings. For instance, including a detailed timeline and reward information for the behavioral paradigms shown in Figures 1C and D would offer crucial context for the study. More importantly, clearly presenting the analysis temporal windows and providing comprehensive statistical analysis details would greatly improve reader comprehension.

We appreciate the reviewer’s insightful comment regarding the need for clearer and more cohesive explanations of the data analysis and experimental design. We recognize that a well-structured presentation of this information is essential for the reader to fully understand and interpret our findings. To address this, we have made the following revisions:

(1) Detailed Explanation of Experimental Design:

We have included a more detailed explanation of the experimental design, particularly for the behavioral paradigms shown in Figures 1C and 1D. This includes a comprehensive timeline of the experiments, along with explicit information about the reward structure and timing. By providing this context upfront, we aim to give readers a clearer understanding of the conditions under which the neuronal recordings were obtained.

(2) Cohesive Presentation of Data Analysis:

Key information regarding data analysis, which was previously dispersed throughout the Results section, has been consolidated and moved to a dedicated subsection within the Methods. This subsection now provides a step-by-step description of the analysis process, including the temporal windows used for examining neuronal activity, as well as the specific statistical methods employed.

We have also ensured that the temporal windows used for different analyses (e.g., onset window, late window, etc.) are clearly defined and consistently referenced throughout the manuscript. This will help readers track the use of these windows across different figures and analyses.

(3) Enhanced Statistical Analysis Details:

We have expanded the description of the statistical analyses performed in the study, including the rationale behind the choice of tests, the criteria for significance, and any corrections for multiple comparisons. These details are now presented in a clear and accessible format within the Methods section, with relevant information also highlighted in the Result section or the figure legends to facilitate understanding.

We believe these changes will significantly improve the clarity and comprehensibility of the manuscript, allowing readers to better follow the experimental design, data analysis, and the conclusions drawn from our findings. Thank you for this valuable feedback, which has helped us to enhance the rigor and transparency of our presentation.

(3) Reward Prediction Analysis:

The conclusion regarding the IC's role in reward prediction is underdeveloped. While the manuscript presents evidence that IC neurons can encode reward prediction, this is only demonstrated with two example neurons in Figure 6. A more comprehensive analysis of the relationship between IC neuronal activity and reward prediction is necessary. Providing population-level data would significantly strengthen the findings concerning the IC's complex functionalities. Additionally, the discussion of reward prediction in lines 437-445, which describes IC neuron responses in control experiments, does not sufficiently demonstrate that IC neurons can encode reward expectations. It would be valuable to include the responses of IC neurons during trials with incorrect key presses or no key presses to better illustrate this point.

We deeply appreciate the detailed feedback provided regarding the conclusions on the inferior colliculus (IC)'s role in reward prediction within our manuscript. We acknowledge the importance of a robust and comprehensive presentation of our findings, particularly when discussing complex neural functionalities.

In response to the reviewers' concerns, we have made the following revisions to strengthen our manuscript:

(1) Inclusion of Population-Level Data for IC Neurons:

In the revised manuscript, we have included population-level results for IC neurons in a supplementary figure. Initially, we focused on two example neurons that did not exhibit motor-related responses to key presses to isolate reward-related signals. However, most IC neurons exhibit motor responses during key presses (as indicated in Fig.7), which can complicate distinguishing between reward-related activity and motor responses. This complexity is why we initially presented neurons without motor responses. To clarify this point, we have added sentences in the Results section to explain the rationale behind our selection of neurons and to address the potential overlap between motor and reward responses in the IC.

(2) Addition of Data on Key Press Errors and No-Response Trials:

In response to the reviewer’s suggestion, we have demonstrated Peri-Stimulus Time Histograms (PSTHs) for two example neurons during error trials as below, including incorrect key presses and no-response trials. Given that the monkeys performed the task with high accuracy, the number of error trials is relatively small, especially for the control condition (as shown in the top row of the figure). While we remain cautious in drawing definitive conclusions from this limited trials, we observed that no clear reward signals were detected during the corresponding window (typically centered around 150 ms after the end of the sound). It is important to note that the experiment was initially designed to explore decision-making signals in the IC, rather than focusing specifically on reward processing. However, the data in Fig. 6 demonstrated intriguing signals of reward prediction error, which is why we believe it is important to present them.

When combined with the results from our reward experiment (Fig. 5), we believe these findings provide compelling evidence of reward prediction errors being processed by IC neurons. Additionally, we observed that the reward prediction error in the IC appears to be signed, meaning that IC neurons showed robust responses to unexpected rewards but not to unexpected no-reward scenarios. However, the sign of the reward prediction error should be explored in greater depth with specifically designed experiments in future studies.

Author response image 1.

(A) PSTH of the neuron from Figure 6a during a key press trial under control condition. The number in the parentheses in the legend represents the number of trials for control condition. (B) PSTHs of the neuron from Figure 6a during non-key press trials under experimental conditions. The numbers in the parentheses in the legend represent the number of trials for experimental conditions. (C-D) Equivalent PSTHs as in A-B but from the neuron in Figure 6b.

We are grateful for the reviewer's insightful suggestions, which have allowed us to improve the depth and rigor of our analysis. We believe these revisions significantly enhance our manuscript's conclusions regarding the complex functionalities of IC.

eLife assessment

In this report, the authors present valuable findings identifying a novel worm-specific protein (sdg-1) that is induced upon loss of dsRNA import via SID-1, but is not required to mediate SID-1 RNA regulatory effects. The genetic and genomic approaches are well-executed. The existing data are solid, but the study would benefit from additional supporting evidence. The manuscript's central findings could also be refined to avoid overstating the results. These findings will be of interest to those working in the germline epigenetic inheritance field.

Reviewer #1 (Public review):

Summary:

In the manuscript "Intergenerational transport of double-stranded RNA limits heritable epigenetic changes" Shugarts and colleagues investigate intergenerational dsRNA transport in the nematode C. elegans. They induce oxidative damage in worms, blocking dsRNA import into cells (and potentially affecting the worms in other ways). Oxidative stress inhibits dsRNA import and the associated heritable regulation of gene expression in the adult germline (Fig. 2). The authors identify a novel gene, sid-1-dependent gene-1 (sdg-1), which is induced upon inhibition of SID-1 (Fig. 3). Both transient inhibition and genetic depletion of SID-1 lead to the upregulation of sdg-1 and a second gene, sdg-2 (Fig. 5). The expression of SDG-1 is variable, potentially indicating buffering regulation. While the expression of Sdg-1 could be consistent with a role in intergenerational transport of dsRNA, neither its overexpression nor loss-of-function impacts dsRNA-mediated silencing (Fig. 7) in the germline. It would be interesting to test if sdg-2 functions redundantly.

In summary, the authors have identified a novel worm-specific protein (sdg-1) that is induced upon loss of dsRNA import via SID-1, but is not required to mediate SID-1 RNA regulatory effects.

Remaining Questions:

• The authors use an experimental system that induces oxidative damage specifically in neurons to release dsRNAs into the circulation. Would the same effect be observed if oxidative damage were induced in other cell types?

• Besides dsRNA, which other RNAs and cellular products (macromolecules and small signalling molecules) are released into the circulation that could affect the observed changes in germ cells?

• SID-1 modifies RNA regulation within the germline (Fig. 7) and upregulates sdg-1 and sdg-2 (Fig. 5). However, SID-1's effects do not appear to be mediated via sdg-1. Testing the role of sdg-2 would be intriguing.

• Are sdg-1 or sdg-2 conserved in other nematodes or potentially in other species? Sdg-1 appears to be encoded or captured by a retro-element in the C. elegans genome and exhibits stochastic expression in different isolates. Is this a recent adaptation in the C. elegans genome, or is it present in other nematodes? Does loss-of-function of sdg-1 or sdg-2 have any observable effect?

Clarification for Readability:

To enhance readability and avoid misunderstandings, it is crucial to specify the model organism and its specific dsRNA pathways that are not conserved in vertebrates:

• In the first sentence of the paragraph "Here, we dissect the intergenerational transport of extracellular dsRNA ...", the authors should specify "in the nematode C. elegans". Unlike vertebrates, which recognise dsRNA as a foreign threat, worms and other invertebrates pervasively use dsRNA for signalling. Additionally, worms, unlike vertebrates and insects, encode RNA-dependent RNA polymerases that generate dsRNA from ssRNA substrates, enabling amplification of small RNA production. Especially in dsRNA biology, specifying the model organism is essential to avoid confusion about potential effects in humans.

• Similarly, the authors should specify "in C. elegans" in the sentence "Therefore, we propose that the import of extracellular dsRNA into the germline tunes intracellular pathways that cause heritable RNA silencing." This is important because C. elegans small RNA pathways differ significantly from those in other organisms, particularly in the PIWI-interacting RNA (piRNA) pathways, which depend on dsRNA in C. elegans but uses ssRNA in vertebrates. Specification is crucial to prevent misinterpretation by the reader. It is well understood that mechanisms of transgenerational inheritance that operate in nematodes or plants are not conserved in mammals.

• The first sentence of the discussion, "Our analyses suggest a model for ...", would also benefit from specifying "in C. elegans". The same applies to the figure captions. Clarification of the model organism should be added to the first sentence, especially in Figure 1.

Reviewer #2 (Public review):

Summary:

RNAs can function across cell borders and animal generations as sources of epigenetic information for development and immunity. The specific mechanistic pathways how RNA travels between cells and progeny remains an open question. Here, Shugarts, et al. use molecular genetics, imaging, and genomics methods to dissect specific RNA transport and regulatory pathways in the C. elegans model system. Larvae ingesting double stranded RNA is noted to not cause continuous gene silencing throughout adulthood. Damage of neuronal cells expressing double stranded target RNA is observed to repress target gene expression in the germline. Exogenous supply of short or long double stranded RNA required different genes for entry into progeny. It was observed that the SID-1 double-stranded RNA transporter showed different expression over animal development. Removal of the sid-1 gene caused upregulation of two genes, the newly described sid-1-dependent gene sdg-1 and sdg-2. Both genes were observed to also be negatively regulated by other small RNA regulatory pathways. Strikingly, loss then gain of sid-1 through breeding still caused variability of sdg-1 expression for many, many generations. SDG-2 protein co-localizes with a Z-granule marker, an intracellular site for heritable RNA silencing machinery. Collectively, sdg-1 presents a model to study how extracellular RNAs can buffer gene expression in germ cells and other tissues.

Strengths:

(1) Very clever molecular genetic methods and genomic analyses, paired with thorough genetics, were employed to discover insights into RNA transport, sdg-1 and sdg-2 as sid-1-dependent genes, and sdg-1's molecular phenotype.

(2) The manuscript is well cited, and figures reasonably designed.

(3) The discovery of the sdg genes being responsive to the extracellular RNA cell import machinery provides a model to study how exogenous somatic RNA is used to regulate gene expression in progeny. The discovery of genes within retrotransposons stimulates tantalizing models how regulatory loops may actually permit the genetic survival of harmful elements.

Weaknesses:

(1) As presented, the manuscript is incredibly broad, making it challenging to read and consider the data presented. This concern is exemplified in the model figure, that requires two diagrams to summarize the claims made by the manuscript.

(2) The large scope of the manuscript denies space to further probe some of the ideas proposed. The first part of the manuscript, particularly Figures 1 and 2, presents data that can be caused by multiple mechanisms, some of which the authors describe in the results but do not test further. Thus, portions of the results text come across as claims that are not supported by the data presented.

(3) The manuscript focuses on the genetics of SDGs but not the proteins themselves. Few descriptions of the SDGs functions are provided nor is it clarified why only SDG-1 was pursued in imaging and genetic experiments. Additionally, the SDG-1 imaging experiments could use additional localization controls.

eLife assessment

The fruit fly brain hosts neurosecretory neurons (Insulin Producing Cells or IPCs) that integrate many inputs and release insulin directly into the hemolymph. In this fundamental study, the population of IPCs are shown to be heterogeneous in their receptor diversity, exhibiting a range of responses to neuromodulation. The authors convincingly demonstrate, using a battery of experimental techniques and relying on the mapped whole brain connectome, how the heterogeneity in the responses across individual IPCs occur simultaneously and together modulate insulin release to maintain metabolic homeostasis. This work will be of interest to neuroscientists and physiologists, in particular for how cellular diversity results in a better control of homeostasis in short time scales.

Reviewer #1 (Public review):

Summary:

Insulin is crucial for maintaining metabolic homeostasis, and its release is regulated by various pathways, including blood glucose levels and neuromodulatory systems. The authors investigated the role of neuromodulators in regulating the dynamics of the adult Drosophila IPC population. They showed that IPCs express various receptors for monoaminergic and peptidergic neuromodulators, as well as synaptic neurotransmitters with highly heterogeneous profiles across the IPC population. Activating specific modulatory inputs, e.g. dopaminergic, octopaminergic or peptidergic (Leucokinin) using an optogenetic approach coupled with in vivo electrophysiology unveiled heterogeneous responses of individual IPCs resulting in excitatory, inhibitory or no responses. Interestingly, calcium imaging of the entire IPC population with or without simultaneous electrophysiological recording of individual cells showed highly specific and stable responses of individual IPCs suggesting their intrinsic properties are determined by the expressed receptor repertoire. Using the adult fly connectome they further corroborate the synaptic input of excitatory and inhibitory neuronal subsets of IPCs. The authors conclude that the heterogeneous modulation of individual IPC activity is more likely to allow for flexible control of insulin release to adapt to changes in metabolic demand and environmental cues.

Strengths:

This study provides a comprehensive, multi-level analysis of IPC properties utilizing single-nucleus RNA sequencing, anatomical receptor expression mapping, connectomics, electrophysiological recordings, calcium-imaging and an optogenetics-based 'intrinsic pharmacology' approach. It highlights the heterogeneous receptor profiles of IPCs, demonstrating complex and differential modulation within the IPC population. The authors convincingly showed that different neuromodulatory inputs exhibit varied effects on IPC activity and simultaneous occurrence of heterogeneous responses in IPCs with some populations exciting a subset of IPCs while inhibiting others, showcasing the intricate nature of IPC modulation and diverse roles of IPC subgroups. The temporal dynamic of IPC modulation showed that polysynaptic and neuromodulatory connections play a major role in IPC response. The authors demonstrated that certain neuromodulatory inputs, e.g. dopamine, can shift the overall IPC population activity towards either an excited or inhibited state. The study thus provides a fundamental entry point to understanding the complex influence of neuromodulatory inputs on the insulinergic system of Drosophila.

Weakness:

GPCRs are typically expressed at low levels and while the transcriptomic and reporter expression analysis was comprehensive, both approaches have the caveat that they do not allow validating protein level expression. Thus, some receptors might have been missed while others might be false positives. The authors acknowledged the challenges in accurately accessing receptor expression in complex modulatory systems indicating there are limitations in full understanding of the receptor profiles of IPCs.

While this study provides valuable insights into the heterogeneity of IPC responses and receptor expression, it will require future studies to elucidate how these modulatory inputs affect insulin release and transcriptional long-term changes.<br /> The authors further analyzed male and female snRNAseq data and claimed that the differences in receptor expression were minimal. The experimental analyses used mated females only and while the study is very complete in this respect, it would have been extremely interesting to compare male flies in terms of their response profiles.<br /> Lastly as also pointed out by the authors, their approach of using optogenetically driven excitation of modulatory neuronal subsets limits the interpretation of the results due to the possibly confounding direct or indirect effect of fast synaptic transmission on IPC excitation/inhibition, and the broad expression of some neuromodulatory lines used in this analysis.

Overall, however, the conclusions of this study are well supported by the data provided by the authors. Moreover, their detailed and thorough analysis of IPC modulation will have a significant impact on the field of metabolic regulation to understand the complex regulatory mechanism of insulin release, which can now be studied further to provide insight about metabolic homeostasis and neural control of metabolic processes.

Reviewer #2 (Public review):

Summary:

Held et al. investigated the distinct activities of Insulin-Producing Cells (IPCs) by electrophysiological recordings and calcium imaging. In the brain of the fruit fly Drosophila melanogaster, there are approximately 14 IPCs that are analogous to mammalian pancreatic beta cells and provide a good model system for monitoring their activities in vivo. The authors performed single-nucleus RNA sequencing analysis to examine what types of neuromodulatory inputs are received by IPCs. A variety of neuromodulatory receptors are expressed heterogeneously in IPCs, which would explain the distinct activities of IPCs in response to the activations of neuromodulatory neurons. The authors also conducted the connectome analysis and G-protein prediction analysis to strengthen their hypothesis that the heterogeneity of IPCs may underlie the flexible insulin release in response to various environmental conditions.

Strengths:

The authors succeeded patch-clamp recordings and calcium imaging of individual IPCs in living animals at a single-cell resolution, which allows them to show the heterogeneity of IPCs precisely. They measured IPC activities in response to 9 types of neurons in patch-clamp recordings and 5 types of neurons in calcium imaging, comparing the similarities and differences in activities between two methods. These results support the idea that the neuromodulatory system affects individual IPC activities differently in a receptor-dependent manner.

Weaknesses:

One concern is how much extent the heterogeneity of IPC activities in a short time scale is relevant to the net output, a release of insulin-like peptides in response to metabolic demands in a relatively longer time scale. The authors can test their hypothesis by manipulating the heterogeneous expressions of receptor genes in IPCs and examining IPC activities on a longer time scale. Moreover, while the authors focus on IPC activities, they did not show the activation of the neuromodulatory inputs and the net output of insulin levels in the data. The readers might want to know which neurons are indeed activated to send signals to IPCs and how IPC activities result in the secretion of insulin peptides.

Reviewer #2 (Public Review):

As discussed in the original review, this manuscript is an important contribution to a mechanistic understanding of LRRK2 kinase. Kinetic parameters for the GTPase activity of the ROC domain have been determined in the absence/presence of kinase activity. A feedback mechanism from the kinase domain to GTP/GDP hydrolysis by the ROC domain is convincingly demonstrated through these kinetic analyses. However, a regulatory mechanism directly linking the T1343 phospho-site and a monomer/dimer equilibrium is not fully supported. The T1343A mutant has reduced catalytic activity and can form similar levels of dimer as WT. The revised manuscript does point out that other regulatory mechanisms can also play a role in kinase activity and GTP/GDP hydrolysis (Discussion section). The environmental context in cells cannot be captured from the kinetic assays performed in this manuscript, and the introduction contains some citations regarding these regulatory factors. This is not a criticism, the detailed kinetics here are rigorous, but it is simply a limitation of the approach.

Author Response:

We would like to thank the reviewers for their constructive feedback and for acknowledging that our approach offers a simple yet powerful framework with the potential to serve as a comprehensive and intuitive tool for analyzing functional activity and connectivity.

In response to the reviewers’ recommendations, we will aim to improve and clarify the following aspects of our work in an upcoming revision:

Scope and limitations of the “fcHNN projection” (R#1 and R#2):

Both reviewers have correctly noted that the interpretability and explanatory power of the simplistic, two-dimensional fcHNN-based projection is limited. In the revised manuscript, we will clarify that, indeed, attractors are in a close mathematical relationship with the principal components of the raw data (i.e., the eigenvectors of the connectome) within our framework. The fcHNN-projection was introduced solely to establish a link between the proposed framework and concepts with which the reader may be more familiar.

We will enhance the presentation and discussion of our results to emphasize that – as the reviewers also kindly pointed out - the value of our approach lies in modelling how different facets of brain activity dynamically emerge from a common space of functional (ghost) attractors, rather than studying in the static attractor patterns themselves.

Motivations and Rationale for Using the Functional Connectome (R#2):

We agree with Reviewer #2 that a deeper mechanistic explanatory power could be achieved by modeling structure-function coupling, and that our framework is well-suited for this challenge. In our revision, we will highlight this as one of the promising future applications of our framework. We will, furthermore, clarify that the scope of the present work was deliberately restricted to functional connectivity to demonstrate that our framework also allows us to “bypass” the significant challenge of structure-function coupling. This enables us to focus on understanding the origins of canonical resting-state networks, the dynamic responses of the system to perturbations and the complex relationship between task-induced activity and resting-state connectivity, without first solving the structure-function coupling problem.

Additionally, we will mathematically justify the use of linear measures of the functional connectome to reconstruct the underlying non-linear dynamic system, thereby clearly delineating which results can and cannot be considered circular when starting from the functional connectome.

Improvements in Overall Clarity of Presentation (R#1):

In line with the above points and in general, we will strive to enhance the overall clarity of the presentation of our results, including figures, wording, and statistical analysis.

eLife assessment

This study presents a useful approach for revealing large-scale brain attractor dynamics during resting states, task processing, and disease conditions using insights from Hopfield neural networks. The evidence supporting the findings is solid across the many datasets analysed. The work will be of broad interest to neuroscientists using neuroimaging data with interest in computational modelling of brain activity.

Reviewer #1 (Public Review):

Summary:

Englert et al. proposed a functional connectome-based Hopfield artificial neural network (fcHNN) architecture to reveal attractor states and activity flows across various conditions, including resting state, task-evoked, and pathological conditions. The fcHNN can reconstruct characteristics of resting-state and task-evoked brain activities. Additionally, the fcHNN demonstrates differences in attractor states between individuals with autism and typically developing individuals.

Strengths:

(1) The study used seven datasets, which somewhat ensures robust replication and validation of generalization across various conditions.

(2) The proposed fcHNN improves upon existing activity flow models by mimicking artificial neural networks, thereby enhancing the representational ability of the model. This advancement enables the model to more accurately reconstruct the dynamic characteristics of brain activity.

(3) The fcHNN projection offers an interesting visualization, allowing researchers to observe attractor states and activity flow patterns directly.

Weaknesses:

(1) The fcHNN projection can offer low-dimensional dynamic visualizations, but its interpretability is limited, making it difficult to make strong claims based on these projections. The interpretability should be enhanced in the results and discussion.

(2) The presentation of results is not clear enough, including figures, wording, and statistical analysis, which contributes to the overall difficulty in understanding the manuscript. This lack of clarity in presenting key findings can obscure the insights that the study aims to convey, making it challenging for readers to fully grasp the implications and significance of the research.

Reviewer #2 (Public Review):

Summary:

Englert et al. use a novel modelling approach called functional connectome-based Hopfield Neural Networks (fcHNN) to describe spontaneous and task-evoked brain activity and the alterations in brain disorders. Given its novelty, the authors first validate the model parameters (the temperature and noise) with empirical resting-state function data and against null models. Through the optimisation of the temperature parameter, they first show that the optimal number of attractor states is four before fixing the optimal noise that best reflects the empirical data, through stochastic relaxation. Then, they demonstrate how these fcHNN-generated dynamics predict task-based functional activity relating to pain and self-regulation. To do so, they characterise the different brain states (here as different conditions of the experimental pain paradigm) in terms of the distribution of the data on the fcHNN projections and flow analysis. Lastly, a similar analysis was performed on a population with autism condition. Through Hopfield modeling, this work proposes a comprehensive framework that links various types of functional activity under a unified interpretation with high predictive validity.

Strengths:

The phenomenological nature of the Hopfield model and its validation across multiple datasets presents a comprehensive and intuitive framework for the analysis of functional activity. The results presented in this work further motivate the study of phenomenological models as an adequate mechanistic characterisation of large-scale brain activity.

Following up on Cole et al. 2016, the authors put forward a hypothesis that many of the changes to the brain activity, here, in terms of task-evoked and clinical data, can be inferred from the resting-state brain data alone. This brings together neatly the idea of different facets of brain activity emerging from a common space of functional (ghost) attractors.

The use of the null models motivates the benefit of non-linear dynamics in the context of phenomenological models when assessing the similarity to the real empirical data.

Weaknesses:

While the use of the Hopfield model is neat and very well presented, it still begs the question of why to use the functional connectome (as derived by activity flow analysis from Cole et al. 2016). Deriving the functional connectome on the resting-state data that are then used for the analysis reads as circular. If the fcHNN derives the basins of four attractors that reflect the first two principal components of functional connectivity, it perhaps suffices to use the empirically derived components alone and project the task and clinical data on it without the need for the fcHNN framework.

As presented here, the Hopfield model is excellent in its simplicity and power, and it seems suited to tackle the structure-function relationship with the power of going further to explain task-evoked and clinical data. The work could be strengthened if that was taken into consideration. As such the model would not suffer from circularity problems and it would be possible to claim its mechanistic properties. Furthermore, as mentioned above, in the current setup, the connectivity matrix is based on statistical properties of functional activity amongst regions, and as such it is difficult to talk about a certain mechanism. This contention has for example been addressed in the Cole et al. 2016 paper with the use of a biophysical model linking structure and function, thus strengthening the mechanistic claim of the work.

eLife assessment

This paper provides a useful analysis of the variation of the burden of strokes across geographic regions, finding differences in the relationship between strokes and their comorbidities. This dataset and the correlations found within will be a resource for directing the focus of future investigations. The results are technically solid, but there are cases where statistical analyses are yet to be carried out to support statements of statistical significance.

Reviewer #2 (Public Review):

Summary:

The authors have analyzed ethnogeographic differences in the comorbidity factors, such as a diabetes and heart disease, for the incidences of stroke and whether it leads to mortality.

Strengths:

The idea is interesting and data are compelling. The results are technically solid when presented, but in many cases statistical analyses are yet to be carried out to support statements of statistical significance.

The authors identify specific genetic loci that increase the risk of a stroke and how they differ by region.

Weaknesses:

The presentation is not focused. It is important to include p-values for all comparisons and focus the presentation on the main effects from the dataset analysis.

Reviewer #1 (Public Review):

Kreeger and colleagues have explored the balance of excitation and inhibition in the cochlear nucleus octopus cells of mice using morphological, electrophysiological, and computational methods. On the surface, the conclusion, that synaptic inhibition is present, does not seem like a leap. However, the octopus cells have been in the past portrayed as devoid of inhibition. This view was supported by the seeming lack of glycinergic fibers in the octopus cell area and the lack of apparent IPSPs. Here, Kreeger et al. used beautiful immunohistochemical and mouse genetic methods to quantify the inhibitory and excitatory boutons over the complete surface of individual octopus cells and further analysed the proportions of the different subtypes of spiral ganglion cell inputs. I think the analysis stands as one of the most complete descriptions of any neuron, leaving little doubt about the presence of glycinergic boutons.

Kreeger et al then examined inhibition physiologically, but here I felt that the study was incomplete. Specifically, no attempt was made to assess the actual, biological values of synaptic conductance for AMPAR and GlyR. Thus, we don't really know how potent the GlyR could be in mediating inhibition. Here are some numbered comments:

(1) "EPSPs" were evoked either optogenetically or with electrical stimulation. The resulting depolarizations are interpreted to be EPSPs. However previous studies from Oertel show that octopus cells have tiny spikes, and distinguishing them from EPSPs is tricky. No mention is made here about how or whether that was done. Thus, the analysis of EPSP amplitude is ambiguous.

(2) For this and later analysis, a voltage clamp of synaptic inputs would have been a simple alternative to avoid contaminating spikes or shunts by background or voltage-gated conductances. Yet only the current clamp was employed. I can understand that the authors might feel that the voltage clamp is 'flawed' because of the failure to clamp dendrites. But that may have been a good price to pay in this case. The authors should have at least justified their choice of method and detailed its caveats.

(3) The modeling raised several concerns. First, there is little presentation of assumptions, and of course, a model is entirely about its assumptions. For example, what excitatory conductance amplitudes were used? The same for inhibitory conductance? How were these values arrived at? The authors note that EPSGs and IPSGs had peaks at 0.3 and 3 ms. On what basis were these numbers obtained? The model's conclusions entirely depend on these values, and no measurements were made here that could have provided them. Parenthetical reference is made to Figure S5 where a range of values are tested, but with little explanation or justification.

(4) In experiments that combined E and I stimulation, what exactly were time timecourses of the conductance changes, and how 'synchronous' were they, given the different methods to evoke them? (had the authors done voltage clamp they would know the answers).

(5) Figure 4G is confusing to me. Its point, according to the text, is to show that changes in membrane properties induced by a block of Kv and HCN channels would not be expected to alter the amplitudes of EPSCs and IPSCs across the dendritic expanse. Now we are talking about currents (not shunting effects), and the presumption is that the blockers would alter the resting potential and thus the driving force for the currents. But what was the measured membrane potential change in the blockers? Surely that was documented. To me, the bigger concern (stated in the text) is whether the blockers altered exocytosis, and thus the increase in IPSP amplitude in blockers is due BOTH to loss of shunting and increase in presynaptic spike width. Added to this is that 4AP will reduce the spike threshold, thus allowing more ChR2-expressing axons to reach the threshold. Figure 4G does not address this point.

(6) Figure 5F is striking as the key piece of biological data that shows that inhibition does reduce the amplitude of "EPSPs" in octopus cells. Given the other uncertainties mentioned, I wondered if it makes sense as an example of shunting inhibition. Specifically, what are the relative synaptic conductances, and would you predict a 25% reduction given the actual (not modeled) values?

(7) Some of the supplemental figures, like 4 and 5, are hardly mentioned. Few will glean anything from them unless the authors direct attention to them and explain them better. In general, the readers would benefit from more complete explanations of what was done.

Reviewer #2 (Public Review):

Summary:

Kreeger et.al provided mechanistic evidence for flexible coincidence detection of auditory nerve synaptic inputs by octopus cells in the mouse cochlear nucleus. The octopus cells are specialized neurons that can fire repetitively at very high rates (> 800 Hz in vivo), yield responses dominated by the onset of sound for simple stimuli, and integrate auditory nerve inputs over a wide frequency span. Previously, it was thought that octopus cells received little inhibitory input, and their integration of auditory input depended principally on temporally precise coincidence detection of excitatory auditory nerve inputs, coupled with a low input resistance established by high levels of expression of certain potassium channels and hyperpolarization-activated channels.

In this study, the authors used a combination of numerous genetic mouse models to characterize synaptic inputs and enable optogenetic stimulation of subsets of afferents, fluorescent microscopy, detailed reconstructions of the location of inhibitory synapses on the soma and dendrites of octopus cells, and computational modeling, to explore the importance of inhibitory inputs to the cells. They determined through assessment of excitatory and inhibitory synaptic densities that spiral ganglion neuron synapses are densest on the soma and proximal dendrite, while glycenergic inhibitory synaptic density is greater on the dendrites compared to the soma of octopus cells. Using different genetic lines, the authors further elucidated that the majority of excitatory synapses on the octopus cells are from type 1a spiral ganglion neurons, which have low response thresholds and high rates of spontaneous activity. In the second half of the paper, the authors employed electrophysiology to uncover the physiological response of octopus cells to excitatory and inhibitory inputs. Using a combination of pharmacological blockers in vitro cellular and computational modeling, the authors conclude that glycine in fact evokes IPSPs in octopus cells; these IPSPs are largely shunted by the high membrane conductance of the cells under normal conditions and thus were not clearly evident in prior studies. Pharmacological experiments point towards a specific glycine receptor subunit composition. Lastly, Kreeger et. al demonstrated with in vitro recordings and computational modeling that octopus cell inhibition modulates the amplitude and timing of dendritic spiral ganglion inputs to octopus cells, allowing for flexible coincidence detection.

Strengths:

The work combines a number of approaches and complementary observations to characterize the spatial patterns of excitatory and inhibitory synaptic input, and the type of auditory nerve input to the octopus cells. The combination of multiple mouse lines enables a better understanding of and helps to define, the pattern of synaptic convergence onto these cells. The electrophysiology provides excellent functional evidence for the presence of the inhibitory inputs, and the modeling helps to interpret the likely functional role of inhibition. The work is technically well done and adds an interesting dimension related to the processing of sound by these neurons. The paper is overall well written, the experimental tests are well-motivated and easy to follow. The discussion is reasonable and touches on both the potential implications of the work as well as some caveats.

Weaknesses:

While the conclusions presented by the authors are solid, a prominent question remains regarding the source of the glycinergic input onto octopus cells. In the discussion, the authors claim that there is no evidence for D-stellate, L-stellate, and tuberculoventral cell (all local inhibitory neurons of the ventral and dorsal cochlear nucleus) connections to octopus cells, and cite the relevant literature. An experimental approach will be necessary to properly rule out (or rule in) these cell types and others that may arise from other auditory brainstem nuclei. Understanding which cells provide the inhibitory input will be an essential step in clarifying its roles in the processing of sound by octopus cells.

The authors showed that type 1a SGNs are the most abundant inputs to octopus cells via microscopy. However, in Figure 3 they compare optical stimulation of all classes of ANFs, then compare this against stimulation of type 1b/c ANFs. While a difference in the paired-pulse ratio (and therefore, likely release probability) can be inferred by the difference between Foxg1-ChR2 and Ntng1-ChR2, it would have been preferable to have specific data with selective stimulation of type 1a neurons.

eLife assessment

This valuable study, which implicates a specific Wolbachia gene in driving the male-killing phenotype in a moth, contributes to a growing body of literature from the authors in which they have nicely teased apart the loci responsible for male killing across diverse insects. Solid evidence supports the conclusions, though improvements to the statistical analysis for certain assays would strengthen the inferences further.

Reviewer #1 (Public review):

Summary:

Insects and their relatives are commonly infected with microbes that are transmitted from mothers to their offspring. A number of these microbes have independently evolved the ability to kill the sons of infected females very early in their development; this male killing strategy has evolved because males are transmission dead-ends for the microbe. A major question in the field has been to identify the genes that cause male killing and to understand how they work. This has been especially challenging because most male-killing microbes cannot be genetically manipulated. This study focuses on a male-killing bacterium called Wolbachia. Different Wolbachia strains kill male embryos in beetles, flies, moths, and other arthropods. This is remarkable because how sex is determined differs widely in these hosts. Two Wolbachia genes have been previously implicated in male-killing by Wolbachia: oscar (in moth male-killing) and wmk (in fly male-killing). The genomes of some male-killing Wolbachia contain both of these genes, so it is a challenge to disentangle the two.

This paper provides strong evidence that oscar is responsible for male-killing in moths. Here, the authors study a strain of Wolbachia that kills males in a pest of tea, Homona magnanima. Overexpressing oscar, but not wmk, kills male moth embryos. This is because oscar interferes with masculinizer, the master gene that controls sex determination in moths and butterflies. Interfering with the masculinizer gene in this way leads the (male) embryo down a path of female development, which causes problems in regulating the expression of genes that are found on the sex chromosomes.

Strengths:

The authors use a broad number of approaches to implicate oscar, and to dissect its mechanism of male lethality. These approaches include:<br /> (1) Overexpressing oscar (and wmk) by injecting RNA into moth eggs.<br /> (2) Determining the sex of embryos by staining female sex chromosomes.<br /> (3) Determining the consequences of oscar expression by assaying sex-specific splice variants of doublesex, a key sex determination gene, and by quantifying gene expression and dosage of sex chromosomes, using RNASeq.<br /> (4) Expressing oscar along with masculinizer from various moth and butterfly species, in a silkmoth cell line.

This extends recently published studies implicating oscar in male-killing by Wolbachia in Ostrinia corn borer moths, although the Homona and Ostrinia oscar proteins are quite divergent. Combined with other studies, there is now broad support for oscar as the male-killing gene in moths and butterflies (i.e. order Lepidoptera). So an outstanding question is to understand the role of wmk. Is it the master male-killing gene in insects other than Lepidoptera and if so, how does it operate?

Weaknesses:

I found the transfection assays of oscar and masculinizer in the silkworm cell line (Figure 4) to be difficult to follow. There are also places in the text where more explanation would be helpful for non-experts (see recommendations).

Reviewer #2 (Public review):

Summary:

Wolbachia are maternally transmitted bacteria that can manipulate host reproduction in various ways. Some Wolbachia induce male killing (MK), where the sons of infected mothers are killed during development. Several MK-associated genes have been identified in Homona magnanima, including Hm-oscar and wmk-1-4, but the mechanistic links between these Wolbachia genes and MK in the native host are still unclear.

In this manuscript, Arai et al. show that Hm-oscar is the gene responsible for Wolbachia-induced MK in Homona magnanima. They provide evidence that Hm-Oscar functions through interactions with the sex determination system. They also found that Hm-Oscar disrupts sex determination in male embryos by inducing female-type dsx splicing and impairing dosage compensation. Additionally, Hm-Oscar suppresses the function of Masc. The manuscript is well-written and presents intriguing findings. The results support their conclusions regarding the diversity and commonality of MK mechanisms, contributing to our understanding of the mechanisms and evolutionary aspects of Wolbachia-induced MK.

Strengths/weaknesses:

(1) The authors found that transient overexpression of Hm-oscar, but not wmk-1-4, in Wolbachia-free H. magnanima embryos induces female-biased sex ratios. These results are striking and mirror the phenotype of the wHm-t infected line (WT12). However, Table 1 lists the "male ratio," while the text presents the "female ratio" with standard deviation. For consistency, the calculation term should be uniform, and the "ratio" should be listed for each replicate.

(2) The error bars in Figure 3 are quite large, and the figure lacks statistical significance labels. The authors should perform statistical analysis to demonstrate that Hm-oscar-overexpressed male embryos have higher levels of Z-linked gene expression.

(3) The authors demonstrated that Hm-Oscar suppresses the masculinizing functions of lepidopteran Masc in BmN-4 cells derived from the female ovaries of Bombyx mori. They should clarify why this cell line was chosen and its biological relevance. Additionally, they should explain the rationale for evaluating the expression levels of the male-specific BmIMP variant and whether it is equivalent to dsx.

(4) Although the authors show that Hm-oscar is involved in Wolbachia-induced MK in Homona magnanima and interacts with the sex determination system in lepidopteran insects, the precise molecular mechanism of Hm-oscar-induced MK remains unclear. Further studies are needed to elucidate how Hm-oscar regulates Homona magnanima genes to induce MK, though this may be beyond the scope of the current manuscript.

Reviewer #3 (Public review):

Summary:

Overall, this is a clearly written manuscript with nice hypothesis testing in a non-model organism that addresses the mechanism of Wolbachia-mediated male killing. The authors aim to determine how five previously identified male-killing genes (encoded in the prophage region of the wHm Wolbachia strain) impact the native host, Homona magnanima moths. This work builds on the authors' previous studies in which:<br /> (1) They tested the impact of these same wHm genes via heterologous expression in Drosophila melanogaster.<br /> (2) They examined the activity of other male-killing genes (e.g., from the wFur Wolbachia strain in its native host: Ostrinia furnacalis moths).

Advances here include identifying which wHm gene most strongly recapitulates the male-killing phenotype in the native host (rather than in Drosophila), and the finding that the Hm-Oscar protein has the potential for male-killing in a diverse set of lepidopterans, as inferred by the cell-culture assays.

Strengths:

Strengths of the manuscript include the reverse genetics approaches to dissect the impact of specific male-killing loci, and the use of a "masculinization" assay in Lepidopteran cell lines to determine the impact of interactions between specific masc and oscar homologs.

Weaknesses:

My major comments are related to the lack of statistics for several experiments (and the data normalization process), and opportunities to make the manuscript more broadly accessible.

The manuscript I think would be much improved by providing more details regarding some of the genes and cross-lineage comparisons. I know some of this is reported in previous publications, but some summary and/or additional analysis would make this current manuscript much more approachable for a broader audience, and help guide readers to specific important findings. For example, a graphic and/or more detail on how the wmk/oscar homologs (within and across Wolbachia strains) differ (e.g., domains, percent divergence, etc) would be helpful for contextualizing some of the results. I recognize the authors discuss this in parts (e.g., lines 223-227), but it does require some bouncing between sections to follow. Similarly, the experiments presented in Figure 4 indicate that Hm-oscar has broad spectrum activity: how similar are the masc proteins from these various lepidopterans? Are they highly conserved? Rapidly evolving? Do the patterns of masc protein evolution provide any hints at how Oscar might be interacting with masc?

It is clear from Figure 1 that the combinations of wmk homologs do not cause male killing on their own. Did the authors test if any of the wmk homologs impact the MK phenotype of oscar? It looks like a previous study tested this in wFur (noted in lines 250-252), but given that the authors also highlight the differences between the wFur-oscar and Hm-oscar proteins, this may be worth testing in this system. Related to this, what is the explanation for why there would be 4 copies of wmk in Hm?

Why are some of the broods male-biased (2/3) rather than ~50:50? (Lines 170-175, Figure 2a). For example, there is a strong male bias in un-hatched oscar-injected and naturally infected embryos, whereas the control uninfected embryos have normal 50:50 sex ratios. It is difficult to interpret the rate of male-killing given that the sex ratios of different sets of zygotes are quite variable.

Figure 2b - it appears there are both male and female bands in the HmOsc male lane. I think this makes sense (likely a partial phenotype due to the nature of the overexpression approach), but this is worth highlighting, especially in the context of trying to understand how much of the MK phenotype might be recapitulated through these methods. Related, there is no negative control for this PCR.

It appears the RNA-seq analysis (Figure 3) is based on a single biological replicate for each condition. And, there are no statistical comparisons that support the conclusions of a shift in dosage compensation. Finally, it is unclear what exactly is new data here: the authors note "The expression data of the wHm-t-infected and non-infected groups were also calculated based on the transcriptome data included in Arai et al. (2023a)" - So, are the data in Figure 3c and 3d a re-print of previous data? The level of dosage compensation inferred by visually comparing the control conditions in 3b and 3d does not appear consistent. With only one biological replicate library per condition, what looks like a re-print of previous data, and no statistical comparisons, this is a weakly supported conclusion.

In Figure 4: There are no statistics to support the conclusions presented here. Additionally, the data have gone through a normalization process, but it is difficult to follow exactly how this was done. The control conditions appear to always be normalized to 100 ("The expression levels of BmImpM in the Masc and Hm-Oscar/Oscar co-transfected cells were normalized by setting each Masc-transfected cell as 100"). I see two problems with this approach:<br /> (1) This has eliminated all of the natural variation in BmImpM expression, which is likely not always identical across cells/replicates.<br /> (2) How then was the percentage of BmImpM calculated for each of the experimental conditions? Was each replicate sample arbitrarily paired with a control sample? This can lead to very different outcomes depending on which samples are paired with each other. The most appropriate way to calculate the change between experimental and control would be to take the difference between every single sample (6 total, 3 control, 3 experimental) and the mean of the control group. The mean of the control can then be set at 100 as the authors like, but this also maintains the variability in the dataset and then eliminates the issue of arbitrary pairings. This approach would also then facilitate statistical comparisons which is currently missing.

Author response:

Reviewer #2 (Public Review):

In this manuscript, Kafri and colleagues assess the contribution of protein degradation to the cell size-dependent accumulation of total protein. This is an interesting line of research that has not previously been explored. Most of the focus on the size-dependence of protein accumulation has been on the synthesis part of the equation. As cells get too big, the efficiency of cell growth (mass accumulation per unit mass) decreases. It is argued that this is not due to the loss of the efficiency in protein synthesis, but rather is due to the increased protein degradation in larger cells. It is an interesting hypothesis, that might well be true, but there are some issues with key aspects of the data and other supporting data are quite indirect. More work needs to be done to support the central claims.

We thank the reviewer for appreciating the work is interesting and previously unexplored.

The major issue is that the data supporting the proportional increase in protein synthesis with cell size need to be strengthened. Protein synthesis is measured by the amount of a methionine analog that is incorporated in a fixed amount of time. Fig. 2 then plots this amount as a function of cell size, which is presumably measured using a total protein dye (this information is not included; incidentally the axis labels should note what the measurement is 'total protein' or 'forward scatter' rather than the more ambiguous 'cell size'). In any case, something is wrong with the cell size measurements in Figure 2 because many cells basically have almost negligible size (near 0) while others have sizes up to 5 or 6 arbitrary units. It makes no sense that there should be a 10-fold or even 100-fold range in cell sizes. For this reason, I can't interpret the data in Figure 2, which is unfortunate since that is a crucial figure for the authors' argument.

The data supporting higher rates of protein degradation per unit mass in large cells suffers from a similar problem as Figure 3E has the same issue as Figure 2 with too many tiny 'cells'.

Yes, the reviewer is correct that we are using a total protein dye (Alexa fluorophore-conjugated succinimidyl ester, abbreviated as SE) to measure cell size. We have included details regarding the methods of cell size (total protein content) measurement in both the Methods (line 463-466) and Results (line 100-102) sections.

Regarding the reviewer’s concern on the cell size range, we apologize for the confusion the figures may have caused. These cell size measurements are within reasonable range and not 10-fold or 100-fold. Please refer to our detailed response above to essential point #1.

Moreover, the reliance on cycloheximide to treat cells and measure reduction in mass isn't ideal since shutting off all protein synthesis is a pretty drastic perturbation. It would have been better to shut off synthesis of a specific protein and measure its degradation in large and small cells while keeping the cells otherwise intact.

We acknowledge that relying on cycloheximide to measure changes in mass has limitations, as acute inhibition in protein synthesis is a significant perturbation. Ideally, we would measure the degradation of specific proteins in large and small cells while keeping the rest of the cellular processes intact. However, this presents considerable technological challenges. While our evidence clearly shows increased protein degradation and compensatory growth slowdown in large cells, we have not yet identified the specific proteins/genes being targeted. Implementing the reviewer's suggestion would require first screening for a suitable protein/gene to serve as a reporter for compensatory degradation. A significant proteomics screen may allow identification of potential targets, but further validation would necessitate substantial effort, including the generation and validation of a reporter system. We agree that this is a valuable experiment to pursue, but it will likely be part of a follow-up study focused on characterizing the specific protein targets and E3 ligases involved in these processes. In the revised manuscript, we discuss these open questions and future directions in line 380-410.

Reviewer #3 (Public Review):

The authors report a previously undocumented role for UPS-mediated protein turnover in size control in human cells. The study builds on previous observations made by the Kafri group that large cells undergo size compensation by reducing their rate of growth. In particular, recent published work by Ginzberg et al showed that CDK2 inhibition is accompanied by long term size compensation in the form of reduced cell growth whereas CDK6 inhibition is not. The authors investigate the basis for this effect and demonstrate in both unperturbed and perturbed growth/division contexts, using both fixed cells and time lapse microscopy, that the rate of protein synthesis increases proportionately in large cells that undergo size compensation even though mass accumulation is attenuated. The authors show that this effect appears to be mediated by increased proteasomal activity, as demonstrated by proteasome-dependent K48-ubiquitin chain turnover. Intriguingly, this degradation-mediated size compensation mechanism appears to be most active at the G1/S transition, the primary point at which size control operates. The experiments are well controlled, and the conclusions of the study are in general well supported by the data. The authors present an interesting set of discussion points that relate their observations to size control mechanisms in dividing and non-dividing cells. While specific mechanisms are not pursued, this study nevertheless adds an important new insight into the still unsolved problem of size control.

We thank the reviewer for appreciating the novelty of the work.

eLife assessment

This important study advances our understanding of the mechanisms underlying chromatin-mediated gene regulation by SET DOMAIN-CONTAINING PROTEIN 7 (SDG7). The evidence supporting the author's claims – centered on a combination of imaging approaches with molecular and genetic experiments – is convincing, although certain aspects can be improved. The work will be of broad interest to molecular biologists studying epigenetic regulation of gene expression.

eLife assessment

The authors conduct a valuable GWAS meta-analysis for COVID-19 hospitalization in admixed American populations and prioritized risk variants and genes. The evidence supporting the claims of the authors is solid. The work will be of interest to scientists studying the genetic basis of COVID pathogenesis.

Reviewer #1 (Public Review):

Summary:

This paper conducted a GWAS meta-analysis for COVID-19 hospitalization among admixed American populations. The authors identified four genome-wide significant associations, including two novel loci (BAZ2B and DDIAS), and an additional risk locus near CREBBP using cross-ancestry meta-analysis. They utilized multiple strategies to prioritize risk variants and target genes. Finally, they constructed and assessed a polygenic risk score model with 49 variants associated with critical COVID-19 conditions.

Strengths:

Given that most of the previous studies were done in European ancestries, this study provides unique findings about the genetics of COVID-19 in admixed American populations. The GWAS data would be a valuable resource for the community. The authors conducted comprehensive analyses using multiple different strategies, including Bayesian fine mapping, colocalization, TWAS, etc., to prioritize risk variants and target genes. The polygenic risk score (PGS) result demonstrated the ability of cross-population PGS model for COVID-19 risk stratification.

Weaknesses:

(1) One of the major limitations of this study is that the GWAS sample size is relatively small, which limits its power.<br /> (2) Lack of replication cohort.<br /> (3) Colocalization and TWAS used eQTL data from GTEx data, which are mainly from European ancestries.

Comments on latest version:

The authors addressed most of my concerns.

Reviewer #2 (Public Review):

This is a genome-wide association study of COVID-19 in individuals of admixed American ancestry (AMR) recruited from Brazil, Colombia, Ecuador, Mexico, Paraguay and Spain. After quality control and admixture analysis, a total of 3,512 individuals were interrogated for 10,671,028 genetic variants (genotyped + imputed). The genetic association results for these cohorts were meta-analyzed with the results from The Host Genetics Initiative (HGI), involving 3,077 cases and 66,686 controls. The authors found two novel genetic loci associated with COVID-19 at 2q24.2 (rs13003835) and 11q14.1 (rs77599934), and other two independent signals at 3p21.31 (rs35731912) and 6p21.1 (rs2477820) already reported as associated with COVID-19 in previous GWASs. Additional meta-analysis with other HGI studies also suggested risk variants near CREBBP, ZBTB7A and CASC20 genes.

Strengths:

These findings rely on state-of-the-art methods in the field of Statistical Genomics and help to address the issue of low number of GWASs in non-European populations, ultimately contributing to reduce health inequalities across the globe.

Weaknesses:

There is no replication cohort, as acknowledged by the authors (page 29, line 587) and no experimental validation to assess the biological effect of putative causal variants/genes. Thus, the study provides good evidence of association, rather than causation, between the genetic variants and COVID-19.

Comments on latest version:

The issues identified in the first round of review were well addressed by the authors in the revised version of the manuscript.

Reviewer #3 (Public Review):

Summary:

In the context of the SCOURGE consortium's research, the authors conduct a GWAS meta-analysis on 4,702 hospitalized individuals of admixed American descent suffering from COVID-19. This study identified four significant genetic associations, including two loci initially discovered in Latin American cohorts. Furthermore, a trans-ethnic meta-analysis highlighted an additional novel risk locus in the CREBBP gene, underscoring the critical role of genetic diversity in understanding the pathogenesis of COVID-19.

Strengths:

(1) The study identified two novel severe COVID-19 loci (BAZ2B and DDIAS) by the largest GWAS meta-analysis for COVID-19 hospitalization in admixed Americans.<br /> (2) With a trans-ethnic meta-analysis, an additional risk locus near CREBBP was identified.

Weaknesses:

(1) The GWAS power is limited due to the relatively small number of cases.

(2) There is no replication study for the novel severe COVID-19 loci, which may lead to false positive findings.

(3) The variants selected for the PGS appear arbitrary and may not leverage the GWAS findings.

(4) The TWAS models were predominantly trained on European samples, and there is no replication study for the findings as well.

Author response:

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

Public Reviews:

Reviewer #1 (Public Review):

Summary:

This paper conducted a GWAS meta-analysis for COVID-19 hospitalization among admixed American populations. The authors identified four genome-wide significant associations, including two novel loci (BAZ2B and DDIAS), and an additional risk locus near CREBBP using cross-ancestry meta-analysis. They utilized multiple strategies to prioritize risk variants and target genes. Finally, they constructed and assessed a polygenic risk score model with 49 variants associated with critical COVID-19 conditions.

Strengths:

Given that most of the previous studies were done in European ancestries, this study provides unique findings about the genetics of COVID-19 in admixed American populations. The GWAS data would be a valuable resource for the community. The authors conducted comprehensive analyses using multiple different strategies, including Bayesian fine mapping, colocalization, TWAS, etc., to prioritize risk variants and target genes. The polygenic risk score (PGS) result demonstrated the ability of the cross-population

PGS model for COVID-19 risk stratification.

Thank you very much for the positive comments and the willingness to revise this manuscript.

Weaknesses:

(1) One of the major limitations of this study is that the GWAS sample size is relatively small, which limits its power.

(2) The fine mapping section is unclear and there is a lack of information. The authors assumed one causal signal per locus, and only provided credible sets, but did not provide posterior inclusion probabilities (PIP) for the variants to be causal.

(3) Colocalization and TWAS used eQTL data from GTEx data, which are mainly from European ancestries. It is unclear how much impact the ancestry mismatch would have on the result. The readers should be cautious when interpreting the results and designing follow-up studies.

We agree with that the sample size is relatively small. Despite that, it was sufficient to reveal novel risk loci supporting the robustness of the main findings. We have indicated this limitation at the end of the discussion section.

Thank you for rising this point. As suggested, we have also used SuSIE, which allows to assume more than one causal signal per locus. However, in this case the results were not different from those obtained with the original Bayesian colocalization performed with corrcoverage. Regarding the PIP, at the fine mapping stage we are inclined to put more weight on the functional annotations of the variants in the credible set than on the statistical contributions to the signal. This is the reason why we prefer not to put weight on the PIP of the variants but prioritize variants that were enriched functional annotations.

This is a good point regarding the lack of diversity in GTEx data. We have also used data from AMR populations (GALA II-SAGE models), although it was only available for blood tissue. Regarding the ancestry mismatch between datasets, several studies have attempted to explore the impact. Gay et al. (PMID: 32912333) studied local ancestry effects on eQTLs from the GTEx consortium and concluded that adjustment of eQTLs by local ancestry only yields modest improvement over using global ancestry (as done in GTEx). Moreover, the colocalization results between adjusting by Local Ancestry and Global Ancestry were not significantly different. Besides, Mogil et al. (PMID: 30096133) observed that genes with higher heritability share genetic architecture between populations. Nevertheless, both studies have evidenced decreased power and poorer predictive performances regarding gene expression because of reduced diversity in eQTL analyses. As consequence of the ancestry mismatch, we now warn the readers that this may compromise signal detection (Discussion, lines 531-533). 

Reviewer #2 (Public Review):

This is a genome-wide association study of COVID-19 in individuals of admixed American ancestry (AMR) recruited from Brazil, Colombia, Ecuador, Mexico, Paraguay, and Spain. After quality control and admixture analysis, a total of 3,512 individuals were interrogated for 10,671,028 genetic variants (genotyped + imputed). The genetic association results for these cohorts were meta-analyzed with the results from The Host Genetics Initiative (HGI), involving 3,077 cases and 66,686 controls. The authors found two novel genetic loci associated with COVID-19 at 2q24.2 (rs13003835) and 11q14.1 (rs77599934), and other two independent signals at 3p21.31 (rs35731912) and 6p21.1 (rs2477820) already reported as associated with COVID-19 in previous GWASs. Additional meta-analysis with other HGI studies also suggested risk variants near CREBBP, ZBTB7A, and CASC20 genes.

Strengths:

These findings rely on state-of-the-art methods in the field of Statistical Genomics and help to address the issue of a low number of GWASs in non-European populations, ultimately contributing to reducing health inequalities across the globe.

Thank you very much for the positive comments and the willingness to revise this manuscript.

Weaknesses:

There is no replication cohort, as acknowledged by the authors (page 29, line 587), and no experimental validation to assess the biological effect of putative causal variants/genes. Thus, the study provides good evidence of association, rather than causation, between the genetic variants and COVID-19. Lastly, I consider it crucial to report the results for the SCOURGE Latin American GWAS, in addition to its meta-analysis with HGI results, since HGI data has a different phenotype scheme (Hospitalized COVID vs Population) compared to SCOURGE (Hospitalized COVID vs Non-hospitalized COVID).

We essentially agree with the reviewer in that one of the main limitations of the study is the lack of a replication stage because of the use of all available datasets on a one-stage analysis. To contribute to the interpretation of the findings in the absence of a replication stage, we now assessed the replicability of the novel loci using the Meta-Analysis Model-based Assessment of replicability (MAMBA) approach (PMID: 33785739) and included the posterior probabilities of replication in Table 2. We also explored further the potential replicability of signals in other populations. We agree that the results should be interpreted in terms of associations given the lack of functional validation of main findings, so we have slightly modified the discussion.

As suggested, the SCOURGE Latin American GWAS summary is now accessible by direct request to the Consortium GitHub repository (https://github.com/CIBERER/Scourge-COVID19) (lines 797-799). We have also included the results from the SCOURGE GWAS analysis for the replication of the 40 lead variants in the Supplementary Table 12. Results from the SCOURGE GWAS for the lead variants in the AMR meta-analysis with HGI were already included in the Supplementary Table 2. As note, we have not been able to conduct the meta-analysis with the same hospitalization scheme as in the HGI study since the population-specific results for those analyses were not publicly released. However, sensitivity analyses included within the supplementary material from the COVID-19 Host Genetics Initiative (2021) stated that there were no significant differences in effects (Odds Ratios) between analyses using population controls or just non-hospitalized COVID-19 patients.

Reviewer #3 (Public Review):

Summary:

In the context of the SCOURGE consortium's research, the authors conduct a GWAS meta-analysis on 4,702 hospitalized individuals of admixed American descent suffering from COVID-19. This study identified four significant genetic associations, including two loci initially discovered in Latin American cohorts. Furthermore, a trans-ethnic meta-analysis highlighted an additional novel risk locus in the CREBBP gene, underscoring the critical role of genetic diversity in understanding the pathogenesis of COVID-19.

Strengths:

(1) The study identified two novel severe COVID-19 loci (BAZ2B and DDIAS) by the largest GWAS meta-analysis for COVID-19 hospitalization in admixed Americans.

(2) With a trans-ethnic meta-analysis, an additional risk locus near CREBBP was identified.

Thank you very much for the positive comments and the willingness to revise this manuscript.

Weaknesses:

(1) The GWAS power is limited due to the relatively small number of cases.

(2) There is no replication study for the novel severe COVID-19 loci, which may lead to false positive findings.

We agree with that the sample size is relatively small. Despite that, it was sufficient to reveal novel risk loci supporting the robustness of the main findings. We have indicated this limitation at the end of the discussion section.

Regarding the lack of a replication study, we now assessed the replicability of the novel loci using the Meta-Analysis Model-based Assessment of replicability (MAMBA) approach (PMID: 33785739). We have included the posterior probabilities of replication in Table 2.

(3) Significant differences exist in the ages between cases and controls, which could potentially introduce biased confounders. I'm curious about how the authors treated age as a covariate. For instance, did they use ten-year intervals? This needs clarification for reproducibility.

Thank you for rising this point. Age was included as a continuous variable. This has been now indicated in line 667 (within Material and Methods).

(4)"Those in the top PGS decile exhibited a 5.90-fold (95% CI=3.29-10.60, p=2.79x10-9) greater risk compared to individuals in the lowest decile". I would recommend comparing with the 40-60% PGS decile rather than the lowest decile, as the lowest PGS decile does not represent 'normal controls'.

Thank you. In the revised version, the PGS categories was compared following the recommendation (lines 461-463).

(5) In the field of PGS, it's common to require an independent dataset for training and testing the PGS model. Here, there seems to be an overfitting issue due to using the same subjects for both training and testing the variants.

We are sorry for the misunderstanding. In fact, we have followed the standard to avoid overfitting of the PGS model and have used different training and testing datasets. The training data (GWAS) was the HGI-B2 ALL meta-analysis, in which our AMR GWAS was not included. The PRS model was then tested in the SCOURGE AMR cohort. However, it is true that we did test the combination of the PRS adding the new discovered variants in the SCOURGE cohort. To avoid potential overfitting by adding the new loci, we have excluded from the manuscript the results on which we included the newly discovered variants.

(6) The variants selected for the PGS appear arbitrary and may not leverage the GWAS findings without an independent training dataset.

Again, we are sorry for the misunderstanding. The PGS model was built with 43 variants associated with hospitalization or severity within the HGI v7 results and 7 which were discovered by the GenOMICC consortium in their latest study and were not in the latest HGI release. The variants are included within the Supplementary Table 14, but we have now annotated the discovery GWAS.

(7) The TWAS models were predominantly trained on European samples, and there is no replication study for the findings as well.

This is a good point regarding the lack of diversity in GTEx data. We have also used data from AMR populations (GALA II-SAGE models), although it was only available for blood tissue. Regarding the ancestry mismatch between datasets, several studies have attempted to explore the impact. Gay et al. (PMID: 32912333) studied local ancestry effects on eQTLs from the GTEx consortium and concluded that adjustment of eQTLs by local ancestry only yields modest improvement over using global ancestry (as done in GTEx). Moreover, the colocalization results between adjusting by Local Ancestry and Global Ancestry were not significantly different. Besides, Mogil et al. (PMID: 30096133) observed that genes with higher heritability share genetic architecture between populations. Nevertheless, both studies have evidenced decreased power and poorer predictive performances regarding gene expression because of reduced diversity in eQTL analyses. As consequence of the ancestry mismatch, we now warn the readers that this may compromise signal detection (Discussion, lines 531-533). 

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) The authors mentioned the fine mapping method did not converge for the locus in chr 11. I would consider trying a different fine-mapping method (such as SuSiE or FINEMAP). It would be helpful to provide posterior inclusion probabilities (PIP) for the variants in fine mapping results and plot the PIP values in the regional association plots.

As suggested, we have also used SuSIE, which allows to assume more than one causal signal per locus. However, in this case the results were not different from those obtained with the original Bayesian colocalization performed with corrcoverage. SuSIE’s fine-mapping for chromosome 11 prioritized a single variant, which is likely due to the rare frequency. Thus, we have maintained the fine-mapping as it was originally indicated in the previous version of the manuscript but have now included the credible set in Supplementary Table 6.

Regarding the PIP, at the fine mapping stage we are inclined to put more weight on the functional annotations of the variants in the credible set than on the statistical contributions to the signal. This is the reason why we prefer not to put weight on the PIP of the variants but prioritize variants that were enriched functional annotations.

(2) Please provide more detailed information about the VEP and V2G analysis and how to interpret those results. My understanding of V2G is that it includes different sources of information (such as molecular QTLs and chromatin interactions from different tissues/cell types, etc.). It is unclear what sources of information and weight settings were used in the V2G model.

Thank you for rising this point. As suggested, we have clarified the basis for VEP and V2G and the interpretation (lines 732-743).

(3) The authors identified multiple genes with different strategies, e.g. FUMA, V2G, COLOC, TWAS, etc. How many genes were found/supported by evidence provided by multiple methods? It could be helpful to have a table summarizing the risk genes found by different strategies, and the evidence supporting the genes. e.g. which genes are found by which methods, and the biological functions of the genes, etc.

Thank you for rising this point. As suggested, we now added a new figure (Figure 5) to summarize the findings with the multiple methods used.

(4) It would be helpful to make the code/scripts available for reproducibility.

As suggested, the SCOURGE Latin American GWAS summary and the analysis scripts (https://github.com/CIBERER/Scourge-COVID19/tree/main/scripts/novel-risk-hosp-AMR-2024) are now accessible in the Consortium GitHub repository (https://github.com/CIBERER/Scourge-COVID19) (lines 806-807).

(5) The fonts in some of the figures (e.g. Figure 2) are hard to read.

Thank you. We have now included the figures as SVG files.

Reviewer #2 (Recommendations For The Authors):

- The abstract lacks a conclusion sentence.

Thank you. As suggested, we have included two additional sentences with broad conclusions from the study. We preferred to avoid relying on conclusions related to known or new biological links of the prioritized genes given the lack of functional validation of main findings.

- Regarding the association analysis (page 27, line 677), I wonder if some of the 10 principal components (PCs) are capturing information about the recruitment areas (countries). It may be relevant to test for multicollinearity among these variables.

Since we acknowledge that some of the categories might be correlated with a certain PC but not all of them do, we have calculated GVIF values for the main variables to assess the categorical variable as a single entity. The scaled GVIF^1(1/2*Df)) value for the categorical variable is 1.52. Thus, if we square this value, we obtain 2.31, which can be then used for applying usual rule-of-thumb for VIF values.

- Still on the topic of association analysis, did the authors adjust the logistic model for comorbidities variables from Table 1? Given these comorbidities also have a genetic component and their distribution differs between non-hospitalized vs hospitalized, I am concerned that comorbidities might be confounding the association between genetic variants and COVID.

We did not adjust by comorbidities since HGI studies were not adjusted either and we aimed to be as aligned as possible with HGI. However, as suggested, we have now tested the association between each of the comorbidities in Table 1 and each of the variants in Table 2, using the comorbidities as dependent variables and adjusting for the main covariables (age, sex, PCs and country of recruitment). None of the variants were significantly associated to the comorbidities (line 333).

- If I understood correctly, the 49 genetic variants used to develop the polygenic risk score model (PRS) were based on the HGI total sample size (data release 7), which is predominantly of European ancestry. I am concerned about the prediction accuracy in the AMR population (PRS transferability issue).

We have explored literature in search of other PRS to compare the associated OR in our cohort with ORs calculated in European populations. Horowitz et al. (2022) reported an OR of 1.38 for the top 10% with respect to hospitalization risk in European individuals using a GRS with 12 variants.

We acknowledge that this might be an issue and is now explained in discussion of the revised version (lines 561-568). However, as this is the first time a PRS for COVID-19 is applied to a relatively large AMR cohort, we believe that this analysis will be of value for further analyses regarding PRS transferability, providing a source for comparison in further studies.    

- On page 23, line 579, the authors acknowledge their "GWAS is underpowered". This sentence requires a sample/power calculation, otherwise, I suggest using "is likely underpowered".

Thanks for the input. We have modified the sentence as suggested.

Reviewer #3 (Recommendations For The Authors):

I wonder if the authors have an approximate date when the GWAS summary statistic will be available. I reviewed some manuscripts in the past, and the authors claimed they would deposit the data soon, but in fact it would not happen until 2 years later.

The summary statistics are already available from the SCOURGE Consortium repository https://github.com/CIBERER/Scourge-COVID19 (lines 806-807).

eLife assessment

This fundamental study provides convincing evidence for pectin modification as a requirement for RALF peptide signalling altering the apoplastic pH, adding further support for a key role of RALF peptides in linking the assembly and dynamics of the extracellular matrix with cellular activity and function. Data that have been added in comparison to a previous version have enhanced the study. The study should be of interest to anyone studying signaling and specifically to plant cell biologists.

Reviewer #1 (Public review):

Summary:

Rößling et al., report in this study that the perception of RALF1 by the FER receptor is mediated by the association of RALF1 with deesterified pectin, contributing to the regulation of the cell wall matrix and plasma membrane dynamics. In addition, they report that this mode of action is independent from the previously reported cell wall sensing mechanism mediated by the FER-LRX complex.

This manuscript reproduces and aligns with the results from a recently published study (Liu et al., Cell) where they also report that RALF1 can interact with deesterified pectin, forming coacervates and promoting the recruitment of LLG-FER at the membrane.

Reviewer #2 (Public review):

Summary:

The study by Rößling et al. addresses the link between the biochemical constitution of the cell wall, in particular the methylesterification state of pectin with signalling induced by the extracellular RALF peptide. The work suggests that only in the presence of demethylesterifies pectin, RALF is able to trigger activation of its receptor FERONIA (FER).<br /> Remarkably, the application of RALF peptides leads to rather dramatic FER-dependent changes in wall integrity and plasma membrane invaginations not observed before. Interestingly, RALF can be out-titrated from the wall by short pectin fragments. In addition, the study provides further evidence for multiple FER-dependent pathways by showing the presence of LRX proteins is not required for the pectin/RALF mediated signalling.

Strengths:

This work provides fundamental insight into a complex emerging pathway, or perhaps several pathways, linking pectin sensing, pectin structure and RALF/FER signalling. The study provides convincing evidence that pectin methylesterase activity is required for RALF sensing, indicating that the physical interaction of RALFs with the cell wall is important for signalling. Beyond that, the study documents very clearly how profoundly RALF signalling can affect cell wall integrity and membrane topology.

Weaknesses:

Not a weakness per se, as it cannot be avoided, but drawing conclusions from genetic material with altered pectin always suffers from the possibility of secondary effects as this cell wall component is under heavy surveillance and able to respond plastically to different cues. However, the authors take that into account and have performed adequate controls to minimize that possibility.

Reviewer #3 (Public review):

In this important work, the authors show compelling evidence that the Rapid Alkalinisation Factor1 (RALF1) peptide acts as an interlink between pectin methyl esterification status and FERONIA receptor-like kinase in mediating extracellular sensing. Moreover, the RALF1-mediated pectin perception is surprisingly independent of LRX-mediated extracellular sensing in roots. The authors also show that the peptide directly binds demethylated pectin and the positively charged amino acids are required for pectin binding as well as for its physiological activity.

Some present findings are surprising; previously, the FERONIA extracellular domain was shown to bind pectin directly, and the mode of operation in the pollen tube involves the LRX8-RALF4 complex, which seems not the case for RALF1 in the present study. Although some aspects remain controversial, this work is a very valuable addition to the ongoing debate about this elusive complex regulation and signaling.

The authors drafted the manuscript well, so I do not have a lot of criticism or suggestions. The experiments are well-designed, executed, and presented, and they solidly support the authors' claims.

Author response:

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

Reviewer #1 (Recommendations for the Authors):

Major 

(a) In the study the authors focus on the RALF1 peptide. But according to expression data and the study from Abarca et al., 2021, RALF1 is not the only peptide expressed in the root and also having an impact in root growth effect. Similarly, looking at the primary sequence from RALF1 it does not differ much chemically from other RALFs such as RALF33, RALF23, RALF22, etc. So, does the cell wall pectin methylation status also have an impact on the effect of other RALFs on root growth or is that specific of RALF1? 

(b) In addition, is the internalization of FER depending only on RALF1 upon the methylation status of cell wall pectins? Or can other RALFs cause a similar effect potentially?

(c) The authors propose that RALF1 associates with deesterifed pectin, through electrostatic interactions. To do that they perform Biolayer interferometry assays using a buffer with pH 7.4. Is that a relevant pH at the cell wall? Is possible that the authors thought that this may not change the charges of R and K residues, however, it will affect the overall charge of the peptide given the fact that it contains quite some N and Q in the exposed surface. The authors may want to consider that.

(d) Moreover, the authors do not use their peptide RALF1KR, suggested as a peptide not binding OGs, as a control in their OG binding assays. That biochemical experiment should also be included to validate their results and conclusions.

We thank reviewer #1 for these comments. In this work, we focused on RALF1 but the majority of AtRALF peptides, when applied exogenously as synthetic peptides, induce RALF1like effects in Arabidopsis (Abarca et al., 2021; PMID: 34608971). Moreover, all RALF peptides display clusters of R and K residues and are negatively charged (Abarca et al., 2021; PMID: 34608971). In comparison to RALF1, we now also use RALF34 because it was suggested to interact also via the Catharanthus roseus receptor-like kinase 1-like (CrRLK1L) THESEUS1 (THE1). Notably, RALF34 also induced the internalization of FER-GFP. Moreover, the interference with PME also disrupted this activity of RALF34. Therefore, we assume that other RALF peptides display the same or similar signalling modalities. Nevertheless, it remains to be addressed if all RALF family members require PME activity. 

We appreciated these comments and incorporated this aspect in the revised version of the manuscript. The pH was chosen for technical reasons associated with the used BLI buffer. As requested, we also included the RALF1-KR peptide in our OG binding assays. Under these conditions, the mutated peptides were not able to interact with the OGs anymore. Accordingly, we conclude that the K and R residues in RALF1 are crucial for its binding to demethylesterified OGs.  

(e) Another important aspect is regarding their design RALF1KR mutant and its effect in planta. The authors report the following: "RALF1-KR peptides are not bioactive, because they did neither affect root growth, nor cell wall integrity, nor did they induce the ligand-induced endocytosis of FER in epidermal root cells (Figure 5D-I). These findings suggest that the positively charged residues in RALF1 are essential for its activity in roots." According to the structure published by Xiao at el. 2019, the R in the alpha helix from RALF peptides (YISYQSLKR... in RALF1 seq) is directly involved in the interaction with LLGs. So, a mutation in that R may impair the interaction of RALF1 with LLG and therefore the complex formation with FER. So, it is well possible that the effect that the authors are seeing on FER signaling and endocytosis, using this peptide variant, may not be due to the impaired capacity of the peptide to bind deesterified pectin but to not be able to be sensed by the membrane complex directly. To verify that the authors should test, either biochemically or by CoIP in planta, that their RALF1KR variant can still be perceived by the LLG-FER complex. 

We agree with reviewer #1 and do not doubt that the positive charges in RALF1 likely interact with several entities. The respective sites were also covered in Liu et al., 2024 (Cell). It would be interesting to understand how the charge-dependent interaction with pectin modulates the RALF binding to the LLG-FER complex, but these experiments are beyond the scope of this manuscript. We confirmed that the negative charges in RALF1 are essential for OG binding as well as for its bioactivity. We however do not rule out that they bear additional structural functions beyond pectin binding. We clarified this aspect in the revised version. It is conceivable that the pectin and receptor complex binding of RALF1 is molecularly and mechanistically related. 

(f) The authors propose in this study that this effect of RALF1-pectin mode of action on FER is independent from LRXs. That is a very interesting observation which also aligns with similar observations from other independent studies (Moussu et al., 2020; Schoenaers et al. Nat Plants, 2024; Franck et al., 2018). However, that seems to be in conflict with the previous mode of action that the authors had described in Dunser et al., 2019. In that last study the authors had described that FER constitutively interacts with LRX proteins in a direct way to sense cell wall changes. In my view the authors do not critically elaborate to explain these two contradicting results, which are key to understand the mode of action they are describing. This relevant aspect should be addressed more in depth by the authors in their discussion.

Thank you for the comment. We do not see that our findings contradict our previous work (from Dünser et al., 2019). There we concluded that LRX and FER directly interact to sense cell wall characteristics. However, the loss of LRX function abolished the cell wall sensing mechanism, but the respective loss-of-function and dominant negative lines were still able to detect RALF peptides. We hence proposed that the LRX/FER function is at least partially independent of the FER function in RALF perception. This is in agreement with our current study where we conclude again that FER shows LRX-dependent but also -independent modes of action. 

Minor

(g) In the introduction (first page), the authors write the following sentence: "RALF peptides are involved in multiple physiological and developmental processes, ranging from organ growth and pollen tube guidance to modulation of immune responses (Stegmann et al., 2017; Abarca et al., 2021)". RALFs are not involved in pollen tube guidance but pollen tube growth.

So, that should be changed in the Introduction sentence. Also, in addition, the authors could cite additional references here to support the sentence such as Mecchia et al., 2017 or Ge et al. , 2017, in addition. 

Thank you for pointing this out and we apologize for our flaw. We corrected the statement in the revised version of the manuscript and added the citations as requested.

(h) The new study of Schoenaers et al. Nat Plants, 2024 should now be included in the revised version.

Thank you. We implemented this reference in the revised manuscript.

Reviewer #2 (Public Review):

The genetic material used by the authors to strengthen the connection of RALF signalling and

PME activity might not be as suitable as an acute inhibition of PME activity.  The PMEI3ox line generated by Peaucelle et al., 2008 is alcohol-inducible. Was expression of the PMEI induced during the experiments? As ethanol inducible systems can be rather leaky, it would not be surprising if PME activity would be reduced even without induction, but maybe this would warrant testing whether PMEI3 is actually overexpressed and/or whether PME activity is decreased. On a similar note, the PMEI5ox plants do not appear to show the typical phenotype described for this line. I personally don't think these lines are necessary to support the study. Short-term interference with PME activity (such as with EGCG) might be more meaningful than life-long PMEI overexpression, in light of the numerous feedback pathways and their associated potential secondary effects. This might also explain why EGCG leads to an increase in pH, as one would expect from decreased PME activity, while PMEI expression (caveats from above apply) apparently does not (Fig 3A-D).

We agree with reviewer #2. The PMEI3ox line from Peaucelle et al., 2008 is ethanolinducible, but we observed a strong phenotype (at seedling and adult stage) without ethanol induction. We performed all experiments (root growth assays and confocal observations) with as well as without induction using ethanol, leading to similar results. We concluded from that, that the line is either leaky or that overexpression of PMEI3 is already induced upon seed sterilisation with ethanol. Accordingly, we did not intend to use the lines as acute inhibition of PME but rather used the lines to genetically confirm our data derived from acute pharmacological inhibition. We do show in Figure 1G that the levels of de-methylesterified pectin is decreased in the PMEI3ox mutant compared to WT seedlings. It is exactly this alteration that we are exploiting to assess the necessity of charged pectin for RALF1 signalling. Since the apoplastic pH in the PMEI3ox line is not altered compared to WT, we can conclude that the observed effect on RALF1 signalling is entirely due to the altered pectin charge.

We would like to note that the PMEI5ox line indeed shows the reported root-bending phenotype when grown on plates. We started to perform RALF application assays in liquid medium, because EGCG does not show activity on MS plates. Moreover, it allows us to perform the assays with low amounts of synthetic peptides. The seedling images in our root growth assay might be hence misleading since the assay was done in liquid MS medium and the seedlings were carefully straightened on MS plates before imaging. This transfer makes it difficult to observe the root-bending or -curling phenotype, which is typical for PMEI5ox. 

At least at first sight, the observation that OGs are able to titrate RALF from pectin binding seems at odds with the idea of cooperative binding with low affinity, leading to high avidity oligomers. Perhaps the can provide a speculative conceptual model of these interactions?

We added a high concentration of OGs in the media and observed a strong repression of RALF1 activity at the root surface. We assume the OGs form oligomers with RALF peptides in the media, preventing them from penetrating the roots.

I could not find a description of the OG treatment/titration experiments, but I think it would be important to understand how these were performed with respect to OG concentration, timing of the application, etc.

Thank you for pointing this out. The description of the OG RALF titration is added in the methods section.

Reviewer #2 (Recommendations for the Authors):

Page 3: „and can bind to extracellular pectin" Liu et al, 2024 should maybe also be cited here. 

Amended.

I am not so sure about the use of "conceptualizing" in the last sentence of the abstract and elsewhere in the manuscript.

I would suggest adding a few sentences that describe and differentiate what this study and other recently published works (e.g. Dünser, Liu, Mossou, Lin) have revealed about the pectin association of RALFs, LRXs, and FER to help the non-expert reader to navigate this increasingly complex area. May also be worth mentioning that the previously described pectin sensing function of FER is physically separated from the RALF binding domain (Gronnier et al., 2022)

Thank you for your constructive comments. We followed your suggestions and further improved the discussion in the revised version of our manuscript.

Reviewer #3 (Recommendations for the Authors): 

(1) The authors claim that pectin is something like an extracellular signaling scaffold. In other fields, signalling scaffold refers to proteins that tether the signalling components and regulate/are involved in the signal transduction. Here, pectin is a cell wall structural component whose molecular status is sensed and perceived rather than a functional signaling component. To me, it is FERONIA to be called a signalling scaffold in this case. However, this is my view, and the authors may present their concept. 

RALF peptides as well as FERONIA bind to de-methylesterified pectin, which is essential for its signalling output. Albeit not being a protein, we propose that pectin functions like a scaffold tethering both signalling components and thereby enabling signalling. FERONIA has been indeed also proposed to function as a scaffold when tethering other signalling components.

(2) I have no problem with authors using the more general term pectin instead of homogalacturonan throughout the text. Still, authors should, at some point in the text, specify that by pectin, they mean homogalacturonan; the authors did not analyze other pectic types on binding. 

We followed your suggestion.

(3) The authors show that RALF1 binds to OGs with a high avidity. Given the fact that OGs released from homogalacturonan upon pathogen infection are Damage-Associated Molecular Patterns (DAMPs), this opens the possibility that this particular activity of RALF1 might actually function in modulation of immune response. I suggest that authors should not exclude this possibility. 

We fully agree to this possibility for FER-dependent signalling.

(4) Are there any indications that a similar mechanism can be extrapolated to other FERONIA homologs, such as THESEUS or HERCULES? Although it is not essential to comment, I think this could enrich the discussion.

This is a highly interesting research question, which we may follow up in our upcoming studies. RALF34, which is considered a ligand for THESEUS, also induced FER internalization, which was also sensitive to PME inhibition. While this requires further investigation, this finding hints at a common mechanism for FER- and THE-dependent RALF peptides.

(5) I suggest using the model scheme currently in the supplement as a main figure to provide an immediate accessible summary of the findings.

Thank you for the suggestion to add the summary scheme in the main figures. We followed your suggestion.

eLife assessment

This valuable study provides new insight into the disassembly of vimentin filaments and the dependence of this mechanism on net charge, albeit with incomplete evidence. In particular, the experimental replicates are limited (in most cases n=1), there is a lack of quantitative analysis to substantiate claims, inconsistency of the proposed mechanisms with previously published work, and lack of biochemical evidence supporting the observations in cells. Addressing these concerns would strengthen the manuscript and help support the proposed hypothesis on vimentin disassembly.

Reviewer #1 (Public review):

Summary:

The authors investigate the mechanism behind the widely observed but poorly understood phenomenon of reversible vimentin disassembly upon hypotonic challenge. Using permeabilized COS-7 cells expressing vimentin-mEos3.2, the authors demonstrate that vimentin disassembly is not due to lower osmotic pressure but rather due to decreased intracellular ionic strength. They propose a model in which vimentin filament stability is predicted by the protein's net charge and support this idea through approaches that involve (i) manipulating buffer ionic strength, (ii) manipulating buffer pH, or (iii) introducing charged amino acids into the linker of the exogenously expressed vimentin-mEos3.2.

Strengths & Weaknesses:

While the discovery is intriguing and presents an interesting concept, significant shortcomings in experimental design and numerous inconsistencies prevent it from reaching the high standards expected. The lack of reproducibility, inadequate controls, and insufficient quantification make the findings feel very preliminary. Additionally, the authors need to address the apparent discrepancies between their current results and their previous work implicating calpains and altered calcium levels in vimentin disassembly upon hypotonic challenge (which has led to much confusion in the field). This discrepancy should be thoroughly addressed in the discussion with the authors citing their prior work and explaining why it was incorrect.

An additional concern is the relevance of the findings to vimentin biology inside cells. The most important insight in this work is the observation that an isotonic buffer balanced with non-electrolytes (glucose or sorbitol) is sufficient to drive vimentin disassembly. The authors show that vimentin disassembly is not due to changes in osmotic pressure but rather due to a change in the concentration of critical dissolved ions, specifically the number of charged states on vimentin. What is missing is when and how this is controlled within cells under physiological conditions - not just when cells are permeabilized with detergents (conditions that cells rarely survive). Without this deeper dive into vimentin states within cells and how it is controlled, the paper seems very narrow in its focus.

Reviewer #2 (Public review):

The reviewed manuscript "Hypersensitivity of the vimentin cytoskeleton to net-charge states and Coulomb repulsion" presents exciting results on the mechanisms governing the assembly and disassembly of the vimentin cytoskeleton. They show, using live-cell imaging, that changes in the intracellular ionic strength induce rapid and dramatic changes to the integrity of the vimentin cytoskeleton. Interestingly, mutants of vimentin with net positive or negative charges display notably different responses to hypotonic stress (and thus changes to the intracellular ionic strength). Even more interesting, the ionic strength-driven mechanism seems to generalize to the several other intermediate filaments explored here. These results are of high interest to the broader cytoskeleton field. A major caveat is that essentially every experiment in the paper is n=1, showing example images of a single cell. The experiments were not repeated, and the results were not quantified. Purported differences between experimental variables/conditions lack statistical significance. Generalization of the ionic strength-based mechanism is hindered by the fact that only one cell type was tested for each cytoskeletal protein. Another caveat is that the fluorescently tagged vimentin used thoroughly in this work is exogenous and overexpressed; it is unclear if the observed effects would also occur at endogenous concentrations of vimentin. As it is currently presented, it is my opinion that all four main figures in this work - although interesting and quite likely correct - should be interpreted as preliminary data by readers.

Reviewer #3 (Public review):

Summary:

This report analyzes the structure of vimentin, GFAP, and keratin intermediate filament networks in cells that have been subjected to hypotonic stress and other treatments that either alter the ionic strength of the cytoplasm or change the charge density of the intermediate filament.

Strengths:

These experiments expand on the work of references 8 and 9, which showed that the vimentin network rapidly dissociates after hypotonic shock. The cellular imaging uses sophisticated super-resolution techniques and produces some striking images.

Weaknesses:

A fundamental weakness of this study lies in the interpretation, and lack of biochemical evidence for the provocative hypothesis raised that the assembly state of intermediate filaments in a cell is governed by coulomb interactions between charged filaments. Several essential experiments need to be done before this striking hypothesis can be plausibly supported.

(1) First the assembly and disassembly of vimentin filaments needs to be done in vitro systems. If the hypothesis is correct then these same effects will happen with purified vimentin intermediate filaments. These proteins are readily purified from bacterial expression systems and there's a wealth of biophysical data on them, so verifying the predictions of this cell-based model can be realistically tested with purified proteins.

(2) Interpretation of these results explicitly avoids a role for post-translational modifications of vimentin, which are well described and related to filament assembly state. For example, reference 9, which is one of the first discoveries that the vimentin network dissociates under osmotic stress explicitly shows that the vimentin network remains intact if either calcium influx or calpain proteolytic activity is prevented. Hypoosmotic stress will still lead to the same dilution, but the filaments remain intact, apparently contradicting the major interpretation of this paper.

(3) The third issue is that the role of polyelectrolyte effects and especially the importance of divalent cations is scarcely mentioned in the interpretation. There are many studies of how strongly vimentin intermediate filaments interact with divalent cations, in a manner that is relatively insensitive to ionic strength, and these effects need to be taken into account to interpret the cellular data or the hypothesis that coulomb repulsion is the major driver for vimentin disassembly.

eLife assessment

The fundamental study by Ding and colleagues identifies subpopulations of neurons recorded in the monkey subthalamic nucleus (STN) with distinct activity profiles and causal contributions during perceptual decision-making. The combination of neuronal recording, microstimulation, and computational methods provides convincing evidence for a heterogenous neural population that could support multifaceted roles in decision formation. This study should be of wide interest to computational and experimental neuroscientists interested in cognitive function.

Reviewer #2 (Public review):

This study uses single-unit recordings in the monkey STN to examine the evidence for three theoretical models that propose distinct roles for the STN in perceptual decision-making. Importantly, the proposed functional roles are predictive of unique patterns of neural activity. Using k-means clustering with seeds informed by each model's predictions, the current study identified three neural clusters with activity dynamics that resembled those predicted by the described theoretical models. The authors are thorough and transparent in reporting the analyses used to validate the clustering procedure and the stability of the clustering results. To further establish a causal role for the STN in decision-making, the researchers applied microsimulation to the STN and found effects on response times, choice preferences, and latent decision parameters estimated with a drift-diffusion model. Overall, the study provides strong evidence for a functionally diverse population of STN neurons that could indeed support multiple roles involved in perceptual decision-making. The manuscript would benefit from stronger evidence linking each neural cluster to specific decision roles in order to strengthen the overall conclusions.

The interpretation of the results, and specifically, the degree to which the identified clusters support each model, is largely dependent on whether the artificial vectors used as model-based clustering seeds adequately capture the expected behavior under each theoretical model. The manuscript would benefit from providing further justification for the specific model predictions summarized in Figure 1B. Further, although each cluster's activity can be described in the context of the discussed models, these same neural dynamics could also reflect other processes not specific to the models. That is, while a model attributing the STN's role to assessing evidence accumulation may predict a ramping up of neural activity, activity ramping is not a selective correlate of evidence accumulation and could be indicative of a number of processes, e.g., uncertainty, the passage of time, etc.. This lack of specificity makes it challenging to infer the functional relevance of cluster activity and should be acknowledged in the discussion.

Additionally, although the effects of STN microstimulation on behavior provide important causal evidence linking the STN to decision processes, the stimulation results are highly variable and difficult to interpret. The authors provide a reasonable explanation for the variability, showing that neurons from unique clusters are anatomically intermingled such that stimulation likely affects neurons across several clusters. It is worth noting, however, that a substantial body of literature suggests that neural populations in the STN are topographically organized in a manner that is crucial for its role in action selection, providing "channels" that guide action execution. The authors should comment on how the current results, indicative of little anatomical clustering amongst the functional clusters, relates to other reports showing topographical organization.

Overall, the association between the identified clusters and the function ascribed to the STN by each of the models is largely descriptive and should be interpreted accordingly. For example, Figure 3 is referenced when describing which cluster activity is choice/coherence dependent, yet it is unclear what specific criteria and measures are being used to determine whether activity is choice/coherence "dependent." Visually, coherence activity seems to largely overlap in panel B (top row). Is there a statistically significant distinction between low and high coherence in this plot? The interpretation of these plots and the methods used to determine choice/coherence "dependence" needs further explanation.

In general, the association between cluster activity and each model could be more directly tested. At least two of the models assume coordination with other brain regions. Does the current dataset include recordings from any of these regions (e.g., mPFC or GPe) that could be used to bolster claims about the functional relevance of specific subpopulations? For example, one would expect coordinated activity between neural activity in mPFC and Cluster 2 according to the Ratcliff and Frank model. Additionally, the reported drift-diffusion model (DDM) results are difficult to interpret as microsimulation appears to have broad and varied effects across almost all the DDM model parameters. The DDM framework could, however, be used to more specifically test the relationships between each neural cluster and specific decision functions described in each model. Several studies have successfully shown that neural activity tracks specific latent decision parameters estimated by the DDM by including neural activity as a predictor in the model. Using this approach, the current study could examine whether each cluster's activity is predictive of specific decision parameters (e.g., evidence accumulation, decision thresholds, etc.). For example, according to the Ratcliff and Frank model, activity in cluster 2 might track decisions thresholds.

Review of revision

The authors have sufficiently addressed the concerns raised in the initial reviews and have revised their manuscript accordingly. We commend the authors for these efforts and feel that the revisions have strengthened the major claims of the manuscript.

Reviewer #3 (Public review):

Summary:

The authors provide compelling evidence for the causal role of the subthalamic nucleus (STN) in perceptual decision-making. By recording from a large number of STN neurons and using microstimulation, they demonstrate the STN's involvement in setting decision bounds, scaling evidence accumulation, and modulating non-decision time.

Strengths:

The study tested three hypotheses about the STN's function and identified distinct STN subpopulations whose activity patterns support predictions from previous computational models. The experiments are well-designed, the analyses are rigorous, and the results significantly advance our understanding of the STN's multi-faceted role in decision formation.

Weaknesses:

While the study provides valuable insights into the STN's role in decision-making, there are a few areas that could be improved. First, the interpretation of the neural subpopulations' activity patterns in relation to the computational models should be clarified, as the observed patterns may not directly correspond to the specific signals predicted by the models. Second, a neural population model could be employed to better understand how the STN population jointly contributes to decision-making dynamics.

Author response:

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

Review #1:

(1) It would be helpful to explain the criteria for choosing a given number of clusters and for accepting the final clustering solution more clearly. The quantitative results (silhouette plots, Rand index) in Supplementary Figure 2 should perhaps be included in the main figure to justify the parameter choices and acceptance of specific clustering solutions.

We revised the text and added labels to the original Supplementary Figure 2 (now main Figure 4) to clarify how we arrived at the best settings for random-seed clustering. 

(2) It would be helpful to show how the activity profiles in Figure 3 would look like for 3 or 5 (or 6) clusters, to give the reader an impression of how activity profiles recovered using different numbers of clusters would differ.

We added a new figure (Supplementary Figure 4) that shows 5- and 6-cluster results. Note that the same three subpopulations in Figure 3 were reliably identified as distinct clusters even with alternative settings, corroborating the results in the tSNE space (Supplementary Figure 3). 

(3) The authors attempt to link the microstimulation effects to the presence of functional neuron clusters at the stimulation site. How can you rule out that there were other, session-specific factors (e.g., related to the animal's motivation) that affected both neuronal activity and behavior? For example, could you incorporate aspects of the monkey's baseline performance (mean reaction time, fixation breaks, error trials) into the analysis?

We tested the potential influences of monkeys’ motivational states on our observations using two sets of analysis. First, we examined whether motivational state modulated the likelihood of observing a specific type of neural activity in STN. We focused on three measurements of motivational states: the rate of fixation break, the overall error rate, and mean RT. We found that none of these measurements differed significantly among sessions when we encountered different subpopulations (new Supplemental Figure 7), suggesting that motivational state alone cannot explain the differences in activity patterns of the four subpopulations. 

Second, we examined how motivational state may be reflected in the microstimulation results. To clarify, because we interleaved trials with and without microstimulation, the microstimulation effects cannot be solely explained by session-specific factors. However, it is possible that motivational state can modulate the magnitude of microstimulation effects. We performed correlation analysis between microstimulation effects (difference in each fitted DDM parameter between trials with and without microstimulation) and motivational state (fixation break, error rate, mean RT on trials without microstimulation). We did not find significant correlation for any combination (Supplemental Table 1). These results suggest that the motivational state of the monkey had little influence on our recording and microstimulation results. However, because our monkeys operated within a narrow range of strong engagement on the task, we cannot rule out the possibility that STN activity or microstimulation effects could change significantly if the monkeys were not as engaged. We have added these results in a new section titled “Heterogeneous activity patterns and microstimulation effects cannot be explained by variations in motivational state”. 

(4) Line 84: What was the rationale for not including both coherence and reaction time in one multiple regression model?

On the task we used, RT depends strongly on coherence in a nonlinear fashion (e.g., example behavior in now Figure 5). We thus performed regressions using coherence and RT separately. We revised the text in Methods to clarify our rationale (lines 470-473):

“To quantitatively measure each neuron’s task-related modulation, we performed two multiple linear regressions for each running window, separately for coherence and RT because monkeys’ RT strongly depends on coherence on our task:”

Review #2:

The interpretation of the results, and specifically, the degree to which the identified clusters support each model, is largely dependent on whether the artificial vectors used as model-based clustering seeds adequately capture the expected behavior under each theoretical model. The manuscript would benefit from providing further justification for the specific model predictions summarized in Figure 1B.

We added information on the original figure/equations that were the basis of the artificial vectors we constructed for clustering analysis and their abbreviated summary in Figure 1B (first paragraph in section “STN subpopulations can support previously theorized functions”). These vectors were meant to capture prominent features of the predicted activity patterns, in the forms of choice, time, and motion strength dependencies. We also emphasize that we obtained very similar results using random clustering seeds.

Further, although each cluster's activity can be described in the context of the discussed models, these same neural dynamics could also reflect other processes not specific to the models. That is, while a model attributing the STN's role to assessing evidence accumulation may predict a ramping up of neural activity, activity ramping is not a selective correlate of evidence accumulation and could be indicative of a number of processes, e.g., uncertainty, the passage of time, etc. This lack of specificity makes it challenging to infer the functional relevance of cluster activity and should be acknowledged in the discussion.

We thank the reviewer for pointing out the alternative interpretation of these modulation patterns. We have added this caveat in the Discussion (lines 398-401): “It is also possible that the ramping activity reflects alternative roles for the STN in the evaluation of the decision process, the tracking of elapsed time, or both. How these possible roles relate to those of caudate neurons awaits further investigation (Fan et al., 2024)”. 

Additionally, although the effects of STN microstimulation on behavior provide important causal evidence linking the STN to decision processes, the stimulation results are highly variable and difficult to interpret. The authors provide a reasonable explanation for the variability, showing that neurons from unique clusters are anatomically intermingled such that stimulation likely affects neurons across several clusters. It is worth noting, however, that a substantial body of literature suggests that neural populations in the STN are topographically organized in a manner that is crucial for its role in action selection, providing "channels" that guide action execution. The authors should comment on how the current results, indicative of little anatomical clustering amongst the functional clusters, relate to other reports showing topographical organization.

We thank the reviewer for raising this important point. We have added the following text in the Discussion:

“The intermingled subpopulations may appear at odds with the conventional idea of topography in how the STN is organized. For example, the “tripartite model” suggests that STN is segregated by motor, associative, and limbic functions (Parent and Hazrati, 1995); afferents from motor cortices and neurons related to different types of movements are largely somatotopically organized in the STN (DeLong et al., 1985; Nambu et al., 1996); and certain molecular markers are expressed in an orderly pattern in the STN (reviewed in Prasad and Wallén-Mackenzie, 2024). Because we focused on STN neurons that were responsive on a single oculomotor decision task, our sampling was likely biased toward STN subdivisions related to associative function and oculomotor movements. As such, our results do not preclude the presence of topography at a larger scale. Rather, our results underscore the importance of activity patternbased analysis, in addition to anatomy-based analysis, for understanding the functional organization of the STN.”

Figure 3 is referenced when describing which cluster activity is choice/coherence dependent, yet it is unclear what specific criteria and measures are being used to determine whether activity is choice/coherence "dependent." Visually, coherence activity seems to largely overlap in panel B (top row). Is there a statistically significant distinction between low and high coherence in this plot? The interpretation of these plots and the methods used to determine choice/coherence "dependence" needs further explanation.

We added a new figure (Sup Figure 3) that shows the summary of choice and coherence modulation, based on multiple linear regression analysis, for each subpopulation separately. We also updated the description of these activity patterns in Results (lines 122-130):

In general, the association between cluster activity and each model could be more directly tested. At least two of the models assume coordination with other brain regions. Does the current dataset include recordings from any of these regions (e.g., mPFC or GPe) that could be used to bolster claims about the functional relevance of specific subpopulations? For example, one would expect coordinated activity between neural activity in mPFC and Cluster 2 according to the Ratcliff and Frank model.

We agree completely that simultaneous recordings of STN and its afferent/efferent regions (such as mPFC, GPe, SNr, and GPi) would provide valuable insights into the specific roles of STN and the basal ganglia as a whole. Such recordings are outside the scope of the current study but are in our future plans. 

Additionally, the reported drift-diffusion model (DDM) results are difficult to interpret as microstimulation appears to have broad and varied effects across almost all the DDM model parameters. The DDM framework could, however, be used to more specifically test the relationships between each neural cluster and specific decision functions described in each model. Several studies have successfully shown that neural activity tracks specific latent decision parameters estimated by the DDM by including neural activity as a predictor in the model. Using this approach, the current study could examine whether each cluster's activity is predictive of specific decision parameters (e.g., evidence accumulation, decision thresholds, etc.). For example, according to the Ratcliff and Frank model, activity in cluster 2 might track decision thresholds.

We thank the reviewer for the suggested analysis. Because including the neural activity in the model substantially increases model fitting time, we performed a preliminary round of model fitting for 15 neurons (5 neurons closest to each of the cluster centroids). For each neuron, we measured the average firing rates in three windows: 1) a 350 ms window starting from dots onset (“Dots”), 2) a 350 ms window ending at saccade onset (“Presac”), and 3) a variable window starting from dots onset and ending at 100 ms before saccade onset (“Fullview”). For each window, the firing rates were z-scored across trials.  We incorporated the firing rates into two model types. In the “DV” type, the firing rates were assumed to influence three DDM parameters related to evidence accumulation: k, me, and z. In the “Bound” type, the firing rates were assumed to influence three DDM parameters related to decision bound: a, B_alpha, and B_d. In total, we fitted six combinations of firing rates and model types to each neuron. For comparison, we also fitted the standard model without incorporating firing rates. 

As shown in Author response image 1, firing rates of single STN neurons had minimal contributions to the fits. With the exception of one neuron, AIC values were greater for model variants including firing rates than the standard model (Author response image 1A), indicating that including firing rate did not improve the fits. For all neurons, the actual fitted coefficients for firing rates were several degrees of magnitude smaller than the corresponding DDM parameter (Author response image 1B; note the range of y axis), indicating that the trial-by-trial variation in firing rate had little influence on the evidence accumulation- or decision bound-related parameters. Based on these preliminary fitting results, we believe that a single STN neuron does not have strong enough influence on the overall evidence accumulation or decision bound to be detected with the model fitting method.  We therefore did not expand the fitting analysis to all neurons. 

Author response image 1.

Firing rates of a single STN neuron did not substantially influence decision-related DDM parameters. A, Differences in AIC between DDM variants that included firing rate-dependent terms and the standard DDM. Red dahsed line: difference = -3. Each column represents results from one unit. B, Fitted coefficients for firing rate-related terms were near zero. Note the range of y axis. Values for the top and bottomw panels were obtained from "DV"- and "Bound"-type models, respectively. See text for more details.

We emphasize, however, that the apparent negative results do not necessarily argue against a causal role of the STN in decision making, rather, these results more likely reflect the methodological limitation: because we used a single task context, the monkeys’ natural trial-by- trial variations in the DDM components may be too small. A better design would be to manipulate task contexts to induce larger changes in evidence accumulation or decision bounds and then test for a correlation between single-neuron firing rates and these changes. We are currently using such a design in a follow-up study. 

The table in Figure 1B nicely outlines the specific neural predictions for each theoretical model but it would help guide the reader if the heading for each column also included a few summary words to remind the reader of the crux of each theory, e.g. "Ratcliff+Frank 2012 (adjusted decision-bounds)"

We thank the reviewer for this suggestion. We considered implementing this but eventually decided not to add more headings to the column, because the predicted STN functions of the three models cannot all be succinctly summarized. We thus prefer to include more detailed descriptions in the main text, instead of in the figure. 

The authors frequently refer to contralateral vs. ipsilateral decisions but never explicitly state what this refers to, i.e. contralateral relative to what (visual field, target direction, recording site, etc.)? The reader can eventually deduce that this means contralateral to the recording site but this should be explicitly stated for clarity.

We added in Methods: 

Line 483: “Contralateral/ipsilateral choices refer to saccades toward the targets contralateral/ipsilateral to the recording sites, respectively.” 

Line 535: Contralateral/ipsilateral choices refer to saccades toward the targets contralateral/ipsilateral to the microstimulation sites, respectively.”

Again, for clarity, it would be helpful to explicitly define what the authors mean by "sensitive to choice" when referring to Figure 1B as this could be interpreted to mean left/right or ipsilateral/contralateral.

In the context of Figure 1B, “sensitive to choice” means showing different responses for the two choices in our 2AFC task, regardless of the task geometry. We added explanation in the figure caption.

Color bar labels would be helpful to include in all figures that include plots with color bars.

We apologize for omitting the labels. They are added to Figure 2B and C, Supplemental Fig. 1.  

The authors should briefly note what a "lapse term" is when describing the logistic function results.

We revised the text in Results (lines 184-186) and Methods (line 527) to clarify that lapse terms were used to capture errors independent of motion strength.

Are the 3 example sessions in Figure 4 stimulating the same STN site and/or the same monkey? This information should be noted in the caption or main text.

We revised the caption: “A-C, Monkey’s choice (top) and RT (bottom) performance for trials with (red) and without (black) microstimulation for three example sessions (A,B: two sites in monkey C; C: monkey F).”

Figure 3B the authors note that "the last cluster shows little task-related modulation" - what criteria are they using to make this conclusion? By eye, the last cluster and cluster 1 seem to show a similar degree of modulation when locked to motion onset.

We added a new figure (Suppl Figure 2) that shows the summary of choice and coherence modulation, based on multiple linear regression analysis, for each subpopulation separately. 

Reviewer #3:

We have grouped the reviewer’s public and specific comments by content. 

First, the interpretation of the neural subpopulations' activity patterns in relation to the computational models should be clarified, as the observed patterns may not directly correspond to the specific signals predicted by the models. The authors claim that the first subpopulation of STN neurons reflects the normalization signal predicted by the model of Bogacz and Gurney (2007). However, the observed activity patterns only show choice- and coherence-dependent activity, which may represent the input to the normalization computation rather than its output. The authors should clarify this point and discuss the limitations of their interpretation. 

We agree with the reviewer that the choice- and coherence-dependent activity pattern does not sufficiently indicate a normalization computation. We interpreted such activity as satisfying a necessary condition for, and therefore consistent with, the theoretical model proposed by Bogacz and Gurney. We have reviewed the text to ensure that we never made the claim that the first subpopulation mediates the normalization.   

Second, the authors could consider using a supervised learning method to more explicitly model the pattern correlations between the three profiles. The authors used k-means clustering to identify STN subpopulations. Given the clear distinction between the three types of neural firing patterns, a supervised learning method (e.g., a generalized linear model) could be used as a more explicit encoding model to account for the pattern correlations between the three profiles.

We used two approaches to examine the different response profiles. The “random-seed” approach used non-supervised clustering to probe the functional organization of STN neurons, with no a priori assumption about how many subpopulations may be present. The “model-seed” approach is similar in spirit to what the reviewer suggested: we defined artificial vectors, akin to regressors in a generalized linear model, that showed key modulation features as predicted by previous theoretical models. We then projected the neurons’ activity profiles onto these vectors, akin to performing a regression analysis.   

Third, a neural population model could be employed to better understand how the STN population jointly contributes to decision-making dynamics. The single-neuron encoding analysis reveals mixed effects from multiple decision-related functions. To better understand how the STN population jointly contributes to the decision-making process, the authors could consider using a neural population model (e.g., Wang et al., 2023) to quantify the population dynamics.

We agree with the reviewer that a neural population model would be helpful for testing our understanding of the roles of STN. However, we believe that this is premature at the moment because we have no knowledge about how these different subpopulations interact with each other within STN, nor how they interact with other basal ganglia nuclei. We hope our results provide a foundation for future experiments that can provide more specific insights in the roles of each subpopulation, which can then be tested in a neural population model as the reviewer suggested.  

Finally, the added value of the microstimulation experiments should be more directly addressed in the Results section, as the changes in firing patterns compared to the original patterns are not clearly evident. The microstimulation results (Figure 7A) do not show significant changes in firing patterns compared to the original patterns (Figure 3B). As microstimulation is used to identify the hypothetical role of the STN beyond the correlational analysis, the authors should more directly address the added value of these experiments in the Results section.

We apologize for the confusion. The average firing rates at the top of original Figure 7A (now Figure 8A) were obtained in recordings just before microstimulation, to document which neuron subpopulation was near the stimulation electrode. We were not able to obtain recordings from the same neurons during microstimulation.  

The ordering of the three hypotheses in the Introduction (1) adjusting decision bounds, (2) computing a normalization signal, (3) implementing a nonlinear computation to improve decision bound adjustment, is inconsistent with the order in which they are addressed in the Results section (2, 1, 3). To improve clarity and readability, the authors should consider presenting the hypotheses and their corresponding results in a consistent order throughout the manuscript.

We thank the reviewer for this suggestion. We have reordered the text in Introduction to be consistent.

Reviewer #3 (Public Review):

Summary:

The manuscript by Hallam et al describes the analysis of various biomarkers in patients undergoing complement factor I supplementation treatment (PPY988 gene therapy) as part of the FOCUS Phase I/II clinical trial. The authors used validated methods (multiplexed assays and OLINK proteomics) for measuring multiple soluble complement proteins in the aqueous humour (AH) and vitreous humour (VH) of 28 patients over a series of time points, up to and including 96 weeks. Based on biomarker comparisons, the levels of FI synthesised by PPY988 were believed to be insufficient to achieve the desired level of complement inhibition. Subsequent comparative experiments showed that PPY988-delivered FI was much less efficacious than Pegceptacoplan (FDA-approved complement inhibitor under the name SYFORVE) when tested in an artificial VH matrix.

Strengths:

The manuscript is well written with data clearly presented and appropriate statistics used for the analysis itself. It's great to see data from real clinical samples that can help support future studies and therapeutic design. The identification that complement biomarker levels present in the AH do not represent the levels found in the VH is an important finding for the field, given the number of complement-targeting therapies in development and the desperate need for good biomarkers for target engagement. This study also provides a wealth of baseline complement protein measurements in both human AH and VH (and companion measurements in plasma) that will prove useful for future studies.

Weaknesses:

Perhaps the conclusions drawn regarding the lack of observed efficacy are not fully justified. The authors focus on the hypothesis that not enough FI was synthesised in these patients receiving the PPY988 gene therapy, suggesting a delivery/transduction/expression issue. But beyond rare CFI genetic variants, most genetic associations with AMD imply that it is a FI-cofactor disease. A hypothesis supported by the authors' own experiments when they supplement their artificial VH matrix with FH and achieve a significantly greater breakdown of C3b than achieved with PPY988 treatment alone. Justification around doubling FI levels driving complement turnover refers to studies conducted in blood, which has an entirely different complement protein profile than VH. In Supplemental Table 5 we see there is approx. 10-fold more FH than FI (533ug/ml vs 50ug/ml respectively) so increasing FI levels will have a direct effect. Yet in Supplemental Table 3 we see there is more FI than FH in VH (608ng/ml vs 466ng/ml respectively). Therefore, adding more FI without more co-factors would have a very limited effect. Surely this demonstrates that the study was delivering the wrong payload, i.e. FI, which hit a natural ceiling of endogenous co-factors within the eye?

eLife assessment

This important work advances our understanding of factors influencing efficacy assessments and biomarker viability for complement-directed gene therapy against age-related macular degeneration. The data presented is convincing and offers insights and teachings for the design of gene therapy and complement-targeted therapeutics in the eye and more broadly for future ocular biomarker studies.

Reviewer #1 (Public Review):

Summary:

This study analyzed biomarker data from 28 subjects with geographic atrophy (GA) in a Phase I/II clinical trial of PPY988, a subretinal AAV2 complement factor I (CFI) gene therapy, to evaluate pharmacokinetics and pharmacodynamics. Post-treatment, a 2-fold increase in the vitreous humor (VH) FI was observed, correlating with a reduction in FB breakdown product Ba but minimal changes in other complement factors. The aqueous humor (AH) was found to be an unreliable proxy for VH in assessing complement activation. In vitro assays showed that the increase in FI had a minor effect on the complement amplification loop compared to the more potent C3 inhibitor pegcetacoplan. These findings suggest that PPY988 may not provide enough FI protein to effectively modulate complement activation and slow GA progression, highlighting the need for a thorough biomarker review to determine optimal dosing in future studies.

Strengths:

This manuscript provides critical data on the efficacy of gene therapy for the eye, specifically introducing complement FI expression. It presents the results from a halted clinical trial, making sharing this data essential for understanding the outcomes of this gene therapy approach. The findings offer valuable insights and lessons for future gene therapy attempts in similar contexts.

Weaknesses:

No particular weaknesses. The study was carefully performed and limitations are discussed.

I have just some concerns about the methodology used. The authors use the MILLIPLEX assays, which allow for multiplexed detection of complement proteins and they mention extensive validation. How are the measurements with this assay correlating with gold standard methods? Is the specificity and the expected normal ranges preserved with this assay? This also stands for the Olink assay. Some of the proteins are measured by both assay and/or by standard ELISA. How do these measurements correlate?

Reviewer #2 (Public Review):

Summary:

The results presented demonstrate that AAV2-CFI gene therapy delivers long-term and marginally higher FI protein in vitreous humor that results in a concomitant reduction in the FB activation product Ba. However, the lack of clinical efficacy in the phase I/II study, possibly due to lower in vitro potency when compared to currently approved pegcetacoplan, raises important considerations for the utility of this therapeutic approach. Despite the early termination of the PPY988 clinical development program, the study achieved significant milestones, including the implementation of subretinal gene therapy delivery in older adults, complement biomarker comparison between serial vitreous humor and aqueous humor samples and vitreous humor proteomic assessment via Olink.

Strengths:

Long-term augmentation of FI protein in vitreous humor over 96 weeks and reduction of FB breakdown product Ba in vitreous humor suggests modulation of the complement system. Developed a novel in vitro assay suggesting FI's ability to reduce C3 convertase activity is weaker than pegcetacoplan and FH and may suggest a higher dose of FI will be required for clinical efficacy. Warn of the poor correlation between vitreous humor and aqueous humor biomarkers and suggest aqueous humor may not be a reliable proxy for vitreous humor with regard to complement activation/inhibition studies.

Weaknesses:

The vitrectomy required for the subretinal route of administration causes a long-term loss of total protein and may influence the interpretation of complement biomarker results even with normalization. The modified in vitro assay of complement activation suggests a several hundred-fold increase in FI protein is required to significantly affect C3a levels. Interestingly, the in vitro assay demonstrates 100% inhibition of C3a with pegcetacoplan and FH therapeutics, but only a 50% reduction with FI even at the highest concentrations tested. This observation suggests FI may not be rate-limiting for negative complement regulation under the in vitro conditions tested and potentially in the eye. It is unclear if pharmacokinetic and pharmacodynamic properties in aqueous humor and vitreous humor compartments are reliable predictors of FI level/activity after subretinal delivery AAV2-CFI gene therapy.

Reviewer #1 (Public Review):

In this manuscript by Wu et al., the authors present the high resolution cryoEM structures of the WT Kv1.2 voltage-gated potassium channel. Along with this structure, the authors have solved several structures of mutants or experimental conditions relevant to the slow inactivation process that these channels undergo and which is not yet completely understood.

One of the main findings is the determination of the structure of a mutant (W366F) that is thought to correspond to the slow inactivated state. These experiments confirm results in similar mutants in different channels from Kv1.2 that indicate that inactivation is associated with an enlarged selectivity filter.

Another interesting structure is the complex of Kv1.2 with the pore blocking toxin Dendrotoxin 1. The results shown in the revised version indicate that the mechanism of block is similar to that of related blocking-toxins, in which a lysine residue penetrates in the pore. Surprisingly, in these new structures, the bound toxin results in a pore with empty external potassium binding sites.

The quality of the structural data presented in this revised manuscript is very high and allows for unambiguous assignment of side chains. The conclusions are supported by the data. This is an important contribution that should further our understanding of voltage-dependent potassium channel gating. In the revised version, the authors have addressed my previous specific comments.

Reviewer #2 (Public Review):

Cryo_EM structures of the Kv1.2 channel in the open, inactivated, toxin complex and in Na+ are reported. The structures of the open and inactivated channels are merely confirmatory of previous reports. The structures of the dendrotoxin bound Kv1.2 and the channel in Na+ are new findings that will of interest to the general channel community.

Review of the resubmission:

I thank the authors for making the changes in their manuscript as suggested in the previous review. The changes in the figures and the additions to the text do improve the manuscript. The new findings from a further analysis of the toxin channel complex are welcome information on the mode of the binding of dendrotoxin.

Reviewer #3 (Public Review):

Wu et al. present cryo-EM structures of the potassium channel Kv1.2 in open, C-type inactivated, toxin-blocked and presumably sodium-bound states at 3.2 Å, 2.5 Å, 2.8 Å, and 2.9 Å. The work builds on a large body of structural work on Kv1.2 and related voltage-gated potassium channels. The manuscript presents a plethora of structural work, and the authors are commended on the breadth of the studies. The structural studies are well-executed. Although the findings are mostly confirmatory, they do add to the body of work on this and related channels. Notably, the authors present structures of DTx-bound Kv1.2 and of Kv1.2 in a low concentration of potassium (which may contain sodium ions bound within the selectivity filter). These two structures add considerable new information. The DTx structure has been markedly improved in the revised version and the authors arrive at well-founded conclusions regarding its mechanism of block. Overall, the manuscript is well-written, a nice addition to the field, and a crowning achievement for the Sigworth lab.

Author response:

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

Public Reviews: 

Reviewer #1 (Public Review): 

In this manuscript by Wu et al., the authors present the high resolution cryoEM structures of the WT Kv1.2 voltagegated potassium channel. Along with this structure the authors have solved several structures of mutants or experimental conditions relevant to the slow inactivation process that these channels undergo and which is not yet completely understood. 

One of the main findings is the determination of the structure of a mutant (W366F) that is thought to correspond to the slow inactivated state. These experiments confirm results in similar mutants in different channels from Kv1.2 that indicate that inactivation is associated with an enlarged selectivity filter. 

Another interesting structure is the complex of Kv1.2 with the pore blocking toxin Dendrotoxin 1. The results shown in the revised version indicate that the mechanism of block is similar to that of related blocking-toxins, in which a lysine residue penetrates in the pore. Surprisingly, in these new structures, the bound toxin results in a pore with empty external potassium binding sites. 

The quality of the structural data presented in this revised manuscript is very high and allows for unambiguous assignment of side chains. The conclusions are supported by the data. This is an important contribution that should further our understanding of voltage-dependent potassium channel gating. In the revised version, the authors have addressed my previous specific comments, which are appended below. 

(1) In the main text's reference to Figure 2d residues W18' and S22' are mentioned but are not labeled in the insets. 

This has been fixed: line 229, p. 9.

(2) On page 8 there is a discussion of how the two remaining K+ ions in binding sites S3 and S4 prevent permeation K+ in molecular dynamics. However, in Shaker, inactivated W434F channels can sporadically allow K+ permeation with normal single-channel conductance but very reduced open times and open probability at not very high voltages. 

This is noted in the discussion Lines 497-500, p. 18

(3) The structures of WT in the absence of K+ shows a narrower selectivity filter, however Figure 4 does not convey this finding. In fact, the structure in Figure 4B is constructed in such an angle that it looks as if the carbonyl distances are increased, perhaps this should be fixed. Also, it is not clear how the distances between carbonyls given in the text on page 12 are measured. Is it between adjacent or kitty-corner subunits? 

We have changed Fig. 4B to show the same view as in Fig. 4A. In the legend we explain that opposing subunits are shown. We no longer give distances, in view of the lack of detectable carbonyl densities.

(4) It would be really interesting to know the authors opinion on the driving forces behind slow inactivation. For example, potassium flux seems to be necessary for channels to inactivate, which might indicate a local conformational change is the trigger for the main twisting events proposed here. 

We address this in the Discussion, line 506-523, pp. 18-19.

Reviewer #2 (Public Review): 

Cryo_EM structures of the Kv1.2 channel in the open, inactivated, toxin complex and in Na+ are reported. The structures of the open and inactivated channels are merely confirmatory of previous reports. The structures of the dendrotoxin bound Kv1.2 and the channel in Na+ are new findings that will of interest to the general channel community. 

Review of the resubmission: 

I thank the authors for making the changes in their manuscript as suggested in the previous review. The changes in the figures and the additions to the text do improve the manuscript. The new findings from a further analysis of the toxin channel complex are welcome information on the mode of the binding of dendrotoxin. 

A few minor concerns: 

(1) Line 93-96, 352: I am not sure as to what is it the authors are referring to when they say NaK2P. It is either NaK or NaK2K. I don't think that it has been shown in the reference suggested that either of these channels change conformation based on the K+ concentration. Please check if there is a mistake and that the Nichols et. al. reference is what is being referred to. 

Thank you for noticing the error. We meant NaK2K and we have changed this throughout.

(2) Line 365: In the study by Cabral et. al., Rb+ ions were observed by crystallography in the S1, S3 and S4 site, not the S2 site. Please correct. 

Thank you. We have re-written this section, lines 364-381, pp. 13-14.

Reviewer #3 (Public Review): 

Wu et al. present cryo-EM structures of the potassium channel Kv1.2 in open, C-type inactivated, toxin-blocked and presumably sodium-bound states at 3.2 Å, 2.5 Å, 2.8 Å, and 2.9 Å. The work builds on a large body of structural work on Kv1.2 and related voltage-gated potassium channels. The manuscript presents a plethora of structural work, and the authors are commended on the breadth of the studies. The structural studies are well-executed. Although the findings are mostly confirmatory, they do add to the body of work on this and related channels. Notably, the authors present structures of DTx-bound Kv1.2 and of Kv1.2 in a low concentration of potassium (which may contain sodium ions bound within the selectivity filter). These two structures add considerable new information. The DTx structure has been markedly improved in the revised version and the authors arrive at well-founded conclusions regarding its mechanism of block. Regarding the Na+ structure, the authors claim that the structure with sodium has "zero" potassium - I caution them to make this claim. It is likely that some K+ persists in their sample and that some of the density in the "zero potassium" structure may be due to K+ rather than Na+. This can be clarified by revisions to the text and discussion. I do not think that any additional experiments are needed. Overall, the manuscript is well-written, a nice addition to the field, and a crowning achievement for the Sigworth lab. 

Most of this reviewer's initial comments have been addressed in the revised manuscript. Some comments remain that could be addressed by revisions of the text. 

Specific comments on the revised version: 

Quotations indicate text in the manuscript. 

(1) "While the VSD helices in Kv1.2s and the inactivated Kv1.2s-W17'F superimpose very well at the top (including the S4-S5 interface described above), there is a general twist of the helix bundle that yields an overall rotation of about 3o at the bottom of the VSD." 

Comment: This seemed a bit confusing. I assume the authors aligned the complete structures - the differences they indicate seem to be slight VSD repositioning relative to the pore rather than differences between the VSD conformations themselves. The authors may wish to clarify. As they point out in the subsequent paragraph, the VSDs are known to be loosely associated with the pore. 

We aligned the VSDs alone, and it is a twist of the VSD helix bundle.

This is now clarified in lines 269-273, p. 10.

(2) Comment: The modeling of DTx into the density is a major improvement in the revision. Figure 3 displays some interactions between the toxin and Kv1.2 - additional side views of the toxin and the channel might allow the reader to appreciate the interactions more fully. The overall fit of the toxin structure into the density is somewhat difficult to assess from the figure. (The authors might consider using ChimeraX to display density and model in this figure.) 

We have added new panels, and stereo pairs, to Figure 3.

(3) "We obtained the structure of Kv1.2s in a zero K+ solution, with all potassium replaced with sodium, and were surprised to find that it is little changed from the K+ bound structure, with an essentially identical selectivity filter conformation (Figure 4B and Figure 4-figure supplement 1)." 

Comment: It should be noted in the manuscript that K+ and Na+ ions cannot be distinguished by the cryo-EM studies - the densities are indistinguishable. The authors are inferring that the observed density corresponds to Na+ because the protein was exchanged from K+ into Na+ on a gel filtration (SEC) column. It is likely that a small amount of K+ remains in the protein sample following SEC. I caution the authors to claim that there is zero K+ in solution without measuring the K+ content of the protein sample. Additionally, it should be considered that K+ may be present in the blotting paper used for cryo-EM grid preparation (our laboratory has noted, for example, a substantial amount of Ca2+ in blotting paper). The affinity of Kv1.2 for K+ has not been determined, to my knowledge - the authors note in the Discussion that the Shaker channel has "tight" binding for K+. It seems possible that some portion of the density in the selectivity filter could be due to residual K+. This caveat should be clearly stated in the main text and discussion. More extensive exchange into Na+, such as performing the entire protein purification in NaCl, or by dialysis (as performed for obtaining the structure of KcsA in low K+ by Y. Zhou et al. & Mackinnon 2001), would provide more convincing removal of K+, but I suspect that the Kv1.2 protein would not have sufficient biochemical stability without K+ to endure this treatment. One might argue that reduced biochemical stability in NaCl could be an indication that there was a meaningful amount of K+ in the final sample used for cryo-EM (or in the particles that were selected to yield the final high-resolution structure).

We now explain in the Methods section, in more detail the steps taken to avoid any residual Na+ contamination during purification, lines 683-687, pp. 24-25. We have changed the text to point out that the ion species cannot be distinguished in the maps, and note results in NaK2K and KcsA (lines 368-381, pp. 13-14).

We note that the same procedures to remove K+ were used for the Kv1.2sW17’F structure (line 385, p. 14). We qualify the ion replacement to say that Na+ replaces “essentially” all K+ (line 607, p. 21).

(4) Referring to the structure obtained in NaCl: "The ion occupancy is also similar, and we presume that Kv1.2 is a conducting channel in sodium solution." 

Comment: Stating that "Kv1.2 is a conducting channel in sodium solution" and implying that conduction of Na+ is achieved by an analogous distribution of ion binding sites as observed for K+ are strong statements to make - and not justified by the experiments provided. Electrophysiology would be required to demonstrate that the channel conducts sodium in the absence of K+. More complete ionic exchange, better control of the ionic conditions (Na+ vs K+), and affinity measurements for K+ would be needed to determine the distribution of Na+ in the filter (as mentioned above). At minimum, the authors should revise and clarify what the intended meaning of the statement "we presume that Kv1.2 is a conducting channel in sodium solution". As mentioned above, it seems possible/likely that a portion of the density in the filter may be due to K+. 

We now present a more detailed argument (lines 376 to 381, pp. 13-14.)

Recommendations for the authors: 

Reviewing Editor: 

After consultation, the reviewers agree that, although the authors have answered most of the comments raised in the previous review, there remains a concern about the structure obtained in the presence if Na. Given that Kv1.2 is more reluctant to slow inactivation, the conducting structure in Na+ could be due to this fact or that it really has higher affinity for K+ than Na+. In the presence of even a small contamination by K+, this ion could thus occupy the selectivity filter, resulting in an open conformation. The authors should clearly state the steps taken to ensure no contamination by K+. It is also possible that indeed the open structure occurs even in the presence of Na+ in the selectivity filter. This should be also discussed, given that this has been observed in other potassium channel structures. 

Reviewer #1 (Recommendations For The Authors): 

In this revised version of the manuscript, the authors have adequately addressed my previous points and improved the clarity and readability of the manuscript. This is a compelling work that shows inactivated structures if the Kv1.2 potassium channel, especially interesting is a structure in the absence of extracellular potassium ions, that can help understand how a reduction in the availability of these ions speed up entrance into the inactivated state in these ion channels. 

I would just recommend that in the absence of functional data (current recordings) when potassium is removed, the authors just use caution in ascribing this structure to an inactivated state. Also, it should be mentioned that the observed ion densities observed in the pore cannot unambiguously be identified as sodium ions. 

Reviewer #3 (Recommendations For The Authors): 

(1)  "The nearby Leu9 is also important as its substitution by alanine also decreases affinity 1000-fold, but we observe no contacts between this residue and residues of the Kv1.2s channel." 

Comment: It seems early in the text to mention the potential interaction of Leu9 to the channel structure. The authors may wish to discuss Leu9 later in the manuscript - a figure showing the location of Leu9 would strengthen the statement. Any hypothesis on why mutation of it has such a profound effect? 

Add a figure panel showing Leu9 position.

We have rewritten the text as suggested, and have identified Leu9 in several panels of Fig. 3.

(2)  "The X-ray structure of a-DTX (Figure 3A)" 

Comment: The authors may wish to cite a reference to this X-ray structure. 

We now cite Skarzynski (1992) on line 321, p. 12.

eLife assessment

This paper investigates how isoform II of transcription factor RUNX2 promotes cell survival and proliferation in oral squamous cell carcinoma cell lines. The authors used gain and loss of function techniques to provide incomplete evidence showing that RUNX2 isoform silencing led to cell death via several mechanisms including ferroptosis that was partially suppressed through RUNX2 regulation of PRDX2 expression. The study provides useful insight into the underlying mechanism by which RUNX2 acts in oral squamous cell carcinoma, but the conclusions of the authors should be revised to acknowledge that ferroptosis is not the only cause of cell death.

Reviewer #1 (Public Review):

Summary:<br /> In this paper, authors investigated the role of RUNT-related transcription factor 2 (RUNX2) in oral squamous carcinoma (OSCC) growth and resistance to ferroptosis. They found that RUNX2 suppresses ferroptosis through transcriptional regulation of peroxiredoxin-2. They further explored the upstream positive regulator of RUNX2, HOXA10 and found that HOXA10/RUNX2/PRDX2 axis protects OSCC from ferroptosis.

Strengths:<br /> The study is well designed and provides a novel mechanism of HOXA10/RUNX2/PRDX2 control of ferroptosis in OSCC.

Weaknesses:

According to the data presented in (Figure 2F, Figure 3Fand G, Figure 5D and Figure 6E and F), apoptosis seems to be affected in the same amount as ferroptosis by HOXA10/RUNX2/PRDX2 axis, which raises questions on the authors' specific focus on ferroptosis in this study. Reasonably, authors should adapt the title and the abstract in a way that recapitulates the whole data, which is HOXA10/RUNX2/PRDX2 axis control of cell death, including ferroptosis and apoptosis in OSCC.

Comments:

- In the description of the result section related to Figure 3E, the author wrote "In addition, we found that isoform II-knockdown induced shrunken mitochondria with vanished cristae with transmission electron microscopy (Figure 3E). These results suggest that RUNX2 isoform II may suppress ferroptosis." The interpretation provided here is not clear to the reviewer. How shrunken mitochondria and vanished cristae can be linked to ferroptosis?<br /> - The electron microscopy images show more elongated mitochondria in the RUNX2 isoform II-KO cells than in RUNX2 isoform II positive cells, which might result from the fusion of mitochondria. These images should completed with a fluorescent mitochondria staining of these cells.<br /> - What is the oxygen consumption rate in RUNX2 KO cells?<br /> - The increase in cell proliferation after RUNX2 overexpression in Figure 2A is not convincing, is there any differences in their migration or invasion capacity?<br /> - The in vivo study shows 50% reduction in primary tumor growth after RUNX2 inhibition by shRNA in CAL 27 xenografts, but only one shRNA is shown. Is this one shRNA clone? At least 2 shRNA clones should be used.<br /> - Apoptosis and necroptosis seem to be affected in the same amount as ferroptosis by HOXA10/RUNX2/PRDX2 axis. This is evident from experiments in Figure 3E, F and from Figure 6E, F and Figure 3G. Either Fer-1, Z-VAD,or Nec-1 used alone, were not able to fully restore cell proliferation to control cell level, which implies an additive effect of ferroptosis, apoptosis and necrosis. The author should verify potential additive or synergistic effect of the combination of Fer-1 and Z-VAD in these assays after si-RUNX2 in Figure 3 F and G and after si-HOX assays.<br /> - What is the effect of PRDX2 or HOXA10 depletion on tumor growth?<br /> - What is the clinical relevance of HOXA10 in OSCC patients?

Reviewing #2 (Public Review):

This paper reports the role of the Isoform II of RUNX2 in activating PRDX2 expression to suppress ferroptosis in oral squamous cell carcinoma (OSCC).<br /> The following major issues should be addressed.

A major postulate of this study is the specific role of RUNX2 isoform II compared to isoform I.

Figure 1F shows association between patient survival and Iso II expression, but nothing is shown for Iso I, this should be added, in addition the number of patients at risk in each category should be shown.<br /> The authors test Iso I and Iso II overexpression in CAL27 or SCC-9 model cell lines. In Fig. 2A in CAL27, the overexpression of Iso II is much stronger than Iso I so it seems premature to draw any conclusions. More importantly, however, no Iso I silencing is shown in either of the cell lines nor the xenografted tumours. This is absolutely essential for the authors hypothesis and should be tested using shRNA in cells and xenografted tumours.

A major conclusion of this study is that Iso II expression suppresses ferroptosis. To support this idea, the authors use the inhibitor Ferrostatin-1 (Fer-1). While Fer-1 typically does not lead to a 100% rescue, here the effect is only marginal and as shown in Figures 3F and G only marginally better than Z-VAD or Necrostatin 1. These data do not support the idea that the major cause of cell death is ferroptosis. Instead, Iso II silencing leads to cell death through different pathways. The authors should acknowledge this and rephrase the conclusion of the paper accordingly.<br /> Moreover, the authors consistently confound cell proliferation with cell death.

In Fig. 4A the authors investigate GPX1 expression, whereas GPX4 is often the key ferroprosis regulator, this has to be tested. This is important as the authors also test the effect of the GPX4 inhibitor RSL3, however, the authors do not determine IC50 values of the different cell lines with or without Iso II overexpression or silencing or compared to other RSL3 sensitive or resistant cells. Without this information, no conclusions can be drawn.

In summary, while the authors show that RUNX2 Iso II expression enhances cell survival, the idea that cell death is principally via ferroptosis is not fully established by the data. The authors should modify their conclusions accordingly.

eLife assessment

Mark and colleagues developed and validated a valuable method for examining subspace generalization in fMRI data and applied it to understand whether the entorhinal cortex uses abstract representations that generalize across different environments with the same structure. Evidence supporting the empirical findings - which show abstract entorhinal representations of hexagonal associative structures across different stimulus sets - is solid but could be further supported through additional analyses, discussion, and clarifications.

Reviewer #1 (Public review):

Summary:

This study develops and validates a neural subspace similarity analysis for testing whether neural representations of graph structures generalize across graph size and stimulus sets. The authors show the method works in rat grid and place cell data, finding that grid but not place cells generalize across different environments, as expected. The authors then perform additional analyses and simulations to show that this method should also work on fMRI data. Finally, the authors test their method on fMRI responses from the entorhinal cortex (EC) in a task that involves graphs that vary in size (and stimulus set) and statistical structure (hexagonal and community). They find neural representations of stimulus sets in lateral occipital complex (LOC) generalize across statistical structure and that EC activity generalizes across stimulus sets/graph size, but only for the hexagonal structures.

Strengths:

(1) The overall topic is very interesting and timely and the manuscript is well-written.

(2) The method is clever and powerful. It could be important for future research testing whether neural representations are aligned across problems with different state manifestations.

(3) The findings provide new insights into generalizable neural representations of abstract task states in the entorhinal cortex.

Weaknesses:

(1) The manuscript would benefit from improving the figures. Moreover, the clarity could be strengthened by including conceptual/schematic figures illustrating the logic and steps of the method early in the paper. This could be combined with an illustration of the remapping properties of grid and place cells and how the method captures these properties.

(2) Hexagonal and community structures appear to be confounded by training order. All subjects learned the hexagonal graph always before the community graph. As such, any differences between the two graphs could thus be explained (in theory) by order effects (although this is practically unlikely). However, given community and hexagonal structures shared the same stimuli, it is possible that subjects had to find ways to represent the community structures separately from the hexagonal structures. This could potentially explain why the authors did not find generalizations across graph sizes for community structures.

(3) The authors include the results from a searchlight analysis to show the specificity of the effects of EC. A better way to show specificity would be to test for a double dissociation between the visual and structural contrast in two independently defined regions (e.g., anatomical ROIs of LOC and EC).

(4) Subjects had more experience with the hexagonal and community structures before and during fMRI scanning. This is another confound, and possible reason why there was no generalization across stimulus sets for the community structure.

Reviewer #2 (Public review):

Summary:

Mark and colleagues test the hypothesis that entorhinal cortical representations may contain abstract structural information that facilitates generalization across structurally similar contexts. To do so, they use a method called "subspace generalization" designed to measure abstraction of representations across different settings. The authors validate the method using hippocampal place cells and entorhinal grid cells recorded in a spatial task, then perform simulations that support that it might be useful in aggregated responses such as those measured with fMRI. Then the method is applied to fMRI data that required participants to learn relationships between images in one of two structural motifs (hexagonal grids versus community structure). They show that the BOLD signal within an entorhinal ROI shows increased measures of subspace generalization across different tasks with the same hexagonal structure (as compared to tasks with different structures) but that there was no evidence for the complementary result (ie. increased generalization across tasks that share community structure, as compared to those with different structures). Taken together, this manuscript describes and validates a method for identifying fMRI representations that generalize across conditions and applies it to reveal entorhinal representations that emerge across specific shared structural conditions.

Strengths:

I found this paper interesting both in terms of its methods and its motivating questions. The question asked is novel and the methods employed are new - and I believe this is the first time that they have been applied to fMRI data. I also found the iterative validation of the methodology to be interesting and important - showing persuasively that the method could detect a target representation - even in the face of a random combination of tuning and with the addition of noise, both being major hurdles to investigating representations using fMRI.

Weaknesses:

In part because of the thorough validation procedures, the paper came across to me as a bit of a hybrid between a methods paper and an empirical one. However, I have some concerns, both on the methods development/validation side, and on the empirical application side, which I believe limit what one can take away from the studies performed.

Regarding the methods side, while I can appreciate that the authors show how the subspace generalization method "could" identify representations of theoretical interest, I felt like there was a noticeable lack of characterization of the specificity of the method. Based on the main equation in the results section of the paper, it seems like the primary measure used here would be sensitive to overall firing rates/voxel activations, variance within specific neurons/voxels, and overall levels of correlation among neurons/voxels. While I believe that reasonable pre-processing strategies could deal with the first two potential issues, the third seems a bit more problematic - as obligate correlations among neurons/voxels surely exist in the brain and persist across context boundaries that are not achieving any sort of generalization (for example neurons that receive common input, or voxels that share spatial noise). The comparative approach (ie. computing difference in the measure across different comparison conditions) helps to mitigate this concern to some degree - but not completely - since if one of the conditions pushes activity into strongly spatially correlated dimensions, as would be expected if univariate activations were responsive to the conditions, then you'd expect generalization (driven by shared univariate activation of many voxels) to be specific to that set of conditions. A second issue in terms of the method is that there is no comparison to simpler available methods. For example, given the aims of the paper, and the introduction of the method, I would have expected the authors to take the Neuron-by-Neuron correlation matrices for two conditions of interest, and examine how similar they are to one another, for example by correlating their lower triangle elements. Presumably, this method would pick up on most of the same things - although it would notably avoid interpreting high overall correlations as "generalization" - and perhaps paint a clearer picture of exactly what aspects of correlation structure are shared. Would this method pick up on the same things shown here? Is there a reason to use one method over the other?

Regarding the fMRI empirical results, I have several concerns, some of which relate to concerns with the method itself described above. First, the spatial correlation patterns in fMRI data tend to be broad and will differ across conditions depending on variability in univariate responses (ie. if a condition contains some trials that evoke large univariate activations and others that evoke small univariate activations in the region). Are the eigenvectors that are shared across conditions capturing spatial patterns in voxel activations? Or, related to another concern with the method, are they capturing changing correlations across the entire set of voxels going into the analysis? As you might expect if the dynamic range of activations in the region is larger in one condition than the other? My second concern is, beyond the specificity of the results, they provide only modest evidence for the key claims in the paper. The authors show a statistically significant result in the Entorhinal Cortex in one out of two conditions that they hypothesized they would see it. However, the effect is not particularly large. There is currently no examination of what the actual eigenvectors that transfer are doing/look like/are representing, nor how the degree of subspace generalization in EC may relate to individual differences in behavior, making it hard to assess the functional role of the relationship. So, at the end of the day, while the methods developed are interesting and potentially useful, I found the contributions to our understanding of EC representations to be somewhat limited.

Reviewer #3 (Public review):

Summary:

The article explores the brain's ability to generalize information, with a specific focus on the entorhinal cortex (EC) and its role in learning and representing structural regularities that define relationships between entities in networks. The research provides empirical support for the longstanding theoretical and computational neuroscience hypothesis that the EC is crucial for structure generalization. It demonstrates that EC codes can generalize across non-spatial tasks that share common structural regularities, regardless of the similarity of sensory stimuli and network size.

Strengths:

(1) Empirical Support: The study provides strong empirical evidence for the theoretical and computational neuroscience argument about the EC's role in structure generalization.

(2) Novel Approach: The research uses an innovative methodology and applies the same methods to three independent data sets, enhancing the robustness and reliability of the findings.

(3) Controlled Analysis: The results are robust against well-controlled data and/or permutations.

(4) Generalizability: By integrating data from different sources, the study offers a comprehensive understanding of the EC's role, strengthening the overall evidence supporting structural generalization across different task environments.

Weaknesses:

A potential criticism might arise from the fact that the authors applied innovative methods originally used in animal electrophysiology data (Samborska et al., 2022) to noisy fMRI signals. While this is a valid point, it is noteworthy that the authors provide robust simulations suggesting that the generalization properties in EC representations can be detected even in low-resolution, noisy data under biologically plausible assumptions. I believe this is actually an advantage of the study, as it demonstrates the extent to which we can explore how the brain generalizes structural knowledge across different task environments in humans using fMRI. This is crucial for addressing the brain's ability in non-spatial abstract tasks, which are difficult to test in animal models.

While focusing on the role of the EC, this study does not extensively address whether other brain areas known to contain grid cells, such as the mPFC and PCC, also exhibit generalizable properties. Additionally, it remains unclear whether the EC encodes unique properties that differ from those of other systems. As the authors noted in the discussion, I believe this is an important question for future research.

eLife assessment

In this important study, the authors have partially revealed the mechanism behind lip thickening in cichlid fishes, which has evolved independently across three lakes in Africa. To explore this phenomenon, the authors utilized histological comparison, proteomics, and transcriptomics, all of which are well suited for their objectives. With convincing evidence, this contribution holds significant value for the field.

Reviewer #1 (Public Review):

Summary:

Machii et al. reported a possible molecular mechanism underlying the parallel evolution of lip hypertrophy in African cichlids. The multifaceted approach taken in this manuscript is highly valued, as it uses histology, proteomics, and transcriptomics to reveal how phylogenetically distinct thick-lips have evolved in parallel. Findings from histology and proteomics connected to wnt signaling through the transcriptome are very exciting.

Strengths:

There is consistency between the results and it is possible to make a strong argument from the results.

Weaknesses:

The authors do not discuss based on genomic information; the genomes of the cichlids from the three lakes have been decoded and are therefore available. However, indeed, the species in Lake Tanganyika and Lake Malawi/Victoria are genetically distant from each other, so a comparative genome analysis would not have yielded the results presented here. I recommend adding such a discussion to the Discussion.

Reviewer #2 (Public Review):

I have carefully reviewed the manuscript titled "Pronounced expression of extracellular matrix proteoglycans regulated by Ant pathway underlies the parallel evolution of lip hypertrophy in East African cichlids." I commend the authors for their work on elucidating the mechanism underlying lip thickening that has evolved in parallel across three lakes in Africa.

The use of histological comparison, proteomics, and transcriptomics methods to investigate this phenomenon is commendable and adds depth to the study. The findings indicate that the overexpression of proteoglycans is the cause of lip thickening and provides valuable insights into the evolutionary process.

I found the writing style to be clear and the explanations provided are easy to understand. Overall, I did not identify any significant issues with the manuscript.

eLife assessment

This study presents an interesting mechanism for the regulation of RNA levels, establishing an important regulatory connection between protein arginine methyltransferase 1 and the splicing factor SFPQ. While these findings have theoretical implications beyond a single field, the evidence is incomplete, with only partial support for the main claims.

Reviewer #1 (Public review):

The central finding of the current manuscript is that embryonic ablation of PRMT1 results in a craniofacial phenotype that is primarily linked to downstream intron splicing defects. This manuscript is one of several to underscore the relative importance of intron splicing to gene expression regulation during development, and moreover, to recapitulate splicing-related craniofacial defects. Specifically, authors introduce a regulatory axis consisting of PRMT1-SFPQ that directs mechanisms of long intron retention. This finding represents a significant contribution to our understanding of splicing regulation, in the sense that it highlights the regulatory impact that post-translational modification of splicing-related proteins can have on intron processing. Further, it emphasizes the importance of extending the study of splicing regulation beyond core components of the spliceosome, to include their upstream regulators as well.

The significance of neural crest cells in the development of craniofacial structures has long been considered a major contributor to developmental phenotypes. This specific symptomology is heavily associated with spliceosomopathies, wherein disruption of spliceosome components is the primary mechanism of disease pathogenesis. Thus, the PRMT1 associated phenotype is noteworthy. The role of PRMT1 in methylating downstream splicing factors introduces a new avenue of research focused on the mechanisms of spliceosome component activation and their effects on splicing. The strength of the current study lies in their establishing the molecular mechanism through which PRMT1 could alter craniofacial development through regulation of the transcriptome, but the data presented to support the claim that a PRMT1-SFPQ axis directly regulates intron retention of the relevant gene networks should be robust and with multiple forms of clear validation. For example, elevated intron retention findings are based on the intron retention index, and according to the manuscript, are assessed considering the relative expression of exons and introns from a given transcript. However, delineating between intron retention and other forms of alternative splicing (i.e., cryptic splice site recognition) requires a more comprehensive consideration of the intron splicing defects that could be represented in data. A certain threshold of intron read coverage (i.e., the percent of an intron that is covered by mapped reads) is needed to ascertain if those that are proximal to exons could represent alternative introns ends rather than full intron retention events. In other words, intron retention is a type of alternative splicing that can be difficult to analyze in isolation given the confounding influence of cryptic splicing and cryptic exon inclusion. If other forms of alternative splicing were assessed and not detected, more confident retention calls can be made.

While data presented to support the PRMT1-SFPQ activation axis is quite compelling, that this is directly responsible for the elevated intron retention remains enigmatic. First, in characterizing their PRMT1 knockout model, it is unclear whether the elevated intron retention events directly correspond to downregulated genes. Moreover, intron splicing is a well-documented node for gene regulation during embryogenesis and in other proliferation models, and craniofacial defects are known to be associated with 'spliceosomopathies'. However, reproduction of this phenotype does not suggest that the targets of interest are inherently splicing factors, and a more robust assessment is needed to determine the exact nature of alternative splicing in this system. Because there are several known splicing factors downstream of PRMT1 and presented in the supplemental data, the specific attribution of retention to SFPQ would be additionally served by separating its splicing footprint from that of other factors that are primed to cause alternative splicing.

Clarifying the relationship between SFPQ and splicing regulation is important given that the observed splicing defects are incongruous with published data presented by Takeuchi et al., (2018) regarding SFPQ control of neuronal apoptosis in mice. In this system, SFPQ was more specifically attributed to the regulation of transcription elongation over long introns and its knockout did not result in significant splicing changes. Thus, to establish the specificity for the SFPQ in regulating these retention events, authors would need to show that the same phenotype is not achieved by mis-regulation of other splicing factors. That the authors chose SFPQ based on its binding profile is understandable but potentially confounding given its mechanism of action in transcription of long introns (Takeuchi 2018). Because mechanisms and rates of transcription can influence splicing and exon definition interactions, the role of SFPQ as a transcription elongation factor versus a splicing factor is inadequately disentangled by authors.

Reviewer #2 (Public review):

Summary:<br /> The manuscript by Lima et al examines the role of Prmt1 and SFPQ in craniofacial development. Specifically, the authors test the idea that Prmt1 directly methylates specific proteins that results in intron retention in matrix proteins. The protein SFPQ is methylated by Prmt1 and functions downstream to mediate Prmt1 activity. The genes with retained introns activate the NMD pathway to reduce the RNA levels. This paper describes an interesting mechanism for the regulation of RNA levels during development.

Strengths:<br /> The phenotypes support what the authors claim that Prmt1 is involved in craniofacial development and splicing. The use of state-of-the-art sequencing to determine the specific genes that have intron retention and changes in gene expression is a strength.

Weaknesses:<br /> Some of the data seems to contradict the conclusions. And it is unclear how direct the relationships are between Prmt1 and SFPQ.

eLife assessment

This important study shows how a combination of the latest generation of Oxford Nanopore Technology long reads with state-of-the art variant callers enables bacterial variant discovery at an accuracy that matches or exceeds the current "gold standard" with short reads. The work thus heralds a new era, in which Illumina short-read sequencing no longer rules supreme. While the inclusion of a larger number of reference genomes would have enabled an even more fine-grained analysis, the evidence as it is supports the claims of the authors convincingly. The work will be of interest to anyone performing sequencing for outbreak investigations, bacterial epidemiology, or similar studies.

Reviewer #2 (Public review):

Summary:

Hall et al describe the superiority of ONT sequencing and deep learning-based variant callers to deliver higher SNP and Indel accuracy compared to previous gold-standard Illumina short-read sequencing. Furthermore, they provide recommendations for read sequencing depth and computational requirements when performing variant calling.

Strengths:

The study describes compelling data showing ONT superiority when using deep learning-based variant callers, such as Clair3, compared to Illumina sequencing. This challenges the paradigm that Illumina sequencing is the gold standard for variant calling in bacterial genomes. The authors provide evidence that homopolymeric regions, a systematic and problematic issue with ONT data, are no longer a concern in ONT sequencing.

Weaknesses:

The study is limited in the number of samples included, even though it covers different species with divergent genome sequences, likely covering major evolutionary changes. The methods section could be more detailed. A structural variation analysis would be an interesting next step.

Reviewer #3 (Public review):

Hall et al. benchmarked different variant calling methods on Nanopore reads of bacterial samples and compared the performance of Nanopore to short reads produced with Illumina sequencing. To establish a common ground for comparison, the authors first generated a variant truthset for each sample and then projected this set to the reference sequence of the sample to obtain a mutated reference. Subsequently, Hall et al. called SNPs and small indels using commonly used deep learning and conventional variant callers and compared the precision and accuracy from reads produced with simplex and duplex Nanopore sequencing to Illumina data. The authors did not investigate large structural variation, which is a major limitation of the current manuscript. It will be very interesting to see a follow-up study covering this much more challenging type of variation.

In their comprehensive comparison of SNPs and small indels, the authors observed superior performance of deep learning over conventional variant callers when Nanopore reads were basecalled with the most accurate (but also computationally very expensive) model, even exceeding Illumina in some cases. Not surprisingly, Nanopore underperformed compared to Illumina when basecalled with the fastest (but computationally much less demanding) method with the lowest accuracy. The authors then investigated the surprisingly higher performance of Nanopore data in some cases and identified lower recall with Illumina short read data, particularly from repetitive regions and regions with high variant density, as the driver. Combining the most accurate Nanopore basecalling method with a deep learning variant caller resulted in low error rates in homopolymer regions, similar to Illumina data. This is remarkable, as homopolymer regions are (or, were) traditionally challenging for Nanopore sequencing.

Lastly, Hall et al. provided useful information on the required Nanopore read depth, which is surprisingly low, and the computational resources for variant calling with deep learning callers. With that, the authors established a new state-of-the-art for Nanopore-only variant calling on bacterial sequencing data. Most likely these findings will be transferred to other organisms as well or at least provide a proof-of-concept that can be built upon.

As the authors mention multiple times throughout the manuscript, Nanopore can provide sequencing data in nearly real-time and in remote regions, therefore opening up a ton of new possibilities, for example for infectious disease surveillance. In these scenarios, computational resources can be very limited. The highest-performing variant calling method, as established in this study, requires the computationally very expensive sup and/or duplex nanopore basecalling, while the least computationally demanding basecalling method underperforms. To comprehensively guide users through the computational resources required for basecalling and variant calling, the authors provide runtime benchmarks assuming GPU access.

Author response:

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

Public Reviews:

Reviewer #1 (Public Review):

Summary:

The authors assess the accuracy of short variant calling (SNPs and indels) in bacterial genomes using Oxford Nanopore reads generated on R10.4 flow cells from a very similar genome (99.5% ANI), examining the impact of variant caller choice (three traditional variant callers: bcftools, freebayes, and longshot, and three deep learning based variant callers: clair3, deep variant, and nano caller), base calling model (fast, hac and sup) and read depth (using both simplex and duplex reads).

Strengths:

Given the stated goal (analysis of variant calling for reads drawn from genomes very similar to the reference), the analysis is largely complete and results are compelling. The authors make the code and data used in their analysis available for re-use using current best practices (a computational workflow and data archived in INSDC databases or Zenodo as appropriate).

Weaknesses:

While the medaka variant caller is now deprecated for diploid calling, it is still widely used for haploid variant calling and should at least be mentioned (even if the mention is only to explain its exclusion from the analysis). 

We have now added Medaka haploid caller to the benchmark. It performs quite well overall (better than the traditional methods), but not as good as Clair3 or DeepVariant.

Appraisal:

The experiments the authors engaged in are well structured and the results are convincing. I expect that these results will be incorporated into "best practice" bacterial variant calling workflows in the future. 

Thank you for the positive appraisal.

Reviewer #2 (Public Review):

Summary:

Hall et al describe the superiority of ONT sequencing and deep learning-based variant callers to deliver higher SNP and Indel accuracy compared to previous gold-standard Illumina short-read sequencing. Furthermore, they provide recommendations for read sequencing depth and computational requirements when performing variant calling.

Strengths:

The study describes compelling data showing ONT superiority when using deep learning-based variant callers, such as Clair3, compared to Illumina sequencing. This challenges the paradigm that Illumina sequencing is the gold standard for variant calling in bacterial genomes. The authors provide evidence that homopolymeric regions, a systematic and problematic issue with ONT data, are no longer a concern in ONT sequencing.

Weaknesses:

(1) The inclusion of a larger number of reference genomes would have strengthened the study to accommodate larger variability (a limitation mentioned by the authors). 

Our strategic selection of 14 genomes—spanning a variety of bacterial genera and species, diverse GC content, and both gram-negative and gram-positive species (including M. tuberculosis, which is neither)—was designed to robustly address potential variability in our results. Moreover, all our genome assemblies underwent rigorous manual inspection as the quality of the true genome sequences is the foundation this research is built upon. Given this, the fundamental conclusions regarding the accuracy of variant calls would likely remain unchanged with the addition of more genomes.  However, we do acknowledge that a substantially larger sample size, which is beyond the scope of this study, would enable more fine-grained analysis of species differences in error rates.

(2) In Figure 2, there are clearly one or two samples that perform worse than others in all combinations (are always below the box plots). No information about species-specific variant calls is provided by the authors but one would like to know if those are recurrently associated with one or two species. Species-specific recommendations could also help the scientific community to choose the best sequencing/variant calling approaches.

Thank you for highlighting this observation. The precision, recall, and F1 scores for each sample and condition can be found in Supplementary Table S4.

Upon investigation of the outliers in Figure 2 we discovered three things. First, there was a parameter in Longshot we were using that automatically capped coverage and lead to a number of false negatives, leading to its outlier. This has now been rectified and the figure is updated accordingly. Second, the outlier in the simplex sup SNP panel (top left) was the same E. coli sample for most variant callers (though Medaka had no issues). The reasoning for this was a variant dense repetitive region. We have added an in-depth explanation of this, along with figures illustrating the issue in Supplementary Section S2, with a brief statement in the main text. Third, the outlier in the duplex sup SNP panel (top right) is due to a very low (duplex) depth sample. This has also been added briefly to the main text and fully in Section S2.

We have now included a species-segregated version of Figure 2 (Suppl. Figures S5-7) for Clair3 with the sup model (best performer) for a clearer interpretation of how each species performs.

(3) The authors support that a read depth of 10x is sufficient to achieve variant calls that match or exceed Illumina sequencing. However, the standard here should be the optimal discriminatory power for clinical and public health utility (namely outbreak analysis). In such scenarios, the highest discriminatory power is always desirable and as such an F1 score, Recall and Precision that is as close to 100% as possible should be maintained (which changes the minimum read sequencing depth to at least 25x, which is the inflection point).

We agree that the highest discriminatory power is always desirable for clinical or public health applications. In which case, 25x is probably a better minimum recommendation. However, we are also aware that there are resource-limited settings where parity with Illumina is sufficient. In these cases, 10x depth from ONT would provide enough data.

The manuscript previously emphasised the latter scenario, but we have revised the text (Discussion) to clearly recommend 25x depth as a conservative aim in settings where resources are not a constraint, ensuring the highest possible discriminatory power.

(4) The sequencing of the samples was not performed with the same Illumina and ONT method/equipment, which could have introduced specific equipment/preparation artefacts that were not considered in the study. See for example https://academic.oup.com/nargab/article/3/1/lqab019/6193612.

To our knowledge, there is no evidence that sequencing on different ONT machines or barcoding kits leads to a difference in read characteristics or accuracy. To ensure consistency and minimise potential variability, we used the same ONT flowcells for all samples and performed basecalling on the same Nvidia A100 GPU. We have updated the methods to emphasise this.

For Illumina and ONT, the exact machines and kits used for each sample have been added as supplementary table S9 We have also added a short paragraph about possible Illumina error rate differences in the ‘Limitations’ section of the Discussion.

The third limitation is that Illumina sequencing was performed on different models: three samples on the NextSeq 500 and the rest on the NextSeq 2000. While differences in error rates exist between Illumina instruments, no specific assessment has been made between these NextSeq models [42]. However, the absolute differences in error rates are minor and unlikely to impact our study significantly. This is particularly relevant since Illumina's lower F1 score compared to ONT was due to missed calls rather than erroneous ones.

In summary, while there may be specific equipment or preparation artifacts to consider, we took steps to minimise these effects and maintain consistency across our sequencing methods.

Reviewer #3 (Public Review):

Hall et al. benchmarked different variant calling methods on Nanopore reads of bacterial samples and compared the performance of Nanopore to short reads produced with Illumina sequencing. To establish a common ground for comparison, the authors first generated a variant truth set for each sample and then projected this set to the reference sequence of the sample to obtain a mutated reference. Subsequently, Hall et al. called SNPs and small indels using commonly used deep learning and conventional variant callers and compared the precision and accuracy from reads produced with simplex and duplex Nanopore sequencing to Illumina data. The authors did not investigate large structural variation, which is a major limitation of the current manuscript. It will be very interesting to see a follow-up study covering this much more challenging type of variation. 

We fully agree that investigating structural variations (SVs) would be a very interesting and important follow-up. Identifying and generating ground truth SVs is a nontrivial task and we feel it deserves its own space and study. We hope to explore this in the future.

In their comprehensive comparison of SNPs and small indels, the authors observed superior performance of deep learning over conventional variant callers when Nanopore reads were basecalled with the most accurate (but also computationally very expensive) model, even exceeding Illumina in some cases. Not surprisingly, Nanopore underperformed compared to Illumina when basecalled with the fastest (but computationally much less demanding) method with the lowest accuracy. The authors then investigated the surprisingly higher performance of Nanopore data in some cases and identified lower recall with Illumina short read data, particularly from repetitive regions and regions with high variant density, as the driver. Combining the most accurate Nanopore basecalling method with a deep learning variant caller resulted in low error rates in homopolymer regions, similar to Illumina data. This is remarkable, as homopolymer regions are (or, were) traditionally challenging for Nanopore sequencing.

Lastly, Hall et al. provided useful information on the required Nanopore read depth, which is surprisingly low, and the computational resources for variant calling with deep learning callers. With that, the authors established a new state-of-the-art for Nanopore-only variant, calling on bacterial sequencing data. Most likely these findings will be transferred to other organisms as well or at least provide a proof-of-concept that can be built upon.

As the authors mention multiple times throughout the manuscript, Nanopore can provide sequencing data in nearly real-time and in remote regions, therefore opening up a ton of new possibilities, for example for infectious disease surveillance.

However, the high-performing variant calling method as established in this study requires the computationally very expensive sup and/or duplex Nanopore basecalling, whereas the least computationally demanding method underperforms. Here, the manuscript would greatly benefit from extending the last section on computational requirements, as the authors determine the resources for the variant calling but do not cover the entire picture. This could even be misleading for less experienced researchers who want to perform bacterial sequencing at high performance but with low resources. The authors mention it in the discussion but do not make clear enough that the described computational resources are probably largely insufficient to perform the high-accuracy basecalling required. 

We have provided runtime benchmarks for basecalling in Supplementary Figure S23 and detailed these times in Supplementary Table S7. In addition, we state in the Results section (P9 L239-241) “Though we do note that if the person performing the variant calling has received the raw (pod5) ONT data, basecalling also needs to be accounted for, as depending on how much sequencing was done, this step can also be resource-intensive.”

Even with super-accuracy basecalling considered, our analysis shows that variant calling remains the most resource-intensive step for Clair3, DeepVariant, FreeBayes, Medaka, and NanoCaller. Therefore, the statement “the described computational resources are probably largely insufficient to perform the high-accuracy basecalling required”, is incorrect. However, we have made this more prominent in the Results and Discussion.

In the results section we added the underlined section:

“… FreeBayes had the largest runtime variation, with a maximum of 597s/Mbp, equating to 2.75 days for the same genome. In contrast, basecalling with a single GPU using the super-accuracy model required a median runtime of 0.77s/Mbp, or just over 5 minutes for a 4Mbp genome at 100x depth. …”

In the discussion we have added the following statement:

“Basecalling is generally faster than variant calling, assuming GPU access, which is likely considered when acquiring ONT-related equipment.”

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

The colour choices in Figure 3 and Figure 4 c made the illustrations somewhat difficult to read. More substantially, a deeper investigation of the causes of non-homopolymeric-related mistaken indel calls would be useful. 

We have updated Figure 3 so that each line has a different style to aid in discriminating between colours. The colour scheme for Figure 4c has also been updated.

In terms of non-homopolymeric false positive (FP) indel calls, we did an investigation of these for Clair3 and DeepVariant on the simplex sup data as these are the two best performing variant callers and deal the best with homopolymers. For Clair3, there were eight FPs across all samples. Five of these were homopolymers. The remaining three occurred within one or two bases of another insertion which inserted a similar sequence to the FP. For DeepVariant, it was much the same story, with 8/11 FP indels being in homopolymers, and the remaining three being within one or two bases of another insertion with a similar sequence. We have added a couple of sentences to the results explaining this finding.

Reviewer #2 (Recommendations For The Authors):

The paper is well-written and provides evidence for the conclusions. Some issues should be addressed.

Include a section in the Results describing species-specific observations, namely if some samples had recurrently lower SNP and INDEL F1 scores (as observed in Figure 2). 

Please see our response in your second point in the ‘Weaknesses’ section of the public review.

Please provide more details on how the samples were sequenced. Section "Sequencing" in the methods is confusing and not clear enough to be reproduced (provide a supplementary table/figure with the workflow for each sample). Add information about how many samples were multiplexed in each run and what was the output achieved in each.

We have now added a Supplementary Table S9 which outlines which instruments, kits, and multiplexing strategies were used for each sample. In addition, the raw pod5 data that we make available has been segregated by sample, so knowledge of the multiplexing strategy is not necessary for someone attempting to reproduce our results.

The authors acknowledge that structural variation was not evaluated in this manuscript. Since ONT sequencing is often used to reconstruct the sequence of plasmids for outbreak/epidemiology analysis, perhaps they could undertake this analysis on a plasmids dataset (which suffers from constant structural variation).

As noted in our response to Reviewer 3’s public review, we fully agree that investigating structural variations (SVs) would be a very interesting and important follow-up. Identifying and generating ground truth SVs is a nontrivial task and we feel it deserves its own space and study. We hope to explore this in the future.

Reviewer #3 (Recommendations For The Authors):

The manuscript is well organized. However, some sections are a bit long and would benefit from being more concise.

Thank you for your valuable feedback and for acknowledging the organisation of our manuscript. We appreciate your suggestion regarding the length of certain sections. We have gone back through and made the manuscript more concise.

Figure 1: Is the Qscore really the same as identity? Isn't the determination of identity only possible after alignment? 

When we say Qscore we are referring to the Phred-scaled version of the read identity, which is alignment based, not the Qscores of the individual bases in the FASTQ file. We have updated the text and figure legend to make this clearer. “The Qscore is the logarithmic transformation of the read identity,  , where 𝑃 is the read identity.”. We also now explicitly state that read identity is alignment-based.

Abbreviations/terms mentioned but not introduced: <br /> - kmers, P2L57

- ANI, P3L93 

We have updated the text to better introduce these terms.

eLife assessment

Notario Manzano et al. offer a valuable first analysis of proteins within tunneling nanotubes (TNTs), membranous bridges connecting cells. This work distinguishes TNTs from extracellular vesicles, but further experimental and analytical tools are needed to refine the TNT proteome. Solid data supports a role for tetraspanins CD9 and CD81 in TNT function. The proposed model for CD9 and CD81 is over-interpreted and requires additional evidence for stronger support.

eLife assessment

This valuable study demonstrates that the behavior of the cells in the presomitic mesoderm in zebrafish embryos depend on both an intrinsic program and external information, providing new insight into the biology underlying embryo axis segmentation. There is convincing support for the findings with a thorough and quantitative single-cell real-time imaging approach, both in vitro and in vivo, developed by the authors.

Reviewer #1 (Public Review):

Summary:

In this manuscript, Rohde et al. discuss how single cells isolated from the presomitic mesoderm of the zebrafish embryo follow a cell-autonomous differentiation "programme", which is dependent on the initial anteroposterior position in the embryo.

Strengths:

This work and, in particular, the comparison to cellular behaviour in vivo presents a detailed description of the oscillatory system that brings the developmental biology forward in their understanding of somitogenesis.<br /> The main novelty lies in the direct comparison of these isolated single cells to single cells tracked within the developing embryo. This allows them to show that isolated cells follow a similar path of differentiation without direct contact to neighbours or the presence of external morphogen gradients. Based on this, the authors propose an internal timer that starts ticking as cells traverse the presomitic mesoderm, while external signals modify this behaviour.

There are a few direct questions that follow up from this study, for instance, intercellular synchronization influences the variability of the timer. However, I agree with the authors that such experiments are out of the scope of this study.

Author response:

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

Reviewer #1 (Public Review):

My main point of concern is the precision of dissection. The authors distinguish cells isolated from the tailbud and different areas in the PSM. They suggest that the cell-autonomous timer is initiated, as cells exit the tailbud.

This is also relevant for the comparison of single cells isolated from the embryo and cells within the embryo. The dissection will always be less precise and cells within the PSM4 region could contain tailbud cells (as also indicated in Figure 1A), while in the analysis of live imaging data cells can be selected more precisely based on their location. This could therefore contribute to the difference in noise between isolated single cells and cells in the embryo. This could also explain why there are "on average more peaks" in isolated cells (p. 6, l. 7).

This aspect should be considered in the interpretation of the data and mentioned at least in the discussion. (It does not contradict their finding that more anterior cells oscillate less often and differentiate earlier than more posterior ones.)

Reviewer #1 rightly points out that selecting cells in a timelapse is more precise than manual dissection. Manual dissection is inherently variable but we believe in general it is not a major source of noise in our experiments. To control for this, we compared the results of 11 manual dissections of the posterior quarter of the PSM (PSM4) with those of the pooled PSM4 data. In general, we did not see large differences in the distributions of peak number or arrest timing that would markedly increase the variability of the pooled data above that of the individual dissections (Figure 1 – supplement figure 7). We have edited the text in the Results to highlight this control experiment (page 6, lines 13-17).

It is of course possible that we picked up adjacent TB cells when dissecting PSM4, however the reviewer’s assertion that inclusion of TB cells “could also explain why there are "on average more peaks" in isolated cells” is incorrect. Later in the paper we show that cells from the TB have almost identical distributions to PSM4 (mean ± SD, PSM4 4.36 ± 1.44; TB 4.26 ± 1.35; Figure 4 _ supplement 1). Thus, inadvertent inclusion of TB cells while dissecting would in fact not increase the number of peaks.

Here, the authors focus on the question of how cells differentiate. The reverse question is not addressed at all. How do cells maintain their oscillatory state in the tailbud? One possibility is that cells need external signals to maintain that as indicated in Hubaud et al. 2014. In this regard, the definition of tailbud is also very vague. What is the role of neuromesodermal progenitors? The proposal that the timer is started when cells exit the tailbud is at this point a correlation and there is no functional proof, as long as we do not understand how cells maintain the tailbud state. These are points that should be considered in the discussion.

The reviewer asks “How do cells maintain their oscillatory state in the tailbud?”. This is a very interesting question, but as recognized by the reviewer, beyond the scope of our current paper.

We now further emphasize the point “One possibility is that cells need external signals to maintain … as indicated in Hubaud et al. 2014” in the Discussion and added a reference to the review Hubaud and Pourquié 2014 (Signalling dynamics in vertebrate segmentation. Nat Rev Mol Cell Biol 15, 709–721 (2014). https://doi.org/10.1038/nrm3891) (page 18, lines 19-22).

To clarify the definition of the TB, we have stated more clearly in the results (page 15, lines 8-12) that we defined TB cells as all cells posterior to the notochord (minus skin) and analyzed those that survived

>5 hours post-dissociation, did not divide, and showed transient Her1-YFP dynamics.

The reviewer asks: What is the role of neuromesodermal progenitors? In responding to this, we refer to Attardi et al., 2018 (Neuromesodermal progenitors are a conserved source of spinal cord with divergent growth dynamics. Development. 2018 Nov 9;145(21):dev166728. doi: 10.1242/dev.166728).

Around the stage of dissection in zebrafish in our work, there is a small remaining group of cells characterized as NMPs (Sox2 +, Tbxta+ expression) in the dorsal-posterior wall of the TB. These NMPs rarely divide and are not thought to act as a bipotential pool of progenitors for the elongating axis, as is the case in amniotes, rather contributing to the developing spinal cord. How this particular group of cells behaves in culture is unclear as we did not subdivide the TB tissue before culturing. It would be possible to specifically investigate these NMPs regarding a role in TB oscillations, but given the results of Attardi et al., 2018 (small number of cells, low bipotentiality), we argue that it would not be significant for the conclusions of the current work. To indicate this, we included a sentence and a citation of this paper in the results towards the beginning of the section on the tail bud (page 15, lines 8-12).

The authors observe that the number of oscillations in single cells ex vivo is more variable than in the embryo. This is presumably due to synchronization between neighbouring cells via Notch signalling in the embryo. Would it be possible to add low doses of Notch inhibitor to interfere with efficient synchronization, while at the same time keeping single cell oscillations high enough to be able to quantify them?

It is a formal possibility that Delta-Notch signaling may have some impact on the variability in the number of oscillations. However, we argue that the significant amount of cell tracking work required to carry out the suggested experiments would not be justified, considering what has been previously shown in the literature. If Delta-Notch signaling was a major factor controlling the variability of the intrinsic program that we describe, then we would expect that in Delta-Notch mutants the anterior- posterior limits of cyclic gene expression in the PSM would extend beyond those seen in wildtype embryos. Specifically, we might expect to see her1 expression extending more anteriorly in mutants to account for the dramatic increase in the number of cells that have 5, 6, 7 and 8 cycles in culture (Fig. 1E versus Fig. 1I). However, as shown in Holley et al., 2002 (Fig. 5A, B; her1 and the notch pathway function within the oscillator mechanism that regulates zebrafish somitogenesis. Development. 2002 Mar;129(5):1175-83. doi: 10.1242/dev.129.5.1175), the anterior limit of her1 expression in the PSM in DeltaD mutants (aei) is not different to WT. Thus, Delta-Notch signaling may exert a limited control over the number of oscillations, but likely not in excess of one cycle difference.

In the same direction, it would be interesting to test if variation is decreased, when the number of isolated cells is increased, i.e. if cells are cultured in groups of 2, 3 or 4 cells, for instance.

This is a great proposal – however the culture setup used here is a wide-field system that doesn’t allow us to accurately follow more than one cell at a time. Cells that adhere to each other tend to crawl over each other, blurring their identity in Z. This is also why we excluded dividing cells in culture from the analysis. Experiments carried out with a customized optical setup will be needed to investigate this in the future.

It seems that the initiation of Mesp2 expression is rather reproducible and less noisy (+/- 2 oscillation cycles), while the number of oscillations varies considerably (and the number of cells continuing to oscillate after Mesp2 expression is too low to account for that). How can the authors explain this apparent discrepancy?

The observed tight linkage of the Mesp onset and Her1 arrest argue for a single timing mechanism that is upstream of both gene expression events; indeed, this is one of the key implications of the paper. However, the infrequent dissociation of these events observed in FGF-treated cells suggests that more than one timing pathway could be involved, although there are other interpretations. We’ve added more discussion in the text on one vs multi-timers (page 17, lines 19-23 – page 18, line 1 - 8)., see next point.

The observation that some cells continue oscillating despite the upregulation of Mesp2 should be discussed further and potential mechanism described, such as incomplete differentiation.

This is an infrequent (5 out of 54 cells) and interesting feature of PSM4 cells in the presence of FGF. We imagine that this disassociation of clock arrest from mesp on-set timing could be the result of alterations in the thresholds in the sensing mechanisms controlling these two processes. Alternatively - as reviewer 2 argues - it might reflect multiple timing mechanisms at work. We have added a discussion of these alternative interpretations (page 17, lines 19-23 – page 18, line 1 - 8).

Fig. 3 supplement 3 B missing

It’s there in the BioRxiv downloadable PDF and full text – but seems to not be included when previewing the PDF. Thanks for the heads up.

Reviewer #2 (Public Review):

The authors demonstrate convincingly the potential of single mesodermal cells, removed from zebrafish embryos, to show cell-autonomous oscillatory signaling dynamics and differentiation. Their main conclusion is that a cell-autonomous timer operates in these cells and that additional external signals are integrated to tune cellular dynamics. Combined, this is underlying the precision required for proper embryonic segmentation, in vivo. I think this work stands out for its very thorough, quantitative, single-cell real-time imaging approach, both in vitro and also in vivo. A very significant progress and investment in method development, at the level of the imaging setup and also image analysis, was required to achieve this highly demanding task. This work provides new insight into the biology underlying embryo axis segmentation.

The work is very well presented and accessible. I think most of the conclusions are well supported. Here a my comments and suggestions:

The authors state that "We compare their cell-autonomous oscillatory and arrest dynamics to those we observe in the embryo at cellular resolution, finding remarkable agreement."

I think this statement needs to be better placed in context. In absolute terms, the period of oscillations and the timing of differentiation are actually very different in vitro, compared to in vitro. While oscillations have a period of ~30 minutes in vivo, oscillations take twice as long in vitro. Likewise, while the last oscillation is seen after 143 minutes in vivo, the timing of differentiation is very significantly prolonged, i.e.more than doubled, to 373min in vitro (Supplementary Figure 1-9). I understand what the authors mean with 'remarkable agreement', but this statement is at the risk of being misleading. I think the in vitro to in vivo differences (in absolute time scales) needs to be stated more explicitly. In fact, the drastic change in absolute timescales, while preserving the relative ones, i.e. the number of oscillations a cell is showing before onset of differentiation remains relatively invariant, is a remarkable finding that I think merits more consideration (see below).

We have changed the text in the abstract (page 1, line 28) to clarify that the agreement is in the relative slowing, intensity increases and peak numbers.

One timer vs. many timers

The authors show that the oscillation clock slowing down and the timing of differentiation, i.e. the time it takes to activate the gene mesp, are in principle dissociable processes. In physiological conditions, these are however linked. We are hence dealing with several processes, each controlled in time (and thereby space). Rather than suggesting the presence of ‘a timer’, I think the presence of multiple timing mechanisms would reflect the phenomenology better. I would hence suggest separating the questions more consistently, for instance into the following three:

a.  what underlies the slowing down of oscillations?

b.  what controls the timing of onset of differentiation?

c.  and finally, how are these processes linked?

Currently, these are discussed somewhat interchangeably, for instance here: “Other models posit that the slowing of Her oscillations arise due to an increase of time-delays in the negative feedback loop of the core clock circuit (Yabe, Uriu, and Takada 2023; Ay et al. 2014), suggesting that factors influencing the duration of pre-mRNA splicing, translation, or nuclear transport may be relevant. Whatever the identity, our results indicate the timer ought to exert control over differentiation independent of the clock.”(page 14). In the first part, the slowing down of oscillations is discussed and then the authors conclude on 'the timer', which however is the one timing differentiation, not the slowing down. I think this could be somewhat misleading.

To help distinguish the clock’s slowing & arrest from differentiation, we have clarified the text in how we describe our experiments using her1-/-; her7-/- cells (page 10, lines 9-20).

From this and previous studies, we learn/know that without clock oscillations, the onset of differentiation still occurs. For instance in clock mutant embryos (mouse, zebrafish), mesp onset is still occurring, albeit slightly delayed and not in a periodic but smooth progression. This timing of differentiation can occur without a clock and it is this timer the authors refer to "Whatever the identity, our results indicate the timer ought to exert control over differentiation independent of the clock." (page 14). This 'timer' is related to what has been previously termed 'the wavefront' in the classic Clock and Wavefront model from 1976, i.e. a "timing gradient' and smooth progression of cellular change. The experimental evidence showing it is cell-autonomous by the time it has been laid down,, using single cell measurements, is an important finding, and I would suggest to connect it more clearly to the concept of a wavefront, as per model from 1976.

We have been explicit about the connection to the clock & wavefront in the discussion (page 17, line 12-17).

Regarding question a., clearly, the timer for the slowing down of oscillations is operating in single cells, an important finding of this study. It is remarkable to note in this context that while the overall, absolute timescale of slowing down is entirely changed by going from in vivo to in vitro, the relative slowing down of oscillations, per cycle, is very much comparable, both in vivo and in vivo.

We have now pointed out the relative nature of this phenomenon in the abstract, page 1, line 28.

To me, while this study does not address the nature of this timer directly, the findings imply that the cell-autonomous timer that controls slowing down is, in fact, linked to the oscillations themselves. We have previously discussed such a timer, i.e. a 'self-referential oscillator' mechanism (in mouse embryos, see Lauschke et al., 2013) and it seems the new exciting findings shown here in zebrafish provide important additional evidence in this direction. I would suggest commenting on this potential conceptual link, especially for those readers interested to see general patterns.

While we posit that the timer provides positional info to the clock to slow oscillations and instruct its arrest – we do not believe that “the findings imply that the cell-autonomous timer that controls slowing down is, in fact, linked to [i.e., governed by] the oscillations themselves.”. As we show, in her1-/-; her7-/- embryos lacking oscillations, the timing / positional information across the PSM still exists as read-out by Mesp expression. Is this different positional information than that used by the clock? – possibly – but given the tight linkage between Mesp onset and the timing/positioning of clock arrest, both cell-autonomously and in the embryo, we argue that the simplest explanation is that the timing/positional information used by the clock and differentiation are the same. Please see page 10, lines 9-20, as well as the discussion (page 17, lines 19-23; page 18. Lines 1-8 ).

We agree that the timer must communicate to the clock– but this does not mean it is dependent on the clock for positional information.

Regarding question c., i.e. how the two timing mechanisms are functionally linked, I think concluding that "Whatever the identity, our results indicate the timer ought to exert control over differentiation independent of the clock." (page 14), might be a bit of an oversimplification. It is correct that the timer of differentiation is operating without a clock, however, physiologically, the link to the clock (and hence the dependence of the timescale of clock slowing down), is also evident. As the author states, without clock input, the precision of when and where differentiation occurs is impacted. I would hence emphasize the need to answer question c., more clearly, not to give the impression that the timing of differentiation does not integrate the clock, which above statement could be interpreted to say.

As far as we can tell, we do not state that “without clock input, the precision of when and where differentiation occurs is impacted”, and we certainly do not want to give this impression. In contrast, as mentioned above, the her1-/-; her7-/- mutant embryo studies indicate that the lack of a clock signal does not change the differentiation timing, i.e. it does not integrate the clock. Of course, in the formation of a real somite in the embryo, the clock’s input might be expected to cause a given cell to differentiate a little earlier or later so as to be coordinated with its neighbors, for example, along a boundary. But this magnitude of timing difference is within one clock cycle at most, and does not match the large variation seen in the cultured cells that spans over many clock cycles.

A very interesting finding presented here is that in some rare examples, the arrest of oscillations and onset of differentiation (i.e. mesp) can become dissociated. Again, this shows we deal here with interacting, but independent modules. Just as a comment, there is an interesting medaka mutant, called doppelkorn (Elmasri et al. 2004), which shows a reminiscent phenotype "the Medaka dpk mutant shows an expansion of the her7 expression domain, with apparently normal mesp expression levels in the anterior PSM.". The authors might want to refer to this potential in vivo analogue to their single cell phenotype.

Thank you, we had forgotten this result. Although we do not agree that this result necessarily means there are two interacting modules, we have included a citation to the paper, along with a discussion of alternative explanations for the dissociation (page 18, lines 2-14).

One strength of the presented in vitro system is that it enables precise control and experimental perturbations. A very informative set of experiments would be to test the dependence of the cell-autonomous timing mechanisms (plural) seen in isolated cells on ongoing signalling cues, for instance via Fgf and Wnt signaling. The inhibition of these pathways with well-characterised inhibitors, in single cells, would provide important additional insight into the nature of the timing mechanisms, their dependence on signaling and potentially even into how these timers are functionally interdependent.

We agree and in future experiments we are taking advantage of this in vitro system to directly investigate the effect of signaling cues on the intrinsic timing mechanism.

eLife assessment

This study provides valuable information about the microbiome and metabolome, and their correlation with acute myocardial infarction. However, the relationship established between these variables is limited to a correlation, and therefore the strength of the evidence is incomplete.

Reviewer #1 (Public Review):

Summary:

The authors aimed to identify potential biomarkers for acute myocardial infarction (AMI) through blood metabolomics and fecal microbiome analysis. They found that long chain fatty acids (LCFAs) could serve as biomarkers for AMI and demonstrated a correlation between LCFAs and the gut microbiome. Additionally, in silico molecular docking and in vitro thrombogenic assays showed that these LCFAs can induce platelet aggregation.

Strengths:

The study utilized a comprehensive approach combining blood metabolomics and fecal microbiome analysis.

The findings suggest a novel use of LCFAs as biomarkers for AMI.

The correlation between LCFAs and the gut microbiome is a significant contribution to understanding the interplay between gut health and heart disease.

The use of in silico and in vitro assays provides mechanistic insights into how LCFAs may influence platelet aggregation.

Weaknesses:

The evidence is incomplete as it does not definitively prove that gut dysbiosis contributes to fatty acid dysmetabolism.

The study primarily shows an association between the gut microbiome and fatty acid metabolism without establishing causation.

Reviewer #2 (Public Review):

Summary:

Fan et al. investigated the relationship between early acute myocardial infarction (eAMI) and disturbances in the gut microbiome using metabolomics and metagenomics analyses. They studied 30 eAMI patients and 26 healthy controls, finding elevated levels of long-chain fatty acids (LCFA) in the plasma of eAMI patients.

Strengths:

The research attributed a substantial portion of LCFA variance in eAMI to changes in the gut microbiome, as indicated by omics analyses. Computational profiling of gut bacteria suggested structural variations linked to LCFA variance. The authors also conducted molecular docking simulations and platelet assays, revealing that eAMI-associated LCFAs may enhance platelet aggregation.

Weaknesses:

The results should be validated using different assays, and animal models should be considered to explore the mechanisms of action.

Author Response:

Reviewer #1 (Public Review):

Summary:

The authors aimed to identify potential biomarkers for acute myocardial infarction (AMI) through blood metabolomics and fecal microbiome analysis. They found that long chain fatty acids (LCFAs) could serve as biomarkers for AMI and demonstrated a correlation between LCFAs and the gut microbiome. Additionally, in silico molecular docking and in vitro thrombogenic assays showed that these LCFAs can induce platelet aggregation.

Strengths:

The study utilized a comprehensive approach combining blood metabolomics and fecal microbiome analysis.

The findings suggest a novel use of LCFAs as biomarkers for AMI.

The correlation between LCFAs and the gut microbiome is a significant contribution to understanding the interplay between gut health and heart disease.

The use of in silico and in vitro assays provides mechanistic insights into how LCFAs may influence platelet aggregation.

Weaknesses:

The evidence is incomplete as it does not definitively prove that gut dysbiosis contributes to fatty acid dysmetabolism.

We appreciate this reviewer’s insightful comment regarding the causal relationship between gut dysbiosis and fatty acid dysmetabolism. We acknowledge that our study primarily demonstrates a strong association rather than causation. While establishing causality was beyond the scope of the current study, we recognize the importance of addressing this point. In our revised manuscript, we will emphasize the observational nature of our findings and discuss the need for future research, including longitudinal studies and interventional trials, to explore the causal links between gut dysbiosis and fatty acid dysmetabolism. We believe that this clarification strengthens the interpretation of our results and aligns with the reviewer's concern.

The study primarily shows an association between the gut microbiome and fatty acid metabolism without establishing causation.

We agree with the reviewer that our study presents an association rather than definitive proof of causation between the gut microbiome and fatty acid metabolism. To address this, we plan to expand the discussion section to more clearly outline the limitations of our study in establishing causality. We will also propose future research directions, such as the use of animal models and longitudinal human studies, which could help elucidate the causal pathways. By clarifying this aspect, we aim to provide a more balanced perspective on our findings.

Reviewer #2 (Public Review):

Summary:

Fan et al. investigated the relationship between early acute myocardial infarction (eAMI) and disturbances in the gut microbiome using metabolomics and metagenomics analyses. They studied 30 eAMI patients and 26 healthy controls, finding elevated levels of long-chain fatty acids (LCFA) in the plasma of eAMI patients.

Strengths:

The research attributed a substantial portion of LCFA variance in eAMI to changes in the gut microbiome, as indicated by omics analyses. Computational profiling of gut bacteria suggested structural variations linked to LCFA variance. The authors also conducted molecular docking simulations and platelet assays, revealing that eAMI-associated LCFAs may enhance platelet aggregation.

Weaknesses:

The results should be validated using different assays, and animal models should be considered to explore the mechanisms of action.

We appreciate the reviewer’s suggestion to validate our findings using additional assays and animal models. We agree that further validation is crucial to confirm the robustness of our results and to explore the underlying mechanisms in greater detail. While our current study focused on human subjects and in vitro assays to establish initial findings, we acknowledge that additional experimental approaches are necessary. In the revised manuscript, we plan to include a discussion on the potential use of different assays (e.g., advanced metabolomics techniques, multi-omics integration) and animal models to validate and expand upon our findings. Moreover, we are planning to undertake these experiments in future studies to build upon the foundational work presented here.

We believe that our revised responses and the planned manuscript revisions will address the reviewers’ concerns effectively. We are confident that these changes will enhance the overall contribution of our study to the field. Thank you again for your valuable feedback.

eLife assessment

This important study advances our understanding of the mechanisms of neuronal large dense-core vesicle (LDCV) secretion, which mediates neuropeptide and neurotrophin release. It describes a negative regulatory process involving the interaction of the Rab3-effector Rabphilin-3A with the SNARE fusion protein SNAP25, which limits LDCV secretion and neurite growth. The evidence in support of the authors' claims is generally convincing, but some conclusions, e.g regarding the role of Rabphilin-3A-controlled neurotrophin signaling in neurite growth, are incompletely supported. This study will be of interest to the fields of cell biology, cellular neuroscience, and neuroendocrinology.

Joint Public Review

The molecular mechanisms that mediate the regulated exocytosis of neuropeptides and neurotrophins from neurons via large dense-core vesicles (LDCVs) are still incompletely understood. Motivated by their earlier discovery that the Rab3-RIM1 pathway is essential for neuronal LDCV exocytosis, the authors now examined the role of the Rab3 effector Rabphilin-3A in neuronal LDCV secretion. Based on live, confocal, and super-resolution imaging approaches, the authors provide evidence for a synaptic enrichment of Rabphilin-3A and for independent trafficking of Rabphilin-3A and LDCVs. Using an elegant NPY-pHluorin imaging approach, they show that genetic deletion of Rabphilin-3A causes an increase in electrically triggered LDCV fusion events and increased neurite length. Finally, knock-out-replacement studies, involving Rabphilin-3A mutants deficient in either Rab3- or SNAP25-binding, indicate that the synaptic enrichment of Rabphilin-3A depends on its Rab3 binding ability, while its ability to bind to SNAP25 is required for its effects on LDCV secretion and neurite development. The authors conclude that Rabphilin-3A negatively regulates LDCV exocytosis and propose that this mechanism also affects neurite growth, e.g. by limiting neurotrophin secretion. These are important findings that advance our mechanistic understanding of neuronal large dense-core vesicle (LDCV) secretion.

The major strengths of the present paper:

(i) The use of a powerful Rabphilin-3A KO mouse model.<br /> (ii) Stringent lentiviral expression and rescue approaches as a strong genetic foundation of the study.<br /> (iii) An elegant FRAP imaging approach.<br /> (iv) A cutting-edge NPY-pHluorin-based imaging approach to detect LDCV fusion events.

Weaknesses of the present paper:

(i) It remains unclear why a process that affects a general synaptic SNARE fusion protein - SNAP25 - would specifically affect LDCV but not synaptic vesicle fusion.<br /> (iii) The mechanistic links between Rabphilin-3A function, LDCV density in neurites, neurite outgrowth, and the proposed underlying mechanisms involving trophic factor release remain unresolved.

Author response:

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

Public Reviews:

Joint Public Review: 

The molecular mechanisms that mediate the regulated exocytosis of neuropeptides and neurotrophins from neurons via large dense-core vesicles (LDCVs) are still incompletely understood. Motivated by their earlier discovery that the Rab3-RIM1 pathway is essential for neuronal LDCV exocytosis, the authors now examined the role of the Rab3 effector Rabphilin-3A in neuronal LDCV secretion. Based on multiple live and confocal imaging approaches, the authors provide evidence for a synaptic enrichment of Rabphilin-3A and for independent trafficking of Rabphilin-3A and LDCVs. Using an elegant NPY-pHluorin imaging approach, they show that genetic deletion of Rabphilin-3A causes an increase in electrically triggered LDCV fusion events and increased neurite length. Finally, knock-out-replacement studies, involving Rabphilin-3A mutants deficient in either Rab3- or SNAP25-binding, indicate that the synaptic enrichment of Rabphilin-3A depends on its Rab3 binding ability, while its ability to bind to SNAP25 is required for its effects on LDCV secretion and neurite development. The authors conclude that Rabphilin-3A negatively regulates LDCV exocytosis and propose that this mechanism also affects neurite growth, e.g. by limiting neurotrophin secretion. These are important findings that advance our mechanistic understanding of neuronal large dense-core vesicle (LDCV) secretion. 

The major strengths of the present paper are: 

(i) The use of a powerful Rabphilin-3A KO mouse model. 

(ii) Stringent lentiviral expression and rescue approaches as a strong genetic foundation of the study. 

(iii) An elegant FRAP imaging approach. 

(iv) A cutting-edge NPY-pHluorin-based imaging approach to detect LDCV fusion events. 

We thank the reviewers for their positive evaluation of our manuscript.

Weaknesses that somewhat limit the convincingness of the evidence provided and the corresponding conclusions include the following: 

(i) The limited resolution of the various imaging approaches introduces ambiguity to several parameters (e.g. LDCV counts, definition of synaptic localization, Rabphilin-3A-LDCV colocalization, subcellular and subsynaptic localization of expressed proteins, AZ proximity of Rabphilin-3A and LDCVs) and thereby limits the reliability of corresponding conclusions. Super-resolution approaches may be required here. 

We thank the reviewer for their constructive suggestion. We fully agree that super-resolution imaging would produce a more precise localization of RPH3A and co-localization with DCVs. We have now repeated our (co)-localization experiments with STED microscopy. We find that RPH3A colocalized with the pre-synaptic marker Synapsin1 and, to a lesser extent, with the post synaptic marker Homer and DCV marker chromogranin B (new Figure 1). This indicates that RPH3A is highly enriched in synapses, mostly the pre-synapse, and that RPH3A partly co-localizes with DCVs.  

(ii) The description of the experimental approaches lacks detail in several places, thus complicating a stringent assessment. 

We apologize for the lack of detail in explaining the experimental approaches. We have included a more detailed description in the revised manuscript. 

(iii) Further analyses of the LDCV secretion data (e.g. latency, release time course) would be important in order to help pinpoint the secretory step affected by Rabphilin-3A. 

We agree. To address this comment, we have now included the duration of the fusion events (new Figure S2D-F). The start time of the fusion events are shown in the cumulative plots in now Figure 3F and I. The kinetics are normal in the RPH3A KO neurons.

(iv) It remains unclear why a process that affects a general synaptic SNARE fusion protein - SNAP25 - would specifically affect LDCV but not synaptic vesicle fusion. 

We agree that we have not addressed this issue systematically enough in the original manuscript. We have now added a short discussion on this topic in the Discussion of the revised manuscript (p 15, line 380-386). In brief, we do not claim full selectivity for the DCV pathway. Some effects of RPH3A deficiency on the synaptic vesicle cycle have been observed. Furthermore, because DCVs typically do not mix in the synaptic vesicle cluster and fuse outside the active zone (and outside the synapse), DCVs might be more accessible to RPH3A regulation.

(v) The mechanistic links between Rabphilin-3A function, LDCV density in neurites, neurite outgrowth, and the proposed underlying mechanisms involving trophic factor release remain unclear. 

We agree that we have not addressed all these links systematically enough in the original manuscript, although we feel that we have at least postulated the best possible working model to link RPH3A function to DCV exocytosis/neurotrophic factor release and neurite outgrowth (p 15-16, line 396-400). Of course, a single study cannot support all these links with sufficient experimental evidence. We have now added a short text on what we can conclude exactly based on our experiments and how we see the links between RPH3A function, DCV exocytosis/neurotrophic factor release, neurite outgrowth and DCV density in neurites (p 13-14, line 317-325).

Reviewer #1 (Public Review): 

Summary:

The manuscript by Hoogstraaten et al. investigates the effect of constitutive Rabphilin 3A (RPH3A) ko on the exocytosis of dense core vesicles (DCV) in cultured mouse hippocampal neurons. Using mCherry- or pHluorin-tagged NPY expression and EGFP- or mCherry tagged RPHA3, the authors first analyse the colocalization of DCVs and RPH3A. Using FRAP, the authors next analyse the mobility of DCVs and RAB3A in neurites. The authors go on to determine the number of exocytotic events of DCVs in response to high-frequency electrical stimulation and find that RPH3A ko increases the number of exocytotic events by a factor 2-3, but not the fraction of released DCVs in a given cell (8x 50Hz stim). In contrast, the release fraction is also increased in RBP3A KOs when doubling the stimulation number (16x 50Hz). They further observe that RPH3A ko increases dendrite and axon length and the overall number of ChgrB-positive DCVs. However, the overall number of DCVs and dendritic length in ko cells directly correlate, indicating that the number of vesicles per dendritic length remains unaffected in the RPH3A KOs. Lentiviral co-expression of tetanus toxin (TeNT) showed a non-significant trend to reduce axon and dendrite length in RPH3a KOs. Finally, the authors use co-expression of RAB3A and SNAP25 constructs to show that RAB3A but not SNAP25 interaction is required to allow the exocytosis-enhancing effect in RPH3A KOs. 

While the authors' methodology is sound, the microscopy results are performed well and analyzed appropriately, but their results in larger parts do not sufficiently support their conclusions. Moreover, the experiments are not always described in sufficient detail (e.g. FRAP; DCV counts vs. neurite length) to fully understand their claims. 

Overall, I thus feel that the manuscript does not provide a sufficient advance in knowledge. 

Strengths: 

- The authors' methodology is sound, and the microscopy results are performed well and analyzed appropriately. 

- Figure 2: The exocytosis imaging is elegant and potentially very insightful. The effect in the RPH3A KOs is convincing. 

- Figure 4: the logic of this experiment is elegant. It shows that the increased number of DCV fusion events in RPH3A KOs is related to the interaction of RPH3A with RAB3A but not with SNAP25. 

We thank the reviewer for their positive evaluation of our manuscript.

Weaknesses: 

- The results in larger parts do not sufficiently support the conclusions. 

- The experiments are not always described in sufficient detail (e.g. FRAP; DCV counts vs. neurite length) to fully understand their claims. 

- Not of sufficient advance in knowledge for this journal 

- The significance of differences in control experiments WT vs. KO) varies between experiments shown in different figures. 

- Axons and dendrites were not analyzed separately in Figures 1 and 2. 

- The colocalization study in Figure 1 would require super-resolution microscopy. 

To address the reviewers’ comments, we have provided a more detailed explanation of our analysis (p 19-20, line 521-542). In addition, we have repeated our colocalization experiments using STED microscopy, see Joint Public Review item (i).  

Reviewer #2 (Public Review): 

Summary: 

Hoogstraaten et al investigated the involvement of rabphilin-3A RPH3A in DCV fusion in neurons during calcium-triggered exocytosis at the synapse and during neurite elongation. They suggest that RPH3A acts as an inhibitory factor for LDV fusion and this is mediated partially via its interaction with SNAP25 and not Rab3A/Rab27. It is a very elegant study although several questions remain to be clarified. 

Strengths: 

The authors use state-of-the-art techniques like tracking NPY-PHluorin exocytosis and FRAP experiments to quantify these processes providing novel insight into LDCs exocytosis and the involvement of RPH3A. 

We thank the reviewer for their positive evaluation of our manuscript.

Weaknesses: 

At the current state of the manuscript, further supportive experiments are necessary to fully support the authors' conclusions. 

We thank the reviewer for their comments and suggestions. We have performed additional experiments to support our conclusions, see Joint Public Review items (i) – (iv)

Reviewer #3 (Public Review): 

Summary: 

The molecular mechanism of regulated exocytosis has been extensively studied in the context of synaptic transmission. However, in addition to neurotransmitters, neurons also secrete neuropeptides and neurotrophins, which are stored in dense core vesicles (DCVs). These factors play a crucial role in cell survival, growth, and shaping the excitability of neurons. The mechanism of release for DCVs is similar, but not identical, to that used for SV exocytosis. This results in slow kinetic and low release probabilities for DCV compared to SV exocytosis. There is a limited understanding of the molecular mechanisms that underlie these differences. By investigating the role of rabphilin-3A (RPH3A), Hoogstraaten et al. uncovered for the first time a protein that inhibits DCV exocytosis in neurons. 

Strengths: 

In the current work, Hoogstraaten et al. investigate the function of rabphilin-3A (RPH3A) in DVC exocytosis. This RAB3 effector protein has been shown to possess a Ca2+ binding site and an independent SNAP25 binding site. Using colocalization analysis of confocal imaging the authors show that in hippocampal neurons RPH3A is enriched at pre- and post-synaptic sites and associates specifically with immobile DCVs. Using site-specific RPH3A mutants they found that the synaptic location was due to its RAB3 interaction site. They further could show that RPH3A inhibits DCV exocytosis due to its interaction with SNAP25. They came to that conclusion by comparing NPY-pHluorin release in WT and RPH3A KO cells and by performing rescue experiments with RPH3A mutants. Finally, the authors showed that by inhibiting stimulated DCV release, RPH3A controlled the axon and dendrite length possibly through the reduced release of neurotrophins. Thereby, they pinpoint how the proper regulation of DCV exocytosis affects neuron physiology. 

We thank the reviewer for their positive evaluation of our manuscript.

Weaknesses: 

Data context 

One of the findings is that RPH3A accumulates at synapses and is mainly associated with immobile DCVs.

However, Farina et al. (2015) showed that 66% of all DCVs are secreted at synapses and that these DCVs are immobile prior to secretion. To provide additional context to the data, it would be valuable to determine if RPH3A KO specifically enhances secretion at synapses. Additionally, the authors propose that RPH3A decreases DCV exocytosis by sequestering SNAP25 availability. At first glance, this hypothesis appears suitable. However, due to RPH3A synaptic localization, it should also limit SV exocytosis, which it does not. In this context, the only explanation for RPH3A's specific inhibition of DCV exocytosis is that RPH3A is located at a synapse site remote from the active zone, thus protecting the pool of SNAP25 involved in SV exocytosis from binding to RPH3A. This hypothesis could be tested using super-resolution microscopy. 

We thank the reviewer for their suggestion. We have now performed super resolution microscopy, see Joint Public Review item (i). However, these new data do not necessarily explain the stronger effect of RP3A deficiency on DCV exocytosis, relative to SV exocytosis. We have added a short discussion on this topic to the revised manuscript, see Joint Public Review item (iv).

Technical weakness 

One technical weakness of this work consists in the proper counting of labeled DCVs. This is significant since most findings in this manuscript rely on this analysis. Since the data was acquired with epi-fluorescence or confocal microscopy, it doesn't provide the resolution to visualize individual DCVs when they are clumped. The authors use a proxy to count the number of DCVs by measuring the total fluorescence of individual large spots and dividing it by the fluorescence intensity of discrete spots assuming that these correspond to individual DCVs. This is an appropriate method but it heavily depends on the assumption that all DCVs are loaded with the same amount of NPY-pHluorin or chromogranin B (ChgB). Due to the importance of this analysis for this manuscript, I suggest that the authors show that the number of DCVs per µm2 is indeed affected by RPH3A KO using super-resolution techniques such as dSTORM, STED, SIM, or SRRF. 

The reviewer is correct that this is a crucial issue, that we have not addressed optimally until now. We have previously devoted a large part of a previous manuscript to this issue, but have not referred to this previous work clearly enough. We have now clarified this (p 7, line 187-190). In brief, we have previously quantified the ratio between fluorescent intensity of ChgB and NPY-pHluorin in confocal microscopy over the number of dSTORM puncta in sparse areas of WT mouse hippocampal neurons (Persoon et al., 2018). This quantification yielded a unitary fluorescence intensity per vesicle that was very stable of different neurons. Although there might be some underestimation of the total number of DCVs when using confocal microscopy, the study of Persoon et al. (2018) has demonstrated that these parameters correlate well and that the estimations are accurate. Considering that the rF/F0 is similar in RPH3A WT and KO neurons (now Figure S2I), meaning that the intensity of NPY-pHluorin of one fusion event is comparable, we can presume that this correlation also applies for the RPH3A KO neurons.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors): 

Major points: 

(1) The authors perform an extensive analysis regarding the colocalization of RPH3A and DCVs (Figure 1 upper part). This analysis is hampered by the fact that the recorded data has in relation to vesicle size limited resolution (> 1 µm) to allow making strong claims here. In my view, super-resolution microscopy would be required for the co-localization studies shown in Figure 1. 

We fully agree and have now performed super-resolution microscopy, see Joint Public Review item (i)

(2) The FRAP experiments (Figure 1 lower part) cannot be sufficiently understood from what is presented. The methods say that both laser channels were activated during bleaching but NPY-pHluorin is not bleached in Fig.1E. Explanation of the bleaching is not very circumspect. In 1D, it is rather EGFP-RPH3A that is entering the bleached area than the NPY vesicles. These experiments require a more careful explanation of methodology, observed results, and their interpretation. Overall, the observed effects in the original kymograph traces require a better explanation. 

We acknowledge that NPY-pHluorin in Figure 1E (now Figure 2C) is not completely bleached. NPY-pHluorin appeared to be more difficult to bleach than NPY-mCherry. However, it is important to clarify that we merely bleached the neurites to remove the stationary puncta and facilitate our analysis of DCV/RPH3A dynamics. This bleaching step does not affect the interpretation of our results. We apologize that this was not clearly stated in the text and have made the necessary adjustments in legend, results- and methods section, (p 6-7, line 162-163; p 5, line 140-142 and p 19, line 508-513). Additionally, we apologize for the accidental switch of the kymographs for NPY-mCherry and EGFP-RPH3A in Figure 1D (now Figure 2B, C). We greatly appreciate identifying this error.  

(3) Figure 1: The authors need to mention whether axons, dendrites, or both were analyzed throughout the different panels and how they were identified. Is it possible that axons were wrapping around dendrites in their cultures (compare e.g. Shimojo et al., 2015)? Given the limited spatial resolution and because of this wrapping, interpretation of results could be affected. 

We completely agree with the reviewer’s assessment and conclusion. We are unable to distinguish axons from dendrites using this experimental design. We have made sure to specify in the text that our observation that RPH3A does not co-travel with DCVs is true for both dendrites and axons, (p 5, line 150).

(4) Figure 2: The exocytosis imaging is elegant and potentially very insightful. The effect in the RPH3A KOs is convincing. However, the authors determine the efficacy of exocytosis from NPY-pHluorin unquenching of DCVs only. This is only one of several possible parameters to read out the efficiency of exocytosis. Kinetics like e.g. delay between stimulation and start of exocytosis events or release time course of NPY after DCV fusion were not determined. Such analysis could give a better insight into what process before or after the fusion of DCVs is affected by RPH3A ko. 

We fully agree with the reviewer. We have now included the duration of the fusion events (new Figure S2D-F). The start time of the fusion events are shown in the cumulative plots in now Figure 3F and I. The kinetics are normal in the RPH3A KO neurons.

Moreover, it needs to be mentioned whether 2C and D are from WT or ko cultures. It would be best to show representative examples from both genotypes. 

We have now adjusted this in the new figure (now Figure 3C, D).

The number of fusion events is much increased but the release fraction is not significantly changed. While this is consistent with results in Figure 4C it is at variance with 4F. This raises questions about the reliability of the effects in RPH3A KOs. 

The release fraction indicates the number of fusion events normalized to the total DCV pool. In Figure 4D, we observed a slightly bigger pool size, which explains the lack of significance when analyzing the released fraction. In Figure 4G, however, DCV pool sizes are similar between KO and WT, leading to a statistically significant effect on release fraction in KO neurons. Furthermore, Figures 4B and E distinctly show a substantial increase in fusion events in RPH3A KO neurons. This variability in pool size observed could potentially be attributed to variation in culture or inherent biological variability.

Given the increased number of ChgrB-positive DCVs in RPH3A KOs (shown in Figure 2) and that only the cumulative number of exocytosis events were analysed, how can the authors exclude that the RPH3A ko only affects vesicle number but not release, if the % change in released vesicles is not different to WT? Kinetics of release don't seem to be affected. Importantly, what was the density of NPY-pHluorin vesicles in WT vs. ko? 

In Figure 2 (now Figure 5) we show that RPH3A KO neurons are larger and contain more endogenous ChgB+ puncta than WT neurons. This increased number of ChgrB+ puncta scales with their size as puncta density is not increased. A previous study (Persoon et al., 2018) has demonstrated a strong correlation between DCV number and neuron size. Our data show that RPH3A deficiency increased DCV exocytosis, but the released fraction of vesicles depends on the total number of DCVs, which we determined during live recording by dequenching NPY-pHluorin using NH4+. Considering that this is an overexpression of a heterologous DCV-fusion reporter, and not endogenous staining of DCVs, as in the case of ChgrB+ puncta, some variability is not unexpected.

Also in these experiments, the question arises of whether the authors analyse axons, dendrites, or both throughout the different panels and how they were identified. 

In our experimental design we record all fusion events per cell, including both axons and dendrites but excluding the cell soma. We have clarified this in the method section, (p 19, line 508 and p 19, line 521-522).

(5) Figure 3: in D the authors show that ChgrB-pos. DCV density is slightly increased in KOs. How does this relate to the density of NPY-pHluorin DCVS in Figure 2? 

We do not observe a difference in NPY-pHluorin density (see Author response image 1). However, it is important to note that we relied on tracing neurites in live recording images to determine the neuronal size. In contrast, the ChgB density was based on dendritic length using MAP2 (post-hoc) staining was limited. In addition, Chgr+ puncta represent an endogenous DCV staining, NPY-pHluorin quantification is based on overexpression of a heterologous DCV-fusion reporter. These two factors likely contribute some variability.

Author response image 1.

The authors show a non-significant trend of TeNT coexpression to reduce axon and dendrite lengths in RPH3A KOs. While this trend is visible, I think one cannot draw conclusions from that when not reaching significance. The argument of the authors that the increased axon and dendrite lengths are created by growth factor peptide release from DCV during culture time is interesting. However, the fact that TeNT expression shows a trend toward reducing this effect on axons/dendrites is not sufficient to prove the release of such growth factors. 

We agree. We have toned down this speculation in the revised manuscript, (p 15-16, line 395-400).

Lastly, the authors don't provide insight into the mechanisms, of how RPH3A ko increases the number of DCVs per µm dendritic length in the neurons. In my view, there are too many loose ends in this story of how RPH3A ko first increases spontaneous release of DCVs and then enhances neurite growth and DCV density. Did the authors e.g. measure the spontaneous release of DCVs in their cultures? 

We measured spontaneous release of DCVs during the 30s baseline recording prior to stimulation. We observed no difference in spontaneous release between WT and KO neurons (now Figure S2H). However, baseline recording lasted only 30 seconds. It is possible that this was too short to detect subtle effects.

Other points: 

(1) Figure 4: the logic of this experiment is elegant. It shows that the increased number of DCV fusion events in RPH3A KOs is related to the interaction of RPH3A with RAB3A but not with SNAP25. As mentioned above, it is irritating that the reduction of fusion events in KOs and on the release fraction is sometimes reaching significance, but sometimes it does not. Likewise, the absence of significant effects on DCV numbers is not consistent with the results shown in Figures 3C and D. 

DCV numbers in Figure 3 (now Figure 5) are determined by staining for endogenous ChgB, whereas in Figure 4D and G DCV numbers are determined by overexpressing NPY-pHluorin and counting the dequenched puncta following a NH4+ puff.

(2) Figure 1B: truncation of the y-axis needs to be clearly indicated. 

We have replaced this figure with new Figure 1 and have indicated truncations of the y-axis when needed (new Figure 1E). 

(3) Page 10: "Given that neuropeptides are key modulators of adult neurogenesis (Mu et al., 2010), and that RPH3A depletion leads to increased DCV exocytosis, it is coherent that we observed longer neurites in RPH3A KO neurons." I cannot follow the argument of the authors here: what has neurogenesis to do with neurite length? 

We apologize for the confusion. We have clarified this in the revised text, (p 16, line 398-400).

Minor point: 

There are some typos in the manuscript. e.g., page 8: "... may partially dependent on regulated secretion...); page 6: "...to dequence all...". 

Thank you for noticing, we have corrected the typos.

Reviewer #2 (Recommendations For The Authors): 

(1) Supplementary Figure S1A, in my opinion, should be in Figure 1A as it illustrates all the constructs used in this study and helps the reader to follow it up. 

We thank the reviewer for their suggestion. However, we feel that with the adjustments we have made in Figure 1, the illustrations of the constructs fit better in Figure S1, since new Figure 1 shows the localization of endogenous RPH3A and not that of the constructs.  

(2) One of the conclusions of the manuscript is the synaptic localization of the different RPH3A mutants. The threshold for defining synaptic localization is not clear either from the images nor from the analysis: for example, the Menders coefficient for VGut1-Syn1 which is used as a positive control, ranges from 0.65-0.95 and that of RPH3A and Syn1 ranges from 0.5-0.95. These values should be compared to all mutants and the conclusions should be based on such comparison. 

We agree. We have now repeated our initial co-localization experiment with all the RPH3A mutants (now Figure S1D-F).  

(3) Strengthening this figure with STED/SIM/dSTORM microscopy can verify and add a new understanding of the subtle changes of RPH3A localization. 

We fully agree and have now added super-resolution microscopy data, see Joint Public Review item (i).

(4) As RAB3A/RAB27A (ΔRAB3A/RAB27A) loses the punctate distribution, please clarify how can it function at the synapse and not act as a KO. Is it sorted to the synapse and how does it is sorted to the synapse? 

We used lentiviral delivery to introduce our constructs, resulting in the overexpression of ΔRAB3A/RAB27A mutant RPH3A. This overexpression likely compensates for the loss of the punctate distribution of RPH3A, thereby maintaining its limiting effect on DCV exocytosis. It is plausible that under physiological conditions, the mislocalization of RPH3A would lead to increased exocytosis, similar to what we observed in the KO. 

(5) Is RPH3A expressed in both excitatory and inhibitory neurons? 

We agree this is an important question. Single cell RNA-seq already suggests the protein is expressed in both, but we nevertheless decided to test expression of RPH3A protein in excitatory and inhibitory neurons, using immunocytochemistry with VGAT and VGLUT as markers in hippocampal and striatal WT neurons. We found that RPH3A is expressed in both VGLUT+ hippocampal neurons and VGAT+ striatal neurons (new Figure S1A, B).  

(6) The differential use of ChgB and NPY as markers for DCVs should be clarified and compared as these are used at different stages of the manuscript. 

We have previously addressed the comparison between ChgB and NPY-pHluorin (Persoon et al., 2018). We made sure to indicate this more clearly throughout the manuscript to clarify the use of the two markers. 

(7) FRAP experiments- A graph describing NPY recovery should be added as a reference to 2H and discussed. 

We agree. We have made the necessary adjustments (new Figure 2G).

(8) Figure 2E shows some degree of "facilitation" between the 2 8x50 pulses RPH3A KO neurons. Can the author comment on that? What was the reason for using this dual stimulation protocol? 

There is indeed some facilitation between the two 8 x 50 pulses in KO neurons and to a lesser extent also in the WT neurons, which we have observed before in WT neurons (Baginska et al., 2023). Baginska et al. (2023) showed recently that different stimulation protocols can influence certain fusion dynamics, like the ratio of persistent and transient events and event duration. We used two different stimulation protocols to thoroughly investigate the effect of RPH3A on exocytosis, and assess the robustness of our findings regarding the number of fusion events. Fusion kinetics was similar in WT an KO neurons for both stimulation protocols (new Figure 2D-F).

(9) Figure 3 quantifies dendrites length and then moves to quantify both axon and dendrites for the Tetanus toxin experiment. What are the effects of KO on axon length? In the main figures, it is not mentioned but in S3 it seems not to be affected. How does it reconcile with the main conclusion on neurite length? 

Figure 3H (now Figure 6C) shows the effect of the KO on axon length: the axon length is increased in RPH3A KO neurons compared to WT, similar to dendrite length. Re-expressing RPH3A in KO neurons rescues axonal length to WT levels. In Figure S3, we observe a similar trend as in main Figure 3 (new Figure 6), yet this effect did not reach significance. Based on this, we concluded that neurite length is increased upon RPH3A depletion.

(10) For lay readers, please explain the total pool and how you measured it. However, see the next comment. 

We agree. We have now defined this better in the revised manuscript, (p 19, line 524-527 and p 20, line 535-539).

(11) It is a bit hard to understand if the total number of DCV was increased in the KO and if the pool size was increased and in which figure it is quantified. Some sentences like: "A trend towards a larger intracellular DCV pool in KO compared to WT neurons was observed" do not fit with "No difference in DCV pool size was observed between WT and KO neurons (Figure S2D)" or with "During stronger stimulation (16 bursts of 50 APs at 50 Hz), the total fusion and released fraction of DCVs were increased in KO neurons compared to WT". They are not directly supported, or not related to specific figures. Please indicate if the total DCVs pool, as measured by NH4, was increased and based on that, the fraction of the releasable DCVs following the long stimulation. From Figure 2H, the conclusion is an increase in fusion events. In general, NH4 is not quantified clearly- is it quantified in Figure S2C? And if it is a trend, how can it become significant in Figure 3? 

We agree there has been some inconsistency in the way we describe the data on the total number of DCVs. We have addressed this in the revised text to ensure better clarity. The total DCV pool measured by NPY-pHluorin was not significantly increased in KO neurons, we see a trend towards a bigger DCV pool in the 2x8 50 Hz stimulation paradigm (now Figure S2C), therefore the released fraction of vesicles is not increased in Figure 1G (now Figure 3G). The number of DCV in Figure 3 (now Figure 5) is based on endogenous ChgB staining and not overexpression like the DCV pool measured by NPY-pHluorin. In Figure 3 (now Figure 5) we show that RPH3A KO neurons have slightly more ChgB+ puncta compared to WT.

(12) In Figure 3, the quantification is not clear, discrete puncta are not visible but rather a smear of chromogranin staining. How was it quantified? An independent method to count DCV number, size, and distribution like EM is necessary to support and add further understanding. 

We acknowledge that discrete ChgB puncta are not completely visible in Figure 3 (now Figure 5). Besides the inherent limitation in resolution with confocal imaging, we believe that this is due to ChgB accumulation in the KO neurons, as shown in now Figure 5D. Nonetheless, to address this concern of the reviewer, we have selected other images that represent our dataset (now Figure 5A). Furthermore, the number of ChgB+ DCVs was calculated using SynD software (Schmitz et al., 2011; van de Bospoort et al., 2012) (see previous reply). EM would offer valuable independent confirmation on the total DCV number, size and distribution. However, with the current method we already know that vesicle numbers are at least similar. Does that justify the (major) investment in a quantitative EM study? Moreover, this issue does not affect the central message of the current study.

(13) Can the author discuss if the source of DCVs that are released at the synapse is similar or different from the source of DCVs fused while neurites elongate? 

With our current experimental design, we are unable to draw conclusions regarding this aspect. We are not sure how experiments to identify this source (probably the Golgi?) would be crucial to sustain the central message of our study.

(14) An interesting and related question: what are the expression levels of RPH3A during development and neuronal growth during the nervous system development? 

While we have not specifically examined the expression levels of RPH3A over development, public databases show that RPH3A expression increases over time in mice, consistent with other synaptic proteins (Blake et al., 2021; Baldarelli et al., 2021; Krupke et al., 2017). We have now added this to the revised manuscript (p 2, line 55-56).

(15) The conclusion from Figure 4 about the contribution of SNAP25 interaction to RPH3A inhibitory effect is not convincing. The data are scattered and in many neurons, high levels of fusion events were detected. Further or independent experiments are needed to support this conclusion. For example, is the interaction with SNAP25 important for its inhibitory activity in other DCV-releasing systems like adrenal medulla chromaffin cells? 

We agree that further studies in other DCV-releasing systems like chromaffin cells would provide valuable insight into the role of SNAP25 interaction in RPH3A’s inhibitory effect on exocytosis. However, we believe that starting new series of experiments in another model system is outside of the scope of our current study.

(16) Furthermore, the number of DCVs in the KO is similar in this experiment, raising some more questions about the quantification of the number of vesicles, that differ, in different sections of the manuscript (points # 10,11). 

The total DCV pool in the fusion experiments is measured by overexpression NPY-pHluorin, this cannot be directly compared to the number of endogenous ChgB+ DCV in Figure 3 (now Figure 5), see also item (11)

(17) The statement - "RPH3A is the only negative regulator of DCV" is not completely accurate as other DCV inhibitors like tomosyn were described before. 

We agree. By this statement, we intend to convey that RPH3A is the only negative regulator of DCVs without substantial impact on synaptic vesicle exocytosis, unlike Tomosyns. We have clarified this in the revised text, (p 15, line 366-367).

(18) The support for the effect of KO on the "clustering of DCVs" is not convincing. 

The intensity of endogenous ChgB puncta was decreased in RPH3A KO neurons (now Figure 5E). However, the peak intensity induced by single NPY-pHluorin labeled DCV fusion events (quanta) was unchanged (now Figure S2I). This indicates that the decrease in ChgB puncta intensity must be due to a reduced number of DCVs (quanta) in this specific location. We have interpreted that as ‘clustering’, or maybe ‘accumulation’. However, we only put forward this possibility. We are now more careful in our speculations within the text, (p 11 line 271-277).

(19) Final sentence: "where RPH3A binds available SNAP25, consequently restricting the assembly of SNARE complexes" should be either demonstrated or rephrased as no effect of trans or general SNARE complex formation is shown. 

We agree. We have made the necessary adjustments in the text, (p 15, line 387-389).   

(20) A scheme summarizing RPH3A's interaction with synaptic proteins and its effects on DCVs release, maybe even versus its effects on SVs release, should be considered as a figure or graphic abstract. 

We have included a working model in Figure 7.  

(21) Figure 4 logically should come after Figure 2 to summarize the fusion-related chapter before moving to neurite elongation. 

We have placed Figure 4 after Figure 2 (now Figure 3).

Reviewer #3 (Recommendations For The Authors): 

One important finding of this study is that RPH3A downregulates neuron size, possibly by inhibiting DCV release. Additionally, the authors demonstrated that the number of DCVs is directly proportional to the number of DCVs per µm2, and that RPH3A KO reduces DCV clustering. This conclusion was drawn by comparing ChgB with NPY-pHluorin loading of the DCVs. However, this comparison is not valid as ChgB is expressed at an endogenous level and NPY-pHluorin is over-expressed. In the KO situation where DCV exocytosis is enhanced, the available endogenous ChgB may be depleted faster than the overexpressed NPY-pHluorin. Hoogstraaten et al. should either perform a study in which ChgB is overexpressed to test whether the difference in DCV remains or at least provides an alternative interpretation of their data. 

We thank the reviewer for this comment. The reviewer challenges one or two conclusions in our original manuscript (It is not entirely clear to what exactly “This conclusion” refers): (a) “the number of DCVs is directly proportional to the number of DCVs per µm2”, and (b) “that RPH3A KO reduces DCV clustering”. The reviewer probably means that the number of DCVs per neuron is directly proportional to size of the neuron (a) and states this (these) conclusion(s) are “not valid as ChgB is expressed at an endogenous level and NPY-pHluorin is over-expressed” because “endogenous ChgB may be depleted faster than the overexpressed NPY-pHluorin”. We have three arguments to conclude that faster depletion of ChgB cannot affect these two conclusions: (1) DCVs bud off from the Golgi with newly synthesized (fresh) ChgB. Whether or not a larger fraction of DCVs is released does not influence this initial ChgB loading into DCVs (together with over-expressed NPY-pHluorin); (2) in hippocampal neurons merely 1-6% of the total DCV pool undergoes exocytosis (the current study and also extensively demonstrated in Persoon et al., 2018). RPH3A KO neurons release few percent more of the total DCV pool. Hence, “depletion of ChgB” is only marginally different between experimental groups; and (c) the proposed experiment overexpressing ChgB will not help scrutinize our current conclusions as ChgB overexpression is known to affect DCV biogenesis and the total DCV pool, most likely much more than a few percent more release by RPH3A deficiency.

Hoogstraaten et al. conducted a thorough analysis of the impact of RPH3A KO and its rescue using various mutants on dendrite and axon length (see Supplementary Figure 3). However, they did not test the effect of the ΔSNAP25 mutant. The authors demonstrated that this mutant is the least efficient in rescuing DCV exocytosis (Figure 4E). Hence the neurons expressing this mutant should have a similar size to the KO neurons. This finding would strongly support the argument that DCV exocytosis regulates neuron size. Otherwise, it would suggest that RPH3A may have a function in regulating exocytosis at the growth cones that is independent of SNAP25. Since the authors most probably have the data that allows them to measure the neuron size (acquired for Supplementary Figure 2), I suggest that they perform the required analysis. 

We agree this is important and performed new experiments to determine the dendrite length of RPH3A WT, KO and KO neurons expressing the ΔSNAP25 mutant. We observed that the dendrite length of RPH3A KO neurons expressing ΔSNAP25 mutant is indeed similar to KO neurons (new Figure S3C). Although not significant we observe a clear trend towards bigger neurons compared to WT.  This strengthens our conclusion that increased DCV exocytosis contributes to the observed increased neuronal size.

The authors displayed the result of DCV exocytosis in two ways. One is by showing the number of exocytosis events the other is to display the proportion of DCVs that were secreted. They do the latter by dividing the secreted DCV by the total number of DCVs. These are visualized at the end of the experiment through NH4+ application. While this method works well for synaptic secretion as the marker of SV is localized to the SV membrane and remains at the synapse upon SV exocytosis, it cannot be applied in the same manner when it is the DCV content that is labeled as it is released upon secretion. Hence, the total pool of vesicles should be the number of DCV counted upon NH4+ application in addition to those that are secreted. This way of analyzing the total pool of DCV might also explain the difference in this pool size between KO neurons stimulated two times with 8 stimuli instead of one time with 16 stimuli (Sup Fig 2 C and D). This is an important point as it affects the conclusions drawn from Figure 2. 

We thank the reviewed for this comment. We agree, and we have made the necessary adjustments throughout the manuscript. 

The kymogram of DCV exocytic events displayed in Figure 2D shows a majority of persistent (>20s long) events. This is strange as NPY-pHluori corresponds to the released cargo. Previous work using the same labeling and stimulation technique showed that content release occurs in less than 10s (Baginska et al. 2023). The authors should comment on that difference. 

In Baginska et al. (2023), the authors distinguished between persistent and transient events. The transient events are shorter than 10s for the 2x8 and 16x stimulation paradigms, whereas persistent events can last for more than 10s. In our study we did not make this distinction. However, in response to this reviewer, we have now quantified the fusion duration per cell. These new data show that the mean duration is similar between genotypes for both stimulation paradigms. We have added these new data (new Figure S2D-F).

In Figures 1D and E, some puncta in the kymogram appeared to persist after bleaching. This raises questions about the effectiveness of the bleaching procedure for the FRAP experiment. 

The reviewer is correct that NPY-pHluorin in Figure 1E (now Figure 2C) is not fully bleached. NPY-pHluorin was more resistant to bleaching than NPY-mCherry. However, we merely bleached the neurites to facilitate our analysis by reducing fluorescence of the stationary puncta without causing phototoxicity. Some remaining fluorescence after bleaching does not affect our conclusions in any way.

In the discussion, the paragraph titled "RPH3A does not travel with DCVs in hippocampal neurons" is quite confusing and would benefit from a streamlined explanation. 

We thank the reviewed for this comment. We made the necessary adjustments to make this paragraph clearer, (p 14, line 339-351).

First paragraph of page 8 "TeNT expression in KO neurons restored neurite length to WT levels. When compared to KO neurons without TeNT, neurite length was not significantly decreased but displayed a trend towards WT levels (Figure 3G, H)." These two sentences are confusing as they seem contradictory. 

We agree that this conclusion has been too strong. However, we do not see a contradiction. The significant effect between KO and control neurons on both axon and dendrite length is lost upon TeNT expression (which forms the basis for our conclusions cited by the reviewer, now Figure 6B, C). While the difference between KO neurons +/- TeNT did not reach statistical significance. The (strong) trend is clearly in the same direction. We have refined our original conclusion in the revised manuscript, (p 12, line 304-306).

The data availability statement is missing. 

We have added the data availability statement, (p 21, line 571-572).

eLife assessment

This is an important paper on the role of engrams and relevant conditions that influence memory and forgetting. The variety of methods used, namely, behavioural, labeling, interrogation, immunohistochemistry, microscopy, pharmacology, computational, are exemplary and provide convincing evidence for the role of engrams in the dentate gyrus in memory retrieval and forgetting. This examination will be of interest broadly across behavioural and neural science communities.

Reviewer #1 (Public Review):

O'Leary and colleagues sought to understand the factors that underlie memory processes, including formation, retrieval, and forgetting. The present data identify time, environmental enrichment, Rac-1, context reexposure, and brief reminders of the familiar object as factors that alter discrimination between novel and familiar objects. This is complimented with an engram approach to quantify cells that are active during learning to examine how their activation is impacted with each of the above factors at test. There are many strengths in the manuscript, including systematic testing of several factors that contribute to poor discrimination between novel and familiar objects. These results are interesting and outline essential boundaries of incidental, nonaversive memory. With this behavioral data, authors apply a modeling approach to understand the factors that contribute to good and poor object memory recall.

Reviewer #2 (Public Review):

Summary:

The manuscript examines an important question about how an inaccessible, natural forgotten memory can be retrieved through engram ensemble reactivation. It uses a variety of strategies including optogenetics, behavioral and pharmacological interventions to modulate engram accessibility. The data characterize the time course of natural forgetting using an object recognition task, in which animals can retrieve 1 day and 1 week after learning, but not 2 weeks later. Forgetting is correlated with lower levels of cell reactivation (c-fos expression during learning compared to retrieval) and reduction in spine density and volume in the engram cells. Artificial activation of the original engram was sufficient to induce recall of the forgotten object memory while artificial inhibition of the engram cells precluded memory retrieval. Mice housed in an enriched environment had a slower rate of forgetting, and a brief reminder before the retrieval session promoted retrieval of a forgotten memory. Repeated reintroduction to the training context in the absence of objects accelerated forgetting. Additionally, activation of Rac1-mediated plasticity mechanisms enhanced forgetting, while its inhibition prolonged memory retrieval. Authors also reproduce the behavioral findings using a computational model inspired by Rescorla-Wagner model. In essence, the model proposes that forgetting is a form of adaptive learning that can be updated based on prediction error rules in which engram relevancy is altered in response to environmental feedback.

Strengths:

(1) The data presented in the current paper are consistent with the authors claim that seemingly forgotten engrams are, in fact, accessible. This suggests that retrieval deficits can lead to memory impairments rather than a loss of the original engram (at least in some cases).

(2) The experimental procedures and statistics are appropriate, and the behavioral effects appear to be very robust. Several key effects are replicated multiple times in the manuscript.

Comments on revised version:

The authors have adequately addressed my prior concerns.

Reviewer #3 (Public Review):

Summary:

The manuscript by Ryan and colleagues uses a well-established object recognition task to examine memory retrieval and forgetting. They show that memory retrieval requires activation of the acquisition engram in the dentate gyrus and failure to do so leads to forgetting. Using a variety of clever behavioural methods, the authors show that memories can be maintained and retrieval slowed when animals are reared in environmental enrichment and that normally retrieved memories can be forgotten if exposed to the environment in which the expected objects are no longer presented. Using a series of neural methods, the authors also show that activation or inhibition of the acquisition engram is key to memory expression and that forgetting is due to Rac1.

Strengths:

This is an exemplary examination of different conditions that affect successful retrieval vs forgetting of object memory. Furthermore, the computational modelling that captures in a formal way how certain parameters may influence memory provides an important and testable approach to understanding forgetting.<br /> The use of the Rescorla-Wanger model in the context of object recognition and the idea of relevance being captured in negative prediction error are novel (but see below).<br /> The use of gain and loss of function approaches are a considerable strength and the dissociable effects on behaviour eliminate the possibility of extraneous variables such as light artifacts as potential explanations for the effects.

Weaknesses:

A closer examination of the process that governs the behavioural effect in the present investigation would have been of even greater significance. The authors acknowledge the distinction between object familiarity vs object recognition, but a direct assessment would benefit the field's understanding of the current role of engrams on behaviour.

Relatedly, while relevance is an interesting concept that has been operationalized in the paper, it is unclear how distinct it is from extinction. Specifically, in the case where the animals are exposed to the context in the absence of the object, the paper currently expresses this as a process of relevance - the objects are no longer relevant in that context. Another way to think about this is in terms of extinction - the association between the context and the objects is reduced resulting in a disrupted ability of the context to activate the object engram. The authors have noted the potential role of extinction in their studies.

The impact of the paper is somewhat limited by the use of only one sex. Do the authors expect an identical process to be engaged in females in the present set of studies?

Author response:

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

Public Reviews:

Reviewer #1 (Public Review):

Summary:

O'Leary and colleagues present data identifying several procedures that alter discrimination between novel and familiar objects, including time, environmental enrichment, Rac-1, context reexposure, and brief reminders of the familiar object. This is complimented with an engram approach to quantify cells that are active during learning to examine how their activation is impacted following each of the above procedures at test. With this behavioral data, authors apply a modeling approach to understand the factors that contribute to good and poor object memory recall.

We thank the Reviewer for summarizing the scope and depth of our manuscript, and indeed for recognizing our efforts. We engage below with the Reviewer’s specific criticisms.

Strengths:

Authors systematically test several factors that contribute to poor discrimination between novel and familiar objects. These results are extremely interesting and outline essential boundaries of incidental, nonaversive memory.<br /> These results are further supported by engram-focused approaches to examine engram cells that are reactivated in states with poor and good object recognition recall.

We thank the Reviewer for these positive comments.

Weaknesses:

For the environmental enrichment, authors seem to suggest objects in the homecage are similar to (or reminiscent of) the familiar object. Thus, the effect of improved memory may not be related to enrichment per se as much as it may be related to the preservation of an object's memory through multiple retrievals, not the enriching experiences of the environment itself. This would be consistent with the brief retrieval figure. Authors should include a more thorough discussion of this.

This is one of the main issues highlighted by the Editor and the Reviewers. We agree that these results dove-tail with the reminder experiments. We have included additional discussion, see line 510-546.

Authors should justify the marginally increased number of engram cells in the non-enrichment group that did not show object discrimination at test, especially relative to other figures. More specific cell counting criteria may be helpful for this. For example, was the DG region counted for engram and cfos cells or only a subsection?

There was a marginal, but non-significant increase in the number of labelled cells within the standard housed mice in Figure 3f. The cell counting criteria was the same across experimental groups and conditions, where the entire dorsal and ventral blade of the dorsal DG was counted for each animal. This non-statistically significant variance may be due to surgical and viral spread difference between mice. We have clarified this in the manuscript, see line 229-232.

It is unclear why the authors chose a reactivation time point of 1hr prior to testing. While this may be outside of the effective time window for pharmacological interference with reconsolidation for most compounds, it is not necessarily outside of the structural and functional neuronal changes accompanied by reconsolidation-related manipulations.

A control experiment was performed to demonstrated that a brief reminder exposure of 5 mins on its own was insufficient to induce new learning that formed a lasting memory (Supplementary Figure S4a). Mice given only a brief acquisition period of 5 mins, exhibited no preference for the novel object when tested 1 hour after training, suggesting the absence of a lasting object memory (Supplementary Figure S4b & c). We therefore used the 1-hour time point for the brief reminder experiment in Figure 4a. We have clarified this within the manuscript and supplementary data see line 258-264.

Figure 5: Levels of exploration at test are inconsistent between manipulations. This is problematic, as context-only reexposures seem to increase exploration for objects overall in a manner that I'm unsure resembles 'forgetting'. Instead, cross-group comparisons would likely reveal increased exploration time for familiar and novel objects. While I understand 'forgetting' may be accompanied by greater exploration towards objects, this is inconsistent across and within the same figure. Further, this effect is within the period of time that rodents should show intact recognition. Instead, context-only exposures may form a competing (empty context) memory for the familiar object in that particular context.

The Reviewer raises an important question, and we agree with the Reviewer that there should be caution and qualification around interpreting these results as “forgetting”. Indeed, for the context-only rexposures, cross-group comparisons show increased exploration time for familiar and novel objects. As the mice exhibit relatively high exploration of both the novel and familiar objects. An alternative explanation would be that the mice have not truly forgotten the familiar object, but rather as the mouse has not seen the familiar object in the last 6 context only sessions, its reappearance makes it somewhat novel again. Therefore, this change in the object’s reappearance triggers the animal’s curiosity, and in turn drives exploration by the animal. In addition, the context-only exposures may form a competing memory for the familiar object in that particular context. We have highlighted this in the results and also included greater discussion. See lines 306-315.

I am concerned at the interpretation that a memory is 'forgotten' across figures, especially considering the brief reminder experiments. Typically, if a reminder session can trigger the original memory or there is rapid reacquisition, then this implies there is some savings for the original content of the memory. For instance, multiple context retrievals in the absence of an object reminder may be more consistent with procedures that create a distinct memory and subsequently recruit a distinct engram.

These findings raise an important question regarding the interpretation of ‘forgetting’. If a reminder trial or experience can trigger the original memory, or there is rapid reacquisition, then this would suggest there is a degree of savings for the original memory content (85, 86). Previous work has emphasized retrieval deficits as a key characteristic of memory impairment, supporting the idea that memory recall or accessibility may be driven by learning feedback from the environment (7, 8, 14–18). Within our behavioral paradigm, a lack of memory expression would still constitute forgetting due to the loss of learned behavioral response in the presence of natural retrieval cues. The changes in memory expression may therefore underlie the adaptive nature of forgetting. This is consistent with the idea that the engram is intact and available, but not accessible. Here we studied natural forgetting, and our data showing memory retrieval following optogenetic reactivation demonstrates that the original engram persists at a cellular level, otherwise activation of those cells would no longer trigger memory recall. We also agree with the reviewer that multiple context retrievals may indeed lead to the formation of a second distinct engram that competes with the original. Recent work suggests that retroactive interference emerges from the interplay of multiple engrams competing for accessibility (18). We have added clarification and included extra discission of this interpretation. See lines 589-598.

Authors state that spine density decreases over time. While that may be generally true, there is no evidence that mature mushroom spines are altered or that this is consistent across figures. Additionally, it's unclear if spine volume is consistently reduced in reactivated and non-reactivated engram cells across groups. This would provide evidence that there is a functionally distinct aspect of engram cells that is altered consistently in procedures resulting in poor recognition memory (e.g. increased spine density relative to spine density of non-reactivated engram cells and non-engram cells)

We thank the Reviewer for their helpful comments on explaining our engram dendritic spine data. We agree with the Reviewer that an analysis of the changes in spine type, as well as the difference between engram and non-engram spines as well and reactivation and non-reactivated engram spines would be interesting and may help to further illuminate the morphological changes of forgetting and memory retrieval. Indeed, future analysis could determine if spine density is reduced in reactivated and non-reactivated engram cells or indeed across engram non-engram cells within different learning conditions. This avenue of investigation could determine if there is a functionally distinct aspect of engram cells that are altered following forgetting (67). However, such analysis is beyond the scope of this study. We have highlighted this limitation and included its discussion. See lines 493-499.

Authors should discuss how the enrichment-neurogenesis results here are compatible with other neurogenesis work that supports forgetting.

We validated the effectiveness of the enrichment paradigm to enhance neural plasticity by measuring adult hippocampal neurogenesis. The hippocampus has been identified as one of the only regions where postnatal neurogenesis continues throughout life (75). Levels of adult hippocampal neurogenesis do not remain constant throughout life and can be altered by experience (41–43, 57).  In addition, adult born neurons have been shown to contribute to the process of forgetting (74, 78, 79). Although the contribution of adult born neurons to cognition and the memory engram is not fully understood (80, 81). Mishra et al, showed that immature neurons were actively recruited into the engram following a hippocampal-dependent task (67). Moreover, increasing the level of neurogenesis rescued memory deficits by restoring engram activity (67). Augmenting neurogenesis further rescued the deficits in spine density in both immature and mature engram neurons in a mouse model of Alzheimer’s disease (67). Whether neurogenesis alters spine density on differentially for reactivated or non-reactivation engrams cells remains to be investigated (67, 68). This avenue of research would help to illuminate the morphological changes following forgetting and provide evidence if there is a functionally distinct aspect of engram cells that is altered in forgetting (67, 68). Our engram labelling strategy which utilized c-fos-tTA transgenic mice combined with an AAV9-TRE-ChR2-eYFP virus does not necessarily label sufficient immature neurons. Future work could utilize a different engram preparation, such as a genetic labelling strategy (TRAP2) or using a different immediate early gene promoter such as Arc to investigate the contribution of new-born neurons to the engram ensemble. We have added additional discussion of how our work fits with previous literature investigating neurogenesis and forgetting. See lines 547-565.

Reviewer #2 (Public Review):

Summary:

The manuscript examines an important question about how an inaccessible, natural forgotten memory can be retrieved through engram ensemble reactivation. It uses a variety of strategies including optogenetics, behavioral and pharmacological interventions to modulate engram accessibility. The data characterize the time course of natural forgetting using an object recognition task, in which animals can retrieve 1 day and 1 week after learning, but not 2 weeks later. Forgetting is correlated with lower levels of cell reactivation (c-fos expression during learning compared to retrieval) and reduction in spine density and volume in the engram cells. Artificial activation of the original engram was sufficient to induce recall of the forgotten object memory while artificial inhibition of the engram cells precluded memory retrieval. Mice housed in an enriched environment had a slower rate of forgetting, and a brief reminder before the retrieval session promoted retrieval of a forgotten memory. Repeated reintroduction to the training context in the absence of objects accelerated forgetting. Additionally, activation of Rac1-mediated plasticity mechanisms enhanced forgetting, while its inhibition prolonged memory retrieval. The authors also reproduce the behavioral findings using a computational model inspired by Rescorla-Wagner model. In essence, the model proposes that forgetting is a form of adaptive learning that can be updated based on prediction error rules in which engram relevancy is altered in response to environmental feedback.

We thank the Reviewer for summarizing the scope and depth of our manuscript, and for recognizing our efforts. We engage below the Reviewer’s specific criticisms of our interpretations.

Strengths:

(1) The data presented in the current paper are consistent with the authors claim that seemingly forgotten engrams sometimes remain accessible. This suggests that retrieval deficits can lead to memory impairments rather than a loss of the original engram (at least in some cases).

We thank the Reviewer for their positive summary.

(2) The experimental procedures and statistics are appropriate, and the behavioral effects appear to be very robust. Several key effects are replicated multiple times in the manuscript.

We thank the Reviewer for their positive comments.

Weaknesses:

(1) My major issue with the paper is the forgetting model proposed in Figure 7. Prior work has shown that neutral stimuli become associated in a manner similar to conditioned and unconditioned stimuli. As a result, the Rescorla-Wagner model can be used to describe this learning (Todd & Homes, 2022). In the current experiments, the neutral context will become associated with the unpredicted objects during training (due to a positive prediction error). Consequently, the context will activate a memory for the objects during the test, which should facilitate performance. Conversely, any manipulation that degrades the association between the context and object should disrupt performance. An example of this can be found in Figure 5A. Exposing the mice to the context in the absence of the objects should violate their expectations and create a negative prediction error. According to the Rescorla-Wagner model, this error will create an inhibitory association between the context and the objects, which should make it harder for the former to activate a memory of the latter (Rescorla & Wagner, 1972). As a result, performance should be impaired, and this is what the authors find. However, if the cells encoding the context and objects were inhibited during the context-alone sessions (Figure 5D) then no prediction error should occur, and inhibitory associations would not be formed. As a result, performance should be intact, which is what the authors observe.

What about forgetting of the objects that occurs over time? Bouton and others have demonstrated that retrieval failure is often due to contextual changes that occur with the passage of time (Bouton, 1993; Rosas & Bouton, 1997, Bouton, Nelson & Rosas, 1999). That is, both internal (e.g. state of the animal) and external (e.g. testing room, chambers, experimenter) contextual cues change over time. This shift makes it difficult for the context to activate memories with which it was once associated (in the current paper, objects). To overcome this deficit, one can simply re-expose animals to the original context, which facilitates memory retrieval (Bouton, 1993). In Figure 2D, the authors do something similar. They activate the engram cells encoding the original context and objects, which enhances retrieval.

Therefore, the forgetting effects presented in the current paper can be explained by changes in the context and the associations it has formed with the objects (excitatory or inhibitory). The results are perfectly predicted by the Rescorla-Wagner model and the context-change findings of Bouton and others. As a result, the authors do not need to propose the existence of a new "forgetting" variable that is driven by negative prediction errors. This does not add anything novel to the paper as it is not necessary to explain the data (Figures 7 and 8).

We thank the reviewer for clearly explaining their concern about our model. We are very sorry that we did not sufficiently explain that our model is, in fact, based on the classic Rescorla-Wagner model. The key equation of the model that updates “engram strength”  is equivalent to the canonical Rescorla-Wagner model that is commonly used in research on reinforcement learning and decision-making (105). One potential minor difference is that we crucially assume different learning rates for positive and negative prediction errors. However, this variant of the Rescorla-Wagner model is common in the computational literature and is generally not regarded as a qualitatively different kind of model. In fact, it allows us to capture that establishing an object-context association (after a positive prediction error) is faster than the forgetting process (through negative errors).

The other equations that are explained in detail in the Methods are necessary to simulate exploration behavior and render the model suitable for model fitting. Concerning exploration behavior, we use the softmax function, which is commonly used in combination with the Rescorla-Wager model, in order to translate the learned quantity (in our case, engram strength) into behavior (here exploration). The other equations are necessary to fit the model to the data (learning rate α and behavioral variability in exploration behavior).

Therefore, we fully agree with the reviewer that the Rescorla-Wagner can explain our empirical results, in particular by assuming that the different manipulations affect the strength of object-context associations, which, in turn, governs forgetting as behaviorally observed. 

In our previous version of the manuscript, we only referred to the Rescorla-Wagner model directly in the Methods. But to make this important point clearer, we now refer to the origin of the model multiple times in the Results section as well. See lines 81, 386-393.

We also agree with the reviewer that the learning/forgetting process can be described in terms of changes in object-context associations (e.g., inhibitory associations after a negative prediction error). Therefore, we now explicitly refer to the relationship between updated object-context associations and forgetting and highlight that we believe that stronger associations signal higher engram “relevancy”. See lines 386-393.

We have extended Figure 7 (new panels a and b), where we illustrate the idea that (a) object-context associations govern forgetting and (b) show the key Rescorla-Wagner equation, including a simple explanation of the main terms (engram strength, prediction error, and learning rate). Finally, we have also extended our discussion of the model, where we now directly state that the Rescorla-Wagner model captures the key results of our experiments. See lines 573-580.

In order to further support a link between our empirical data and computational modeling, we also added extra experiments that showed the modulation of engram cells within the dentate gyrus can regulate these object-context associations. See Supplementary Figure 12a-f and lines 401-404.

To summarize our reply, we agree with the reviewer’s comment and hope that we have clarified the direct relationship to the Rescorla-Wagner model.

(2) I also have an issue with the conclusions drawn from the enriched environment experiment (Figure 3). The authors hypothesize that this manipulation alleviates forgetting because "Experiencing extra toys and objects during environmental enrichment that are reminiscent of the previously learned familiar object might help maintain or nudge mice to infer a higher engram relevancy that is more robust against forgetting.". This statement is completely speculative. A much simpler explanation (based on the existing literature) is that enrichment enhances synaptic plasticity, spine growth, etc., which in turn reduces forgetting. If the authors want to make their claim, then they need to test it experimentally. For example, the enriched environment could be filled with objects that are similar or dissimilar to those used in the memory experiments. If their hypothesis is correct, only the similar condition should prevent forgetting.

We thank the Reviewer for this alternative perspective on our findings. First of all, we agree that this statement is speculative. The effects of enrichment on neural plasticity are well established and it likely contributes to the enhanced memory recall. It is important to emphasize that this process of updating is not necessarily separate from enrichment-induced plasticity at an implementational level, but part of the learning experience within an environment containing multiple objects. The enrichment or, more generally, experience, may therefore enhance memory through the modification of activity of specific engram ensembles. The idea of enrichment facilitating memory updating is consistent with the results obtained by the reminder experiments and further supported by our analysis with the Rescorla-Wagner computational model, where experience updates the accessibility of existing memories, possibly through reactivation of the original engram ensemble.

We would like to further clarify that our explanation concerns the algorithmic level, in contrast to the neural level. Based on the computational analyses using the Rescorla-Wagner model and in line with the reviewer’s previous comment on the model, we believe that forgetting is governed by the strength of object-context associations (or engram relevancy). Our interpretation is that stronger associations signal that the memory or engram representation is important ("relevant") and should not be forgotten. Accordingly, due to a vast majority of experiences with extra cage objects in the enriched environment, mice might generally learn that such objects are common in their environment and potentially relevant in the future (i.e., the object-context association is strong, preventing forgetting). Our speculation of these results is to help unify our empirical data with the computational model.

We believe that the Reviewer's alternative explanation in terms of synaptic plasticity, spine growth is not mutually exclusive with the modelling work. It is possible that the computational mechanisms that we explore based on the Rescorla-Wagner model are neuronally related to the biological mechanisms that the reviewer suggests at the implementational level. Therefore, ultimately, the two perspectives might even complement each other. We have included additional discussion to clarify this point. See lines 510-546.

(3) It is well-known that updating can both weaken or strengthen memory. The authors suggest that memory is updated when animals are exposed to the context in the absence of the objects. If the engram is artificially inhibited (opto) during context-only re-exposures, memory cannot be updated. To further support this updating idea, it would be good to run experiments that investigate whether multiple short re-exposures to the training context (in the presence of the objects or during optogenetic activation of the engram) could prevent forgetting. It would also be good to know the levels of neuronal reactivation during multiple re-exposures to the context in the absence versus context in the presence of the objects.

We thank the Reviewer for their comments. We agree that additional experiments would be helpful to further support the idea of updating. We have performed additional experiments to test the idea that multiple short re-exposures to the training context, in the presence of objects prevents forgetting. In this paradigm, mice were repeatedly exposed to the original object pair (Supplementary Figure S5a). The results indicate that repeated reminder trials facilitate object memory recall (Supplementary Figure 5b&c). These data indicated that subsequent object reminders over time facilitates the transition of a forgotten memory to an accessible memory. See Supplementary Figure S5 and Lines 279-287.

(4) There are a number of studies that show boundary conditions for memory destabilization/reconsolidation. Is there any evidence that similar boundary conditions exist to make an inaccessible engram accessible?

The Reviewer asks an interesting question about boundary conditions and engram accessibility. Boundary conditions could indeed affect the degree of destabilization and reconsolidation, the salience or strength of the memory, as well as the timing of retrieval cues. Future models could focus on understanding the specific boundary conditions in which a memory becomes retrievable and the degree to which it is sufficiently destabilized and liable for updating and forgetting. We have included additional discussion on the potential role of boundary conditions for engram accessibility. See lines 661-666.

(5) More details about how the quantification of immunohistochemistry (c-fos, BrdU, DAPI) was performed should be provided (which software and parameters were used to consider a fos positive neurons, for example).

We have added additional information for the parameters of quantification of immunohistochemistry. See lines 796-809.

(6) Duration of the enrichment environment was not detailed.

We have highlighted the details for the environmental enrichment duration. See lines 756.

Reviewer #3 (Public Review):

Summary:

The manuscript by Ryan and colleagues uses a well-established object recognition task to examine memory retrieval and forgetting. They show that memory retrieval requires activation of the acquisition engram in the dentate gyrus and failure to do so leads to forgetting. Using a variety of clever behavioural methods, the authors show that memories can be maintained and retrieval slowed when animals are reared in environmental enrichment and that normally retrieved memories can be forgotten if exposed to the environment in which the expected objects are no longer presented. Using a series of neural methods, the authors also show that activation or inhibition of the acquisition engram is key to memory expression and that forgetting is due to Rac1.

We thank the Reviewer for summarizing the scope and depth of our manuscript, and indeed for recognizing our efforts. We engage below the Reviewer’s specific criticisms of our interpretations.

Strengths:

This is an exemplary examination of different conditions that affect successful retrieval vs forgetting of object memory. Furthermore, the computational modelling that captures in a formal way how certain parameters may influence memory provides an important and testable approach to understanding forgetting.

The use of the Rescorla-Wagner model in the context of object recognition and the idea of relevance being captured in negative prediction error are novel (but see below).

The use of gain and loss of function approaches are a considerable strength and the dissociable effects on behaviour eliminate the possibility of extraneous variables such as light artifacts as potential explanations for the effects.

We thank the Reviewer for their positive comments.

Weaknesses:

Knowing what process (object retrieval vs familiarity) governed the behavioural effect in the present investigation would have been of even greater significance.

The Reviewer touches on an important issue of the object recognition task. Understanding how experience alters object familiarity versus object retrieval and its impact on learning would help to develop better models of object memory and forgetting. We have added additional discussion. See lines 666-669.

The impact of the paper is somewhat limited by the use of only one sex.

We agree that using only male mice limits the impact of the paper. Indeed, the field of behavioural neuroscience is moving to include sex as a variable. Future experiments should include both male and female mice.

While relevance is an interesting concept that has been operationalized in the paper, it is unclear how distinct it is from extinction. Specifically, in the case where the animals are exposed to the context in the absence of the object, the paper currently expresses this as a process of relevance - the objects are no longer relevant in that context. Another way to think about this is in terms of extinction - the association between the context and the objects is reduced results in a disrupted ability of the context to activate the object engram.

We thank the reviewer for their insightful comment on the connection between engram relevance and memory extinction. Lacagnina et al., demonstrated that extinction training suppressed the reactivation of a fear engram, while activating a second putative extinction ensemble (59). In another study, these extinction engram cells and reward cells were shown to be functionally interchangeable (92). Moreover, in a study conducted by Lay et al., the balance between extinction and acquisition was disrupted by inhibiting the extinction recruited neurons in the BLA and CN (93). These results suggested that decision making after extinction can be governed by a balance between acquisition and extinction specific ensembles (93). Together, this may suggest that in the present study, when mice are repeatedly exposed to the training context, the association between the context and the objects is reduced, resulting in a disrupted ability of the context to activate the object engram. Therefore, memory relevance and extinction may operate similarly to effect engram accessibility, and in essence ‘forgetting’ of object memories may be due to neurobiological mechanisms similar to that of extinction learning (4). We have included additional discussion on the link between our results and the extinction literature. See lines 642-654.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

Additional measures that may help interpretation of and clarify data are:

A minute-by-minute analysis for training and testing may provide insight about the learning rate and testing temporal dynamics that may shed light substantially on differential levels of exploration. This should be applied across figures and would support conclusions from models in Figures 7-8 as well.

Locomotion/distance travelled measures.

We have included additional analysis for a minute-by-minute analysis of training and testing of the object memory test at 24 hr, 2 weeks as well as under the standard housing and enrichment conditions. The results further support the initial finding that novel object recognition is increased in mice that recall the object at 24 hr. Similarly, mice housed in the enriched housing initially explore the novel object more compared to the familiar object. See Supplementary Figure 1 and 2, as well as lines 103-105 and 211-213.

The appropriate control for the context exposure figure would be to expose to a novel context in one group and the acquisition/testing context for the other.

We agree with the reviewer that an additional control of a novel context would further support our findings. Indeed, this line of investigate may dove-tail with the other reviewer comments on the role of competing engrams and interference. Future work could investigate the degree to which novel contexts and multiple memories can affect the rate of forgetting through engram updating. We have included additional discussion. See lines 643 and 655. However, in our experience it is necessary to pre-expose mice to different contexts before object exposure (e.g. Autore et al ’23), in order to form discriminate object/context associations. Establishing such a paradigm for this study would be at odds with the established paradigms and schedules in this current study. Moreover, the possibility that the effect of object displacement on forgetting requires the familiar context, or not, does not impact the main conclusions of this study. However, we agree that it is a point for expansion in the future.

A control virus+light group vs simply a no-light condition.

For optogenetic experiments. Control mice underwent the same surgery procedure with virus and optic fibre implantation. However, no light was delivered to excite or inhibit the respective opsin. Previous papers have shown laser light delivered to tissue expressing an AAV-TRE-EYFP lacking an light-opsin does cause cellular excitation. We have clarified this in the text. See lines 726-729.

Reviewer #2 (Recommendations For The Authors):

Minor details:

(1) In the pharmacological modification of Rac 1, please specify what percentage of DMSO was used to dissolve Rac1 inhibitor and correct the typo 'DSMO'

Rac1 inhibitor (Ehop016) was reconstituted and prepared in PBS with 1% Tween-80, 1% DMSO and 30% PEG. We have clarified this in the text and corrected the typo, thank you. See lines 767.

(2) In the penultimate paragraph there is a typo 'predication error'

This is now corrected. Thankyou.

Reviewer #3 (Recommendations For The Authors):

I was unable to find information on what the No Light group consisted of. Was there a control virus infused, were the animals implanted with optical fibres (in the presence or absence of a virus), were they surgical controls, etc?

For optogenetic experiments. No Light Control mice underwent the same surgery procedure with virus and optic fibre implantation. However, no light was delivered to excite or inhibit the respective opsin. We have clarified this in the text. See lines 726-729.

The discussion lacked specificity in places. For example, the idea of eluding to 'other variables' is somewhat vague (p. 21, middle paragraph). Some examples of what other variables could be relevant would be helpful in capturing what direction or relevance the model may have going forward.

We have expanded the discussion of other variables which might impact engram relevance and how the model might be developed moving forward. These may include, boundary conditions of destabilization and reconsolidation, the salience or strength of the memory as well as the timing of retrieval cues or updating experience. Future models could focus on understanding the specific boundary conditions in which a memory becomes retrievable and the degree to which it is sufficiently destabilized and liable for updating and forgetting. The role of perceptual learning on memory retrieval and forgetting may also be an avenue of future investigation. Understanding how experience alters object familiarity versus object retrieval and its impact on learning would also help to develop better models of object memory and forgetting. In the current study, only male mice were utilized. Therefore, future work could also include sex as a variable to fully elucidate the impact of experience on the processes of forgetting. See lines 642-669.

In the same paragraph (p. 21, middle paragraph) there is mention of multiple engrams and how they can compete. The authors reference Autore et al (2023), but I thought Lacagina did this really beautifully also in an experimental setting. This idea is also expressed in Lay et al. (2022). So additional references would further strengthen the authors argument here.

We thank the reviewer for the additional references for discussing engram competition. We have included these papers in the discission. See lines 642-654.

Relatedly, environmental enrichment was considered in terms of object relevance. I wonder if the authors may want to consider thinking about their results in terms of effects on perceptual learning.

Indeed, perceptual learning maybe playing a role in environmental enrichment. We have included additional discussion. See lines 666-669.

eLife assessment

This important study evaluates a model for multisensory correlation detection, focusing on the detection of correlated transients in visual and auditory stimuli. Overall, the experimental design is sound and the evidence is compelling. The synergy between the experimental and theoretical aspects of the paper is strong, and the work will be of interest to both neuroscientists and psychologists working in the domain of sensory processing and perception.

Reviewer #1 (Public Review):

The authors present a model for multisensory correlation detection that is based on the neurobiologically plausible Hassenstein Reichardt detector (Parise & Ernst, 2016). They demonstrate that this model can account for human behaviour in synchrony or temporal order judgements and related temporal tasks in two new data sets (acquired in this study) and a range of previous data sets. While the current study is limited to the model assessment for relatively simple audiovisual signals, in future communications, the authors demonstrate that the model can also account for audiovisual integration of complex naturalistic signals such as speech and music.

The significance of this work lies in its ability to explain multisensory perception using fundamental neural mechanisms previously identified in insect motion processing.

Strengths:

(1) The model goes beyond descriptive models such as cumulative Gaussians for TOJ and differences in cumulative Gaussians for SJ tasks by providing a mechanism that builds on the neurobiologically plausible Hassenstein-Reichardt detector.<br /> (2) This model can account for results from two new experiments that focus on the detection of correlated transients and frequency doubling. The model also accounts for several behavioural results from experiments including stochastic sequences of A/V events and sine wave modulations (and naturalistic Av signals such as speech and music as shown in future communications).

Reviewer #2 (Public Review):

Summary:

This is an interesting and well-written manuscript that seeks to detail performance on two human psychophysical experiments designed to look at the relative contributions of transient and sustained components of a multisensory (i.e., audiovisual) stimulus to their integration. The work is framed within the context of a model previously developed by the authors and now somewhat revised to better incorporate the experimental findings. The major takeaway from the paper is that transient signals carry the vast majority of the information related to the integration of auditory and visual cues, and that the Multisensory Correlation Detector (MCD) model not only captures the results of the current study, but is also highly effective in capturing the results of prior studies focused on temporal and causal judgments.

Strengths:

Overall the experimental design is sound and the analyses well performed. The extension of the MCD model to better capture transients make a great deal of sense in the current context, and it is very nice to see the model applied to a variety of previous studies.

Comments on the revised version:

In the revised manuscript, the authors have done an excellent job of responding to the prior critiques. I have no additional concerns or comments.

Author response:

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

Public Reviews:

Reviewer #1 (Public Review):

The authors present a model for multisensory correlation detection that is based on the neurobiologically plausible Hassenstein Reichardt detector. It modifies their previously reported model (Parise & Ernst, 2016) in two ways: a bandpass (rather than lowpass) filter is initially applied and the filtered signals are then squared. The study shows that this model can account for synchrony judgement, temporal order judgement, etc in two new data sets (acquired in this study) and a range of previous data sets.

Strengths:

(1) The model goes beyond descriptive models such as cumulative Gaussians for TOJ and differences in cumulative Gaussians for SJ tasks by providing a mechanism that builds on the neurobiologically plausible Hassenstein-Reichardt detector.

(2) This modified model can account for results from two new experiments that focus on the detection of correlated transients and frequency doubling. The model also accounts for several behavioural results from experiments including stochastic sequences of A/V events and sine wave modulations.

Additional thoughts:

(1) The model introduces two changes: bandpass filtering and squaring of the inputs. The authors emphasize that these changes allow the model to focus selectively on transient rather than sustained channels. But shouldn't the two changes be introduced separately? Transients may also be detected for signed signals.

We updated the original model because our new psychophysical evidence demonstrates the fundamental role of unsigned transient for multisensory perception. While the original model received input from sustained unimodal channels (low-pass filters), the new version receives input from unsigned unimodal transient channels. Transient channels are normally modelled through bandpass filters (to remove the DC and high-frequency signal components) and squaring (to remove the sign). While these may appear as two separate changes in the model, they are, in fact, a single one: the substitution of sustained with unsigned transient channels (for a similar approach, see Stigliani et al. 2017, PNAS). Either change alone would not be sufficient to implement a transient channel that accounts for the present results.

That said, we were also concerned with introducing too many changes in the model at once. Indeed, we simply modelled the unimodal transient channels as a single band-pass filter followed by squaring. This is already a stripped-down version of the unsigned transient detectors proposed by Adelson and Bergen in their classic Motion Energy model. The original model consisted of two biphasic temporal filters 90 degrees out of phase (i.e., quadrature filters), whose output is later combined. While a simpler implementation of the transient channels was sufficient in the present study, the full model may be necessary for other classes of stimuli (including speech, Parise, 2024, BiorXiv). Therefore, for completeness, we now include in the Supplementary Information a formal description of the full model, and validate it by simulating our two novel psychophysical studies. See Supplementary Information “The quadrature MCD model” section and Supplementary Figure S8.

(2) Because the model is applied only to rather simple artificial signals, it remains unclear to what extent it can account for AV correlation detection for naturalistic signals. In particular, speech appears to rely on correlation detection of signed signals. Can this modified model account for SJ or TOJ judgments for naturalistic signals?

It can. In a recent series of studies we have demonstrated that a population of spatially-tuned MCD units can account for audiovisual correlation detection for naturalistic stimuli, including speech (e.g. the McGurk Illusion). Once again, unsigned transients were sufficient to replicate a variety of previous findings. We have now extended the discussion to cover this recent research: Parise, C. V. (2024). Spatiotemporal models for multisensory integration. bioRxiv, 2023-12.

Even Nidiffer et al. (2018) which is explicitly modelled by the authors report a significant difference in performance for correlated and anti-correlated signals. This seems to disagree with the results of study 1 reported in the current paper and the model's predictions. How can these contradicting results be explained? If the brain detects correlation on signed and unsigned signals, is a more complex mechanism needed to arbitrate between those two?

We believe the reviewer here refers to our Experiment 2 (where, like Nidiffer at al. (2018) we used periodic stimuli, not Experiment 1, which consists of step stimuli). We were also puzzled by the difference between our Experiment 2 and Nidiffer et al. (2018): we induced frequency doubling, Nidiffer did not. Based on quantitative simulations, we concluded that this difference could be attributed to the fact that while Nidiffer included on each trial an intensity ramp in their periodic audiovisual stimuli, we did not. As a result, when considering the ramp (unlike in Nidiffer’s analyses), all audiovisual signals used by Nidiffer were positively correlated (irrespective of frequency and phase offset), while our signals in Experiment 2 were sometimes correlated and other times not (depending on the phase offset). This important simulation is included in Supplementary Figure S7; we also have now updated the text to better highlight the role of the pedestal in determining the direction of the correlation.

(3) The number of parameters seems quite comparable for the authors' model and descriptive models (e.g. PSF models). This is because time constants require refitting (at least for some experimental data sets) and the correlation values need to be passed through a response mode (i.e. probit function) to account for behavioural data. It remains unclear how the brain adjusts the time constants to different sensory signals.

This is a deep question. For simplicity, here the temporal constants were fitted to the empirical psychometric functions. To avoid overfitting, whenever possible we fitted such parameters over some training datasets, while trying to predict others. However, in some cases, it was necessary to fit the temporal constants to specific datasets. This may suggest that the temporal tuning of those units is not crystalised to some pre-defined values, but is adjusted based on recent perceptual history (e.g., the sequence of trials and stimuli participants are exposed to during the various experiments).

For transparency, here we show how varying the tuning of the temporal constants of the filters affects the goodness of fit of our new psychophysical experiments (Supplementary Figure S8). As it can be readily appreciated, the relative temporal tuning of the unimodal transient detector was critical, though their absolute values could vary over a range of about 15 to over 100ms. The tuning of the low-pass filters of the correlation detector (not shown here) displayed much lower temporal sensitivity over a range between 0.1s to over 1s.

This simulation shows the impact of temporal tuning in our simulations, however, the question remains as to how such a tuning gets selected in the first place. An appealing explanation relies on natural scene statistics: units are temporally tuned to the most common audiovisual stimuli. Although our current empirical evidence does not allow us to quantitatively address this question, in previous simulations (see Parise & Ernst, 2016, Supplementary Figure 8), by analogy with visual motion adaptation, we show how the temporal constants of our model can dynamically adjust and adapt to recent perceptual history. We hope these new and previous simulations address the question about the nature of the temporal tuning of the MCD units.

(4) Fujisaki and Nishida (2005, 2006) proposed mechanisms for AV correlation detection based on the Hassenstein-Reichardt motion detector (though not formalized as a computational model).

This is correct, Fujisaki and Nishida (2005, 2007) also hypothesized that AV synchrony could be detected using a mechanism analogous to motion detection. Interestingly, however, they ruled out such a hypothesis, as their “data do not support the existence of specialized low-level audio-visual synchrony detectors”. Yet, along with our previous work (Parise & Ernst, 2016, where we explicitly modelled the experiments of Fujisaki and Nishida), the present simulations quantitatively demonstrate that a low-level AV synchrony detector is instead sufficient to account for audiovisual synchrony perception and correlation detection. We now credit Fujusaki and Nishida in the modelling section for proposing that AV synchrony can be detected by a cross-correlator.

Finally, we believe the reviewer is referring to the 2005 and 2007 studies of Fujisaki and Nishida (not 2006); here are the full references of the two articles we are referring to:

Fujisaki, W., & Nishida, S. Y. (2005). Temporal frequency characteristics of synchrony–asynchrony discrimination of audio-visual signals. Experimental Brain Research, 166, 455-464.

Fujisaki, W., & Nishida, S. Y. (2007). Feature-based processing of audio-visual synchrony perception revealed by random pulse trains. Vision Research, 47(8), 1075-1093.

Reviewer #2 (Public Review):

Summary:

This is an interesting and well-written manuscript that seeks to detail the performance of two human psychophysical experiments designed to look at the relative contributions of transient and sustained components of a multisensory (i.e., audiovisual) stimulus to their integration. The work is framed within the context of a model previously developed by the authors and is now somewhat revised to better incorporate the experimental findings. The major takeaway from the paper is that transient signals carry the vast majority of the information related to the integration of auditory and visual cues, and that the Multisensory Correlation Detector (MCD) model not only captures the results of the current study but is also highly effective in capturing the results of prior studies focused on temporal and causal judgments.

Strengths:

Overall the experimental design is sound and the analyses are well performed. The extension of the MCD model to better capture transients makes a great deal of sense in the current context, and it is very nice to see the model applied to a variety of previous studies.

Weaknesses:

My one major issue with the paper revolves around its significance. In the context of a temporal task(s), is it in any way surprising that the important information is carried by stimulus transients? Stated a bit differently, isn't all of the important information needed to solve the task embedded in the temporal dimension? I think the authors need to better address this issue to punch up the significance of their work.

In hindsight, it may appear unsurprising that transient signals carry most information for audiovisual integration. Yet, so somewhat unexpectedly, this has never been investigated using perhaps the most diagnostic psychophysical tools for perceived crossmodal timing; namely temporal order and simultaneity judgments–along with carefully designed experiments with quantitative predictions for the effect of either channel. The fact that the results conform to intuitive expectations further supports the value of the present work: grounding empirically with what is intuitively expected. This offers solid psychophysical evidence that one can build on for future advancements. Importantly, developing a model that builds on our new results and uses the same parameters to predict a variety of classic experiments in the field, further supports the current approach.

If “significance” is intended as shaking previous intuitions or theories, then no: this is not a significant contribution. If instead, by significance we intend to build a solid empirical and theoretical ground for future work, then we believe this study is not significant, it is foundational. We hope that this work's significance is better captured in our discussion.

On a side note, there is an intriguing factor around transient vs. sustained channels: what matters is the amount of change, not the absolute stimulus intensity. Previous studies, for example, have suggested a positive cross modal mapping between auditory loudness and visual lightness or brightness [Odegaard et al., 2004]. This study, conversely, challenges this view and demonstrates that what matters for multisensory integration in time is not the intensity of a stimulus, but changes thereof.

In a more minor comment, I think there also needs to be a bit more effort into articulating the biological plausibility/potential instantiations of this sustained versus transient dichotomy. As written, the paper suggests that these are different "channels" in sensory systems, when in reality many neurons (and neural circuits) carry both on the same lines.

The reviewer is right, in our original manuscript we glossed over this aspect. We have now expanded the introduction to discuss their anatomical basis. However, we are not assuming any strict dichotomy between transient and sustained channels; rather, our results and simulations demonstrate that transient information is sufficient to account for audiovisual temporal integration.

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

(1) Related to point 2 of the public review, can the authors provide additional results showing that the model can also account for naturalistic signals and more complex stochastic signals?

While working on this manuscript, we were also working in parallel on a project related to audiovisual integration of naturalistic signals. A pre-print is available online [Parise, 2024, BiorXiv], and the related study is now discussed in the conclusions.

(2) As noted in the public review, Fujisaki and Nishida (2005, 2006) already proposed mechanisms for AV correlation detection based on the Hassenstein-Reichardt motion detector. Their work should be referenced and discussed.

We have now acknowledged the contribution of Fujisaki and Nishida in the modelling section, when we first introduce the link between our model and the Hassenstein-Reichardt detectors.

(3) Experimental parameters: Was the phase shift manipulated in blocks? If yes, what about temporal recalibration?

To minimise the effect of temporal recalibration, the order of trials in our experiments was randomised. Nonetheless, we can directly assess potential short-term recalibration effects by plotting our psychophysical responses against both the current SOA, and that of the previous trials. The resulting (raw) psychometric surfaces below are averaged across observers (and conditions for Experiment 1). In all our experiments, responses are obviously dependent on the current SOA (x-axis). However, the SOA of the previous trials (y-axis) does not seem to meaningfully affect simultaneity and temporal order judgments. The psychometric curves above the heatmaps represent the average psychometric functions (marginalized over the SOA of the previous trial).

All in all, the present analyses demonstrate negligible temporal recalibration across trials, likely induced by a random sequence of lags or phase shifts. Therefore, when estimating the temporal constants of the model, it seems reasonable to ignore the potential effects of temporal recalibration. To avoid increasing the complexity of the present manuscript, we would prefer not to include the present analyses in the revised version.

Author response image 1.

Effect of previous trial. Psychometric surfaces for Experiments 1 and 2 plotted against the lag in the current vs. the previous trial. While psychophysical responses are strongly modulated by the lag in the last trial (horizontal axis), they are relatively unaffected by the lag in the previous trial (vertical axis).

(4) The model predicts no differences for experiment 1 and this is what is empirically observed. Can the authors support these null results with Bayes factors?

This is a good suggestion: we have now included a Bayesian repeated measures ANOVA to the analyses of Experiment 1. As expected, these analyses provide further, though mild evidence in support for the null hypothesis (See Table S2). For completeness, the new Bayesian analyses are presented alongside the previous frequentist ones in the revised manuscript.

eLife assessment

The authors combined neurophysiological (electroencephalography [EEG]) and neurochemical (magnetic resonance spectroscopy [MRS]) measures to empirically evaluate the neural noise hypothesis of developmental dyslexia. Their results are solid, supported by consistent findings from the two complementary methodologies and Bayesian statistics. Additional analyses, particularly on the neurochemical measures, are necessary to further substantiate the results. This study is useful for understanding the neural mechanisms of dyslexia and neural development in general.

Reviewer #1 (Public Review):

Summary:

"Neural noise", here operationalized as an imbalance between excitatory and inhibitory neural activity, has been posited as a core cause of developmental dyslexia, a prevalent learning disability that impacts reading accuracy and fluency. This study is the first to systematically evaluate the neural noise hypothesis of dyslexia. Neural noise was measured using neurophysiological (electroencephalography [EEG]) and neurochemical (magnetic resonance spectroscopy [MRS]) in adolescents and young adults with and without dyslexia. The authors did not find evidence of elevated neural noise in the dyslexia group from EEG or MRS measures, and Bayes factors generally informed against including the grouping factor in the models. Although the comparisons between groups with and without dyslexia did not support the neural noise hypothesis, a mediation model that quantified phonological processing and reading abilities continuously revealed that EEG beta power in the left superior temporal sulcus was positively associated with reading ability via phonological awareness. This finding lends support for analysis of associations between neural excitatory/inhibitory factors and reading ability along a continuum, rather than as with a case/control approach, and indicates the relevance of phonological awareness as an intermediate trait that may provide a more proximal link between neurobiology and reading ability. Further research is needed across developmental stages and over a broader set of brain regions to more comprehensively assess the neural noise hypothesis of dyslexia, and alternative neurobiological mechanisms of this disorder should be explored.

Strengths:

The inclusion of multiple methods of assessing neural noise (neurophysiological and neurochemical) is a major advantage of this paper. MRS at 7T confers an advantage of more accurately distinguishing and quantifying glutamate, which is a primary target of this study. In addition, the subject-specific functional localization of the MRS acquisition is an innovative approach. MRS acquisition and processing details are noted in the supplementary materials according to the experts' consensus-recommended checklist (https://doi.org/10.1002/nbm.4484). Commenting on the rigor, the EEG methods is beyond my expertise as a reviewer.

Participants recruited for this study included those with a clinical diagnosis of dyslexia, which strengthens confidence in the accuracy of the diagnosis. The assessment of reading and language abilities during the study further confirms the persistently poorer performance of the dyslexia group compared to the control group.

The correlational analysis and mediation analysis provide complementary information to the main case-control analyses, and the examination of associations between EEG and MRS measures of neural noise is novel and interesting.

The authors follow good practice for open science, including data and code sharing. They also apply statistical rigor, using Bayes Factors to support conclusions of null evidence rather than relying only on non-significant findings. In the discussion, they acknowledge the limitations and generalizability of the evidence and provide directions for future research on this topic.

Weaknesses:

Though the methods employed in the paper are generally strong, there are certain aspects that are not clearly described in the Materials & Methods section, such as a description of the statistical analyses used for hypothesis testing.

With regard to metabolite quantification, it is unclear why the authors chose to analyze and report metabolite values in terms of creatine ratios rather than quantification based on a water reference given that the MRS acquisition appears to support using a water reference. GABA is typically quantified using J-editing sequences as lower field strengths (~3T), and there is some evidence that the GABA signal can be reliably measured at 7T without editing, however, the authors should discuss potential limitations, such as reliability of Glu and GABA measurements with short-TE semi-laser at 7T. In addition, MRS measurements of GABA are known to be influenced by macromolecules, and GABA is often denoted as GABA+ to indicate that other compounds contribute to the measured signal, especially at a short TE and in the absence of symmetric spectral editing. A general discussion of the strengths and limitations of unedited Glu and GABA quantification at 7T is warranted given the interest of this work to researchers who may not be experts in MRS.

Further, the single MRS voxel location is a limitation of the study as neurochemistry can vary regionally within individuals, and the putative excitatory/inhibitory imbalance in dyslexia may appear in regions outside the left temporal cortex (e.g., network-wide or in frontal regions involved in top-down executive processes). While the functional localization of the MRS voxel is a novelty and a potential advantage, it is unclear whether voxel placement based on left-lateralized reading-related neural activity may bias the experiment to be more sensitive to small, activity-related fluctuations in neurotransmitters in the CON group vs. the DYS group who may have developed an altered, compensatory reading strategy.

As the authors note in the discussion, sex could serve as a moderator of associations between neural noise and reading abilities and should be considered in future studies.

Appraisal:

The authors present a thorough evaluation of the neural noise hypothesis of developmental dyslexia in a sample of adolescents and young adults using multiple methods of measuring excitatory/inhibitory imbalances as an indicator of neural noise. The authors concluded that there was no support for the neural noise hypothesis of dyslexia in their study based on null significance and Bayes factors. This conclusion is justified, and further research is called for to more broadly evaluate the neural noise hypothesis in developmental dyslexia.

Impact:

This study provides an exemplary foundation for the evaluation of the neural noise hypothesis of dyslexia. Other researchers may adopt the model applied in this paper to examine neural noise in various populations with/without dyslexia, or across a continuum of reading abilities, to more thoroughly examine the evidence (or lack thereof) for this hypothesis. Notably, the lack of evidence here does not rule out the possibility of a role for neural noise in dyslexia, and the authors point out that presentation with co-occurring conditions, such as ADHD, may contribute to neural noise in dyslexia. Dyslexia remains a multi-faceted and heterogenous neurodevelopmental condition, and many genetic, neurobiological, and environmental factors play a role. This study demonstrates one step toward evaluating neurobiological mechanisms that may contribute to reading difficulties.

Reviewer #2 (Public Review):

Summary:

This study utilized two complementary techniques (EEG and 7T MRI/MRS) to directly test a theory of dyslexia: the neural noise hypothesis. The authors report finding no evidence to support an excitatory/inhibitory balance, as quantified by beta in EEG and Glutamate/GABA ratio in MRS. This is important work and speaks to one potential mechanism by which increased neural noise may occur in dyslexia.

Strengths:

This is a well-conceived study with in-depth analyses and publicly available data for independent review. The authors provide transparency with their statistics and display the raw data points along with the averages in figures for review and interpretation. The data suggest that an E/I balance issue may not underlie deficits in dyslexia and is a meaningful and needed test of a possible mechanism for increased neural noise.

Weaknesses:

The researchers did not include a visual print task in the EEG task, which limits analysis of reading-specific regions such as the visual word form area, which is a commonly hypoactivated region in dyslexia. This region is a common one of interest in dyslexia, yet the researchers measured the I/E balance in only one region of interest, specific to the language network. Further, this work does not consider prior studies reporting neural inconsistency; a potential consequence of increased neural noise, which has been reported in several studies and linked with candidate-dyslexia gene variants (e.g., Centanni et al., 2018, 2022; Hornickel & Kraus, 2013; Neef et al., 2017). While E/I imbalance may not be a cause of increased neural noise, other potential mechanisms remain and should be discussed.

Reviewer #3 (Public Review):

Summary:

This study by Glica and colleagues utilized EEG (i.e., Beta power, Gamma power, and aperiodic activity) and 7T MRS (i.e., MRS IE ratio, IE balance) to reevaluate the neural noise hypothesis in Dyslexia. Supported by Bayesian statistics, their results show solid 'no evidence' of EI balance differences between groups, challenging the neural noise hypothesis. The work will be of broad interest to neuroscientists, and educational and clinical psychologists.

Strengths:

Combining EEG and 7T MRS, this study utilized both the indirect (i.e., Beta power, Gamma power, and aperiodic activity) and direct (i.e., MRS IE ratio, IE balance) measures to reevaluate the neural noise hypothesis in Dyslexia.

Weaknesses:

The authors may need to provide more data to assess the quality of the MRS data.

eLife assessment:

This valuable study investigates the oscillatory activity of gonadotropin-releasing hormone (GnRH) neurones in mice using GCaMP fiber photometry. It demonstrates three distinct patterns of oscillatory activity that occur in GnRH neurons comprising low-level rapid baseline activity, abrupt short-duration oscillations that drive pulsatile gonadotropin secretion, and, in females, a gradual and prolonged oscillating increase in activity responsible for the relatively short-lived preovulatory LH surge. The evidence presented in the study is solid, offering theoretical implications for understanding the behaviour of GnRH neurones in the context of reproductive physiology, and will be of interest to researchers in neuroendocrinology and reproductive biology.

Reviewer #1 (Public Review):

Summary:

The authors aimed to investigate the oscillatory activity of GnRH neurones in freely behaving mice. By utilising GCaMP fiber photometry, they sought to record real-time neuronal activity to understand the patterns and dynamics of GnRH neuron firing and their implications for reproductive physiology.

Strengths:

(1) The use of GCaMP fiber photometry allows for high temporal resolution recordings of neuronal activity, providing real-time data on the dynamics of GnRH neurones.

(2) Recording in freely behaving animals ensures that the findings are physiologically relevant and not artifacts of a controlled laboratory environment.

(3) The authors used statistical methods to characterise the oscillatory patterns, ensuring the reliability of their findings.

Weaknesses:

(1) While the study identifies distinct oscillatory patterns in GnRH neurones' calcium dynamics, it falls short in exploring the functional implications of these patterns for GnRH pulsatility and overall reproductive physiology.

(2) The study lacks a broader discussion to include comparisons with existing studies on GnRH neurone activity and pulsatility and highlight how the findings of this study align with or differ from previous research and what novel contributions are made.

(3) The authors aimed to characterise the oscillatory activity of GnRH neurons and successfully identified distinct oscillatory patterns. The results support the conclusion that GnRH neurons exhibit complex oscillatory behaviours, which are critical for understanding their role in reproductive physiology. However, it has not been made clear what exactly the authors mean by "multi-dimensional oscillatory patterns" and how has this been shown.

Reviewer #2 (Public Review):

Summary:

In this manuscript, the authors report GCaMP fiber-photometry recordings from the GnRH neuron distal projections in the ventral arcuate nucleus. The recordings are taken from intact, male and female, freely behaving mice. The report three patterns of neuronal activity:

(1) Abrupt increases in the Ca2+ signals that are perfectly correlated with LH pulses.

(2) A gradual, yet fluctuating (with a slow ultradian frequency), increase in activity, which is associated with the onset of the LH surge in female animals.

(3) Clustered (high frequency) baseline activity in both female and male animals.

Strengths:

The GCaMP fiber-photometry recordings reported here are the first direct recordings from GnRH neurones in vivo. These recordings have uncovered a rich repertoire of activity suggesting the integration of distinct "surge" and "pulse" generation signals, and an ultradian rhythm during the onset of the surge.

Weaknesses:

The data analysis method used for the characterisation of the ultradian rhythm observed during the onset of the surge is not detailed enough. Hence, I'm left wondering whether this rhythm is in any way correlated with the clusters of activity observed during the rest of the cycle and which have similar duration.

eLife assessment

Somatostatin-expressing neurons of the entopeduncular nucleus (EPNSst+) co-release GABA and glutamate in their projection to the lateral habenula, a structure that is key for reward-based learning. Combining fiber photometry and computational modeling, the authors provide compelling evidence that EPNSst+ neural activity represents movement, choice direction, and reward outcomes in a probabilistic switching task but, surprisingly, neither chronic genetic silencing of these neurons nor selective elimination glutamate release affected behavioral performance in well-trained animals. This valuable study shows that despite its representation of key task variables, EPNSst+ neurons are dispensable for ongoing performance in a task requiring outcome monitoring to optimize reward. This study will be of interest to those interested in reward learning and/or reward-related behavior and systems or behavioral neuroscience more broadly.

Reviewer #1 (Public Review):

Summary:

In this series of studies, Locantore et al. investigated the role of SST-expressing neurons in the entopeduncular nucleus (EPNSst+) in probabilistic switching tasks, a paradigm that requires continued learning to guide future actions. In prior work, this group had demonstrated EPNSst+ neurons co-release both glutamate and GABA and project to the lateral habenula (LHb), and LHb activity is also necessary for outcome evaluation necessary for performance in probabilistic decision-making tasks. Previous slice physiology works have shown that the balance of glutamate/GABA co-release is plastic, altering the net effect of EPN on downstream brain areas and neural circuit function. The authors used a combination of in vivo calcium monitoring with fiber photometry and computational modeling to demonstrate that EPNSst+ neural activity represents movement, choice direction, and reward outcomes in their behavioral task. However, viral-genetic manipulations to synaptically silence these neurons or selectively eliminate glutamate release had no effect on behavioral performance in well-trained animals. The authors conclude that despite their representation of task variables, EPN Sst+ neuron synaptic output is dispensable for task performance.

Strengths and Weaknesses:

Overall, the manuscript is exceptionally scholarly, with a clear articulation of the scientific question and a discussion of the findings and their limitations. The analyses and interpretations are careful and rigorous. This review appreciates the thorough explanation of the behavioral modeling and GLM for deconvolving the photometry signal around behavioral events, and the transparency and thoroughness of the analyses in the supplemental figures. This extra care has the result of increasing the accessibility for non-experts, and bolsters confidence in the results. To bolster a reader's understanding of results, we suggest it would be interesting to see the same mouse represented across panels (i.e. Figures 1 F-J, Supplementary Figures 1 F, K, etc i.e via the inclusion of faint hash lines connecting individual data points across variables. Additionally, Figure 3E demonstrates that eliminating the 'reward' and 'choice and reward' terms from the GLM significantly worsens model performance; to demonstrate the magnitude of this effect, it would be interesting to include a reconstruction of the photometry signal after holding out of both or one of these terms, alongside the 'original' and 'reconstructed' photometry traces in panel D. This would help give context for how the model performance degrades by exclusion of those key terms. Finally, the authors claimed calcium activity increased following ipsilateral movements. However, Figure 3C clearly shows that both SXcontra and SXipsi increase beta coefficients. Instead, the choice direction may be represented in these neurons, given that beta coefficients increase following CXipsi and before SEipsi, presumably when animals make executive decisions. Could the authors clarify their interpretation on this point? Also, it is not clear if there is a photometry response related to motor parameters (i.e. head direction or locomotion, licking), which could change the interpretation of the reward outcome if it is related to a motor response; could the authors show photometry signal from representative 'high licking' or 'low licking' reward trials, or from spontaneous periods of high vs. low locomotor speeds (if the sessions are recorded) to otherwise clarify this point?

There are a few limitations with the design and timing of the synaptic manipulations that would improve the manuscript if discussed or clarified. The authors take care to validate the intersectional genetic strategies: Tetanus Toxin virus (which eliminates synaptic vesicle fusion) or CRISPR editing of Slc17a6, which prevents glutamate loading into synaptic vesicles. The magnitude of effect in the slice physiology results is striking. However, this relies on the co-infection of a second AAV to express channelrhodopsin for the purposes of validation, and it is surely the case that there will not be 100% overlap between the proportion of cells infected. Alternative means of glutamate packaging (other VGluT isoforms, other transporters, etc) could also compensate for the partial absence of VGluT2, which should be discussed. The authors do not perform a complimentary experiment to delete GABA release (i.e. via VGAT editing), which is understandable, given the absence of an effect with the pan-synaptic manipulation. A more significant concern is the timing of these manipulations as the authors acknowledge. The manipulations are all done in well-trained animals, who continue to perform during the length of viral expression. Moreover, after carefully showing that mice use different strategies on the 70/30 version vs the 90/10 version of the task, only performance on the 90/10 version is assessed after the manipulation. Together, the observation that EPNsst activity does not alter performance on a well-learned, 90/10 switching task decreases the impact of the findings, as this population may play a larger role during task acquisition or under more dynamic task conditions. Additional experiments could be done to strengthen the current evidence, although the limitation is transparently discussed by the authors.

Finally, intersectional strategies target LHb-projecting neurons, although in the original characterization, it is not entirely clear that the LHb is the only projection target of EPNsst neurons. A projection map would help clarify this point.

Overall, the authors used a pertinent experimental paradigm and common cell-specific approaches to address a major gap in the field, which is the functional role of glutamate/GABA co-release from the major basal ganglia output nucleus in action selection and evaluation. The study is carefully conducted, their analyses are thorough, and the data are often convincing and thought-provoking. However, the limitations of their synaptic manipulations with respect to the behavioral assays reduce generalizability and to some extent the impact of their findings.

Reviewer #2 (Public Review):

Summary:

This paper aimed to determine the role EP sst+ neurons play in a probabilistic switching task.

Strengths:

The in vivo recording of the EP sst+ neuron activity in the task is one of the strongest parts of this paper. Previous work had recorded from the EP-LHb population in rodents and primates in head-fixed configurations, the recordings of this population in a freely moving context is a valuable addition to these studies and has highlighted more clearly that these neurons respond both at the time of choice and outcome.

The use of a refined intersectional technique to record specifically the EP sst+ neurons is also an important strength of the paper. This is because previous work has shown that there are two genetically different types of glutamatergic EP neurons that project to the LHb. Previous work had not distinguished between these types in their recordings so the current results showing that the bidirectional value signaling is present in the EP sst+ population is valuable.

Weaknesses:

(1) One of the main weaknesses of the paper is to do with how the effect of the EP sst+ neurons on the behavior was assessed.

(a) All the manipulations (blocking synaptic release and blocking glutamatergic transmission) are chronic and more importantly the mice are given weeks of training after the manipulation before the behavioral effect is assessed. This means that as the authors point out in their discussion the mice will have time to adjust to the behavioral manipulation and compensate for the manipulations. The results do show that mice can adapt to these chronic manipulations and that the EP sst+ are not required to perform the task. What is unclear is whether the mice have compensated for the loss of EP sst+ neurons and whether they play a role in the task under normal conditions. Acute manipulations or chronic manipulations without additional training would be needed to assess this.

(b) Another weakness is that the effect of the manipulations was assessed in the 90/10 contingency version of the task. Under these contingencies, mice integrate past outcomes over fewer trials to determine their choice and animals act closer to a simple win-stay-lose switch strategy. Due to this, it is unclear if the EP sst+ neurons would play a role in the task when they must integrate over a larger number of conditions in the less deterministic 70/30 version of the task.

The authors show an intriguing result that the EP sst+ neurons are excited when mice make an ipsilateral movement in the task either toward or away from the center port. This is referred to as a choice response, but it could be a movement response or related to the predicted value of a specific action. Recordings while mice perform movement outside the task or well-controlled value manipulations within the session would be needed to really refine what these responses are related to.

(2) The authors conclude that they do not see any evidence for bidirectional prediction errors. It is not possible to conclude this. First, they see a large response in the EP sst+ neurons to the omission of an expected reward. This is what would be expected of a negative reward prediction error. There are much more specific well-controlled tests for this that are commonplace in head-fixed and freely moving paradigms that could be tested to probe this. The authors do look at the effect of previous trials on the response and do not see strong consistent results, but this is not a strong formal test of what would be expected of a prediction error, either a positive or negative. The other way they assess this is by looking at the size of the responses in different recording sessions with different reward contingencies. They claim that the size of the reward expectation and prediction error should scale with the different reward probabilities. If all the reward probabilities were present in the same session this should be true as lots of others have shown for RPE. Because however this data was taken from different sessions it is not expected that the responses should scale, this is because reward prediction errors have been shown to adaptively scale to cover the range of values on offer (Tobler et al., Science 2005). A better test of positive prediction error would be to give a larger-than-expected reward on a subset of trials. Either way, there is already evidence that responses reflect a negative prediction error in their data and more specific tests would be needed to formally rule in or out prediction error coding especially as previous recordings have shown it is present in previous primate and rodent recordings.

(3) There are a lot of variables in the GLM that occur extremely close in time such as the entry and exit of a port. If two variables occur closely in time and are always correlated it will be difficult if not impossible for a regression model to assign weights accurately to each event. This is not a large issue, but it is misleading to have regression kernels for port entry and exits unless the authors can show these are separable due to behavioral jitter or a lack of correlation under specific conditions, which does not seem to be the case.

Reviewer #3 (Public Review):

Summary:

The authors find that Sst-EPN neurons, which project to the lateral habenula, encode information about response directionality (left vs right) and outcome (rewarded vs unrewarded). Surprisingly, impairment of vesicular signaling in these neurons onto their LHb targets did not impair probabilistic choice behavior.

Strengths:

Strengths of the current work include extremely detailed and thorough analysis of data at all levels, not only of the physiological data but also an uncommonly thorough analysis of behavioral response patterns.

Weaknesses:

Overall, I saw very few weaknesses, with only two issues, both of which should be possible to address without new experiments:

(1) The authors note that the neural response difference between rewarded and unrewarded trials is not an RPE, as it is not affected by reward probability. However, the authors also show the neural difference is partly driven by the rapid motoric withdrawal from the port. Since there is also a response component that remains different apart from this motoric difference (Figure 2, Supplementary Figure 1E), it seems this is what needs to be analyzed with respect to reward probability, to truly determine whether there is no RPE component. Was this done?

(2) The current study reaches very different conclusions than a 2016 study by Stephenson-Jones and colleagues despite using a similar behavioral task to study the same Sst-EPN-LHb circuit. This is potentially very interesting, and the new findings likely shed important light on how this circuit really works. Hence, I would have liked to hear more of the authors' thoughts about possible explanations of the differences. I acknowledge that a full answer might not be possible, but in-depth elaboration would help the reader put the current findings in the context of the earlier work, and give a better sense of what work still needs to be done in the future to fully understand this circuit.

For example, the authors suggest that the Sst-EPN-LHb circuit might be involved in initial learning, but play less of a role in well-trained animals, thereby explaining the lack of observed behavioral effect. However, it is my understanding that the probabilistic switching task forces animals to continually update learned contingencies, rendering this explanation somewhat less persuasive, at least not without further elaboration (e.g. maybe the authors think it plays a role before the animals learn to switch?).

Also, as I understand it, the 2016 study used manipulations that likely impaired phasic activity patterns, e.g. precisely timed optogenetic activation/inhibition, and/or deletion of GABA/glutamate receptors. In contrast, the current study's manipulations - blockade of vesicle release using tetanus toxin or deletion of VGlut2, would likely have blocked both phasic and tonic activity patterns. Do the authors think this factor, or any others they are aware of, could be relevant?