409 Matching Annotations
  1. May 2021
    1. Author Response:

      Reviewer #1:

      In this manuscript Lituma et al. provides compelling evidence demonstrating the physiological role of presynaptic NMDA receptors at mossy fiber synapses. The existence of these receptors on the presynaptic site at this synapse was suggested more than 20 years ago based on morphological data, but their functional role was only shown in a single abstract since then (Alle, H., and Geiger, J. R. (2005)). The current manuscript uses a wide variety of complementary technical approaches to show how presynaptic NMDA receptors contribute to shaping neurotransmitter release at this synapse. They show that presynaptic NMDA receptors enhance short-term plasticity and contribute to presynaptic calcium rise in the terminal. The authors use immunocytochemistry, electrophysiology, two-photon calcium imaging, and uncaging to build a very solid case to show that these receptors play a role at synaptic communication at mossy fiber synapses. The authors conclusions are supported by the experimental data provided.

      The study is built on a solid and logical experimental plan, the data is high quality. However, the authors would need to provide stronger evidence to demonstrate the physiological function of these receptors. It is hard to reconcile these experimental conditions with the authors' claim in the abstract: 'Here, we report that presynaptic NMDA receptors (preNMDARs) at hippocampal mossy fiber boutons can be activated by physiologically relevant patterns of activity'. We know that extracellular calcium can have a very significant impact of neurotransmitter release and how short-term plasticity is shaped. For this reason, it would be important to explore how the activity of these receptors at more physiological calcium concentrations contribute to calcium entry and short-term plasticity at these synapses.

      We thank the reviewer for noting our study is “built on a solid and logical experimental plan, the data is of high quality”. We agree with the reviewer that exploring the role of preNMDAR under more physiological conditions is extremely important. In response, we have performed new experiments at 35 ºC and at a more physiological 1.2 mM Ca+2 and 1.2 mM Mg+2 concentrations. Our new results, now included in Figure 4-figure supplement 1, demonstrate that our conclusion that preNMDARs at mossy fiber boutons can be activated by physiologically relevant patterns of activity is also true under more physiological recording conditions.

      Reviewer 2:

      Lituma et al. examined the presence and functions of preNMDARs in dentate gyrus granule cells (GCs) in the hippocampus. The authors found that GluN1+ preNMDARs are indeed present at mossy fiber (mf) terminals with electron microscopy. With pharmacological and genetic approaches, the authors showed that preNMDARs are important in low frequency facilitation (LFF), burst-induced facilitation and information transfer at the mf-CA3 synapse. The authors further demonstrated that this preNMDAR contribution is independent of the somatodendritic compartment of the GCs. With 2-photon calcium imaging, the authors found that preNMDARs contribute to presynaptic Ca2+ transients and can be activated by local glutamate uncaging. Separately, the authors showed that GluN1+ preNMDARs might also contribute to BDNF release at mossy fiber terminals during repetitive stimulation. Lastly, non-postsynaptic NMDARs specifically mediates mf transmission onto mossy cells, similar to mf-CA3 synapses, but not interneurons. The authors concluded that preNMDARs mediate synapse-specific transmission originating from the GCs/mf inputs.

      Overall, the study provides compelling evidence from a battery of techniques, ranging from EM, pharmacology, genetic deletion, electrophysiology to 2-photon imaging/uncaging. The data supports a coherent story on the presence of preNMDARs at mf terminals and that preNMDARs play important roles in LFF.

      In conclusion, this study reveals how NMDA receptors can be found in unexpected locations and how they may have unconventional functions, i.e. outside the narrow textbook view that they primarily serve as coincidence detectors in Hebbian learning. This study thus helps to change the way we think about NMDA receptor functioning, so should be of broad interest.

      We appreciate the reviewer’s comments that our study provides compelling evidence for the presence and role of preNMDARs at mossy fiber terminals. We also agree with the reviewer that our study challenges the way we think about NMDA receptor function.

      Reviewer #3:

      In this manuscript Lituma and colleagues investigate a potential role for presynaptic NMDARs at hippocampal mossy fiber (MF) synapses in regulating synaptic transmission. The combined use of electron microscopy, electrophysiology, optogenetics, calcium imaging, and genetic manipulations expertly employed by the authors yields high quality compelling evidence that presynaptic NMDARs can participate in activity dependent short term facilitation of release onto postsynaptic CA3 pyramid and mossy cell targets but not onto inhibitory interneurons. Moreover, presynaptic NMDAR activation is demonstrated to be particularly effective in promoting BDNF release from MF boutons. The investigation is well designed with a clear hypothesis, appropriate methodological considerations, and logical flow yielding results that fully support he authors conclusions. The manuscript fills an important gap in our understanding of MF regulation by unambiguously confirming a functional role for presynaptic NMDARs that were first described anatomically at MF terminals nearly 30 years ago. Combined with a handful of other studies describing presynaptic NMDARs at various central synapses this study expands the role of NMDARs as critical players in synaptic plasticity on both sides of the cleft.

      We very much appreciate the reviewer’s positive remarks of our study as “well designed with a clear hypothesis, appropriate methodological considerations, and logical flow”. We concur that the manuscript fills an important gap in understanding MF regulation by preNMDARs and expanding the role of NMDARs in synaptic plasticity on both sides of the cleft.

    2. Reviewer #3 (Public Review):

      In this manuscript Lituma and colleagues investigate a potential role for presynaptic NMDARs at hippocampal mossy fiber (MF) synapses in regulating synaptic transmission. The combined use of electron microscopy, electrophysiology, optogenetics, calcium imaging, and genetic manipulations expertly employed by the authors yields high quality compelling evidence that presynaptic NMDARs can participate in activity dependent short term facilitation of release onto postsynaptic CA3 pyramid and mossy cell targets but not onto inhibitory interneurons. Moreover, presynaptic NMDAR activation is demonstrated to be particularly effective in promoting BDNF release from MF boutons. The investigation is well designed with a clear hypothesis, appropriate methodological considerations, and logical flow yielding results that fully support he authors conclusions. The manuscript fills an important gap in our understanding of MF regulation by unambiguously confirming a functional role for presynaptic NMDARs that were first described anatomically at MF terminals nearly 30 years ago. Combined with a handful of other studies describing presynaptic NMDARs at various central synapses this study expands the role of NMDARs as critical players in synaptic plasticity on both sides of the cleft.

    3. Reviewer #2 (Public Review):

      Lituma et al. examined the presence and functions of preNMDARs in dentate gyrus granule cells (GCs) in the hippocampus. The authors found that GluN1+ preNMDARs are indeed present at mossy fiber (mf) terminals with electron microscopy. With pharmacological and genetic approaches, the authors showed that preNMDARs are important in low frequency facilitation (LFF), burst-induced facilitation and information transfer at the mf-CA3 synapse. The authors further demonstrated that this preNMDAR contribution is independent of the somatodendritic compartment of the GCs. With 2-photon calcium imaging, the authors found that preNMDARs contribute to presynaptic Ca2+ transients and can be activated by local glutamate uncaging. Separately, the authors showed that GluN1+ preNMDARs might also contribute to BDNF release at mossy fiber terminals during repetitive stimulation. Lastly, non-postsynaptic NMDARs specifically mediates mf transmission onto mossy cells, similar to mf-CA3 synapses, but not interneurons. The authors concluded that preNMDARs mediate synapse-specific transmission originating from the GCs/mf inputs.

      Overall, the study provides compelling evidence from a battery of techniques, ranging from EM, pharmacology, genetic deletion, electrophysiology to 2-photon imaging/uncaging. The data supports a coherent story on the presence of preNMDARs at mf terminals and that preNMDARs play important roles in LFF.

      In conclusion, this study reveals how NMDA receptors can be found in unexpected locations and how they may have unconventional functions, i.e. outside the narrow textbook view that they primarily serve as coincidence detectors in Hebbian learning. This study thus helps to change the way we think about NMDA receptor functioning, so should be of broad interest.

    4. Reviewer #1 (Public Review):

      In this manuscript Lituma et al. provides compelling evidence demonstrating the physiological role of presynaptic NMDA receptors at mossy fiber synapses. The existence of these receptors on the presynaptic site at this synapse was suggested more than 20 years ago based on morphological data, but their functional role was only shown in a single abstract since then (Alle, H., and Geiger, J. R. (2005)). The current manuscript uses a wide variety of complementary technical approaches to show how presynaptic NMDA receptors contribute to shaping neurotransmitter release at this synapse. They show that presynaptic NMDA receptors enhance short-term plasticity and contribute to presynaptic calcium rise in the terminal. The authors use immunocytochemistry, electrophysiology, two-photon calcium imaging, and uncaging to build a very solid case to show that these receptors play a role at synaptic communication at mossy fiber synapses. The authors conclusions are supported by the experimental data provided.

      The study is built on a solid and logical experimental plan, the data is high quality. However, the authors would need to provide stronger evidence to demonstrate the physiological function of these receptors. It is hard to reconcile these experimental conditions with the authors' claim in the abstract: 'Here, we report that presynaptic NMDA receptors (preNMDARs) at hippocampal mossy fiber boutons can be activated by physiologically relevant patterns of activity'. We know that extracellular calcium can have a very significant impact of neurotransmitter release and how short-term plasticity is shaped. For this reason, it would be important to explore how the activity of these receptors at more physiological calcium concentrations contribute to calcium entry and short-term plasticity at these synapses.

    5. Evaluation Summary:

      This paper will be of interest to a larger neuroscience community as this is the first functional demonstration of presynaptic NMDA receptors at mossy fiber terminals in the hippocampus. NMDA receptors are generally known for being critically involved in learning & memory as coincidence detectors in Hebbian plasticity. Some studies, however, find NMDA receptors that function in more unconventional manners. The present paper provides strong evidence for the existence of such unconventional NMDA receptors at a specific subset of hippocampal mossy-fibre boutons. The combined use of electron microscopy, electrophysiological, optogenetic, calcium imaging, and genetic manipulation approaches expertly employed by the authors yields high quality compelling evidence in full support of the study's main conclusions. Overall, the investigation is well designed with a clear hypothesis, appropriate methodological considerations, and logical flow resulting in a well written manuscript that is sure to be of broad scientific interest.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)

  2. Apr 2021
    1. Reviewer #3 (Public Review):

      Mutations in DHCR7, a key enzyme in cholestrol biosysnthesis have been shown to result in Smith-Lemili-Opitz syndrome. However, the mechanism by which loss of this enzyme alters brain development has not been resolved.

      In this study, the authors demonstrate that DHCR7 depletion results in depletion of cholestrol in the brain and also the accumulation of the substrate 7 dehydrocholestrol. These observations are conserved in both the brain of DHCR7 knockout mice as well as patient derived iPSC differentiated in vitro.

      The authors present evidence that the developmental defects in the brain are a consequence of accelerated differentiation of NSC into neural cells. These defects could be recapitulated by the addition of 7DHC metabolites on wild type cells.

      Throughout the manuscript, the authors demonstrate that their findings are conserved between DHC7 k/o mice and patient derived iPSC for SLO syndrome.

      To explain the mechanism underlying the cellular phenotypes described, authors propose that the accumulated 7DHC metabolites bind to and activate the glucocorticoid receptor leading to transcriptional activity.

      Overall this paper attempts to provide a comprehensive mechanistic explanation for the neurodevelopmental phenotype arising from the loss of a lipid metabolizing enzyme.

    2. Reviewer #2 (Public Review):

      The authors study in this report enzymes and sterols implicated in SLOS. They have performed in-vitro and in-vivo experiments. They show that a major metabilte, DHCEO, mediates the effects in neurogenesis and neuronal localisation. They have studied the mechanism of action of this effect. Pharmacological intervention can rescue the negative effects.

      The Introduction is clearly written and provides nice background information on the disorder, the implicated enzymes and sterols.

      The authors analyse extensively cell survival, neurogenesis, proliferation, several progenitor markers in both cell culture and in the Dhcr7-KO mice. In vivo they study several developmental stages.

      They have generated SLOS hiPSCs and studied those too.

      The analysis of sterol and oxysterol levels in WT vs Dhcr7-KO is very interesting and informative.

      The Dhcr7 shRNA experiments show clear effects on neurogenesis and cycling precursor cell population number.

      The RNAseq experiments also give interesting gene expression results and possible signaling pathways involved.

    3. Reviewer #1 (Public Review):

      The study is focused on neural deficits in Smith-Lemli-Opitz syndrome (SLOS) that is caused by loss of function of 3b-hydroxysterol-D7 -reductase (DHCR7) and results in lower cholesterol. Individuals with SLOS have cognitive impairment and the authors use mouse models and human iPSCs to investigate the effects of the SLOS mutation on neural progenitor proliferation and neurogenesis. Data show that the loss of DHCR7 leads to premature differentiation of cortical progenitors and altered cortical development. However, the work offers little mechanistic insight.

    4. Evaluation Summary:

      This paper will be of interest to developmental biologists and neuroscientists as it aims to resolve the unknown mechanism by which loss of a key enzyme in cholesterol biosynthesis results in neurodevelopmental defects. It provides a conceptual framework for understanding how altered lipid metabolism can impact brain development. Many of the key claims of the paper are well-supported, but reasonable alternative explanations remain.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Author Response:

      For the reader, we specifically want to highlight the following new data (Figures 7, 8 and accompanying supplements) that were added:

      1) To directly compare physiological (Wntoff/Wnton) and oncogenic (i.e. constitutively active) signaling, we generated a second cell line using CRISPR/Cas9 genome editing, harboring an oncogenic point mutant form of CTNNB1 (SGFP2-CTNNB1S45F).

      2) To further quantify the levels, complex state and multimerization status when WNT/CTNNB1 signaling is hyperactivated, we performed additional FCS and N&B experiments in the new mutant cell line and upon GSK3B inhibition by CHIR99021 treatment (as requested by reviewers 1 and 2).

      3) We use these same perturbations to strengthen the link between our experimental data and the computational model (as suggested by reviewer 2) and provide access to the model in the form of an interactive app (available at https://wntlab.shinyapps.io/WNT_minimal_model/).

      While we are in the process of further revising our manuscript, we do want to take this opportunity to briefly reply to two of the points made by Reviewer #1:

      The authors have concluded with FCS that the diffusion coefficient of free β-catenin to be 14.9 um2/s (line 259) and the complexed β-catenin to be 0.17 um2/s (line 327). Similar to the authors' argument in the manuscript, this difference means about a 100-fold change of the complex length scale. If the complex is linear, this means a 100-fold change in molecule size, but if the complex is spherical, this means a one-million-fold increase of the molecule size.

      To clarify: We indeed measure a 14.9/0.17 = 87-fold change in speed. IF we assume Einstein-Stokes relation, this would be indicative of a 87^3 change in molecular size. However, the Einstein-Stokes equation is only valid when specific conditions are met (including the assumption that we are dealing with perfectly spherical particles in a homogeneous environment). Therefore, we noted the following in the material and methods section: “It must be noted that, especially for larger protein complexes, the linearity between the radius of the protein and the speed is not ensured, if the shape is not globular, and due to other factors such as molecular crowding in the cell and hindrance from the cytoskeletal network. We therefore did not estimate the exact size of the measured CTNNB1 complexes, but rather compared them to measurements from other FCS studies.” Put differently: The most important take home message is not an absolute size estimate of the CTNNB1 complex (which is why were careful not to make that point explicitly, although it is unlikely that this complex only contains one copy of the ‘standard’ destruction complex components APC, AXIN, GSK3 and CK1), but the fact that this complex is still present after WNT stimulation, although it does undergo a substantial reduction in size (a 3.5-fold change in speed upon WNT stimulation), and thus changes its identity. We will take care to ensure that a future revision leaves no room for further confusion on this point.

      From the biology point of view, APC is the backbone of the destruction complex, which has several β-catenin binding sites by itself. Additionally, APC also contains several Axin1 binding sites where each Axin1 can also recruit one β-catenin. It is unlikely that one APC complex contains only one β-catenin, not mentioning the potential oligomerization of APC.

      Here we can only agree with the reviewer: We were equally surprised by the findings from our N&B analysis, which is why we extensively discuss possible explanations in our manuscript. Future follow-up by ourselves and others will reveal in how far our interpretation of these measurements stands the test of time.

    2. Reviewer #2 (Public Review):

      Wnt signaling plays critical roles in cell fate determination in essentially every tissue in all animals, regulates tissue homeostasis in many adult tissues, and is inappropriately activated in many human cancers. It has been the focus of research for decades, and we have an outline of signal transduction. However, remarkably, key questions remain controversial. Central among these are questions about the nature of the negative regulatory destruction complex, its mechanism of action and how it is turned down by Wnt signaling. Here Saskia and colleagues take a novel and very exciting approach to these questions, combining innovative quantitative live-cell imaging and computational modelling.

      What I can say unequivocally is that there is data in this manuscript that will force a re-evaluation of our current models of Wnt signaling, and also serve as the foundation for future research. Particular notable are: 1) precise measurements of the concentrations of beta-catenin in the cytoplasm and nucleus before and after Wnt signaling and after inhibition of GSK3. 2) Definition of a high MW complex, likely the destruction complex, whose assembly state appears to be regulated by Wnt signaling, and 3) Intriguing evidence that at steady state this complex appears not to contain multiple copies of beta-catenin. These data are exceptionally interesting and timely, as controversy continues about the size/assembly state of the destruction complex.

    3. Reviewer #1 (Public Review):

      The authors have done a great job in carefully labeling the β-catenin with fluorescent protein SGFP2 and quantitatively measuring the β-catenin behavior during Wnt pathway activation with advanced biophysical methods. This is an excellent effort on quantitative biological studies. The knock-in constructs, the cell lines the authors made are great resources for the Wnt field. And the quantification like the β-catenin concentration, β-catenin diffusion coefficient are great knowledge for future studies. The finding that S45F mutation lead to higher fraction of the slow-moving complexes is interesting. Other areas could borrow the research ideas and methods used in this manuscript. My primary concern is the difficulty of interpreting some of the quantitative results in the biological context.

      The authors have concluded that β-catenin has two major populations: free population and slow-diffusing complexed population. The authors have concluded with FCS that the diffusion coefficient of free β-catenin to be 14.9 um2/s (line 259) and the complexed β-catenin to be 0.17 um2/s (line 327). Similar to the authors' argument in the manuscript, this difference means about a 100-fold change of the complex length scale. If the complex is linear, this means a 100-fold change in molecule size, but if the complex is spherical, this means a one-million-fold increase of the molecule size. Furthermore, in the next section, with the N&B method, the authors have suggested that "few, if any, of these complexes contain multiple SGFP2-CTNNB1 molecules" (line 366). When combining the two parts of information, it is hard to imagine a complex that contains one thousand to one million molecules only have one or a few β-catenin subunits. From the biology point of view, APC is the backbone of the destruction complex, which has several β-catenin binding sites by itself. Additionally, APC also contains several Axin1 binding sites where each Axin1 can also recruit one β-catenin. It is unlikely that one APC complex contains only one β-catenin, not mentioning the potential oligomerization of APC. The conclusion that most of the β-catenin containing complexes has only one β-catenin could either be real or due to the misinterpretation of experimental data.

    4. Evaluation Summary:

      Wnt signaling plays critical roles in cell fate determination in essentially every tissue in all animals, regulates tissue homeostasis in many adult tissues, and is inappropriately activated in many human cancers. It has been the focus of research for decades, and we have an outline of signal transduction. Remarkably, some of the key questions of Wnt signaling remain controversial. Central among these are questions about the nature of the negative regulatory destruction complex, its mechanism of action and how it is turned down by Wnt signaling. Here Saskia and colleagues take a novel and very exciting approach to these questions, combining innovative quantitative live-cell imaging and computational modelling.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      This is an interesting paper combining several impressive techniques to argue that synaptically released glutamate is allowed to diffuse to and activate receptors at much greater distance than previously thought. iGluSnFR recordings show that glutamate released from single vesicles activates the indicator with a spatial spread (length constant) of 1.2 um, substantially farther than previous estimates based on the time course of glutamate clearance by glial transporters (PMC6725141). Similar parameters are observed with spontaneous and evoked events, large or small, or when glutamate is released via 2P uncaging. Further uncaging experiments show that both AMPARs and especially NMDARs are activated a substantial distance. AMPARs, previously thought to be recruited only within active synapses, are activated with a spatial length constant that compares quite closely with the average distance between synapses in the hippocampus. More heroic experiments and some geometric calculations show that this behavior enables neighboring synapses to interact supralinearly. The results suggest that "crosstalk" between neighboring synapses may be substantially more common than previously thought.

      The experiments in this paper appear carefully performed and are analyzed thoroughly. Despite all of the quantitative rigor and careful thought, however, the authors fail to reconcile convincingly their results with what we know about neuropil structure and the laws of diffusion. There are very good data in the literature regarding the extracellular volume fraction and geometric tortuosity of the neuropil, the diffusion characteristics of glutamate and the time course of glutamate uptake. These data more or less demand that synaptically released glutamate is diluted over a much smaller spatial range than that suggested here. In the Discussion, the authors suggest that this discrepancy might reflect a simplified view of the neuropil as an isotropic diffusion medium (PMC6763864, PMC6792642, PMC6725141), whereas a more realistic network of sheets and tunnels (PMC3540825) might prolong the extracellular lifetime of neurotransmitter. I like this idea in principle, but there is no quantitative support in the paper for the claim - in fact, it seems at odds with the authors' very nice demonstration that diffusion appears to be similar in all directions (Figure 3B). I don't necessarily think a solution is within the scope of this single paper, but I would suggest that the authors acknowledge the present lack of a compelling explanation.

    2. Reviewer #2 (Public Review):

      Matthews, Sun, McMahon et al. addresses the extent of the spread of the neurotransmitter glutamate into the extracellular space. The authors use a combination of imaging techniques, 2-photon glutamate uncaging and electrophysiology to conclude that vesicular glutamate release reaches nearby, adjacent synapses. Although this is an interesting question, and one that has been addressed many times previously, I have several technical concerns about the strength of the conclusions that reduces my enthusiasm.

    3. Reviewer #1 (Public Review):

      This MS combines two-photon glutamate sensing (using the iGluSnFR fluorescent probe), two-photon glutamate uncaging, two-photon calcium imaging and electrophysiology to investigate whether synaptically released glutamate activates receptors outside the synapse of release, and at neighboring synapses. The data themselves are very impressive. The authors arrive at the revolutionary conclusion that synaptically released glutamate is able to activate both NMDA and even AMPA receptors at neighboring synapses, remarkably strongly. I say revolutionary, because previous modelling has yielded diametrically opposite conclusions. The reflex would be to prefer experiment over theory, yet the modelling was based upon quite strongly constrained physical parameters that would be quite incompatible with the interpretations reported here. However, I believe the authors have failed to take into account significant technical limitations inherent in the technologies they apply. These include spatial averaging of fluorescence, possible saturation of iGluSnFR and diffusive exchange of (caged) glutamate during uncaging. As a result, the conclusion is wholly unproven. Indeed, I believe it highly probable that all of the data in favor of distal activation will prove to be consistent with synapse specificity and the presence of technical artifacts related to spatial averaging of fluorescence signals and diffusive exchange of (caged) glutamate during uncaging.

    4. Evaluation Summary:

      The authors address the spatial spread of glutamate outside of synapses, with the surprising conclusion that glutamate released at one synapse can strongly activate receptors at neighboring synapses. This manuscript should interest those studying neural signaling and techniques associated with that field. However, caveats of the advanced techniques used to address this difficult question limit the strength of the main conclusion.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #2 (Public Review):

      This interesting study from Kurashina et al. examines novel postmitotic roles for transcription factors traditionally considered to specify neuronal cell fate. The paper examines a form of synaptic tiling in C. elegans motor neurons to provide evidence that the unc-4 and unc-37 transcription factors, previously implicated in determining cholinergic motor neuron identity, have additional roles in the regulation of synaptic wiring that are at least partially separable from cell fate specification.

      The authors develop new tools for defining the temporal actions of unc-4 and unc-37 and the clean dissection of the spatiotemporal requirements for unc-4/unc-37 transcriptional regulation is a major advance offered by the study. In particular, the authors demonstrate that unc-4 acts at a later development stage to control synaptic wiring compared with its role in cell fate regulation. Overall, the paper is clearly written and offers new insight into how transcription factors that act to define neuronal identity may have additional roles in specifying aspects of synapse organization. The study falls a little short in clearly defining mechanism of action downstream of unc-4/unc-37 and in describing the relationship of these newly described roles for unc-4 and unc-37 to those previously described.

      The authors use a clever strategy to assess tiling of individual cholinergic motor neurons using DA8 and DA9 as a model, but in some cases observe variable degradation of the RAB-3::GFPnovo, presumably due to weak expression of ZIF1 in some of the mutants. This makes it a little difficult to assess the tiling defects in some of the figures. The residual GFPnovo signal seems to be defined based on colocalization with the more broadly expressed mCherry::RAB-3 marker, but no data is shown for the extent of colocalization in the absence of ZIF1. This analysis would benefit from more explanation.

      The analysis of temporal requirements using ts alleles in combination with the AID system is very convincing and quite informative. The authors clearly show a later requirement for proper tiling, at stages when cell fate determination is expected to be complete. However, it is less clear how these newly defined aspects of unc-4 and unc-37 functions relate to their previously defined roles.

      The authors examine PLX-1::GFP subcellular localization in DA neurons (using cell specific itr-1 promoter) of unc-4 mutants but do not directly examine plx-1 expression levels in DA neurons. This analysis would further solidify links between plx-1 and unc-4 transcriptional regulation.

      Did the authors examine whether degradation of unc-4 and/or unc-37 at much later developmental time points also lead to tiling defects? Is there an ongoing requirement to maintain tiling?

      Did the authors examine whether the unc-4::AID and unc-37::AID animals became uncoordinated subsequent to treatment with auxin analog? Do the tiling defects potentially contribute to locomotor changes?

    2. Reviewer #1 (Public Review):

      This manuscript by Kurashina et al. describes a novel post-mitotic role in synaptic patterning for a cell fate determining gene, unc-4, and its co-repressor, unc-37. The DA neurons in C. elegans are cholinergic motor-neurons that exhibit unique synaptic tiling of their dorsal axonal segments. Mizumoto et al has previously shown that Semaphorin-Plexin signaling is required to establish the tiling between DA8 and DA9, by functioning in cis in the DA9 neuron. Using temperature sensitive mutant of unc-4, as well as a combination of CRISPR/Cas9 genome editing with the AID system for specific temporal degradation, the authors nicely examine the spatiotemporal requirement of unc-4, and show that unc-4 is required only post-mitotically for synapse tiling, but not during the development of the DA neurons. Interestingly, activity of the corepressor unc-37 is required both during development and postmitotically for correct tiling. unc-4 and unc-37 are suggested to function by inhibiting the canonical wnt signaling. Overall this is an interesting study which sheds light on our understanding of the post-developmental role of cell fate genes in synapse patterning. I only have one major issue that requires some clarification. The authors present in their introduction the results from Kerk et al, regarding the role of unc-4 as a cell fate determining gene for the VAs and DAs. Kerk et al have shown that UNC-4 is specifically required for the expression of DA genes, without affecting ACh pathway genes. Table 1, however, doesn't fully recapitulate the same results and actually shows that unc-4 and unc-37 mutants do not exhibit significant cell fate defects. The authors use these results to argue in the discussion that the synaptic patterning defects can occur independent of the cell fate transformation. The issue of unc-4 as a cell fate determining gene of A type motor neurons needs to be more clearly addressed. The authors should test whether acr-5 expression is elevated in DAs in unc-4 and unc-37 mutants (Winnier 1999, Kerk 2017). In addition, they should also analyze these DB markers in the temp shift experiments (do the VB-DB markers show up in the post embryonic knockout? And if so, will silencing them specifically in DA neurons rescue the tiling defects?). The discussion, accordingly, should also address these issues.

    3. Evaluation Summary:

      This paper is of potential interest to a broad audience of neuroscientists, as it adds to our growing understanding of transcriptional mechanisms that regulate neural connectivity. Specifically, the paper provides support for the idea that transcriptional pathways previously implicated in neuronal cell fate determination can have independent roles in specifying connectivity between neurons. The study is highly technically innovative and cleverly uses a set of newly developed tools to analyze the developmental time window over which transcriptional activity is required to achieve to proper connectivity. However, the paper falls a little short in defining specific mechanisms involved downstream of the transcription factors themselves.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #4 (Public Review):

      Higashi et al. provide a new "Brownian ratchet" model for DNA loop extrusion mechanism by cohesin, a member of SMC protein family complexes. Based on previous works on crystal structures, cryo-EM structures, and DNA-protein crosslinking experiments, they shed light on two HEAT-repeat DNA binding modules on cohesin - Scc2-head and Scc3-hinge - and their relationships. They hypothesized that the association between Scc2-head and Scc3-hinge modules were dissociated and Scc2-head released DNA upon ATP hydrolysis, driving DNA slipping. By performing FRET experiments, they found that Scc2 and hinge modules indeed come close only in ATP-bound "Gripping" state, while hinge and Scc3 are always close to each other. Therefore, they suggest that, for DNA loop extrusion model, 1) upon ATP binding to the head domains, both Scc2-head and Scc3-hinge modules grip DNA, 2) when ATPs are hydrolyzed, Scc2-head module releases DNA so that DNA-associating Scc3-hinge module pulls DNA depending on stochastic Brownian motion of Scc3-hinge module, then 3) both Scc2-head and Scc3-hinge modules release DNA and go back to the state 1). This "Brownian ratchet" model also provides an explanation of how cohesin entraps DNA by opening the gate between Smc3 and Scc1, which also nicely explains the known facts regarding Scc1 cleavage-dependent DNA release and in vitro behaviors of single cohesin molecules that topologically bound to DNA. In addition, by performing computational modeling, they showed that the Brownian ratchet model well fits all previously reported in vitro loop extrusion assays by cohesin and condensin, making their model rigid and reliable.

      Their model is mostly well supported by data, but several detailed points need to be explained or clarified.

      1) In Figure 2C FRET experiments, proximity of Scc3-C and Scc2-N does not seem to be drastically increased in Gripping state compared to the case of hinge and Scc2-N. This could be because the FRET pairs (Scc3-C and Scc2-N) are still far. If the authors could label internal part in Scc3, this could solve the problem. In addition, if Scc3-C and Scc2-N are always close to each other irrespective of Gripping state, the authors should consider this fact in their modeling.

      2) Major differences between topological loading and loop extrusion is kleisin-gate opening and head gate passage. Even if kleisin-gate wouldn't be opened, DNA should be released after head opening like in the topological loading. In case it happens, DNA and Scc1 would be tangled and it seems to be difficult to come back to next gripping state again. It would be helpful to add the explanation of why such tangling DNAs do not have to be considered in the model.

      3) In the manuscript line 338, the authors mention "After DNA dissociation from the Scc3-hinge module, there is a time without tight contact between the cohesin ring and the DNA loop." However, both in Figure 3B and 4F, it seems that head-Scc2 always associates with DNA. This could be discrepancy. The authors should clarify the point if certain free time without any contact to DNA is assumed in the modeling.

      4) Generally, initial DNA bending is the most challenging part in loop extrusion models. Especially in Figure 3B-a, such a bent DNA seems to be impossible if we consider the persistence length of DNA is 50 nm. The authors should discuss how DNA loop extrusion could be initiated.

    2. Reviewer #3 (Public Review):

      Bifurcation between topological loading and loop extrusion is determined by DNA passing through the N-gate. For loop extrusion to occur processively, this decision needs to be made only once at the beginning. However, the authors also argue that Scc2 dissociation between rounds of ATPase cycles is required for symmetric loop extrusion. In combination, the model requires that N-gate opening is allowed only at the very beginning and cannot occur during loop extrusion, even when the cohesion loader is released. The authors should state whether this interpretation is correct and feasible given the structural data.

      Loop extrusion has never been observed using yeast cohesin. It will be important to learn how the authors reconcile their model and the lack of experimental demonstration of loop extrusion in a reconstituted system.

      The discrepancy in speed and the measured ATPase rate is not discussed. In vitro, loop extrusion rates are about 1000 bp per second and in vivo measurements of gamma-H2AX spreading from a double strand break, ~150kbp per min according to PMID: 32527834, which was proposed to be caused by loop extrusion (PMID: 33597753), also matches that in vitro rate. But the authors model accounts for only about 100 bp extrusion per ATPase cycle whereas the average ATPase rate is 1 per second. They do mention that the model requires 9 ATP hydrolyzed per second but do not make an attempt to explain the discrepancy.

    3. Reviewer #2 (Public Review):

      How the genome chromatin fiber is folded into loops and topologically associating domains (TADs) remains unclear. A recent attractive model is that these genomic structures are formed by a loop extrusion process mediated by cohesin. While the Uhlmann group has proposed an alternative mechanism, the diffusion capture model, to make loops (Cheng et al., 2015; Gerguri et al., 2021), in this paper, Higashi et al. proposed a structure-based model providing mechanistic insight into the reported loop extrusion activity of cohesin. For its topological DNA binding, cohesin inserts DNA into the cohesin ring by sequential passage through a kleisin N-gate and an ATPase head gate. Hisgashi et al. suggested that the gripping state in which DNA has not passed the kleisin N-gate might facilitate the loop extrusion activity reported. This paper is very intriguing, and informative to the chromatin/chromosome field. My specific comments are the following:

      1) Since this paper is primarily based on the detailed structural information on cohesin loading onto DNA, which the Uhlmann group published in Mol Cell (2020), it might be hard for general readers to follow the whole story in this paper. For better understanding, the authors should provide readers with Supplemental Fig. corresponding to the Graphical abstract and Figs. 6E/7G in the Mol Cell paper, and adequately explain it first. Structural models such as Fig. 1 are accurate but might be difficult to capture cohesin's dynamic behavior with DNA.

      2) Although this paper is very intriguing, it looks like a review paper, and the authors' message is not so clear. Given that the Uhlmann group has proposed an alternative mechanism to make loops, I wonder whether the main message might be that the loop extrusion, like reported in vitro, is unlikely to occur in vivo. If so, the authors should clearly state the point and shorten the Discussion part to enhance the paper's impact.

      3) Page 24. The critical issue of the loop extrusion mechanism proposed is "not opening" of kleisin N-gate. The authors discussed that the low salt condition in vitro could be a reason: " For instance, electrostatic interactions contribute to keeping the kleisin N-gate closed and these are augmented in a low salt buffer." However, I assume that the condition also helps the topological loading, and this explanation is not so convincing.

      4) While I agree with the authors' loop extrusion mechanism, there are other models to explain cohesin loading onto DNA (e.g., Shi et al., 2020; Collier et al.). They might want to discuss its compatibility with them.

    4. Reviewer #1 (Public Review):

      Higashi et al. present a molecular mechanism of how the cohesin complex, a supramolecular assembly of several proteins, can topologically embrace DNA and actively extrude DNA into loops. The loop-extruding activity of cohesin and of related condensin complexes have been proposed to represent a cornerstone of genome organization. While recent in vitro studies demonstrate that cohesin and condensin complexes are capable of extruding loops, the molecular mechanisms driving loop extrusion, i.e. how ATP energy is utilized, and what underlies the processivity of the loop extrusion, remains enigmatic. The cohesin complex consist of two long flexible protein "arms" connected at the 'hinge' ends. The other, 'head' ends are linked by the kleisin protein and also can dimerize to form an ATP-binding chamber. Defining how transitions in the cohesin complex structure and its ATPase activity underlies known cohesin functions has been the object of numerous studies for over two decades. Here, the authors build upon these studies.

      The authors start by analyzing available structural data for cohesin domains and associated loading factors. First, by combining the structure of the cohesin-head-domain complex engaged with DNA in ATP-bound state and the corresponding free crystal structures, they show that the 'head module' in the ATP-bound state can tightly wrap around DNA, and upon ATP hydrolysis the DNA-embracing cavity will dilate. In other words, the complex transitions from a 'DNA-gripping state' into a 'DNA-slipping' state after ATP is hydrolyzed. Next, they show that the other DNA-binding module, the 'hinge module', does not change its interaction with the DNA after ATP hydrolysis. The authors also conclude that ATP hydrolysis weakens the interaction between the head and hinge modules, suggesting that the cohesin ring alternates between folded (with head and hinge closed) and unfolded ('free' hinge) states. The authors next carried out FRET experiments to provide experimental evidence for the predicted change in spatial arrangement between the head and hinge modules. Based on this structural analysis, they propose that whether DNA is passed (or not) through the 'kleisin gate' before binding to the head module (into the gripping state) determines if the DNA will be released inside the cohesin ring (i.e. 'topological entry') or if the DNA will remain loosely associated with the head module (i.e. 'loop extrusion') upon ATP hydrolysis. In the latter case, repetitive simultaneous binding of DNA to the head and hinge modules in a folded state followed by relaxation of the cohesin ring while DNA remains bound to the hinge module, may result in a overall 'inward' directed motion of the DNA thread relative to the head domain, i.e. loop extrusion. Stochastic simulations of a coarse-grain model further support that such a model can give rise to loop extrusion.

      The real strength of the paper is in its combination of several pieces of structural and biophysical data that results in a compelling mechanism for cohesin function. The outcome is a united model for cohesin's two characteristic activities - topological engagement of the DNA and DNA loop extrusion. Importantly, the authors explore the role of ATP hydrolysis in driving conformational changes, and, thus, the translocation of DNA, as well as the role of the DNA binding kinetics. The authors go on to relate these findings to the consequences for cohesin function inside cells, where it must content with chromatized substrates. For example, they suggest that while a single nucleosome probably can be bypassed by the cohesin complex, an array of the nucleosome may present a significant hindrance.

      Given its interdisciplinary nature and important conclusions, I believe that this paper will be of broad interest to scientists across disciplines and will influence and stimulate future consideration of how cohesin contributions to the spatiotemporal organization of chromatin.

    5. Evaluation Summary:

      This work combines experiments and simulations together with previously reported biophysical and structural observations to develop a structure-based model that provides mechanistic insight into the two functions of cohesin: cohesion and loop extrusion. This intriguing and informative manuscript will be of broad interest to those working in the fields of chromatin structure, chromosome biology and molecular machines. While the data and analysis support the authors' conclusions, the presentation of the work can be improved for clarity.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Author Response:

      Reviewer #1 (Public Review):

      Redmond et al. use single-cell and single-nucleus RNA-sequencing to reveal the molecular heterogeneity that underlies regional differences in neural stem cells in the adult mouse V-SVZ. The authors generated two datasets: one which was whole cell RNA-seq of whole V-SVZ and one which consisted of nuclear RNA-seq of V-SVZ microdissected into anterior-posterior and dorsal-ventral quadrants. The authors first identified distinct subtypes of B cells and showed that these B cell subtypes correspond to dorsal and ventral identities. Then, they identified distinct subtypes of A cells and classified them into dorsal and ventral identities. Finally, the authors identified a handful of genes that they conclude constitute a conserved molecular signature for dorsal or ventral lineages. The text of the manuscript is well written and clear, and the figures are organized and polished. The datasets generated in this manuscript will be a great resource for the field of adult neurogenesis. However, the arguments and supporting data used to assign dorsal/ventral identities to B cells and A cells could be strengthened, and more rigorous data analysis could result in new biological insights into stem and progenitor cell heterogeneity in the V-SVZ.

      We thank the Reviewer for their feedback on our manuscript. As suggested by Reviewer #1, we are performing additional analyses in the following areas:

      1) Performing additional analyses to further strengthen the dorsal/ventral scRNA-Seq B cell marker analysis and its relationship to our sNucRNA-Seq B data.

      2) Performing additional analyses to identify potential novel biological insights into stem & progenitor cell heterogeneity and text edits to discuss how differentially-expressed sets of genes among B cells and A cells are related to biological processes and/or signaling pathways.

      Reviewer #2 (Public Review):

      The paper is well written, and the data are well analyzed and presented. My concerns centre on terminology and alternative explanations of some of the data, which the authors might deal with in the introduction or discussion.

      We thank Reviewer #2 for their positive reception of our manuscript and the data, and for the constructive suggestions, which we have addressed by changes to the manuscript and in our responses below:

      1) I am slightly confused about some of the data shown in Figure 1. If B cells are defined as GFAP expressing cells, then why do only 25% of the B cells in the plot in Figure 1C express GFAP? I may be missing something here, as other readers may as well. Similarly in the same panel, only 25% of astrocytes seem to be expressing GFAP or GFP driven by a GFAP promotor.

      Importantly, among all cells captured in our scRNA-Seq, only B cells (51.86%), a subpopulation of parenchymal astrocytes (25%) and a small subpopulation of ependymal cells (E cells) had GFAP expression. This is consistent with immunocytochemical staining (Ponti et al. 2013) and other studies of scRNA-Seq expression (Xie et al. 2020). Similarly, Gfp (under the control of hGFAP promoter) is not expected to be expressed in all B cells (here 31.08% of B cells are Gfp+).

      Note that previous work has shown that B cells express different levels of GFAP protein, and some B1 cells were negative (Ponti et al. 2013). This supports the notion that this intermediate filament is a good marker of the V-SVZ primary progenitors, but also present in a subpopulation of parenchymal astrocytes and ependymal cells. However, a negative signal for GFAP does not imply that a cell is not a B cell. This highlights the importance of our clustering analysis revealing additional genes associated with B cells. Our analysis suggests that a combination of Gfap, Thrombospondin 4, Slc1a3 (GLAST) and S100a6 provide a better marker combination to identify B cells.

      The reason for the variability among B cells in the expression of GFAP remains unknown. It could be associated with the normal regulation of intermediate filaments as B cells transit the cell cycle or different stages of their activation or quiescence. It could also be linked to technical aspects of scRNA-Seq analysis: e.g gene dropout; detection limits; sequencing saturation. Since on our dot plot the actual proportion is only graphically shown, to clarify this issue in the text we have added the specific percentages and the following sentences:

      “A fraction of both populations expressed GFAP: 51.85% of B cells (clusters 5,13,14 & 22), 24.37% of parenchymal astrocytes (clusters 21, 26 & 29). This is consistent with previous reports (Chai et al. 2017; Xie et al. 2020; Ponti et al. 2013). Note that across all cells captured in our scRNAseq analysis, only B cells, parenchymal astrocytes or ependymal cells expressed GFAP. Among these three cell types, B cells had the highest average expression of GFAP (4.41 for B cells, 1.00 for astrocytes, 0.29767 for Ependymal cells, values in Pearson residuals). Other markers, like S100a6 (Kjell et al. 2020) (88.9% of B cells; 54% of parenchymal astrocytes and 80% of ependymal cells) and Thbs4 (Zywitza et al. 2018) (45% of B cells; 28.77% in parenchymal astrocytes, 2.88 % in ependymal cells) are also expressed preferentially in B cells and parenchymal astrocytes, but they alone do not distinguish these two cell populations.”

      2) The authors term the germinal zone of the adult mouse brain - the ventricular-subventricular zone. They should discuss the evidence that the adult germinal zone is made up of cells from both the ventricular zone and the sub ventricular zone in the late embryo, where those zones are described clearly on the basis of morphology. Many of the early embryonic neural stem cells are present in the ventricular zone before the sub ventricular zone has developed and continue to be present into the adult. If there is not clear mouse evidence that the progeny of embryonic sub ventricular cells are present in the adult germinal zone independent of embryonic ventricular zone progeny, then the authors might consider calling the zone - the adult ventricular zone, or alternatively terming the neurogenic area around the lateral ventricle the adult germinal zone or by a more straightforward descriptive term - the adult subependymal zone or the adult periventricular zone. Also, I think the first word in line 6 on page 3 should be neural rather than neuronal.

      We agree that the terminology in the field is confusing and multiple names have been used to describe the same region. In order to clarify that we are referring to the same adult periventricular germinal region, we have added a short sentence in the introduction to indicate that the V-SVZ is also referred by other authors as the SVZ, the subependyma or subependymal zone: We have added in the text: “This neurogenic region has also been referred to as the SVZ or the subependymal zone (Kazanis et al. 2017; Morshead et al. 1994)”.

      This reviewer argues that the adult V-SVZ should only be called V-SVZ if a lineage relationship could be established with the embryonic SVZ. To our knowledge there is no need to link the adult SVZ to the embryo, as this structure, like the embryonic SVZ, anatomically sits beneath the VZ (the area next to the ventricle). However, a lineage relationship does exist between the adult V-SVZ and the embryonic VZ, established in previous studies showing that PreB1 cells around E15.5 became quiescent and give rise to adult B cells in the V-SVZ (Fuentealba et al., 2015; Furutachi et al., 2015). In addition, developmental studies show a continuum in the gradual transformation of the embryonic periventricular germinal layers, including the SVZ. Importantly, B1 cells are derived from VZ radial glia (RG), maintain RG markers and retain RG-like interkinetic behavior establishing that functionally and anatomically a VZ is retained in the adult (Merkle et al., 2004; Mirzadeh et al., 2008). Therefore the adult periventricular epithelium is not made of a pure layer of ependymal cells with progenitor cells underneath, as previously thought. Moreover, recent work indicates that just like in the embryo, the more basal adult SVZ progenitors (B2 cells) can be derived from adult VZ progenitors (B1 cells) (Obernier et al. 2018). This transformation of apical to basal cells begins to occur in embryonic stages further suggesting equivalences between the adult and the embryonic progenitor cells. For all the above reasons we prefer to use the term V-SVZ.

      In line 6, page 3, We have changed neuronal cell types to “neural cell types”, as suggested.

      3) The authors refer to their molecularly described B cells as stem cells. Certainly, their lab and others have shown that adult olfactory bulb neurons are the progeny of those B cells, however the classic definition of stem cells (in the blood or intestine for example) require demonstration that single stem cells can make all of the differentiated cells in that tissue. Is their evidence that a single adult B1 cell can make astrocytes, neurons and oligodendrocytes? Indeed, what percentage of the single adult B cells characterized here on the bases of RNA expression can be shown to be multipoint for both macroglial and neuron lineages in vivo or in vitro? Perhaps progenitor or precursor cells might be a better term for a B cells that appears to give rise to neurons primarily.

      This is also an issue of definitions. We modified the text to refer to the primary progenitors in the V-SVZ as adult neural stem cells, or progenitor cells “NSPCs”. We agree that this needs to be clarified and in the introduction we modified one paragraph to indicate:

      “From the initial interpretation that adult NSPCs are multipotent and able to generate a wide range of neural cell types (Reynolds and Weiss 1992; van der Kooy and Weiss 2000; Morshead et al. 1994), more recent work suggest that the adult NSPCs in vivo are heterogeneous and specialized, depending on their location, for the generation of specific types of neurons, and possibly glia (Merkle et al. 2014; Fiorelli et al. 2015; Chaker, Codega, and Doetsch 2016; Merkle, Mirzadeh, and Alvarez-Buylla 2007; Tsai et al. 2012; Delgado et al. 2020).”

      Under normal in vivo conditions, a primitive state for NSCs capable of generating all neuronal and glial cell types of the CNS may only exist at very early stages of development and even their regional specification seems to occur very early (as early as E10.5; Fuentealba et al. 2015). Note that recent work in the hematopoietic system suggests that stem cells there also become restricted embryonically (Carrelha et al., 2018) and in adults their potential to generate lymphoid or myeloid lineages changes dramatically with age, yet at all these ages they are referred as HSCs. We are well aware of the work from the van der Kooy lab, suggesting the existence in the V-SVZ of rare “primitive” Oct4+/GFAP- cells that may be pluripotent and earlier in the lineage from B cells (Reeve et al., 2017). However, as indicated above lineage tracing from the embryo indicates that adult NSPC are specified in the embryo and are already in place and regionally specified between E11.5 and E15. We have investigated whether we could detect Oct4+/Gfap- cells in our datasets. However, we did not detect Oct4 expression in B cells or other cell types. We now indicate in the discussion:

      “It has been suggested that in the adult V-SVZ a more primitive population of Oct4+/GFAP- NSCs may be present and that these cells may be earlier in the lineage from the “definitive” GFAP+ B cells (Reeve et al. 2017). However, regionally specified NSPCs can be lineage traced to the embryo (Fuentealba et al. 2015; Furutachi et al. 2015), and we could not detect a population of Oct4+ cells in our datasets. We, however, cannot exclude that rare primitive OCT4+ NSPCs were not captured in our analysis for technical reasons.” ……. “This underscores the early embryonic regional specification of adult V-SVZ NSPCs and how these primary progenitors maintain a memory of their regions of origin.”

      4) This may be more than a semantic issue, as the rare clonal neurophere forming neural stem cells that do make all three neural cell types in vitro, and also maintain their AP and DV positional identity through clonal passaging in vitro (Hitoshi et al, 2002). However, Emx1 expressing cortical neural stem cells can be lineage traced as they migrate from the embryonic cortical germinal zone to the striata germinal zone in the perinatal period (Willaime-Morawek et al, 2006). Surprisingly, in their new striatal home the Emx1 lineage cortical neural stem cells will turn down Emx1 expression and turn up Dlx2 striatal germinal zone expression. The switch in positional identities of clonal neural stem cells can be seen also in vitro when the stem cells are co-cultured with an excess of cells from a different region and then regrown as clonal neural stem cells. This may suggested that Emx1 expressing neural stem cells (the clonal neurosphere forming cells), may switch their positional identities in vivo as they migrate into the striatal germinal zone, but the downstream neuron producing precursor B cells studied in this paper may maintain their Emx1 expression into the adult germinal zone. This raises an interesting issue concerning which cells in the neural stem cell lineage can be regionally re-specified.

      The interesting question about plasticity and respecification is not addressed by our current manuscript that focuses on the gene expression profile of unmanipulated cells from adult samples. However, regional re-specification is controversial. While work from van der Kooy lab suggests that striatal Emx1+ NSPCs originate in the pallium and migrate into the striatum in the perinatal brain (Willaime-Morawek et la., 2006), other studies suggest that rare Emx1 cells are already present in the developing LGE from embryonic stages as early as E12.5 (Gorski et al. 2002). In addition, we have labeled neonatal radial glial cells in the pallium, when this migration has been suggested to occur, and do not see migration of cells ventrally into the striatal wall. We have also transplanted dorsal NSPCs into ventral locations -- and vice versa -- and do not observe evidence of regional re-specification (Merkle, Mirzadeh, and Alvarez-Buylla 2007; Delgado et al. 2020).

      5) The authors nicely show dorsal versus ventral germinal zone lineages are marked by some of the same positional genes from B cells to A cells, suggesting complete dorsal versus ventral neurogenic lineages giving rise to different types of olfactory bulb neurons. Indeed, they nicely test this idea with dissection of the dorsal versus ventral germinal zones, followed by nuclear RNA sequencing. However, they don't discuss the broader issues concerning the embryological origins of the dorsal versus ventral germinal zones. Emx1 is one of the genes the authors use to mark dorsal lineages. The authors reference papers (Young et al, 2007; Willaime-Morawek et al, 2006;2008) that use Emx1 lineage tracing to show that certain classes of olfactory bulb neurons originate from embryonic cortical neural stem cells that migrate perinatally from the cortical germinal zone into the dorsal subcortical germinal zone. Could cortical versus subcortical embryonic origins of the dorsal versus ventral adult germinal zone explain the origin of different sets of adult olfactory bulb neurons? Further, the authors report that one of the GO terms for their dorsal lineages in cortical regionalization.

      This is a very interesting question that unfortunately we cannot answer. The dorsal domain includes both pallial and subpallial components, but the specific origin of B cells in this dorsal domain and the contribution of the pallium and subpallium remains unresolved.

      We went back to our data to try to find evidence of pallial vs. subpallial components in the dorsal B clusters (5 & 22). Indeed, there are some hints that cluster 22 may be more pallial and 5 more dorsal subpallial. However, when we try to confirm differential distribution of markers associated with these two dorsal subdomains anatomically, it is not possible to determine segregation, likely due to the intermixing of cells as the wedge is formed. We also looked for Dbx1, a relatively specific marker of the border region between pallium and subpallium that has been termed ventral pallium, but unfortunately our scRNA-Seq dataset did not capture this marker. Further, targeted lineage tracing of this region is required to determine the subdivisions of the dorsal V-SVZ. We have added as requested a short discussion on this issue:

      “The dorsal V-SVZ domain is likely further subdivided into multiple subdomains. In the current analysis we pooled together clusters B(5) and B(22) as dorsal. However, largely pallial marker Emx1 and dorsal lateral ganglionic eminence marker Gsx2 were differentially enriched in clusters B(22) and B(5), respectively, suggesting that these two clusters may also represent different sets of regionally specified B cells with distinct embryonic origins. These regions become blurred by cells intermixing in the formation of the wedge region in the postnatal V-SVZ making it difficult to confirm their origin based on expression patterns. In addition to pallial and dorsal subpallial markers, this wedge region likely also includes what has been termed the ventral pallium (Puelles et al. 2016), which is characterized in the embryo by the expression of Dbx1. Unfortunately, our scRNA-Seq analysis did not detect this marker. Further lineage tracing experiments will help determine the precise embryonic origin and nature of the dorsal V-SVZ, including the wedge region.”

      6) The percentages of dividing cells based on gene expression is given for some clusters of cells but not others. It might be useful to have a chart showing the percentages of cells in cycle (ki67 expression) for each cluster. This might be especially useful in characterizing some fo the differences between various subclusters of B, A and C cells. On page 9 it is suggested that the heterogeneity amongst C cell clusters was driven by cell cycle genes. However, it is possible to remove the cell cycle genes from the data analysis to see if this then produces clearer dorsal versus ventral positional identities. This may be an important issue as the dorsal versus ventral positional identity genes appear to be expressed more in less dividing A and B cells, than in the more dividing C cells. This leads to a potentially alternative conclusion - that dorsal/ventral regional identity genes are primarily expressed in the non-dividing post mitotic cells in their resident dorsal or ventral region, and not in precursor cells in the lineage.This could be easiy tested by removing the cell cycle genes from the analysis of highly dividing clusters to see if they then break down into doral versus ventral clusters.

      We now provide a table indicating the fraction of proliferating cells (defined as in S phase or G2-M phase) for all scRNA-Seq clusters.

      Concerning whether dorsal and ventral identities are maintained during the period of proliferation we have analyzed our data looking at dorsal and ventral signature levels over pseudotime (Figure 6-Supplement 1F). Here we do not observe a reduction in either dorsal or ventral score at the proliferative cell stages (pseudotime ~0.75, Figure 2L). This is in contrast to gene signatures that show clear up- or down-regulation over pseudotime, such as Gfap, Egfr & Dcx (Figure 2M). To understand how cell clustering is affected in the absence of proliferative gene influence, and whether clearer dorsal/ventral signatures are observed in proliferating cells, we are performing additional analyses using our scRNA-Seq dataset that is clustered after cell-cycle gene regression.

      References Cited:

      Chaker, Zayna, Paolo Codega, and Fiona Doetsch. 2016. “A Mosaic World: Puzzles Revealed by Adult Neural Stem Cell Heterogeneity.” Wiley Interdisciplinary Reviews. Developmental Biology 5 (6): 640–58.

      Delgado, Ryan N., Benjamin Mansky, Sajad Hamid Ahanger, Changqing Lu, Rebecca E. Andersen, Yali Dou, Arturo Alvarez-Buylla, and Daniel A. Lim. 2020. “Maintenance of Neural Stem Cell Positional Identity by.” Science 368 (6486): 48–53.

      Fiorelli, Roberto, Kasum Azim, Bruno Fischer, and Olivier Raineteau. 2015. “Adding a Spatial Dimension to Postnatal Ventricular-Subventricular Zone Neurogenesis.” Development 142 (12): 2109–20.

      Fuentealba, Luis C., Santiago B. Rompani, Jose I. Parraguez, Kirsten Obernier, Ricardo Romero, Constance L. Cepko, and Arturo Alvarez-Buylla. 2015. “Embryonic Origin of Postnatal Neural Stem Cells.” Cell 161 (7): 1644–55.

      Furutachi, Shohei, Hiroaki Miya, Tomoyuki Watanabe, Hiroki Kawai, Norihiko Yamasaki, Yujin Harada, Itaru Imayoshi, et al. 2015. “Slowly Dividing Neural Progenitors Are an Embryonic Origin of Adult Neural Stem Cells.” Nature Neuroscience 18 (5): 657–65.

      Gorski, Jessica A., Tiffany Talley, Mengsheng Qiu, Luis Puelles, John L. R. Rubenstein, and Kevin R. Jones. 2002. “Cortical Excitatory Neurons and Glia, but Not GABAergic Neurons, Are Produced in the Emx1-Expressing Lineage.” The Journal of Neuroscience: The Official Journal of the Society for Neuroscience 22 (15): 6309–14.

      Kazanis, Ilias, Kimberley A. Evans, Evangelia Andreopoulou, Christina Dimitriou, Christos Koutsakis, Ragnhildur Thora Karadottir, and Robin J. M. Franklin. 2017. “Subependymal Zone-Derived Oligodendroblasts Respond to Focal Demyelination but Fail to Generate Myelin in Young and Aged Mice.” Stem Cell Reports 8 (3): 685–700.

      Kooy, D. van der, and S. Weiss. 2000. “Why Stem Cells?” Science 287 (5457): 1439–41.

      Merkle, Florian T., Luis C. Fuentealba, Timothy A. Sanders, Lorenza Magno, Nicoletta Kessaris, and Arturo Alvarez-Buylla. 2014. “Adult Neural Stem Cells in Distinct Microdomains Generate Previously Unknown Interneuron Types.” Nature Neuroscience 17 (2): 207–14.

      Merkle, Florian T., Zaman Mirzadeh, and Arturo Alvarez-Buylla. 2007. “Mosaic Organization of Neural Stem Cells in the Adult Brain.” Science 317 (5836): 381–84.

      Morshead, C. M., B. A. Reynolds, C. G. Craig, M. W. McBurney, W. A. Staines, D. Morassutti, S. Weiss, and D. van der Kooy. 1994. “Neural Stem Cells in the Adult Mammalian Forebrain: A Relatively Quiescent Subpopulation of Subependymal Cells.” Neuron 13 (5): 1071–82.

      Ponti, Giovanna, Kirsten Obernier, Cristina Guinto, Lingu Jose, Luca Bonfanti, and Arturo Alvarez-Buylla. 2013. “Cell Cycle and Lineage Progression of Neural Progenitors in the Ventricular-Subventricular Zones of Adult Mice.” Proceedings of the National Academy of Sciences of the United States of America 110 (11): E1045–54.

      Puelles, Luis, Loreta Medina, Ugo Borello, Isabel Legaz, Anne Teissier, Alessandra Pierani, and John L. R. Rubenstein. 2016. “Radial Derivatives of the Mouse Ventral Pallium Traced with Dbx1-LacZ Reporters.” Journal of Chemical Neuroanatomy 75 (Pt A): 2–19.

      Reeve, Rachel L., Samantha Z. Yammine, Cindi M. Morshead, and Derek van der Kooy. 2017. “Quiescent Oct4 Neural Stem Cells (NSCs) Repopulate Ablated Glial Fibrillary Acidic Protein NSCs in the Adult Mouse Brain.” Stem Cells 35 (9): 2071–82.

      Reynolds, B. A., and S. Weiss. 1992. “Generation of Neurons and Astrocytes from Isolated Cells of the Adult Mammalian Central Nervous System.” Science 255 (5052): 1707–10.

      Tsai, Hui-Hsin, Huiliang Li, Luis C. Fuentealba, Anna V. Molofsky, Raquel Taveira-Marques, Helin Zhuang, April Tenney, et al. 2012. “Regional Astrocyte Allocation Regulates CNS Synaptogenesis and Repair.” Science 337 (6092): 358–62.

      Xie, Xuanhua P., Dan R. Laks, Daochun Sun, Asaf Poran, Ashley M. Laughney, Zilai Wang, Jessica Sam, et al. 2020. “High Resolution Mouse Subventricular Zone Stem Cell Niche Transcriptome Reveals Features of Lineage, Anatomy, and Aging.”Cold Spring Harbor Laboratory. https://doi.org/10.1101/2020.07.27.223602.

    2. Reviewer #2 (Public Review):

      The paper is well written, and the data are well analyzed and presented. My concerns centre on terminology and alternative explanations of some of the data, which the authors might deal with in the introduction or discussion.

      1) I am slightly confused about some of the data shown in Figure 1. If B cells are defined as GFAP expressing cells, then why do only 25% of the B cells in the plot in Figure 1C express GFAP? I may be missing something here, as other readers may as well. Similarly in the same panel, only 25% of astrocytes seem to be expressing GFAP or GFP driven by a GFAP promotor.

      2) The authors term the germinal zone of the adult mouse brain - the ventricular-subventricular zone. They should discuss the evidence that the adult germinal zone is made up of cells from both the ventricular zone and the sub ventricular zone in the late embryo, where those zones are described clearly on the basis of morphology. Many of the early embryonic neural stem cells are present in the ventricular zone before the sub ventricular zone has developed and continue to be present into the adult. If there is not clear mouse evidence that the progeny of embryonic sub ventricular cells are present in the adult germinal zone independent of embryonic ventricular zone progeny, then the authors might consider calling the zone - the adult ventricular zone, or alternatively terming the neurogenic area around the lateral ventricle the adult germinal zone or by a more straightforward descriptive term - the adult subependymal zone or the adult periventricular zone. Also, I think the first word in line 6 on page 3 should be neural rather than neuronal.

      3) The authors refer to their molecularly described B cells as stem cells. Certainly, their lab and others have shown that adult olfactory bulb neurons are the progeny of those B cells, however the classic definition of stem cells (in the blood or intestine for example) require demonstration that single stem cells can make all of the differentiated cells in that tissue. Is their evidence that a single adult B1 cell can make astrocytes, neurons and oligodendrocytes? Indeed, what percentage of the single adult B cells characterized here on the bases of RNA expression can be shown to be multipoint for both macroglial and neuron lineages in vivo or in vitro? Perhaps progenitor or precursor cells might be a better term for a B cells that appears to give rise to neurons primarily.

      4) This may be more than a semantic issue, as the rare clonal neurophere forming neural stem cells that do make all three neural cell types in vitro, and also maintain their AP and DV positional identity through clonal passaging in vitro (Hitoshi et al, 2002). However, Emx1 expressing cortical neural stem cells can be lineage traced as they migrate from the embryonic cortical germinal zone to the striata germinal zone in the perinatal period (Willaime-Morawek et al, 2006). Surprisingly, in their new striatal home the Emx1 lineage cortical neural stem cells will turn down Emx1 expression and turn up Dlx2 striatal germinal zone expression. The switch in positional identities of clonal neural stem cells can be seen also in vitro when the stem cells are co-cultured with an excess of cells from a different region and then regrown as clonal neural stem cells. This may suggested that Emx1 expressing neural stem cells (the clonal neurosphere forming cells), may switch their positional identities in vivo as they migrate into the striatal germinal zone, but the downstream neuron producing precursor B cells studied in this paper may maintain their Emx1 expression into the adult germinal zone. This raises an interesting issue concerning which cells in the neural stem cell lineage can be regionally re-specified.

      5) The authors nicely show dorsal versus ventral germinal zone lineages are marked by some of the same positional genes from B cells to A cells, suggesting complete dorsal versus ventral neurogenic lineages giving rise to different types of olfactory bulb neurons. Indeed, they nicely test this idea with dissection of the dorsal versus ventral germinal zones, followed by nuclear RNA sequencing. However, they don't discuss the broader issues concerning the embryological origins of the dorsal versus ventral germinal zones. Emx1 is one of the genes the authors use to mark dorsal lineages. The authors reference papers (Young et al, 2007; Willaime-Morawek et al, 2006;2008) that use Emx1 lineage tracing to show that certain classes of olfactory bulb neurons originate from embryonic cortical neural stem cells that migrate perinatally from the cortical germinal zone into the dorsal subcortical germinal zone. Could cortical versus subcortical embryonic origins of the dorsal versus ventral adult germinal zone explain the origin of different sets of adult olfactory bulb neurons? Further, the authors report that one of the GO terms for their dorsal lineages in cortical regionalization.

      6) The percentages of dividing cells based on gene expression is given for some clusters of cells but not others. It might be useful to have a chart showing the percentages of cells in cycle (ki67 expression) for each cluster. This might be especially useful in characterizing some fo the differences between various subclusters of B, A and C cells. On page 9 it is suggested that the heterogeneity amongst C cell clusters was driven by cell cycle genes. However, it is possible to remove the cell cycle genes from the data analysis to see if this then produces clearer dorsal versus ventral positional identities. This may be an important issue as the dorsal versus ventral positional identity genes appear to be expressed more in less dividing A and B cells, than in the more dividing C cells. This leads to a potentially alternative conclusion - that dorsal/ventral regional identity genes are primarily expressed in the non-dividing post mitotic cells in their resident dorsal or ventral region, and not in precursor cells in the lineage.This could be easiy tested by removing the cell cycle genes from the analysis of highly dividing clusters to see if they then break down into doral versus ventral clusters.

    3. Reviewer #1 (Public Review):

      Redmond et al. use single-cell and single-nucleus RNA-sequencing to reveal the molecular heterogeneity that underlies regional differences in neural stem cells in the adult mouse V-SVZ. The authors generated two datasets: one which was whole cell RNA-seq of whole V-SVZ and one which consisted of nuclear RNA-seq of V-SVZ microdissected into anterior-posterior and dorsal-ventral quadrants. The authors first identified distinct subtypes of B cells and showed that these B cell subtypes correspond to dorsal and ventral identities. Then, they identified distinct subtypes of A cells and classified them into dorsal and ventral identities. Finally, the authors identified a handful of genes that they conclude constitute a conserved molecular signature for dorsal or ventral lineages. The text of the manuscript is well written and clear, and the figures are organized and polished. The datasets generated in this manuscript will be a great resource for the field of adult neurogenesis. However, the arguments and supporting data used to assign dorsal/ventral identities to B cells and A cells could be strengthened, and more rigorous data analysis could result in new biological insights into stem and progenitor cell heterogeneity in the V-SVZ.

    4. Evaluation Summary:

      Redmond et al. use single-cell and single-nucleus RNA-sequencing to reveal the molecular heterogeneity that underlies regional differences in neural stem cells in the adult mouse. Prior work had separate subventricular stem cells as type A and B. By generating bulk and single cell transcriptome sequence data, the authors identified a distinct subtype of both A and B cells that differentiate into dorsal and ventral identities. They also identify a set of genes that constitute a conserved molecular signature of these cell types.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Author Response:

      Reviewer #1 (Public Review):

      The energy released upon zippering of SNARE complexes from the N-terminus to the membrane-proximal C-terminus is widely believed to provide the driving force for membrane fusion, and the cis-SNARE complexes resulting after fusion are disassembled by formation of a 20S complex with Sec17 and Sec18, followed by ATP-hydrolysis by Sec18. This paper now shows that membrane fusion still occurs when the hydrophobic interactions that drive C-terminal zippering of the yeast vacuolar SNARE complex is completely prevented by C-terminal truncation of two of the SNAREs and replacement of the hydrophobic residues at the C-terminus of the SNARE domain of a third SNARE with polar residues, and that such fusion requires Sec17, Sec18 and non-hydrolyzable ATP homologues, in addition to the HOPS tethering complex, which mediates SNARE complex assembly. The results also show that Sec17 plays a key role in fusion through hydrophobic residues in an N-terminal loop that are known to interact with membranes. These results suggest that the core membrane fusion machinery is formed by the SNAREs, Sec17 and Sec18 rather than by the SNAREs alone, and that fusion is driven by a combination of SNARE C-terminal zippering and perturbation of the lipid bilayers by the hydrophobic loops of Sec17. These conclusions are strongly supported by a variety of membrane fusion experiments. FRET assays to SNARE complex assembly also support the conclusions but are less convincing.

      Thank you. We feel that the FRET assays of HOPS-dependent, Sec17-driven zippering are important, as they indicate one of the 3 vital Sec17 functions: (1) promote SNARE zippering which, when it can occur, makes an important contribution to fusion. This is the first demonstration of Sec17-induced zippering in the context of HOPS. (2) promote fusion directly even when zippering contributes no energy. and (3) support Sec18 in disassembly of SNARE complexes.

      Reviewer #2 (Public Review):

      This manuscript shows that the Sec17/18 machine can do more than we might have expected, and places new constraints on models for how this works. As the field expects from the Wickner lab, the work is creative and beautifully executed. I do still have some reservations, however, about whether the manuscript ultimately forwards our mechanistic understanding enough to merit publication in eLife. Some of the outstanding mechanistic questions articulated by the authors include:

      1) Why is HOPS required for Sec17/18/ATPγS activity? The authors suggest that HOPS and Sec17 bind to one another, but the assay (Figure 4) is rather non-physiological and the result does not really answer the question.

      We have not established why HOPS can work with Sec17/Sec18 while other nonspecific tethers cannot, but future explorations of this question will be founded on a knowledge of which components bind directly to the others, including HOPS:Sec17. Each demonstration of binding between two purified proteins is of course non-physiological, yet contributes to our understanding. Still, we "hear you", and we've moved this to a supplemental figure.

      2) What is the mechanistic role of Sec18? An intricate inhibitor experiment (Figure 9) suggests that Sec18 acts later than Vps33. This is consistent with current thinking on the early role of SM proteins, but does not further delineate the mechanistic role of Sec18.

      Exactly. This is a fascinating question which is reinforced, but not answered, by our current work. We note that Sec17 is now seen to perform 3 functions (working with Sec18 for ATP-driven SNARE complex disassembly, driving completion of SNARE zippering, and supporting fusion per se) and Sec18 performs 2 functions (SNARE complex disassembly and supporting Sec17 for fusion per se).

      3) Does "entropic confinement" explain the role of Sec17? This very interesting question was not, so far as I could tell, directly addressed. My understanding is that the concept of entropic confinement comes from studies of chaperonins such as GroEL/ES, which entirely enclose their substrates in what Paul Sigler memorably described as "a temple for protein folding". Here, it's much less clear that Sec17 could sufficiently constrain the presumably-unfolded juxtamembrane regions of the truncated and/or mutant SNAREs to drive membrane fusion. Indeed, Schwartz et al. (2017) noted "open portals" between adjacent Sec17 molecules that would "allow SNARE residues spanning the partially-zipped helical bundle and the transmembrane anchors to pass cleanly between pairs of adjacent Sec17 subunits".

      We have removed the term "entropic confinement", and in deference to Schwartz et al. (which we cite) we refer to the Sec17 open cage and the folding environment it may create for SNARE complex assembly.

      4) What is the mechanistic role of the "hydrophobic loop" at the N-terminus of Sec17? Previous work from the Wickner lab (Song et al., 2017) concluded that its main function under normal circumstances was to promote Sec17 membrane association, but when zippering was incomplete it might act as a wedge to perturb the bilayers. These experiments made use of artificially membrane-anchored Sec17, either wild-type or the "FSMS" hydrophobic loop mutant. This approach was extended here (Figure 8) but did not, so far as I could tell, greatly advance our mechanistic understanding.

      Agreed. Each of your points 2-4 reinforce central questions which our lab, and others, will strive to answer: What does Sec18 do? How does Sec17 oligomerization around the SNAREs relate to those SNAREs? What is the role of the Sec17 apolar loop? We do find though that Sec17 and Sec18, however they act, are so important as contributors toward driving fusion that they can compensate for only partial zippering when tested to do so.

    2. Reviewer #3 (Public Review):

      In this work, the authors used in vitro binding and liposome fusion assays to study how Sec17 and Sec18 regulate SNARE-driven fusion. In previous studies, it was found that deletion of the C-terminal layers of the Qc SNARE involved in yeast vacuole fusion blocks fusion but the inhibition can be partially bypassed by addition of Sec17 and Sec18. This work extended the finding and showed that Sec17 and Sec18 can even restore fusion when two Q SNAREs are C-terminally truncated and the third chain bears point mutations. The authors conclude that HOPS and membrane anchored Rabs first promote the tethering of vacuole membranes. Subsequently, HOPS promotes membrane docking - the initial assembly of the SNAREs, likely through the SM protein in the HOPS complex. Then Sec17 and Sec18 kick in to activate the zippering of membrane-proximal regions of SNAREs. This function seems to require interactions of Sec17 with HOPS. The findings are unexpected and raise important questions.

    3. Reviewer #2 (Public Review):

      This manuscript shows that the Sec17/18 machine can do more than we might have expected, and places new constraints on models for how this works. As the field expects from the Wickner lab, the work is creative and beautifully executed. I do still have some reservations, however, about whether the manuscript ultimately forwards our mechanistic understanding enough to merit publication in eLife. Some of the outstanding mechanistic questions articulated by the authors include:

      1) Why is HOPS required for Sec17/18/ATPγS activity? The authors suggest that HOPS and Sec17 bind to one another, but the assay (Figure 4) is rather non-physiological and the result does not really answer the question.

      2) What is the mechanistic role of Sec18? An intricate inhibitor experiment (Figure 9) suggests that Sec18 acts later than Vps33. This is consistent with current thinking on the early role of SM proteins, but does not further delineate the mechanistic role of Sec18.

      3) Does "entropic confinement" explain the role of Sec17? This very interesting question was not, so far as I could tell, directly addressed. My understanding is that the concept of entropic confinement comes from studies of chaperonins such as GroEL/ES, which entirely enclose their substrates in what Paul Sigler memorably described as "a temple for protein folding". Here, it's much less clear that Sec17 could sufficiently constrain the presumably-unfolded juxtamembrane regions of the truncated and/or mutant SNAREs to drive membrane fusion. Indeed, Schwartz et al. (2017) noted "open portals" between adjacent Sec17 molecules that would "allow SNARE residues spanning the partially-zipped helical bundle and the transmembrane anchors to pass cleanly between pairs of adjacent Sec17 subunits".

      4) What is the mechanistic role of the "hydrophobic loop" at the N-terminus of Sec17? Previous work from the Wickner lab (Song et al., 2017) concluded that its main function under normal circumstances was to promote Sec17 membrane association, but when zippering was incomplete it might act as a wedge to perturb the bilayers. These experiments made use of artificially membrane-anchored Sec17, either wild-type or the "FSMS" hydrophobic loop mutant. This approach was extended here (Figure 8) but did not, so far as I could tell, greatly advance our mechanistic understanding.

    4. Reviewer #1 (Public Review):

      The energy released upon zippering of SNARE complexes from the N-terminus to the membrane-proximal C-terminus is widely believed to provide the driving force for membrane fusion, and the cis-SNARE complexes resulting after fusion are disassembled by formation of a 20S complex with Sec17 and Sec18, followed by ATP-hydrolysis by Sec18. This paper now shows that membrane fusion still occurs when the hydrophobic interactions that drive C-terminal zippering of the yeast vacuolar SNARE complex is completely prevented by C-terminal truncation of two of the SNAREs and replacement of the hydrophobic residues at the C-terminus of the SNARE domain of a third SNARE with polar residues, and that such fusion requires Sec17, Sec18 and non-hydrolyzable ATP homologues, in addition to the HOPS tethering complex, which mediates SNARE complex assembly. The results also show that Sec17 plays a key role in fusion through hydrophobic residues in an N-terminal loop that are known to interact with membranes. These results suggest that the core membrane fusion machinery is formed by the SNAREs, Sec17 and Sec18 rather than by the SNAREs alone, and that fusion is driven by a combination of SNARE C-terminal zippering and perturbation of the lipid bilayers by the hydrophobic loops of Sec17. These conclusions are strongly supported by a variety of membrane fusion experiments. FRET assays to SNARE complex assembly also support the conclusions but are less convincing.

    5. Evaluation Summary:

      This is a very important paper that challenges the generally accepted dogma that full zippering of SNARE complexes is essential for intracellular membrane fusion. Previous work had already shown that C-terminal truncation of one SNARE arrested liposome fusion mediated by the yeast vacuolar SNARE complex and that Sec17/Sec18 could rescue fusion, but it was argued that such rescue could arise because Sec17/Sec18 restored C-terminal zippering. This paper now shows that Sec17/Sec18 rescue fusion even when three SNAREs are crippled -by truncation or mutation- to definitively prevent zippering, thus showing that Sec17/18 have a direct, positive role in membrane fusion.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #2 (Public Review):

      In this manuscript Ma et al., sought to investigate the breadth of genetic mechanisms available across various lineages of clinical isolates of Klebsiella pneumoniae, with a specific focus on carbapenem resistance evolution. The authors systematically evaluated how different carbapenems and genetic backgrounds affect the rate of evolution by measuring mutation frequencies. The authors found three major observations: First, that a higher mutational frequency is dependent on genetic background and high-level transposon activity affecting porins associated to carbapenem resistance. Importantly transposon activity was not only higher than SNP acquisition rates in distinct backgrounds, but was also reversible, thus emphasizing that resistance evolution via this mechanism might impart less of a cost than by the accumulation of mutations in other genetic backgrounds. Second, that CRISPR-cas systems have the potential to restrict the horizontal acquisition of resistance elements. Importantly, determining the presence or absence of such systems alone is not enough to determine wether a strain is "resistant" to certain foreign elements, but specific sequences within the different spacers can be more informative of the exact range of plasmids or genetic elements to which the system is restrictive. Third, pre-selection with ertapenem increases the likelihood of resistance evolution against other carbapenems both via de novo mutation and HGT.

      Altogether, these results emphasize the importance of additional factors, other than MIC values, such as genetic background, plasmid/transposon activity, and drug identity and choice in determining the rate at which resistance can evolve in K. pneumoniae. I consider that the data generally supports the authors conclusions and provides relevant observations to the field. I do not have any major concern and think the authors have done a very complete and systematic evaluation of the data necessary to answer their questions.

      My only minor concern is regarding the authors emphasis in their introduction and discussion on how these kind of data is relevant for clinical decision making. It remains unclear to me exactly how. While I completely agree that genomic information and drug choice play a major role in the evolution of antibiotic resistance, it is unclear to me how to efficiently and promptly translate all of this information at the bedside. Genome sequencing, however economical it has become in the recent years, is still not affordable to be implemented at the scales needed for diagnosis at the clinic. Perhaps the authors could expand on how they envision this could be implemented?

    2. Reviewer #1 (Public Review):

      In this manuscript, Ma, Hung and colleagues rewind the tape to explore the genetic landscape that precedes carbapenem resistance of Klebsiella pneumoniae strains. The importance of this work is underscored by the paucity of new drugs to treat CPO (carbapenemase producing organisms). 'Given the need for 35 greater antibiotic stewardship, these findings argue that in addition to considering the current 36 efficacy of an antibiotic for a clinical isolate in antibiotic selection, considerations of future 37 efficacy are also important.' And so I would say the major weakness of the paper is the aspirational nature of how this work could be used by clinicians in antibiotic selection or treatment of the patient.

      The strains selected for these experiments and the evolutionary in vitro models are both well considered. One idea that has stuck with me from the figures of a review article by Kishony (https://pubmed.ncbi.nlm.nih.gov/23419278/, figure 4) is the concept of constraining the evolutionary pathways or fitness landscape for antibiotic resistance. Are there any peaks that a microbial strain reaches that optimize resistance to one AbX but basically leave it inherently unable to evolve resistance to another AbX? This could have application for dual drug therapy or pulsed therapy. When you sequence the isolates that have increased their MIC do you find 'unrelated' mutations in genes that would control protein synthesis or other functions that might be compensatory mutations. Developing a clearer understanding of the rewiring of the bacterium's basic processes might also elucidate both integrated functions and potential weaknesses. You mention mutations in wzc, ompA, resA, bamD.

      Point of discussion. Classic ST258 carries blaKPC on pKpQIL plasmid. Your ST258 strain (UCI38) carries blaSHV-12 on pESBL. Am I to assume that pESBL is in lieu of pKpQIL? Transformation of CPO have many variables and in vitro data does not always mirror what is observed in vivo. So the findings of Fig 2f might need to be considered under different laboratory conditions (substrate, temperature) [https://pubmed.ncbi.nlm.nih.gov/27270289/].

    3. Evaluation Summary:

      This manuscript is of interest to several fields in biology and medicine including evolutionary genomics and antibiotic stewardship. Ma et al. sought to investigate the breadth of genetic mechanisms for evolution of carbapenem resistance across various lineages of the bacterial pathogen Klebsiella pneumoniae. The authors performed systematic and thorough bioinformatic and genetic analyses to identify how transposon activity and CRISPR-Cas systems facilitate the evolution of antibiotic resistance and restriction of horizontally acquired genetic elements, respectively. The study's results emphasize the importance of additional factors, other than MIC values, such as genetic background, plasmid/transposon activity, and drug identity and choice in determining the rate at which resistance can evolve in K. pneumoniae.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Reviewer #2 (Public Review):

      In this manuscript, the authors investigate the role of Relish in the Drosophila lymph gland (LG). They establish that relish is expressed in PSC cells and that reducing its expression in these cells (by expressing relish RNAi with a PSC-gal4 driver) leads to an enlarged PSC, increased plasmatocyte differentiation, no effect on crystal cell numbers, and fewer progenitors in the medullary zone (MZ). In the PSC, Relish controls Wingless levels that in turn control PSC cell proliferation and thus PSC size. This study also establishes that the knock down of relish in the PSC leads to increased levels of several actin binding proteins, reduced filopodia formation in PSC cells and a decrease in Hh (HhExt) release from the PSC. In addition, relish knock-down in the PSC leads to the activation of the JNK pathway in the PSC. Epistasis experiments establish that JNK acts downstream of Relish to control filopodia formation and HhExt. Under normal conditions, Relish levels in the PSC are under the control of ecdysone. Finally, in response to an E.coli infection, a decrease in Relish levels in the PSC is observed together with increased plasmatocyte differentiation.

      This is an important study describing a yet unknown regulation of Drosophila LG hematopoiesis.

    2. Reviewer #1 (Public Review):

      Mandal and colleagues identified novel functions of Relish in the hematopoietic niche development and its coordinative role in innate immunity. The authors found that Relish is expressed in the PSC, which is essential for various developmental functions, including the maintenance of hematopoietic progenitors, the number of PSC cells, expression of Wg, and the PSC actin cytoskeletal structure. Furthermore, Relish acts as an inhibitor of JNK signaling and functions downstream of the ecdysone pathway in the PSC. The authors moved on to find the developmental and physiological relevance of this phenomenon and discovered that Relish is downregulated upon bacterial infections to accommodate immune responses. These findings show that Relish plays a critical role in hematopoiesis as a downstream of hormonal control and in switching between the developmental and physiological mode of the PSC.

      Conclusions are well-supported by data, and experiments were carefully performed and analyzed. Given that most of the studies on Relish describe its function in innate immunity and that it is the first study showing critical roles of Relish in blood development, this study will draw broad attention and contribute to understanding insect hematopoiesis and immunity.

    3. Evaluation Summary:

      Mandal and colleagues identified novel functions of the Imd pathway transcription factor Relish in the hematopoietic niche development. The authors found that Relish is required for the maintenance of hematopoietic progenitors downstream of hormonal control. This is the first study showing critical roles of Relish in blood development, and therefore, this study will draw broad attention and contribute to understanding of insect hematopoiesis and immunity.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #3 (Public Review):

      This paper demonstrates the additional utility that can be extracted from short-read genome resources such as the genomes from the 1000 Genomes Project by leveraging variant discovery in long-read platforms. These genotyped variants can be used for eQTL studies, or to identify potential signatures of selection. Thus, low-coverage population-scale sequencing datasets such as the 1000 Genomes data can still be of use when coupled with other datasets.

      One of the challenges I have with this manuscript however is clearly understanding the novel aspects of the reported results in the context of previous work in this field. Initially, it is unclear how many of the genotyped variants are already in the 1000 Gnomes dataset, this should be clearly reported. Comparisons of LD to nearby SNPs does not take into account that the SV discovery in the 1000-genomes project was done separately from the SNP calling. Thus, while it is suggested as presented that most of these variants were previously intractable, this is insufficiently explored. Additionally, discussion of low LD with SVs is well documented in 1KG and elsewhere. Subsequently, the eQTL analyses are "broadly consistent" with previously reported eQTL analyses from both the 1000 genomes project and GTEx, but no direct comparison is performed. If the overall goal is to point out that using additional datasets can identify new variants that can be genotyped, it is important to perform comparisons to other population-scale datasets such as HGDP and SGDP (Almarri et al Cell, Hseih et al Science, etc). In these cases, higher coverage sequencing allowed discovery of variants which could then be genotyped, similar to this paper's assertion that long-read sequencing provided a new discovery set for subsequent genotyping. Indeed, the two highly stratified variants selected for follow up are reported in gnomAD. The paper mostly focusses on the identification of highly stratified loci. Again, comparison to previously reported highly stratified loci (1KG, Sudmant et al 2015, and Almarri 2020, Hseih et al) is necessary here.

      Furthermore, while the analyses of the IGH hapotype are clearly presented and interesting, as noted in the manuscript, these have already been identified. The authors mention that this locus was already identified but suggest it was "not further examined," due to "stringent filtering" however this locus was reported as one of 11 "high frequency introgressed regions" thus this description seems to mischaracterize Browning et al's recognition of the importance of this locus. The strongest part of the manuscript is the ABC modelling of the IGH haplotype elucidating the putatively extremely strong selective signatures at this locus. More focus on these results and the importance of following up and fully understanding such loci would benefit the manuscript. Broadly, this paper is well written and clearly presented however would be very much strengthened by placing it more broadly in the context of previous work and focusing more on the novel modelling analyses of specific loci that are performed.

    2. Reviewer #2 (Public Review):

      The technical challenges of identifying and quantifying the frequency of structural variants (SV) on a population scale has been a major limitation to the study of recent human adaptation. This manuscript applies a recent graph-based genotyping method that leverages a library of SVs identified by long-read sequencing to identify SVs in large short-read based cohorts. This is a sensible and powerful approach that highlights several examples of likely adaptive SV evolution in different human populations. The key findings and examples are well supported by the data and methods used. However, the manuscript would benefit from: 1) testing more hypotheses rather than listing examples and 2) more framing of how the results and methods expand on several recent studies of SVs across populations. In addition to providing novel examples of adaptive SV evolution, I anticipate this analysis can serve as a template for future analyses that merge long-read and short-read datasets.

    3. Reviewer #1 (Public Review):

      Yan et al. take a comprehensive look at structural variants in the 1000 Genomes Project high-coverage dataset, using recent developments that can link short- and long-read data. Combined with genomic simulations, they identify and characterize the timing and origin of a likely selected region in Southeast Asian populations. The combination of multiple data types adds depth to the interpretation.

      The study is timely, combing recently released data and methods, and had interesting biological implications. Tree main areas would help interpretation and robustness of the paper:

      1) Further context and interpretation of the original SV set found is needed, for example comparisons to previous work to identify clearer "positive controls" or sanity checks on the method, and to understand what the contribution of the method/dataset/paper is.

      2) The above is particularly important across ancestries/populations which differ in their LD levels. How does population-specific LD patterns impact the ability to detect these SV patterns? and therefore to make cross-population comparisons or infer differences in frequency that are central to the selection scan and the 220 highly differentiated SVs of interest. Perhaps this is in the original methods paper, but is central to this paper so should at least be explained or analyzed.

      3) The genomic simulations to infer the strength selection was a nice addition, a step beyond common empirically-driven work. It would help to know how to interpret the ABC model in the context of the later finding that the region was introgressed from Neanderthals--the model seems to not include this aspect.

    4. Evaluation Summary:

      The technical challenges of identifying and quantifying the frequency of structural variants (SV) on a population scale has been a major limitation to the study of recent human adaptation. This manuscript applies a recent graph-based genotyping method that leverages a library of SVs identified by long-read sequencing to identify SVs in large short-read based cohorts. This is a sensible and powerful approach that highlights several examples of likely adaptive SV evolution in different human populations. The key findings and examples are well supported by the data and methods used. However, the manuscript would benefit from further comparisons and context from previous studies, and deeper exploration of the biological significance. In addition to providing novel examples of adaptive SV evolution, this analysis may serve as a template for future analyses that merge long-read and short-read datasets.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #3 (Public Review):

      The article by Sprenger et al. uses the power of yeast genetics to generate mutants of the ESCRT-III subunits, and study their impact on the formation of a functional ESCRT-III complex. By using functional FP tags of subunits Snf7, Vps2 and Vps24, and of the CPS cargo, they essentially follow recruitment of subunits to the vacuolar membrane, formation of Class E compartments and sorting of CPS as readouts of the endosomal ESCRT-III function. They found that recruitment of Vps2, Vps24 and Snf7 is unaffected by deletions of other subunits (Did2, Ist1, Vps60), supporting the view that Vps2-Vps24 and Snf7 form an initial subcomplex.

      To decipher molecular interactions between Vps2-Vps24 and Snf7 subunits, they use point mutants to replace well-chosen hydrophobic residues in two subunits by cysteines, and cross-link them to probe the interactions of those residues in the functional case. They also change hydrophobic residue pairs into charged residue pairs to replace the hydrophobic interaction by an electrostatic interaction, and restore functionality (only when both mutants were used).

      Overall, it is an elegant study, with very clear and well executed experiments, and which give strong support to a so far hypothetical architecture of the Vps2-Vps24-Snf7 as a double-strand filament, one of which is Snf7 only, and the other is an alternative repeat of Vps2 and Vps24.

    2. Reviewer #2 (Public Review):

      ESCRT-III is a filament-forming machinery that is necessary for a variety of physiological and pathophysiological membrane remodelling events. These events are linked to an ability of an ESCRT-III filament to assemble and remodel cellular membranes. In recent years, it has become clear that whilst the ESCRT-III component Snf7 is likely the major component of ESCRT-III, individual filaments can form lateral interactions with alternate filaments, that remodelling the composition of ESCRT-III subunits within a filament likely allows its geometric changes and that it is unclear what role the Vps2/Vps24 subunits of ESCRT-III have alongside the major Snf7 filament. Building upon a previous publication in eLIFE, in which the authors used advanced microscopical approaches to quantitatively document the assembly kinetics of ESCRT-III upon endosomes (demonstrating transient co-assembly of Snf7, Vps2 and Vps24), Sprenger et al have now used biochemical and microscopical approaches to understand individual interactions within the ESCRT-III holo-filament.

      Protein-protein interactions are typically driven by two different modes that rely upon the physicochemical properties of the amino acids involved (namely electrostatics, or the shielding of hydrophobic residues by mutual interaction). Using published and modelled structural data, Sprenger et al., identify hydrophobic interactions governing longitudinal interaction of ESCRT-III monomers and electrostatic interactions that govern lateral interactions. They make elegant use of targeted mutagenesis to switch the interaction mode between individual monomers, and employ pairwise mutagenesis to rescue the disrupted interactions. They also employ chemical crosslinking to stabilise these transient interactions, and integrate this with an analysis of cargo sorting to the vacuole lumen, which is the archetypal function of ESCRT-III in yeast. In contrast to models proposing the Vps2/Vps24 unit as a 'cap' for a Snf7 filament, the authors propose that these subunits instead form a parallel filament that has important implications for our understanding of how Vps4 can access subunits within the ESCRT-III holo-filament.

      The strengths of this manuscript are the integration of molecular and biochemical data with clear functional readouts of vacuolar sorting and the use of knock-in techniques bearing functionally tagged versions of the ESCRT-III proteins to analyse phenotypes. I think some improvement could be made to the description of the author's selection of residues for mutagenesis and to the degree of quantification of the data throughout the manuscript. I also wonder if there are different interpretations of the cross-linking experiments that could be integrated into their discussion.

    3. Reviewer #1 (Public Review):

      In this work the authors address inter-subunit interactions leading to ESCRT-III function during MVB sorting in a yeast model system. ESCRTs mediate function in multiple biological processes, however the fundamental question of how ESCRT-III mediates membrane remodeling is not understood. As such this represents a topic of considerable interest despite significant technical limitations surrounding the issue. Random and rational mutagenic strategies, including compensatory mutations, are combined with protein-protein interaction studies and in vivo functional assays to identify residues within Vps24 and Vps2 mediating associations with each other and Snf7. Based on these analyses the authors put forth a series of "rules" governing ESCRT-III assembly and function. While beneficial to our conceptual understanding of ESCRT-III these rules fall short in explicitly defining the structural basis of assembly and function including explaining requisite heterodimerization of Vps24-Vps2. This work represents a significant step forward in addressing this challenging question, the experimental design and implementation are convincing, however the limitations of this work could be conveyed more clearly.

    4. Evaluation Summary:

      ESCRT-III is a conserved hetero-oligomeric membrane remodeling machine known to impact a number of cellular phenomena, yet mechanistic details of its function have remained enigmatic. This work identifies critical inter-subunit contact sites critical for ESCRT-III assembly and function.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

    1. Reviewer #3 (Public Review):

      In this manuscript, Takaine et. al. leveraged their QUEEN ATP biosensor to ask an interesting and important question: how and why cells maintain high and stable ATP concentrations in Saccharomyces cerevisiae.

      The strength of their approach is to obtain single-cell quantification of ATP concentration over time. They use the technology to demonstrate the importance of the AMP kinase, and two other proteins involved in ATP synthesis/homeostasis (the adenylate kinase, ADK1, and the transcription factor, BAS1) in the maintenance of stable and high levels of ATP.

      The main novelty of their findings with respect to ATP homeostasis is the detection of sudden, transient decreases in ATP concentration in mutants. The main claim in the title and abstract of the paper is that "High and stable ATP levels prevent aberrant intracellular protein aggregation". In our opinion, the data do not yet support this claim.

      Essential issues:

      1) The most important missing experiment, which would be required to support the title, is to image both ATP levels and protein aggregation events in the same cell. The current dataset shows that the mutants under study have both decreased ATP levels and suggest that these levels are less stable, and finally that complete ATP depletion leads to protein aggregation, but it is not possible to extrapolate these observations to the current conclusions.

      2) The second most important issue is a lack of statistics with respect to spontaneous drops in ATP concentration. A couple of examples are shown, but it should be possible to obtain data for hundreds of cells. Do the examples in figure 2 represent 90% of cells? 1% of cells? 1/1000? We need to be given a more complete sense of the penetrance of these effects.

    2. Reviewer #2 (Public Review):

      Takaine et al., use a fluorescent reporter to quantify ATP levels within single yeast cells with high temporal resolution. With this approach, they aim to understand the molecular components required to maintain cytoplasmic ATP levels at a constant 4 mM concentration. They identify two enzymes (ADK, AMPK) and one transcription factor (Bas1) that cooperate in buffering cellular ATP levels. Without these proteins, yeast cells experience transient depletions of ATP, which the authors term "ATP catastrophes". These stochastic events are sometimes reversed, but sometimes not, leading to death of the cell. Such ATP catastrophes also make the cell prone to aggregation of neuropathic peptides, which could explain why protein aggregates occur in aging neurons (which experience declines in ATP levels). Their experiments provide strong in vivo evidence that cells maintain high levels of ATP to keep proteins soluble in a crowded cytoplasm.

      Strengths:

      1) This work moves the field forward by providing a single-cell approach. Previous studies of ATP levels analyzed extracts taken from cell populations, which could hide cell-to-cell variability. Indeed, using their ATP reporter, Takaine et al. demonstrate how ATP levels are dynamic, different between cells, and can even undergo dramatic stochastic changes.

      2) The authors use a variety of orthogonal approaches to test their hypotheses. They use the ATP probe QUEEN as their primary approach, but back it up with biochemical analysis of ATP levels in cell populations. Furthermore, they use genetic knockouts, acute insults (chemicals to deplete ATP), and rescue experiments to corroborate their results.

      3) The paper is well written and the logic is easy to follow.

      Weaknesses/Criticisms:

      1) Possible indirect effects due to knock outs of AMPK, ADK, and Bas1. These proteins are involved in many biochemical pathways, including lipid homeostasis, mitophagy, and ATP regulation. How do we know that snf1 KO (AMPK knock out) directly effects ATP levels? Also, it is possible that these yeast have acquired suppressor mutations that let them survive at reduced ATP levels, which could confound interpretation of the results.

      2) Lack of wild-type controls in Figure 2. The authors do quote their previous paper, but I want to see the controls done the exact same way. I need to know that transient changes in ATP levels are due to the mutations and not to user error or a different microscope setup. This is really important, since observation of the "ATP catastrophe" is a major finding of this paper.

      3) Insufficient quantification of the ATP catastrophe phenotype. Figure 2 shows only two cells, so I'm not sure how representative these data are. This is an important discovery, so it deserves better quantification and characterization. It would be important to quantify: a) how many cells in a population experience ATP catastrophe, b) the average time interval of depressed ATP levels before restoration, c) frequency of ATP catastrophes in a single cell, and d) how long can ATP levels be suppressed before the cell dies.

    3. Reviewer #1 (Public Review):

      For bacteria, yeast and mammalian cells, energy depletion has been linked to a vitrification of cytosol and protein aggregation. Previous studies have postulated this is in part due to acidification and the shift in pH to match a large set of proteins pIs resulting in large-scale protein aggregation as well as changes in crowding of the cytosol. Additionally, a more direct role for ATP in protein aggregation has been proposed through its chemical properties as a hydrotrope. The appeal of this hypothesis is that the steady-state levels of ATP far exceed the Kd of most enzymes pointing to a potential non-enzymatic role for the high levels.

      In this study, the authors take advantage of a FRET-reporter for ATP that they developed previously called "Queen". They then manipulate ATP levels using mutants in AMP kinase(Snf1) or Adenyl kinase (Adk1) and find null mutants indeed have lower concentrations of ATP and experience sudden drops in ATP levels which the authors term ATP catastrophe. These mutants also show genetic interactions with protein folding/glycosylation pathways and are sensitive to conditions that generate proteotoxic stress. Hsp104 forms foci in the genetically induced lowered ATP levels as well as exogenous ectopic aggregation prone proteins such as alpha-synuclein. The authors attempt to show that the cause of aggregation is due to limiting ATP directly by adding excess adenosine to the media and showing this diminished the formation of foci, potentially due to the ability of increased exogenous to raise ATP levels according to previous reports.

      The issue of whether ATP levels play a direct or indirect role in preventing protein aggregation is extraordinarily challenging to address. While ATP can act as a hydrotrope, the formation of aggregates could be due to limitations of the activity of chaperones and helicases which would not be surprising role for ATP in the cell. While the experiments are carefully performed, well analyzed and fairly interpreted; questions still remain about the impact of these experiments on understanding how ATP impacts cytosol.

    4. Evaluation Summary:

      Over the past decade, the role of ATP levels in the material properties of cells has gathered substantial interest in part because of the potential role of ATP in solubilizing biomolecular condensates. This study uses a quantitative imaging-based measurement of ATP levels in live cells to assess the impact of mutants in ATP homeostasis on ATP levels and protein aggregation. The strength of this paper is the quantitative, single cell analysis, and the manipulation of ATP using native control pathways. The authors suggest that fluctuations in ATP concentrations can lead to protein aggregation, which would be of broad interest to many fields, including cell biology, aging and neurodegeneration.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      Cellular quiescence, the reversible exit from the cell cycle, is essential for long-term cell survival. One feature of quiescent cells is transcription inactivation and this paper examines gene reactivation during quiescence exit and the accompanying changes to chromatin structure. Using a variety of genome-wide analyses, including 4tU-seq, ChIP-seq, and MNase-seq, the authors show that transcription occurs within minutes of quiescence exit, and for most genes, the initial rate of transcription exceeds that of normal cycling cells. Moreover, this work shows that gene repression during quiescence, and activation upon quiescence exit, are associated with distinct chromatin organization, particularly over promoters. Finally, the authors uncover a role for the RSC chromatin-remodeling complex in establishing a chromatin organization that facilitates normal gene expression during quiescence exit. To support the above findings, the authors generated an impressive amount of sequencing datasets that robustly support their findings and will undoubtedly be of great use to many yeast transcription researchers. Although more transparent and consistent bioinformatic analyses of these data would better communicate the findings, this work enhances our understanding of gene expression changes during the transition between key cell states and thus will be of interest to a broad spectrum of readers ranging from molecular to developmental biologists.

    2. Reviewer #2 (Public Review):

      In this manuscript, Cucinotta et al investigate the role of the conserved RSC chromatin remodeler in preparing cells for hypertranscription during exit from quiescence using cellular perturbations and a range of genomic techniques. They find that upon exit from quiescence there is a large and rapid increase in transcription (within 5 minutes) and this hypertranscription cannot be explained solely by alterations to histone acetylation. Therefore, the authors investigated what is driving this process and identified that RSC, a well describe chromatin remodeler with activities in altering chromatin structure to promote transcription, has altered binding profiles within quiescent cells relative to log cells, and loss of RSC results in altered nucleosome positioning within gene bodies and increased histone occupancy within nucleosome depleted regions (NDRs). They find that RSCs biochemical activity is important for promoting transcription and is required for appropriate RNAPII occupancy during exit. Finally, they find that RSC is required for appropriate transcription as depletion of RSC results in increase aberrant transcription, leading to the model that RSC is important for regulating chromatin structure for appropriate binding of RNAPII throughout the genome during exit from quiescence. The conclusions of this paper are well supported by data, but some aspects of data analysis need to be extended.

      Strengths:

      To my knowledge, this is the first mechanistic description of quiescent exit, adding to the many roles of the important RSC chromatin remodeling complex. The data are extensive to support the claims made by the authors. Data are also clearly described within the text and put into great context within the field.

      Weaknesses:

      Correlations are not directly drawn across the datasets, and aspects of data presentation could be clarified. For example, there is little comparison between the expression data (4tU-seq) and the localization (ChIP-seq) or nucleosome positioning (MNase-seq) datasets. Direct comparisons of where locations have altered factor occupancy and/or nucleosome changes with the expression changes or aberrant transcription increases would help facilitate a mechanistic description.

    3. Reviewer #1 (Public Review):

      Cucinotta et al. examine the widespread, transient transcription of genes that occurs within minutes of refeeding quiescent Saccharomyces cerevisiae cells, focusing on the role of the RSC remodeler complex in this process. A range of appropriate genomic approaches are used to characterize the initial burst of transcription, changes in localization of RSC and RNA Pol II, and changes in the occupancy and positioning of nucleosomes during the first minutes after nutrient repletion. Several new insights are reported including the role of RSC in maintaining promoters in state that is ready to respond rapidly to nutrient repletion, the relocalization of RSC into genes following initiation of transcription, a role for TFIIS in exiting quiescence that was not apparent in log phase, the timing of histone acetylation in response to transcription, changes in chromatin architecture during the exit from quiescence, and the effects of chromatin changes on transcription start site selection and repression of antisense transcription from downstream nucleosome depleted regions. Given how little is known about the emergences of cells from quiescence and how common and important this transition is in long-term viability, development, and carcinogenesis, these insights are certain to have broad impact. The data are of high quality and the manuscript is very clearly written, with good correlation between the level of support provided by the data and the strength of the conclusions drawn. Only minor issues remain to be addressed.

    4. Evaluation Summary:

      Cucinotta et al. describe mechanisms that support an intense burst of transcription from many genes within minutes of nutrient repletion as Saccharomyces cerevisiae cells emerge from quiescence. They focus primarily on the role of the nucleosome remodeler RSC in managing chromatin architecture over promoters during quiescence and as cells re-enter the cell cycle using a broad range of genome-wide measurements that strongly support the conclusions. This important process of cell cycle re-entry from quiescence is understudied but impacts areas as diverse as development and carcinogenesis in multicellular organisms to long-term survival and adaptation of microorganisms to environmental cues, so the results will be of interest to a broad audience.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      Volatile anesthetics (VA) are thought to cause developmental defects in newborns and the authors previously studied the metabolic consequences of VA on newborn mice. Surprisingly, they found VA exposure rapidly and dramatically dropped circulating levels of the ketone beta-hydoxybutyrate (BHB). Newborn mice use ketones as energetic substrates (compared to glucose in weaned animals) so perturbing ketone metabolism could underpin some of the detrimental side-effects of VA. Therefore, the authors sought to determine why VA cause this drop in ketone availability in newborns.

      The authors first found that multiple VAs rapidly (half maximal effect occurs in ~10min) and at subanesthetic doses decrease BHB levels from ~2mM to <1mM in newborn but not in older (older than P19) mice. Extended VA exposure (>60min) also caused a decrease in circulating glucose. BHB levels could be rescued by IP injection prior to anesthesia. Why do VAs cause this effect? Ketones are known to be produced by fatty acid oxidation in the liver. The authors therefore indirectly assessed fatty acid oxidation by measuring levels of acylcarnitines (an intermediate metabolite in fatty acid oxidation) in newborn livers after VA treatment and found lower levels of acylcarnitines consistent with lower levels of fatty acid oxidation in the liver upon VA treatment. Pharmacologically inhibiting fatty acid oxidation could also drop BHB levels in newborn plasma as well. Thus, the authors provide compelling evidence that VA exposure blocks fatty acid oxidation and ketogenesis in the liver of newborns and this underlies the drop in BHB in the circulation.

      The authors next asked why VAs decreased fatty acid oxidation. VAs are thought to inhibit the electron transport chain (ETC) which would cause redox imbalances (particularly in the NAD/NADH ratio) that could lead to altered TCA cycle metabolic activity that could potentially impact fatty acid oxidation. The authors therefore indirectly tested this hypothesis by measuring TCA cycle intermediates and did by VA exposure altered newborn liver levels of several TCA cycle metabolites including citrate. Citrate is metabolized by the enzyme ACLY to generate cytosolic Ac-CoA which is used by the enzyme ACC to produce malonyl-CoA, an intermediate in lipid synthesis. Malonyl-CoA is also known to inhibit the production of acylcarnitines and fatty acid oxidation. Therefore, the higher levels of citrate in VA exposed livers prompted the authors to determine if VA exposure specifically in neonates increased malonyl-CoA and if this blocked fatty acid oxidation and ketogenesis. The authors measured malonyl-CoA in newborn livers and observed an increased upon VA exposure. ACC inhibitions have been developed and the authors found that ACC inhibition (which presumably would prevent malonyl-CoA formation) could partially rescue the drop in BHB brought on by VA exposure in newborns. Thus, this study delineates how altered fatty acid oxidation and ketogenesis in the liver underlies the drop in BHB elicited upon VA exposure and opens the door to future studies determining if the drop in BHB contributes to newborn sensitivity to VAs and future studies elucidating exactly how VA exposure alters the TCA cycle and citrate metabolism to block fatty acid oxidation.

    2. Reviewer #2 (Public Review):

      Here the author reported that Volatile anesthetics VA induce a rapid depletion of circulating ß-HB and the induction of hypoglycemia by VA in neonates, but not in adults. The phenomenon is very interesting and robust, however it has already been described. Whats new here is that through a metabolomics analysis they demonstrate a role of ACC and CPT1 in this phenomenon. Intermediates of the TCA cycle are reduced as would be expected and this is interesting, but chiefly descriptive, and not mechanistic. The key question what causes these derangements in TCA cycle and for sure it's altered enzymatic activity but again what accounts for these and that questions answered would get at the mechanism, but this study here remains descriptive. Is this a cell autonomous effect? For example could you replicate this in a dish with isolated hepatocyte or myotubes from neonates versus adults?

    3. Reviewer #1 (Public Review):

      Stokes, et al. describe the effects of isoflurane on metabolism in post-natal day 7 mice, and older mice. They demonstrate that blood levels of glucose and ß-hydroxybutyrate fall quickly in response to isoflurane, and that the magnitude of the decrease increases with the length of the exposure. Mice 30 days post-natal do not exhibit these changes in response to isoflurane. The authors document the much higher circulating levels of ß-hydroxybutyrate in the post-natal day 7 mice, highlighting the importance of this substrate for supporting the energetics of the developing brain. Important control experiments, administering 100% oxygen without anesthetic to post-natal day 7 mice, as well as administering anesthetics to 30 day old mice on a ketogenic diet, did not result in significant decreases in glucose and ß-hydroxybutyrate blood levels. Remarkably, they observed significant decreases in response to very small, subanesthetic doses of isoflurane, halothane and sevoflurane in post-natal day 7 mice. Administration of bolus glucose corrects the glucose level for these mice under anesthesia, but not the level of ß-hydroxybutyrate, while administration of bolus ß-hydroxybutyrate corrects both levels.

      The authors then proceed to a series of measurements in an attempt to determine a direct target of volatile anesthetics on metabolism, focusing on hepatic metabolism. This is something of a Procrustean bed, given that the there is ample evidence that volatile anesthetics affect a large number of different membrane bound processes. Nonetheless, these experiments provided valuable data demonstrating anesthetic induced decreases in fatty acid oxidation. This reviewer finds the arguments regarding impairment of the citric acid cycle a bit unconvincing: 7 and 30 day old mice exhibit the same increase in citrate and isocitrate levels, yet only the 7 day old mice show elevated lactate levels. Rather than exhibiting increased metabolic flexibility, as the authors suggest, this finding seems to argue that 7 day old mice have less metabolic flexibility. The authors demonstrate that several perturbations of fatty acid metabolism can result in depression of ß-hydroxybutyrate, leading them to focus on carnitine palmitoyl transferase-1. They demonstrate that inhibition of this enzyme produces a decrease in ß-hydroxybutyrate; however, they also find that mice with a knockout of this enzyme do not have decreased ß-hydroxybutyrate levels.

      The authors are circumspect in their conclusions regarding the targets responsible for the metabolic changes observed in neonatal mice in response to anesthetics. They do correctly highlight the potential importance of these metabolic effects. It will be crucial for future research to determine whether these effects can be directly correlated to measures of cerebral function during anesthesia, e.g., EEG or evoked potentials, and to measures of neuropathological change. Of great interest to clinicians will be demonstration of whether co-administration of glucose or ß-hydroxybutyrate together with anesthetics can abrogate such changes.

    4. Evaluation Summary:

      This manuscript reports that Volatile anesthetics VA induce a rapid depletion of circulating ß-HB and the induction of hypoglycemia by VA in neonates, but not in adults. The phenomenon is very interesting and robust, however it has already been described. Whats new here is that through a metabolomics analysis they demonstrate a role of ACC and CPT1 in this phenomenon.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

  3. Mar 2021
    1. Reviewer #3 (Public Review):

      Zilova et al. investigate cell differentiation in aggregates made from cells of early medaka and zebrafish embryos upon culture in defined media. Using reporter lines and immunostaining, they find evidence for retinal differentiation and morphogenesis in these aggregates, the extent of which depends on the size of the aggregates. This dependence of patterning and morphogenesis on aggregate size indicates that these processes are at least partially controlled by cell-cell interactions in the population. The authors also perform experiments with cells from genetic mutants that indicate similar genetic control of retinal morphogenesis in aggregates and intact fish embryos.

      This work is a nice example of morphogenesis of differentiated cell types upon dissociation and re-aggregation of early embryonic cells. The similar behaviour of aggregates from evolutionarily distant species reported in the manuscript underscores the generality of the findings. Organoid formation from teleost cells recapitulates species-specific timescales and is therefore faster than organoid generation from mammalian cells which constitutes a potential technical advantage of this system. The major advance of this work lies in providing a clear example that organoids consisting of early neural and retinal cells can be formed in non-mammalian species. Such an approach can open up new avenues for describing basic principles of cell differentiation and pattern formation during embryogenesis, and can thereby be useful to the community.

      While the reported observations are highly interesting, the level of quantitative analysis currently does not fully support all of the author's interpretations and conclusions.

      1) The authors variably interpret their observations as the result of self-assembly or self-organization. At the moment, the data does not allow distinguishing whether the observed phenomena result from cells following largely cell-autonomous differentiation paths and come together through cell sorting, or whether dissociation and aggregation generates a condition that leads to (spatially restricted) retinal differentiation in cells that would not normally adopt this fate. I would say that the first scenario is consistent with self-assembly, while the second one is more self-organized in the sense that the new cell-cell interactions resulting from the aggregation result in emergent cellular behaviours. A first step to distinguish between these possibilities would be to quantitatively demonstrate that aggregation biases cell differentiation towards neural and retinal fates at the expense of other cell types, compared to the intact embryo. The examples shown in Figure 2 and 3d seem to indicate an overrepresentation of neural cells, but it would be good to see a quantitative comparison to the embryo.

      2) The authors use the term "primary embryonic stem cells" for the early embryonic cells that they aggregate. I find this problematic as some cells in this population may already have lineage bias and not have true multi-lineage potential. I also understand there is a difference between the cells that are used in this study and the teleost embryonic stem cells referenced in lines 49 and 50, in the sense that the latter were established as true self-renewing cell lines. But correct me if I have missed something here.

      3) The authors claim that their system is highly reproducible. Unfortunately, they do not give an indication of the success rate of aggregate formation in figure 1. Figure 4 shows the most complex patterns, but I realize that there is quite a bit of variability in between the aggregates - they are just as likely to have one or two Rx2-expressing areas (panel b). I also could not find information how many aggregates show the patterns in panels e and f, and from how many aggregates the data in panels g - i has been collected.

    2. Reviewer #2 (Public Review):

      This study shows that dissociated blastula cells from teleost fishes (medaka and zebrafish) reaggregate to form optic vesicle-like organoids if cultured in the presence of extracellular matrix molecules. Notably, cell number is critical for a reaggregation with movements that resemble those observed in vivo. These organoids acquire dorso-ventral polarity and can differentiate into different retinal cell types.

      This is well written manuscript describing a technological advance: the generation of an organoid from teleost cells. Some of the images are impressive as since blastula cells seem to reproduce an organized forebrain with bilateral optic vesicles. Still these vesicles are rudimentary when compared with those obtained from mouse or human cells (see work from Eiraku team).

      There are no critiques to the work per se, which is technically impeccable, well illustrated and quantified. However, one wonder what happens to the RPE cells in the differentiation process. In Fig 4, the authors show that the optic vesicle organoids are organized as in vivo with cells expressing RPE markers. These cells are no longer present in Fig 5. What happens to them? There is no mention of this problem in the text.

      The discussion is generally informative but somehow fails to provide real advantages of using teleost organoids vs the fish per se or vs for example human organoids. Indeed, obtaining a fish organoid is faster that a human one, but more expensive and time consuming than using fish embryos.

    3. Reviewer #1 (Public Review):

      In this manuscript, Zilova et al. show that primary embryonic cells derived from blastula-stage Medaka and Zebrafish embryos can self-organize into retinal organoids. When aggregates of 1000-2000 primary embryonic cell are embedded in Matrigel addition, they form a neuroepithelium under the control of Rx3 which develops into a retinal organoid. The process mirrors some aspects of embryo development. Moreover, another interesting finding is that Rx3 expression is initiated in the absence of Matrigel at day 0, which indicates that the retinal fate occurs by default and is not dependent on extracellular matrix components. The authors compare the ability of cells from Mesaka and zebra fish and show that both are competent to form organoids, though each does it with the time scale of the embryo of origin. The authors show that by reducing the number of Medaka cells to aggregate (500-800 cells), Rx2 and Rx3 are expressed only in restricted regions of the small aggregates, presumably where they organize into discrete circular Rx2 and Rx3 positive neuroepithelial units that develop into structure resembling retinal epitjhelia with some diversity of retinal cell types including amacrine, ganglion, photoreceptor, bipolar and horizontal cells.

      This is a novel and original piece of work that reveals the capacity of fish primary embryonic pluripotent cells to behave like mammalian embryonic stem cells and organize optic cup organoids.

    4. Evaluation Summary:

      The experimental system characterized in this paper opens up new avenues for studying mechanisms of retinal patterning and morphogenesis. The data presented make a compelling case for the emergence of complex multicellular structures upon re-aggregation of embryonic teleost cells, but open questions remain regarding the basic underlying principles.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      In the manuscript, "Dynamic persistence of UPEC intracellular bacterial communities in a human bladder-chip model of urinary tract infection" by Sharma, et al., the authors develop a bladder-on-chip model and provide evidence that this is a useful model for mimicking in vivo infections. The focus is on intracellular infection structures created by uropathogenic Escherichia coli (UPEC) seen in experimental mice infections and "real" human infections; such structures have been most extensively characterized in mouse models for obvious reasons. The authors focus on three key aspects: development of a structure known as an intracellular bacterial community (IBC), the neutrophil response to infecting UPEC, and the bacterial response to antibiotic treatment. There is a minor point about the ability to apply mechanical stretch to the model to mimic bladder filling and voiding.

      In my assessment, key strengths of this work are:

      1) Integration of both epithelial and vascular endothelial cell types, allowing for multiple fluid spaces and studies of neutrophil migration

      2) Ability to apply mechanical stretch to the entire system to mimic changes in bladder volume

      3) Extensive microscopic characterization of the model (a key feature enabled by this system) including live microscopy, immunostaining, and electron microscopy

      I believe there is one key underlying issue with this paper: as a report on the technical development of a new system / device / technique, the authors have what amounts to a very strong hypothesis, namely that their new system is a good model for the in vivo infection. This leads to a general bias in both the presentation and, in my opinion, the interpretation of the data, to make the system sound "as good as possible". Key manifestations of this bias and overinterpretation include:

      1) The immediate interpretation of all intracellular structures as IBCs.

      2) The immediate interpretation of all data in Figure 2 as neutrophil swarms and NETs.

      3) Some odd behaviors in response to ampicillin, which should not penetrate host cells and has been shown using the same cell types to not affect intracellular UPEC.

      4) The claim that a 10% linear change in dimension is "physiologically relevant" and "a significant proportion" of that seen in vivo.

      To clarify point 1 (which applies as well to point 2), IBC is an abbreviation for "intracellular bacterial community", and these were first described in mice. There has been very sparing molecular characterization of IBCs, which makes a morphological classification very tricky - I believe the field generally thinks that IBCs refer to a specific structure that is formed (at least) in mice and humans in vivo. Somewhat similar structures have been seen previously in vitro but rightfully are more carefully described with different terms or as "structures resembling IBCs". I think similar care needs to be taken with this model as well.

      Overall, the authors have done quite a complete job in characterizing their model and have good data to argue for a morphological similarity to key steps that have been previously described to happen in vivo. I believe they get ahead of themselves both in data interpretation and in the writing of the manuscript, which leads to some oddness where it seems the authors begin to talk as if their model has already been validated. This occurs throughout the manuscript in the use of the IBC abbreviation and also largely in the section on neutrophil responses (in particular swarms and NETs). There are occasional sentences where the appropriate care is taken (i.e. that the data is being collected to argue that the structures seen are indeed NETs), but this is interspersed with writing that is assuming the point is already proven (for example, see lines 286 (appropriate) and 287-289 (not); and 471-476 (appropriate), 477-479 (not), 481-483 (appropriate)).

      Regarding the ampicillin data, the odd behaviors are:

      1) Apparent elimination of intracellular UPEC (particularly for large collections)

      2) Apparent indifference for some intracellular UPEC (they continue to grow)

      3) Ampicillin is generally thought to not cross host membranes, and in Blango & Mulvey 2010 it does not affect UPEC harbored within 5637 cells. The authors collect #1 and #2 under "dynamic heterogeneity" and then claim in the discussion that they can "realistically model antibiotic treatment regimens". Given these discrepancies listed above, I do not believe they can yet support this claim.

      Finally, the ability to apply mechanical stretch is only used in one pilot experiment at the end, producing a suggestive result (that UPEC burden increases when a duty cycle of stretching and relaxing is used). This is a key advantange of their model that gets a proportionately larger share of attention in the introduction and discussion. It also may provide an explanation for the ability of ampicillin to enter the host cells, or to access intracellular bacteria (through vesicular uptake during contraction, as UPEC themselves are thought to do).

    2. Reviewer #2 (Public Review):

      Sharma et al established a bladder on a chip model for studies of E. coli infection using a co-culture HTB9 bladder epithelial cells and primary human bladder microvascular endothelial cells in an organ-on-a-chip device. The two cell types expressed cell-specific markers when cultivated on-a-chip. Linear strain was applied to the sides of the device up to 19% to mimic stretching during bladder filling. The bladder chip was perfused with the diluted human urine during the experiments. The authors also observed formation of neutrophil extracellular traps by neutrophils in the infected bladder chip. They also demonstrate that the planktonic bacteria are eliminated upon application of antibiotics on a chip, with intracellular bacteria retaining the ability to grow after a lag period. The strength of the system is its fine imaging capability. It is necessary to consider if another antibiotic would enable clearance of intracellular bacteria.

    3. Reviewer #1 (Public Review):

      The complexity of the infection model developed by the authors is to be praised as it allows the dissection of host-pathogen interactions with multiple players coming together, namely human epithelial cells, endothelial cells and neutrophils, UPEC, urine, antibiotics and mechanical forces at play during bladder filling and micturition. This is truly a tour de force and should provide the authors (and other labs potentially able to recapitulate it) with an unprecedented model to study UTIs and their response to antibiotics. Notably, authors have been able to document the formation of NETs in response to UPEC infection in this model. One small caveat was the choice of antibiotics used to treat the infection in their model. Is Ampicillin really a drug of choice, both because of its inability to reach intracellular niches and it not being a drug of choice in the clinic?

    4. Evaluation Summary:

      Reviewers value the development and characterization of a bladder-on-chip infection model for recapitulating the multiples factors involved in UPEC driven UTIs. Notably, it consists of human bladder epithelial cells, bladder microvascular endothelial cells, neutrophils and urine that are also subjected to mechanical changes mimicking those occurring during bladder filling and micturition. This model is a lot more complex than in vitro tissue culture models and more amenable to analysis such as imaging than animal models and therefore constitute a distinct advance for in vitro modeling of UTI that has potential to reveal key aspects of UTIs and reasons for the difficulty to clear these infections with antibiotics.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Author Response:

      Reviewer #1 (Public Review):

      In this project, the authors set out to create an easy to use piece of software with the following properties: The software should be capable of creating immersive (closed loop) virtual environments across display hardware and display geometries. The software should permit easy distribution of formal experiment descriptions with minimal changes required to adapt a particular experimental workflow to the hardware present in any given lab while maintaining world-coordinates and physical properties (e.g. luminance levels and refresh rates) of visual stimuli. The software should provide equal or superior performance for generating complex visual cues and/or immersive visual environments in comparison with existing options. The software should be automatically integrated with many other potential data streams produced by 2-photon imaging, electrophysiology, behavioral measurements, markerless pose estimation processing, behavioral sensors, etc.

      To accomplish these goals, the authors created two major software libraries. The first is a package for the Bonsai visual programming language called "Bonsai.Shaders" that brings traditionally low-level, imperative OpenGL programming into Bonsai's reactive framework. This library allows shader programs running on the GPU to seamlessly interact, using drag and drop visual programming, with the multitude of other processing and IO elements already present in numerous Bonsai packages. The creation of this library alone is quite a feat given the complexities of mapping the procedural, imperative, and stateful design of OpenGL libraries to Bonsai's event driven, reactive architecture. However, this library is not mentioned in the manuscript despite its power for tasks far beyond the creation of visual stimuli (e.g. GPU-based coprocessing) and, unlike BonVision itself, is largely undocumented. I don't think that this library should take center stage in this manuscript, but I do think its use in the creation of BonVision as well as some documentation on its operators would be very useful for understanding BonVision itself.

      We have added a reference to the Shaders package at multiple points in the manuscript including lines 58-59 and in Supplementary Details. We will be adding documentation of key Shaders nodes that are important for the creation of BonVision stimuli to the documentation on the BonVision website.

      Following the creation of Bonsai.Shaders, the authors used it to create BonVision which is an abstraction on top of the Shaders library that allows plug and play creation of visual stimuli and immersive visual environments that react to input from the outside world. Impressively, this library was implemented almost entirely using the Bonsai visual programming language itself, showcasing its power as a domain-specific language. However, this fact was not mentioned in the manuscript and I feel it is a worthwhile point to make.

      Thank you - we have now added clarification on this in Supplementary details (section Customised nodes and new stimuli)

      The design of BonVision, combined with the functional nature of Bonsai, enforces hard boundaries between the experimental design of visual stimuli and (1) the behavioral input hardware used to drive them, (2) the dimensionality of the stimuli (i.e. 2D textures via 3D objects), (3) the specific geometry of 3D displays (e.g. dual monitors, versus spherical projection, versus head mounted stereo vision hardware), and (4) automated hardware calibration routines. Because of these boundaries, experiments designed using BonVision become easy to share across labs even if they have very different experimental setups. Since Bonsai has integrated and standardized mechanisms for sharing entire workflows (via copy paste of XML descriptions or upload of workflows to publicly accessible Nuget package servers), this feature is immediately usable by labs in the real world.

      After creating these pieces of software, the authors benchmarked them against other widely used alternatives. IonVisoin met or exceeded frame rate and rendering latency performance measures when compared to other single purpose libraries. BonVision is able to do this while maintaining its generality by taking advantage of advanced JIT compilation features provided by the .NET runtime and using bindings to low-level graphics libraries that were written with performance in mind. The authors go on to show the real-world utility of BonVision's performance by mapping the visual receptive fields of LFP in mouse superior colliculus and spiking in V1. The fact that they were able to obtain receptive fields indicates that visual stimuli had sufficient temporal precision. However, I do not follow the logic as to why this is because the receptive fields seem to have been created using post-hoc aligned stimulus-ephys data, that was created by measuring the physical onset times of each frame using a photodiode (line 389). Wouldn't this preclude any need for accurate stimulus timing presentation?

      We thank the reviewer for this suggestion. We now include receptive field maps calculated using the BonVision timing log in Figure5 – figure supplement 1. Using the BonVision timing alone was also effective in identifying receptive fields.

      Finally the authors use BonVision to perform one human psychophysical and several animal VR experiments to prove the functionality of the package in real-world scenarios. This includes an object size discrimination task with humans that relies on non-local cues to determine the efficacy of the cube map projection approach to 3D spaces (Fig 5D). Although the results seem reasonable to me (a non-expert in this domain), I feel it would be useful for the authors to compare this psychophysical discrimination curve to other comparable results. The animal experiments prove the utility of BonVision for common rodent VR tasks.

      The psychometric test we performed on human subjects was primarily to test the ability of BonVision to present VR stimuli on a head-mounted display. We have edited the text to reflect this. The efficacy of the cube map approach for 3D spaces is well-established in computer graphics and gaming and is currently the industry standard, which was the reason for our choice.

      In summary, the professionalism of the code base, the functional nature of Bonsai workflows, the removal of overhead via advanced JIT compilation techniques, the abstraction of shader programming to high-level drag and drop workflows, integration with a multitude of input and output hardware, integrated and standardized calibration routines, and integrated package management and workflow sharing capabilities make Bonsai/BonVision serious competitors to widely-used, closed-source visual programming tools for experiment control such as LabView and Simulink. BonVision showcases the power of the Bonsai language and package management ecosystem while providing superior design to alternatives in terms of ease of integration with data sources and facilitation of sharing standardized experiments. The authors exceeded the apparent aims of the project and I believe BonVision will become a widely used tool that has major benefits for improving experiment reproducibility across laboratories.

      Reviewer #2 (Public Review):

      BonVision is a package to create virtual visual environments, as well as classic visual stimuli. Running on top of Bonsai-RX it tries and succeeds in removing the complexity of the above mentioned task and creating a framework that allows non-programmers the opportunity to create complex, closed loop experiments. Including enough speed to capture receptive fields while recording different brain areas.

      At the time of the review, the paper benchmarks the system using 60Hz stimuli, which is more than sufficient for the species tested, but leaves an open question on whether it could be used for other animal models that have faster visual systems, such as flies, bees etc.

      Thank you for prompting us to do this - we have now added new benchmarks for a faster refresh rate (144 Hz; new Figure 4 - figure supplement 1).

      The authors do show in a nice way how the system works and give examples for interested readers to start their first workflows with it. Moreover, they compare it to other existing software, making sure that readers know exactly what "they are buying" so they can make an informed decision when starting with the package.

      Being written to run on top of Bonsai-RX, BonVision directly benefits from the great community effort that exists in expanding Bonsai, such as its integration with DeepLabCut and Auto-pi-lot. Showing that developing open source tools and fostering a community is a great way to bring research forward in an additive and less competitive way.

      Reviewer #3 (Public Review):

      Major comments:

      While much of the classic literature on visual systems studies have utilized egocentrically defined ("2D") stimuli, it seems logical to project that present and future research will extend to not only 3D objects but also 3D environments where subjects can control their virtual locations and viewing perspectives. A single software package that easily supports both modalities can therefore be of particular interest to neuroscientists who wish to study brain function in 3D viewing conditions while also referencing findings to canonical 2D stimulus responses. Although other software packages exist that are specialized for each of the individual functionalities of BonVision, I think that the unifying nature of the package is appealing for reasons of reducing user training and experimental setup time costs, especially with the semi-automated calibration tools provided as part of the package. The provisions of documentation, demo experiments, and performance benchmarks are all highly welcome and one would hope that with community interest and contributions, this could make BonVision very friendly to entry by new users.

      Given that one function of this manuscript is to describe the software in enough detail for users to judge whether it would be suited to their purposes, I feel that the writing should be fleshed out to be more precise and detailed about what the algorithms and functionalities are. This includes not shying away from stating limitations -- which as I see it, is just the reality of no tool being universal, but because of that is one of the most important information to be transmitted to potential users. My following comments point out various directions in which I think the manuscript can be improved.

      We thank the reviewer for this suggestion. We have added a major new section, “Supplementary Details”, where we have highlighted known limitations and available workarounds. We also added new rows in the Supplementary Table that make these limitations transparent (eg. web-based deployment).

      The biggest point of confusion for me was whether the 3D environment functionality of BonVision is the same as that provided by virtual spatial environment packages such as ViRMEn and gaming engines such as Unity. In the latter software, the virtual environment is specified by geometrically laying out the shape of the traversable world and locations of objects in it. The subject then essentially controls an avatar in this virtual world that can move and turn, and the software engine computes the effects of this movement (i.e. without any additional user code) then renders what the avatar should see onto a display device. I cannot figure out if this is how BonVision also works. My confusion can probably be cured by some additional description of what exactly the user has to do to specify the placement of 3D objects. From the text on cube mapping (lines 43 and onwards), I guessed that perhaps objects should be specified by their vectorial displacement from the subject, but I have very little confidence in my guess and also cannot locate this information either in the Methods or the software website. For Figure 5F it is mentioned that BonVision can be used to implement running down a virtual corridor for a mouse, so if some description can be provided of what the user has to do to implement this and what is done by the software package, that may address my confusion. If BonVision is indeed not a full 3D spatial engine, it would be important to mention these design/intent differences in the introduction as well as Supplementary Table 1.

      Thank you for prompting us to do this. BonVision does indeed essentially render the view of an avatar in a virtual world (or multiple views, of multiple avatars), without any additional coding required by the user. We have now included in the new “Supplementary Details” specific pathways to the construction and rendering of 3D scenes. We have avoided the use of the terminology ‘game-engine’ as it has a particular definition that most softwares do not satisfy.

      More generally, it would be useful to provide an overview of what the closed-loop rendering procedure is, perhaps including a Figure (different from Supplementary Figure 2, which seems to be regarding workflow but not the software platform structure). For example, I imagine that after the user-specified texture/object resources have been loaded, then some engine runs a continual loop where it somehow decides the current scene. As a user, I would want to know what this loop is and how I can control it. For example, can I induce changes in the presented stimuli as a function of time, whether this time-dependence has to be prespecified before runtime, or can I add some code that triggers events based on the specific history of what the subject has done in the experiment, and so forth. The ability to log experiment events, including any viewpoint changes in 3D scenes, is also critical, and most experimenters who intend to use it for neurophysiological recordings would want to know how the visual display information can be synchronized with their neurophysiological recording instrumental clocks. In sum, I would like to see a section added to the text to provide a high-level summary of how the package runs an experiment loop, explaining customizable vs. non-customizable (without directly editing the open source code) parts, and guide the user through the available experiment control and data logging options.

      We have now added a brief paragraph regarding the basic structure of a BonVision program, and how to ‘close the loop’ in the new “Supplementary Details”.

      Having some experience myself with the tedium (and human-dependent quality) of having to adjust either the experimental hardware or write custom software to calibrate display devices, I found the semi-automated calibration capabilities of BonVision to be a strong selling point. However I did not manage to really understand what these procedures are from the text and Figure 2C-F. In particular, I'm not sure what I have to do as a user to provide the information required by the calibration software (surely it is not the pieces of paper in Fig. 2C and 2E..?). If for example, the subject is a mouse head-fixed on a ball as in Figure 1E, do I have to somehow take a photo from the vantage of the mouse's head to provide to the system? What about the augmented reality rig where the subject is free to move? How can the calibration tool work with a single 2D snapshot of the rig when e.g. projection surfaces can be arbitrarily curved (e.g. toroidal and not spherical, or conical, or even more distorted for whatever reasons)? Do head-mounted displays require calibration, and if so how is this done? If the authors feel all this to be too technical to include in the main text, then the information can be provided in the Methods. I would however vote for this as being a major and important aspect of the software that should be given air time.

      We have a dedicated webpage going through the step-by-step protocol for the automated screen calibration. We now explicitly point to this page in the new Supplementary Details section.

      As the hardware-limited speed of BonVision is also an important feature, I wonder if the same ~2 frame latency holds also for the augmented reality rendering where the software has to run both pose tracking (DeepLabCut) as well as compute whole-scene changes before the next render. It would be beneficial to provide more information about which directions BonVision can be stressed before frame-dropping, which may perhaps be different for the different types of display options (2D vs. 3D, and the various display device types). Does the software maintain as strictly as possible the user-specified timing of events by dropping frames, or can it run into a situation where lags can accumulate? This type of technical information would seem critical to some experiments where timings of stimuli have to be carefully controlled, and regardless one would usually want to have the actual display times logged as previously mentioned. Some discussion of how a user might keep track of actual lags in their own setups would be appreciated.

      We now provide this as part of the Supplementary Details, specifically animation and timing lags.

      On the augmented reality mode, I am a little puzzled by the layout of Figure 3 and the attendant video, and I wonder if this is the best way to showcase this functionality. In particular, I'm not entirely sure what the main scene display is although it looks like some kind of software rendering — perhaps of what things might look like inside an actual rig looking in from the top? One way to make this Figure and Movie easier to grasp is to have the scene display be the different panels that would actually be rendered on each physical panel of the experiment box. The inset image of the rig should then have the projection turned on, so that the reader can judge what an actual experiment looks like. Right now it seems for some reason that the walls of the rig in the inset of the movie remain blank except for some lighting shadows. I don't know if this is intentional.

      Because we have had limited experimental capacity in this period, we only simulated a real-time augmented reality environment off-line, using pre-existing videos of animal behaviour. We think that the comment above reflects a misunderstanding of what the Figure and associated Supplementary Movie represents, and we realise that their legends were not clear enough. We have now made sure that these legends make clear that these are based on simulations (new legends for Figure 3 and Figure 3 - video supplement 1).

    1. Reviewer #3 (Public Review):

      By means of in vitro reconstitution, the authors find that the microtubule associated protein Sjögren's Syndrome Nuclear Autoantigen 1 (SSNA1), know to form fibrils binding longitudinally along microtubules, modulates microtubule instability by reducing dynamicity and inducing rescues, prevents catastrophes in absence of free tubulin or in presence of the tubulin-sequestering protein stathmin, inhibits microtubule severing of spastin and detects spastin-induced damage sites. SSNA1, thus, is revealed as a very potent microtubule stabilizing factor.

      The reconstitution of microtubule dynamics is sound and well performed, and the parameters of dynamicity are thoroughly analyzed. The observed intensity of SSNA1 fluorescence demonstrates that the proteins do not bind uniformly along microtubules. Consequently, the rates of microtubule dynamics are not affected globally. Instead the observed rates are affected at different times for individual microtubules and, importantly, directly correlate with locally accumulating SSNA1. The authors thus validly conclude that nucleotide state recognition is not the primary mechanism of SSNA1 localization and activity. Clues towards the mechanism of SSNA1 activity are provided by the observation that SSNA1 detects spastin-induced damage sites, indicating that SSNA1 binds to partial, open microtubule structures and then stabilizes them, which is consistent with cryo-electron-tomograms available in the literature. To me it is not clear, however, if SSNA1 localize-to and act-on distinct sites of microtubule damage exclusively, or if these sites rather serve as positions of initiation or nucleation of cooperative SSNA1 binding, which the kymographs and movies seem to suggest.

      The presented observations nicely explain how the microtubule severing enzyme spastin, which directly interacts with SSNA1 and thus recruits it to the very sites of immediate damage, promotes regrowth of microtubules and increases their number and mass in vivo. The manuscript would benefit from further investigation-into and quantifications-of the "progressive accumulation" of SSNA1 on the dynamic microtubules, which, thus far are presented only by way of representational example.

    2. Reviewer #2 (Public Review):

      The manuscript provides some long awaited follow-up work to a controversial publication implicating SSNA1/NA14 in microtubule branching (Basnet et al. NCB 2018). The authors have strong expertise in in-vitro microtubule dynamic behaviour. While the experiments are technically strong, the authors use unphysiological amounts of the SSNA1, making interpretations about biological function hard.

      The authors take a rigorous approach to analyze details of microtubule dynamic behaviour presented in Figure 1. While I recognize the enormous amount of work that went into Figure 1, in my opinion these experiments shows that SSNA1 has no effect on microtubule dynamics at physiological concentration (sub 100 nM). That finding is i) very publishable and ii) should not take away from SSNA1 as an important molecule, but rather open up alternative ways of thinking about the protein.

      I believe similar conclusions should be applied to microtubule slow-down in Figure 2 and the stabilization against tubulin loss by dilution/sequestration in Figure 3. If 5 uM (the only concentration shown) are required to achieve above effects, these observations are likely not relevant to SSNA1's biological function.

      Taking into account SSNA1's cellular localization at centrosomes, midbodies, and branch points etc., I am not sure a major effect on microtubule dynamics other than nucleation should be expected.

      The authors pursue an alternative and very interesting avenue in Figure 4, by examining the interplay between spastin and SSNA1 with regards to microtubules. Here, (1 uM) SSNA1 has protective effects against severing by spastin.

      The discussion could use a direct contrast to differences in findings between the current work and the branched nucleation. It is not stated in the manuscript, though presumably no branching has been observed in several thousands of microtubule growth events? I would find a lot of value in such a potential statement.

    3. Reviewer #1 (Public Review):

      In their manuscript, Lawrence et al. investigate the direct effects of the microtubule-associated protein, SSNA1, on microtubule (MT) dynamics and damage using purified proteins and TIRF microscopy. Prior work on this protein showed that SSNA1 self-assembles into higher-order filaments and binds longitudinally along stabilised MTs, inducing MT branching and nucleation. In this study, they find that SSNA1 promotes templated MT nucleation, consistent with prior results, but further define the effect of SSNA1 on MT dynamics. SSNA1 overall dampens MT dynamics by reducing both growth and shrinkage rates, suppressing catastrophe frequency, and increasing rescues. The authors also quantify SSNA1 on GMPCPP over a timecourse both at single-molecule and multi-molecule concentrations. On dynamic MTs, SSNA1 recognizes the growing end and promotes end curvature, but it did not recognize the curves of taxol-stabilised MTs, leading the authors to conclude that it likely induces curvature, rather than recognizes it. Perhaps this is the mechanism by which SSNA1 prevents catastrophe, a role which the authors demonstrate for SSNA1 after both tubulin dilution or stathmin sequestration of tubulin. The most interesting part of this study is found in Figure 4, where the authors show that SSNA1 prevents MT severing by spastin and also localizes to sites of lattice damage. The authors conclude that SSNA1 is a MT stabilizing protein and a sensor of MT damage. The results on MT dynamics do not provide much insight into the mechanism of this protein, which isn't even found to colocalize with MTs in vivo (SSNA1 instead accumulates at branchpoints in neurons). The role of SSNA1 in lattice damage recognition is the highlight of this paper, and also correlates well with its in vivo localization pattern, indicating this could be a true function of this protein. This damage recognition ability could potentially be the first step that leads to SSNA1-induced MT nucleation and branching from an existing MT.

    4. Evaluation Summary:

      In this manuscript, Lawrence et al. investigate the direct effects of the microtubule-associated protein, SSNA1, on microtubule dynamics and damage using purified proteins and TIRF microscopy. The authors conclude that SSNA1 is a microtubule stabilizing protein and a sensor of microtubule damage. This paper is of high interest to scientists within the field of microtubule mechanics and of broad interest to scientists studying cilia, cell division and neuronal development.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      Magnusson et al., do an excellent job of defining how the repeated separator sequence of Wild Type Cas12a CRISPR arrays impacts the relative efficacy of downstream crRNAs in engineered delivery systems. High-GC content, particularly near the 3' end of the separator sequence appears to be critically important for the processing of a downstream crRNA. The authors demonstrated naturally occurring separators from 3 Cas12a species also display reduced GC content. The authors use this important new information to construct a synthetic small separator DNA sequence which can enhance CRISPR/Cas12a-based gene regulation in human cells. The manuscript will be a great resource for the synthetic biology field as it shows an optimization to a tool that will enable improved multi-gene transcriptional regulation.

      Strengths:

      • The authors do an excellent job in citing appropriate references to support the rationale behind their hypotheses.
      • The experiments and results support the authors' conclusions (e.g., showing the relationship between secondary structure and GC content in the spacers).
      • The controls used for the experiments were appropriate (e.g., using full-length natural separator vs single G or 1 to 4 A/T nucleotides as synthetic separators).
      • The manuscript does a great job assessing several reasons why the synthetic separator might work in the discussion section, cites the relevant literature on what has been done and restates their results to argument in favor or against these reasons.
      • This paper will be very useful for research groups in the genome editing and synthetic biology fields. The data presented (specially the data concerning the activation of several genes) can be used as a comparison point for other labs comparing different CRISPR-based transcriptional regulators and the spacers used for targeting.
      • This paper also provides optimization to a tool that will be useful for regulating several endogenous genes at once in human cells thus helping researchers studying pathways or other functional relationships between several genes.

      Opportunities for Improvement:

      • The authors have performed all the experiments using LbCas12a as a model and have conclusively proven that the synSeparator enhances the performance of Cas12a based gene activation. Is this phenomenon will be same for other Cas12a proteins (such as AsCas12a)? The authors should perform some experiments to test the universality of the concept. Ideally, this would be done in HEK293T cells and one other human cell type.
    2. Reviewer #2 (Public Review):

      Type V CRISPR-Cas systems are used in a variety of biotechnology applications, which rely on the association of a Cas12a-CRISPR RNA complex association with a complementary target DNA sequence. One advantage of the Cas12a system over other CRISPR-Cas systems is the ability to multiplex by expressing multiple CRISPR RNAs in an array, with the individual RNAs processed from a longer transcript by Cas12a. Magnusson et al. show that the activity of CRISPR RNAs in this system is enhanced by including a short, A/T-rich sequence between each encoded CRISPR RNA. The authors propose that these separator sequences reduce the potential for secondary structure, thereby promoting RNA processing. This is an exciting idea, with obvious applications wherever Cas12a is used. However, while the presented data are consistent with the model, I think the conclusions are too preliminary, and require (i) a more targeted assessment of the importance of RNA secondary structure for RNA processing, (ii) direct measurement of RNA processing, and (iii) a more extensive assessment of the effect of adding spacer sequences to CRISPR arrays in a functional assay.

    3. Reviewer #1 (Public Review):

      The authors interrogated an underexplored feature of CRISPR arrays to enhance multiplexed genome engineering with the CRISPR nuclease Cas12a. Multiplexing represents one of the many desirable features of CRISPR technologies, and use of highly compact CRISPR arrays from CRISPR-Cas systems allows targeting of many sites at one time. Recent work has shown though that the composition of the array can have a major impact on the performance of individual guide RNAs encoded within the array, providing ample opportunities for further improvements. In this manuscript, the authors found that the region within the repeat lost through processing, what they term the separator, can have a major impact on targeting performance. The effect was specifically tied to upstream guide sequences with high GC content. Introducing synthetic separator sequences shorter than their natural counterparts but exhibiting similarly low GC content boosted targeted activation of a reporter in human cells. Applying one synthetic separator to a seven-guide array targeting chromosomal genes led to consistent though more modest targeted activation. These findings introduce a distinct design consideration for CRISPR arrays that can further enhance the efficacy of multiplexed applications. The findings also suggest a selective pressure potentially influencing the repeat sequence in natural CRISPR arrays.

      Strengths:

      The portion of the repeat discarded through processing normally has been included or discarded when generating a CRISPR-Cas12a array. The authors clearly show that something in between-namely using a short version with a similarly low GC content-can enhance targeting over the truncated version. A coinciding surprising result was that the natural separator completely eliminated any measurable activation, necessitating the synthetic separator.

      The manuscript provides a clear progression from identifying a feature of the upstream sequences impacting targeting to gaining insights from natural CRISPR-Cas12a systems to applying the insights to enhance array performance.

      With further support, the use of synthetic separators could be widely adopted across the many applications of CRISPR-Cas12a arrays.

      Weaknesses:

      The terminology used to describe the different parts of the CRISPR array could better align with those in the CRISPR biology field. For one, crRNAs (abbreviated from CRISPR RNAs) should reflect the final processed form of the guide RNA, whereas guide RNAs (gRNAs) captures both pre-processed and post-processed forms. Also, "spacers" should reflect the natural spacers acquired by the CRISPR-Cas system, whereas "guides" better capture the final sequence in the gRNA used for DNA target recognition.

      A running argument of the work is that the separator specifically evolved to buffer adjacent crRNAs. However, this argument overlooks two key aspects of natural CRISPR arrays. First, the spacer (~30 nts) is normally much longer than the guide used in this work (20 nts), already providing the buffer described by the authors. This spacer also undergoes trimming to form the mature crRNA. Second, the repeat length is normally fixed as a consequence of the mechanisms of spacer acquisition. At most, the beginning of each repeat sequence may have evolved to reduce folding interactions without changing the repeat length, although some of these repeats are predicted to fold into small hairpins.

      Prior literature has highlighted the importance of a folded hairpin with an upstream pseudoknot within the repeat (Yamano Cell 2016), where disrupting this structure compromises DNA targeting by Cas12a (Liao Nat Commun 2019, Creutzburg NAR 2020). This structure is likely central to the authors' findings and needs to be incorporated into the analyses.

      Many claims could better reflect the cited literature. For instance, Creutzburg et al. showed that adding secondary structures to the guide to promote folding of the repeat hairpin enhanced rather than interfered with targeting. Liu et al. NAR 2019 further showed that the pre-processed repeat actually enhanced rather than reduced performance compared to the processed repeat. Finally, the complete loss of targeting with the unprocessed repeat appears represent an extreme example given multiple studies that showed effective targeting with this repeat (e.g. Liu NAR 2019, Zetsche Nat Biotechnol 2016).

      Relating to the above point, the vast majority of the results relied on a single guide sequence targeting GFP. While the seven-guide CRISPR array did involve other sequences, only the same GFP targeting guide yielded strong gene activation. Therefore, the generalizability of the conclusions remains unclear.

    4. Evaluation Summary:

      This manuscript is of broad interest to those performing multiplexed genome engineering and related applications with CRISPR Cas12a technologies. While the proposed use of synSeparators is promising, the paper would benefit from further investigation of the mechanism by which synSeparators function to promote Cas12a activity. Additional data would be required to support the current conclusions regarding the generalizability of the findings.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

    1. Reviewer #3 (Public Review):

      This manuscript sheds new light on the regulation and function of a signaling network comprised of the adaptor molecules Cas and BCAR3. The data presented in the manuscript are generated through rigorous experimentation, frequently with the use of multiple approaches to arrive at the stated conclusions.

      Minor concerns:

      1) Figure 3e. The authors state that "SOCS6 binds BCAR3 and Cas independently" (bottom of page 7). However, while they show that the EE BCAR3 mutant binds to SOCS6 under conditions when it does not bind to Cas, they do not show the reciprocal interaction in this paper. Their previous paper (J Cell Sci 2014) suggests that SOCS6 binding to Cas may be independent of BCAR3 but neither that paper nor the current manuscript explicitly examine that. Unless there is direct evidence that SOCS6 can bind to Cas in the absence of BCAR3, perhaps it would be more accurate for the authors to limit their conclusion by saying that "SOCS6 binds to BCAR3 independently of Cas."

      2) Figure 8a and c. Without showing a Western blot to address total pools of phosphorylated Cas, it is not clear whether the depletion in pY165 is targeted to the pool of Cas present in adhesions or to a diminution in phosphorylation of the total pool of Cas in the cell. At a minimum, the authors would need to clarify that phosphorylation at Y165 of Cas in the pool of Cas that is localized to adhesions is reduced in the presence of Y117F, R177K, or the EE mutant of BCAR3.

    2. Reviewer #2 (Public Review):

      This is a well-written paper describing the co-recruitment of p117-BCAR3 and Cas to adhesion sites for activation of lamellipodial ruffling and the subsequent ubiquitin-dependent degradation. The completeness of the description of the cycle is a major success of this article and warrants publication. I didn't find major holes in their arguments and they did document that this pathway was not universal but there were possibly analogous signaling processes with other players.

    3. Reviewer #1 (Public Review):

      Summary: The study by Steenkiste focuses on the formation of adaptor protein complexes at sites of integrin receptor adhesion in the modulation of in vitro membrane ruffling and cell movement. The authors are studying the role of BCAR3 (also termed AND34 or NSP1) protein regulation by post-translational mechanisms (ubiquitin degradation and tyrosine phosphorylation). This is one of many adaptor proteins localized to adhesion sites. Studies are being performed on MCF10A or Hela cells to knockdown (siRNA) or over-express tagged protein constructs. By proteomics, a new phosphorylation site was identified (BCAR3 Y117). Mutagenesis showed that BCAR3 Y117 is important for enhancement of in vitro cell movement under conditions where the cullin-5 E3 ligase has also been reduced by siRNA expression.

      Opinion: The authors provide support for a "co-regulatory" model whereby the recruitment of BCAR3 to adhesions acts in part to modulate another adaptor protein tyrosine phosphorylation, p130Cas. This is associated with enhanced cell migration. The data presented are generally supportive of the conclusions and consistent with previous studies of BCAR3 and p130Cas. However, an unresolved issue is why cell phenotypes are dependent on cullin-5 knockdown or otherwise investigated by BCAR3 mutant over-expression. Cul5 loss can alter multiple aspects of cell signaling and the transient knockdown or inducible over-pression assays are a limited primary means of investigation. As multiple protein domains and post-translational modifications modulate the BCAR3-p130Cas complex, the authors did not establish a strong mechanistic linkage between newly-identified BCAR3 Y117 phosphorylation, SOCS6 binding, and a CUL5-dependent cell phenotype. Additionally, some of the experimental conditions (+/- EGF in growth media) are difficult to connect to EGF receptor activation and or signaling.

    4. Evaluation Summary:

      This study focuses on the formation of adaptor protein complexes at adhesion sites and their links to in vitro membrane ruffling and cell movement. Specifically, the authors study the role of the adaptor BCAR3 protein which is regulated by post-translational mechanisms (ubiquitin degradation and tyrosine phosphorylation). The authors propose a "co-regulatory" model whereby the recruitment of BCAR3 to adhesions acts to modulate p130Cas tyrosine phosphorylation and cell migration. This manuscript would be of particular interest to cell and cancer biologists interested in the molecular regulation of cell migration.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Reviewer #2 (Public Review):

      Thank you for the opportunity to review the short report "Regional sequencing collaboration reveals persistence of the T12 Vibrio cholerae O1 lineage in West Africa" by Ekeng and colleagues. The authors report an analysis of 46 new Vibrio cholerae genomes in context of 1280 published genomes. The goal of their analysis was to establish a recent snapshot of VC population genomics in West Africa and assess the occurrence importations of new lineages. From their analysis, they infer that the recent cases were endemic.

      Overall, this report presents findings from a region with little genomic surveillance, and as such these data are valuable for the understanding of endemic cholera in the region. The authors' analysis is technically sound, and the figures are well constructed. However, the depth of the analysis is relatively shallow, even for a short report, and the conclusions drawn from the data appear more subjective then based on the analysis at hand. These weaknesses could be addressed by a more in-depth analysis and clarification of the points below. Last, I did appreciate that this study was conducted in the context of a regional training. This could be an effective model for future analyses of regional importance. I feel like that wasn't the main focus of the report. If they were to shift their focus, I would want to know:

      1) Where did the isolates come from (e.g., cholera treatment centers, hospitals, or broader active surveillance)?

      2) Do they conduct environmental sampling and could this be part of future efforts?

      3) Who attended the training? Were they members from regional ministry of health labs, academic institutes etc?

      4) Were the attendees laboratorians, bioinformaticians, clinicians etc?

      5) Was there an effort to analyze the data, particularly the bioinformatics portion, locally or did the rely 100% on the collaborators at JHU? If the latter, then I don't think this is a good sustainable model for ongoing genomic epidemiology. If the prior, then were local or regional computing resources used? 6) Are they continuing to sequence isolates after the training?

    2. Reviewer #1 (Public Review):

      Ekeng et al. have sequenced and analyzed 46 Vibrio cholerae whole genome sequence data. The authors demonstrated a predominant lineage (T12) where all isolates from 2018-2019 fall. Their analysis suggest continuous transmission through repeated reintroduction of the same lineage back into the population. The work is interesting and the conclusions of this paper are mostly well supported by data. The present study reinforce the need of more genome sequence data and a strong surveillance network to interpret the data.

      This is a successful model of regional coordination to have genomic surveillance data from a region where surveillance data was inadequate. The manuscript should be modified to focus on strengthening of genomic surveillance further.

    3. Evaluation Summary:

      The paper "Regional sequencing collaboration reveals persistence of the T12 Vibrio cholerae O1 lineage in West Africa" presents results from sequencing and analyzing 46 Vibrio cholerae whole genome sequence data. The paper presents findings from a region with little genomic surveillance, and as such these data are valuable. While the analysis doesn't provide much novelty in terms of understanding cholera transmission, the study was conducted in the context of a regional training, and as such adds value as a potential model for regionally coordinated genomic surveillance efforts in areas where surveillance is limited. However even though it seems that the authors aim to present this as a surveillance model, the current focus of the paper is on the somewhat limited inference made about transmission.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

    1. Reviewer #3 (Public Review):

      This manuscript by Koiwai et al. described the single-cell RNA-seq analysis of shrimp hemocytes and was submitted as a Resource Paper in eLife. In this study, they identified 9 cell types in shrimp hemocytes based on their transcriptional profiles and identified markers for each subpopulation. They predicted different immune roles among these subpopulations from differentially expressed immune-related genes. They also identified cell growth factors that might play important roles in hemocyte differentiation. This study helps to understand the immune system of shrimp and maybe useful for improving the control of the pathogen infections. The analysis of the data and interpretation is overall good but there are also some concerns:

      1) The number of UMI and genes detected per cell after mapping to the in-house reference genome does not appear to be presented, and the similarities or differences between the three replicated samples are not discussed, as well as the low number of genes detected per cell (~300 in this study) .

      2) The correlation between the morphology and the expression of marker genes demonstrated in Figure 6 is questionable. Cells of the same size could express totally different genes. On the other hand, cells that are different in size can express nearly identical genes. The evidence presented in this manuscript is not enough to support a correlation between cell size and gene expression. Therefore, the author would either need to provide more evidence to support this correlation, or not make such correlation.

      3) There are many spindle-shaped cells in Figure 6B, but none of them appeared in Figure 6C and D after sorting, and the reason for this is unclear.

      4) The hemocyte differentiation model in Figure 7 is not supported by any experimental data.

    2. Reviewer #2 (Public Review):

      In this manuscript Koiwai et al. used single cell RNA sequencing of hemocytes from the shrimp Marsupenaeus japonicus. Due to lack of complete genome information for this species, they first did a de novo assembly of transcript data from shrimp hemocytes, and then used this as reference to map the scRNA results. Based on expression of the 3000 most variable genes, and a subsequent cluster analysis, nine different subpopulations of hemocytes were identified, named as Hem1-Hem9. They used the Seurat marker tool to find in total 40 cluster specific marker transcripts for all cluster except for Hem6. Based upon the predicted markers the authors suggested Hem1 and Hem2 to be immature hemocytes. In order to determine differentiation lineages they then used known cell-cycle markers from Drosophila melanogaster and could confirm Hem1 as hemocyte precursors. While genes involved in the cell cycle could be used to identify hemocyte precursors, the authors concluded that immune related genes from the fly was not possible to use to determine functions or different lineages of hemocytes in the shrimp. This is an important (and known) fact, since it is often taught that the fruit fly can be used as a general model organism for invertebrate immunologists which obviously is not the case. Even among arthropods, animals are different. The authors suggest four lineages based upon a pseudo temporal analysis using the Drosophila cell-cycle genes and other proliferation-related genes. Further, they used growth factor genes and immune related genes and could nicely map these into different clusters and thereby in a way validating the nine subpopulations. This paper will provide a good framework to detect and analyze immune responses in shrimp and other crustaceans in a more detailed way.

      Strengths:

      The determination of nine classes of hemocytes will enable much more detailed studies in the future about immune responses, which so far have been performed using expression analysis in mixed cell populations. This paper will give scientists a tool to understand differential cell response upon an injury or pathogen infection. The subdivision into nine hemocyte populations is carefully done using several sets of markers and the conclusions are on the whole well supported by the data.

      Weaknesses:

      One obvious drawback of the paper is first the low number of UMIs. A total number of 2704 cells gave a median UMI as low as 718 which is very low. Especially shrimp no. 2 has an average far below 500 and should perhaps be omitted. Therefore, one question is about cell viability prior to the drop-seq analysis. The fact of this low number of UMIs should be discussed more thoroughly.

      Details about how quality control (QC) was performed would be needed, for example the cutoff values for number of UMI per cell, and also one important information showing the quality is the proportion of mitochondrial genes. The clustering into nine subpopulations seems solid, however the determination of lineages based upon the pseudo time analysis with cell-cycle related genes is not that strong. The authors identify four lineages, all starting from hem1 via hem2-Hem3- Hem4 and then one to Hem5, another through part of Hem 6 to Hem 7, next through part of Hem 6 to Hem 8 and finally through part of Hem 6 to Hem 9. Referring to Figure 3 - supplement 3, it seems as if Hem6 could be subdivided into two clusters, one visible in B and C, while another part of Hem & is added in D. Also, the data in figure 3 - supplement 1 showing expression of cell cycle markers do not convincingly show the lineages. Cluster Hem 3 and 4 seems to express much fewer and lower amount of these markers compared to cluster Hem6 - Hem9.

      It is also clear (from figure 5 - supplement 1) that there are more than one TGase gene and the authors would need to discuss that fact related to differentiation.

      While the part to determine subpopulations is very strong, the part about FACS analysis and qRT-PCR is weaker than the other sections, and doesn't add so much information. Validation of marker genes and the relationship between clusters and morphology shown in figure 6 is not totally convincing. It seems clear that both R1 and R2 contains a mixture of different cell types even if TGase expression is a bit higher in R1. A better way to confirm the results could be to do in situ hybridization (or antibody staining) and show the cell morphology of some selected marker proteins in a mixed hemocyte population. FACS sorting is very crude and does not really separate the shrimp hemocytes in clear groups based on granularity and size. This may be because the size of hemocytes without granules vary a lot. You need cell surface markers to do a good sorting by FACS. Another minor issue is the discussion about KPI. There are a huge number of Kazal-type proteinase inhibitors in crustaceans and it is not clear from this data if the authors discuss a specific KPI-gene, and there is a mistake in referring to reference 65 which is about a Kunitz-type inhibitor.

      In summary, this paper is a very important contribution to crustacean immunology, and although a bit weak in lineage determination it will be of extremely high value.

    3. Reviewer #1 (Public Review):

      Summary and Strength:

      Single-cell RNA sequencing is the most appropriate technique to profile unknown cell types and Koiwai et al. made good use of the suitable tool to understand the heterogeneity of shrimp hemocyte populations. The authors profiled single-cell transcriptomes of shrimp hemocytes and revealed nine subtypes of hemocytes. Each cluster recognizes several markers, and the authors found that Hem1 and Hem2 are likely immature hemocytes while Hem5 to Hem9 would play a role in immune responses. Moreover, pseudotime trajectory analysis discovered that hemocytes differentiate from a single subpopulation to four hemocyte populations, indicating active hematopoiesis in the crustacean. The authors explored cell growth- and immune-related genes in each cluster and suggested putative functions of each hemocyte subtype. Lastly, scRNA-seq results were further validated by in vivo analysis and identified biological differences between agranulocytes and granulocytes. Overall, conclusions are well-supported by data and hemocyte classifications were carefully performed. Given the importance of aquaculture in both biology and industry, this study will be an extremely useful reference for crustacean hematopoiesis and immunity. Moreover, it will be a good example and prototype for cell-type analysis in non-model organisms.

      Weaknesses:

      The conclusions of this paper are mostly well supported by data, but some aspects of data analysis QC and in vivo lineage validation need to be clarified.

      1) It is not a trivial task to perform genome-wide analyses of gene expression on species without sufficient reference genome/transcriptome maps. With this respect, the authors should have de novo assembled a transcriptome map with a careful curation of the resulting transfrags. One of the weaknesses of this study is the lack of proper evaluation for the assembly results. To reassure the results, the authors would need to first assess their de novo transcripts in detail and additional data QC analysis would help substantiate the validity.

      2) The authors applied SCTransform to adjust batch effects and to integrate independent sequencing libraries. SCTransform performs well in general; however, the authors would need to present results on how batch effects were corrected along with before and after analysis. In addition, the authors would need to check if any cluster was primarily originated from a single library, which could be indicative of library-specific bias (or batch effects).

      3) Hem6 cells lack specific markers and some cells in this cluster are scattered throughout the other clusters (Fig. 1 & 2). Based on the pattern, it is possible that these cells are continuous subsets of other clusters. It would be good if the authors could group these cells with Hem7 or other clusters based on transcriptomic similarities or by changing clustering resolution. Additionally, they may also be a result of doublets, and it is unclear whether doublets were removed. Hem6 cells require additional measures to fully categorize as a unique subset.

      4) The authors took advantage of FACS sorting, qRT-PCR, and microscopic observation to verify in silico analyses and defined R1 and R2 populations. While the experiments are appropriate to delineate differences between the two populations, it is not sufficient to determine agranulocytes as a premature population (Hem1-4) and granulocytes as differentiated subsets (Hem5-9). To better understand the two groups (ideally nine subtypes), additional in vivo experiments would be essential. For example, proliferation markers (BrdU or EdU) could be examined after FACS sorting R1 and R2 cells to show R1 cells (immature hemocytes) are indeed proliferating as indicated in the analyses.

      5) FACS-sorted R1 or R2 population does not look homogeneous based on the morphology and having two subgroups under nine hemocyte subtypes may not be the most appropriate way to validate the data. The better way to prove each subtype is to use in situ hybridization to validate marker gene expressions and match with morphology.

    4. Evaluation Summary:

      This study provides identification of different subpopulations of blood cells and gives new insights in putative hemocyte lineage relationships by single cell RNA sequencing. The main conclusions are fairly well supported by the data and this manuscript will be of high interest to crustacean immunologists and readers in the field of aquaculture.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #3 (Public Review):

      Moncla et al. investigated the transmission of mumps virus in Washington, USA during an outbreak in 2016-2017. They sequenced viral genomes from infected individuals in Washington and elsewhere within the United States and used phylogenetic approaches to understand the origins and patterns of spread exhibited by the virus during the outbreak. They observe a large fraction of cases in individuals who are part of the Marshallese community, and identify a link to a similar outbreak in the Marshallese community in Arkansas. They develop a method for determining the role of the Marshallese community in the Washington outbreak that is robust to sampling bias and size. This method is well thought-out and presented and demonstrates that the outbreak in Washington state was sustained by transmission within this particular community. This paper provides a thoughtful approach to dealing with sampling issues that are often overlooked in phylogenetic studies. By consulting with a public health professional from within the affected (Marshallese) community, the authors are able to contextualize their results and demonstrate the underlying issues that may have contributed to mumps spread within the state.

      Working with public health advocates from affected communities is exceptionally important for long term public health impact, and this paper sets an example that should be followed by others in the pathogen genomics field. The methodology used to determine mumps transmission patterns in Washington is sound and the conclusions are well explained. However, some additional context on the issues and potential pitfalls of source-sink analyses based on phylogenetic inference would help improve this already solid paper. Specifically:

      1) The authors seem to assume a somewhat random sample throughout Washington state. They state that given a low sampling proportion they do not expect to have captured infection pairs, which seems reasonable. However, they then go onto assume that their sample is primarily comprised of samples from long, successful transmission chains. This is a reasonable assumption if there is no major difference in accessibility of samples from long transmission chains and shorter ones (for example, decreased access to healthcare). Could this impact the assumption of sampling primarily from long transmission chains? It seems from the data collected in this outbreak that this was not the case for mumps in Washington but addressing this assumption clearly (and potential ways to interrogate it) could make their methodology more applicable to other pathogen studies.

      2) There are many examples of phylogenetic analyses that have led to conclusions about pathogen sources and sinks that were later shown to be wrong because of oversampling or other sampling biases. The authors address unequal sampling between clades, but additional contextualization of the problem and how this approach is different may help strengthen the methodology presented in the paper.

      3) The authors present compelling evidence that the mumps outbreak in Washington state was sustained by the Marshallese community, and state that mumps did not transmit efficiently among the general Washington populace. That said, there were several other mumps outbreaks in the United States in the same 2016-2017 time period. Was there something different about Washington state that prevented mumps transmission outside of the Marshallese community? Were there no other close-knit communities (universities, prisons, other cultural communities, etc.) affected? It just seems surprising that the Marshallese community was the only community sustaining transmission at a time where many different types of communities were affected across the United States.

    2. Reviewer #2 (Public Review):

      In this manuscript, Moncla et al. undertake a large sequencing and phylogenetic study to investigate the underlying epidemiology of the 2016-2017 Washington State Mumps epidemic. The authors generate 110 sequences and include 166 novel sequences in their analysis. This data set represents over a quarter of the publicly available Mumps genomes from North America.

      They then apply a mixture of phylogenetic methods and intuitive data analyses to uncover, that i) Mumps was imported into Washington at least 13 times. ii) A disproportionate amount of transmission occurred in the Marshallese community in WA with limited transmission in the non-Marshallese community. iii) These heterologous transmission dynamics might be explained by historical and current health disparities within the community, but are not due to low vaccination coverage.

      These conclusions are supported by a wide array of carefully controlled phylogenetic methods. The authors explore the sensitivity of their findings to sampling bias. Additionally, the conclusion that transmission occurred disproportionally within the Marshallese community is supported by multiple implementations of the structured coalescent as well as, more coarse but intuitive methods such as the rarefaction analysis and the "descendent" analysis in Figure 4. The "descendent" analysis complements the structured coalescent models and highlights how tips that are close to internal nodes inform the "state" of those unsampled ancestors. Each internal node represents an unsampled ancestor, and if transmission rates are higher in one population, then samples from that population are more likely to be close to those ancestors. The approach captures these processes; however, calling downstream tips "descendants" is unfortunate, as it is unknown if the tips that have "descendants" are direct ancestors of their "descendants" in the transmission chain. Inferring transmission dynamics from divergence trees is difficult, and variants of this approach are likely to be useful in other systems.

      The finding that transmission disproportionally occurred in the Marshallese community leads the authors to propose several possibilities for why this may be. The authors should be commended for reaching out to Marshallese health advocates in this process and including the community in their study. This context is a major strength of the study.

      Both the data generation and data analysis are achievements that advance our understanding of the epidemiology of Mumps. As can be seen in the tree in Figure 1 the 2016-2017 epidemic in North America was seeded by at least two divergent lineages that appear to have all contributed to the same outbreak. The large number of sequences contributed by this study will help future work uncover the dynamics that drive Mumps epidemics at larger scales. The findings also highlight how large outbreaks can persist in highly vaccinated populations and how an array of phylogenetic approaches can be employed to uncover the underlying population heterogeneity behind an outbreak. To have both of these achievements in the same manuscript sets this work apart.

    3. Reviewer #1 (Public Review):

      In this study, Moncla et al. used genomic data to analyse a mumps outbreak in Washington, in order to draw inferences about the epidemiological factors driving the outbreak. Some important strengths of the analysis include sophisticated sequencing and modeling techniques to reconstruct chains of transmission during the outbreak, which support the conclusions that the mumps virus was introduced several times in Washington from other North American regions during the outbreak, and that the Washington Marshallese community was particularly at risk of mumps infection and transmission during this time. Limitations of the analysis include potential for sampling bias, where the sample may not be entirely representative of mumps outbreak cases, and a sample size that is too low to allow sufficient statistical power to assess the impacts of age and vaccination status on transmission. The work has potential public health impacts in terms of identification of a vulnerable community and points to social networks as the primary risk factor for potential future respiratory virus outbreaks. The analysis methods could be potentially applied for the phylodynamic analysis of other infectious disease outbreaks.

    4. Evaluation Summary:

      This interesting phylogenetic analysis of a mumps outbreak in Washington will be of interest to a wide audience, especially those working at the intersection of pathogen genomics and public health. An array of classic and novel phylogenetic approaches supports the conclusions that mumps was introduced several times in Washington during the outbreak, and that the Washington Marshallese community was particularly at risk of mumps infection and transmission despite high vaccination coverage. Inferences regarding the role of age and vaccination status are however less conclusive given the small sample size. Consultation with a community health advocate from the affected communities helps contextualize the results.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #2 (Public Review):

      In this incredibly detailed effort, Hulse, Haberkern, Franconville, Turner-Evans, and coauthors painstakingly and patiently reveal the connectivity of central complex neurons within one "hemibrain" EM-imaged connectome of a fruit fly. This is best read as one of a series of such detailed papers including Scheffer et al., 2020 (which introduces the dataset) and Li et al., 2020 (which focuses on the mushroom body).

      The authors achieve two major goals. First, they present a full account of all neurons (by type) present in the central complex and the connections between them (including to and from regions outside the central complex). By necessity, this work only examines such connections within a single animal from whose brain the hemibrain volume was imaged. Nonetheless, the relatively conserved morphology of fly neurons (at the scale of which regions they form arbors within) allows the authors to confidently relate their neurons to known examples from genetically labeled lines imaged at the light level. (And in some cases, they are able to show that some neurons with similar morphology can then be further subdivided into different types on the basis of their connectivity). Importantly, the hemibrain dataset contains both sides of the central complex, allowing for a complete analysis.

      Secondly, the authors contextualize the observed connectivity patterns within the known functions of the central complex (particularly navigation and sleep/arousal). Appropriately and importantly, they offer detailed explanations for how the circuitry observed can support these functions. In some cases, particularly in their discussion of the fan body, they show how the connectivity patterns can support multiple variations of models of path integration (and more broadly how its architecture supports vector computation in general). These analyses make their central complex connectome a useful map - there is little doubt that it will inspire many future experiments in the fly community.

      The only limitations of this work are rooted in the nature of the source material: it's only one animal's brain and because it's EM-based there's often no way to know whether a given cell type (if new) is even excitatory or inhibitory (though, notably, the authors take care to note where this is the case and to offer alternate interpretations of the circuit function). Synaptic strength is another relative unknown (not to mention plasticity rules or modulatory influences). For EM-based connectomes, the number of synapses made between two neurons is considered the basis for determining whether or not they are meaningfully connected. However, this precise number can vary as a function of how complete the reconstructions are (generally, as proofreading progresses, more synapses are found). This work improves on prior hemibrain studies by carefully demonstrating that it is possible to set a threshold on the relative fraction of synaptic contributions within a region in order to identify meaningful connections. (That is, they find that as the number of synapses discovered increases, the relative contribution remains relatively constant).

      This is a massive work. There are 75 figures, not including supplements, and numerous region and neuron names to keep track of (not to mention visualize). It is impossible to read in a single sitting. So for the purposes of this public review, I highly recommend to any reader that they first find the region of the paper they're interested in and skip to that to view in side-by-side mode. The "generally interested" reader is best served by reading through the Discussion, which has more of the structure-function analyses in it and then referring to the Results as their curiosity warrants.

      Scheffer et al., 2020 is available here: https://elifesciences.org/articles/57443#content Li et al., 2020 is available here: https://elifesciences.org/articles/62576#content

    2. Reviewer #1 (Public Review):

      It is difficult to overestimate the importance of this paper. The full connectome of the Drosophila central complex is both the beginning and the end of an era. It provides the first comprehensive dataset of arguably the most enigmatic brain region in the insect brain. This endeavor has generated ground truth data for years of functional work on the neural circuits the connectome outlines, and constitutes an unparalleled foundation for exploring the structure function relations in nervous systems in general. This will be of great importance far beyond work on the Drosophila brain, and will have far reaching implications for comparative research on insect brains and likely also smoothen the path toward understanding navigation circuits in vertebrate nervous systems. Based on presented data, the paper develops overarching ideas (at exquisite detail) of how sensory information is transformed into head direction signals, how these signals are used to enable goal direction behavior, how goals are represented, and how internal state can modulate these processes. The connectome enables the authors to base these ideas and their detailed models on actual biological data, where earlier work was forced to indirectly infer or speculate. While significantly going beyond models of central-complex function that existed previously, the authors have to be much credited for incorporating huge amounts of existing knowledge and data into their interpretations, not only work from Drosophila, but also from many other insects. This makes this paper not only an invaluable resource on the connectome of the Drosophila central complex, but also a most comprehensive review on the current state of the art in central-complex research. This unifying approach of the paper clearly marks a reset of central-complex research, essentially providing a starting point of hundreds of new lines of enquiry, probably for decades to come.

      Given the type and amount of data presented, the paper is clearly overwhelming. That said, it also clearly needs to be presented in the way it was done, mostly because no single aspect of the function of this neuropil makes as much sense in isolation as it makes sense when viewed in conjunction of all its other functions. The complexity of the neural circuits discussed is clearly reflected in the enormous scope of the paper. Nevertheless, the authors have done a fantastic job in breaking the circuits and their function down into digestible bits. The manuscript is very systematic in its approach and starts with sensory pathways leading to the CX, covering the clearly delineated head direction circuits and then moving on to the more complex and less understood parts, always maintaining a clear link between structure and function. As function is necessarily based on previous work, including that from other species, the results part is interwoven with interpretation, but this is clearly necessary to keep the text readable. The authors have made considerable efforts to provide additional introductions and summaries whenever needed, almost creating nested papers embedded within the overall paper.

      The figures are equally overwhelming as the text at first sight, but when taking the time to digest each one in detail, they present the data in a rich and clear manner. The figures are often encyclopedic and will serve as reference about the central complex for years. The summary graphs that are presented in regular intervals are welcome resting places for the reader, helping to digest all the detailed information that has preceded or that will follow.

      The analysis performed in the paper is excellent, comprehensive and should set the standard for any future work on this topic. Also, the text is very honest about the limits of the conclusions that can be reached based on this kind of data, which is important in generating realistic and feasible hypotheses for future experiments.

    3. Evaluation Summary:

      It is difficult to overestimate the importance of this paper. The full connectome of the Drosophila central complex is both the beginning and the end of an era. It provides the first comprehensive dataset of arguably the most enigmatic brain region in the insect brain. This endeavor has generated ground truth data for years of functional work on the neural circuits the connectome outlines and constitutes an unparalleled foundation for exploring the structure function relations in nervous systems in general. While significantly going beyond models of central-complex function that existed previously, the authors have to be much credited for incorporating huge amounts of existing knowledge and data into their interpretations, not only work from Drosophila, but also from many other insects. This effort makes this paper not only an invaluable resource on the connectome of the Drosophila central complex, but also a most comprehensive review on the current state of the art in central-complex research. This unifying approach of the paper clearly marks a reset of central-complex research, essentially providing a starting point of hundreds of new lines of enquiry, probably for decades to come.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

    1. Reviewer #3 (Public Review):

      The authors investigated pupillary response looking at the changes corresponding to perceptual events (spontaneous or physical changes) and contrasting them with requirements of over reporting (changes were reported or ignored). They demonstrate that the former is associated with a rapid constriction and re-dilation, whereas the latter shows an opposite effect with dilation being followed by re-constriction. The particular strength of the work is in no-report conditions using on OKN-based inference about timing of perceptual events that allowed for this dissociation to be observed, whereas manual report conditions allowed for a direct comparison with prior work. The analysis and control experiments are very thorough showing that reported results are unlikely to be explained other factors such as saccades or blinks.

      The study makes a significant contribution but proposing a no-report paradigm for identifying perceptual events that should work for any multistable display. The fairly rapid pupil constriction event could provide an easy to detect and temporally reliable marker of perceptual switches, expanding ways the multistability data is collected. The same approach could also be useful for no-report studies of visual awareness in general.

      The ability to decompose pupillary response into two components - perception and over manual response - will also be useful for studying neural correlates of spontaneous perceptual switches, as it could help to better understand switch-time activity in various frontal and parietal regions. Here, also some regions are associated with active response, whereas other with perception, distinction that could be potentially better understood based on the idea that only the former involves noradrenaline-affected processing. My main worry methodologically is the under and overestimation of mean switch rate via OKN (figure 1C). OKN estimates are all within .4-.8 range, whereas for self-report rates differ from 0.2 to over 1. Further analysis would be helpful. I think it would be helpful if the authors elaborated on what kind of switches went unreported (or, conversely, what kind of events led to false alarms): switches before very short dominance phases (could be to fast to report via key presses), to return transitions, etc.

    2. Reviewer #2 (Public Review):

      Pupillometry is an increasingly accessible tool for the non-invasive readout of brain activity. However, our understanding of pupil-control circuits and of the relationship between changes in pupil size and perception, cognition or action, is far from complete. Therefore, any measurements that further this understanding are of great interest to a wide audience in the field of psychology and neurobiology.

      This study used pupillometry to explore the neural processing that underlie perception and dissociate those from action-related neural processing. The authors use a novel and comprehensive task design, centered on binocular rivlary, that is likely to find wider use among researchers studying the neural processes that underlie perception and action. They used a non-invasive method (pupillometry) to disscociate putative processes and circuits that might drive perceptual switching. They found changes in pupil size that are reliably different depending on the task: for example - between the conditions that require reporting a perceptual switch versus not reporting it and between rivalrous and explicit changes in the visual stimulus.

      Such approaches can be very useful in deciphering which of the myriad factors that can affect pupil size are in fact active under specific, controlled conditions and thus provide a basis for guided, direct measurements of these specific brain regions.

      Overall, this study is well-conceived and executed. However, I have some questions and concerns about the analyses and conclusions made from the results shown. In general, I would encourage the authors to try and include more of what we do know about neuromodulation and the cortical control of pupil pathways to frame the hypothesis and interpret the results. Further, it is unclear to me whether the constriction/dilation dissociation is tenable with the presented data and analyses.

    3. Reviewer #1 (Public Review):

      Brascamp and colleagues address pupil-size changes around perceptual switches in perceptual multistability. Several previous studies have found pupil dilation around or after the switch and some have found pupil constriction, though the latter was typically less robust. Moreover, while most previous studies included some controls for the effect of reporting and for the physical stimulus change, to my knowledge, so far, no study has fully crossed the factors report/no-report and endogenous/exogeneous switch. In the present study, this gap is filled using a binocular-rivalry stimulus and an OKN-based no-report paradigm. This allows the authors to isolate the constriction component from the dilation component and interestingly they find the constriction more robustly tied to the perceptual switch, while the dilation component is mostly related to the response. Experiments are soundly conducted and analysed and results are interpreted with appropriate care. Since the results challenge frequent interpretations as to why perceptual switches in multistability may cause pupil-size changes, the paper is of high relevance to the fields of pupillometry and multistability, but also to other areas where pupillometry is used as index of perceptual and cognitive processes. I only have some minor questions and requests for clarification with regard to result presentation and interpretation.

    4. Evaluation Summary:

      Pupillometry is an increasingly accessible tool for the non-invasive readout of brain activity. However, our understanding of pupil-control circuits and of the relationship between changes in pupil size and perception, cognition or action is incomplete. Therefore, any measurements that further this understanding are of great interest to a wide audience in psychology and neurobiology. This study used pupillometry to explore the neural processing underlying perception and dissociate them from action-related neural processing. Results reveal changes in pupil size that are reliably different depending on the task. Such approaches can be very useful in deciphering which of the myriad factors that can affect pupil size are active under specific, controlled conditions.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer 3 agreed to share their names with the authors.)

    1. Reviewer #4 (Public Review):

      Using a transgenic line of Platynereis, in which GFP is expressed under the control of cis-regulatory elements for r-opsin, the study isolates r-opsin expressing cells from the head (eye photoreceptors) and trunk region (a population of segmentally repeated r-opsin expressing cells associated with the parapodia) by FACS. Subsequent RNA-Seq establishes that both populations of cells express genes for all components of the rhabdomeric phototransduction cascade, while the population of trunk sensory cells additionally expresses genes encoding proteins involved in mechanosensation. Using heterologous expression in a mammalian cell line, it is shown that the Platynereis r-opsin responds to blue light via coupling to Gαq suggesting that it mediates photoresponses via a canonical rhabdomeric phototransduction cascade. Transcriptomic analysis of an r-opsin mutant created by TALEN mediated gene editing then reveals that expression levels of the mechanosensory Atp2b channel are modulated by protracted exposure to blue light, a response abolished in the mutant. Behavioral analysis further suggests that undulatory movements of the worms are equally altered under these illumination conditions. Taken together this suggests that the r-opsin expressing trunk sensory cells act as both photo- and mechanoreceptors and that their mechanosensory properties are modulated in response to light. In combining the transcriptomic analysis of cell types with experimental studies of gene function and behavioral analyses, this study provides exciting new insights into the evolution of sensory cells. Several prior studies have found co-expression of photosensory and mechanosensory proteins in sensory cells of various bilaterians, and comparative studies suggested that photo- and mechanosensory cells may share a common evolutionary origin. However, the current study goes far beyond these findings in establishing a direct functional link between photo-and mechanosensation in a population of sensory cells suggesting that these sensory cells function as multimodal cells and that their mechanosensory properties are altered in response to light. Furthermore, the behavioral data (based on a novel machine-learning based tool of analysing the animals' movement) suggest that these cells have a behaviorally relevant function. Because r-opsin was found to be expressed in mechanoreceptors not only in lophotrochozoans (including Platynereis) but also in ecdysozoans and vertebrates (although functional studies are lacking here) and r-opsins belong to a large family of opsins, almost all of which are responsive to light, the present study suggests that r-opsins may have an ancestral bilaterian role in modulating mechanosensory function in response to light (in addition to their purely photosensory role in the photoreceptors of the eyes). Light-independent functions of r-opsin as recently revealed in Drosophila may, thus, be secondarily derived.

      The study is very carefully conducted and well presented. The only minor flaw is that in its present form, the discussion of the evolutionary implications of the finding lacks in clarity and specificity. The authors here often refer ambiguously to an "ancient" or "ancestral" role of r-opsins without specifying the lineage referred to (ancestral for lophotrochozoans? bilaterians? eumetazoans? metazoans?). The discussion should, therefore be revised with an explicit phylogenetic framework in mind.

    2. Reviewer #3 (Public Review):

      Opsin proteins are ancient light-sensitive molecules found in photoreceptor cells throughout the animal kingdom. Recent discoveries including those made in the current paper have revealed that besides r-opsins, some classes of photoreceptor cell also express genes that are found in mechanosensory cells, and that r-opsins have both light-dependent and light-independent effects on mechanical force transduction or motion. A question remains as to whether or not: 1) a protosensory cell of animals existed which contained both photoreceptor and mechanoreceptor-like features and, 2) whether the original function of opsin included light-dependent mechanosensory features? The authors consider three competing hypotheses for the cellular evolution of photoreceptor and mechanosensory function. Two of the hypotheses envision either photo- or mechanosensory function for opsins evolving first, the third imagines them evolving simultaneously. The authors note that the majority of what we know about rhabdomeric opsins comes from studying the eye photoreceptors of the fruit fly, Drosophila melanogaster. But might this kind of photoreceptor have functions that are derived compared to the ancestral photoreceptor cell? To investigate this question, the authors turn to the non-model system, Platynereis dumerilii, which has both head and non-head photoreceptors. Here the authors use 1) a fluorescent cell sorting method to perform RNA profiling of eye and trunk photoreceptor cells of a mutant marine worm and find evidence of co-expression of photo- and mechanosensory genes in photoreceptor cells. They also compare the genes that are expressed in Platyneris photoreceptors with genes expressed in Drosophila JO (hearing organ in flies), Zebrafish lateral lines and mouse IEH (inner ear hair) cells, and again they find some commonly-expressed genes. 2) The authors use cell culture to express the opsin, demonstrate that it interacts with G-alphaq, and that it's peak sensitivity is in the blue range. 3) They use in situ hybridization to validate the RNA-seq and detect select enriched transcripts in the photoreceptor tissues. 4) They use a new method, which should be widely useful to other researchers, to detect undulation behavior of the opsin mutant vs. wildtype worms and show that the mutant worm behavior is perturbed in altered light cycles. Taken together, the authors suggest that an ancient light-dependent function of opsin was linked to mechanosensation and that light-independent mechanosensory functions of opsins evolved secondarily. The interpretation is somewhat reasonable given the available data but does not yet entirely rule out other possibilities (see below).

      This paper is a tour-de-force and a really impressive collection of experiments which examines the function of r-opsin in Platyneris. There's lots of innovation here from the use of fluorescent cell sorting and cell-specific RNA-Seq on a non-model system to the deep-learning based approach to examining behavior. Overall, the authors' interpretation of their data seems reasonable however I do believe a even stronger case could be made that what we are talking about is shared ancestry vs. recent recruitment if the authors made phylogenetic trees of the numerous TRE genes that are enriched between Drosophila JO and mouse IEH cells. If a significant number of these genes were true orthologs vs. paralogs across all three species then this would provide stronger evidence of an ancient light-dependent mechanosensory function for r-opsin. GO enrichment terms, while intriguing and suggestive, don't go far enough into the weeds. Also, I think the estimate of there being only 12 genes involved in making a photoreceptor cell able to detect light is probably an underestimate, as this ignores, for example, the understudied molecular machinery required for chromophore metabolism and transport. At the very least, the work should help inspire vigorous debate between vision and auditory neuroscience communities (which do not usually converse with one another) to more carefully consider the ways in which their systems overlap and why.

    3. Reviewer #2 (Public Review):

      Rhabdomeric Opsins (r-Opsins) are well known for their role in photon detection by photosensory cells which are commonly found in eyes. However, r-Opsin expression has also been detected in non-photosensory cells (e.g., mechanosensors), but their function(s) in these other sensory cells is less well understood. To explore the function of r-Opsins outside the context of an eye/head (non-cephalic function) as well as to investigate the potential evolutionary path by which sensory systems that rely on r-Opsins have evolved, Revilla-i-Domingo et al. have investigated gene expression in two distinct subsets of r-Opsin expressing cells in the marine bristle worm Platynereis dumerilii : EP (eye photoreceptor) and TRE (trunk r-opsin1 expressing) cells. The authors also generate two Pdu-r-Opsin1 mutant strains in order to investigate how the loss of r-Opsin function affects gene expression and behavior.

      The question of what role r-Opsins play outside of photoreceptors is an interesting one that remains poorly understood. In this manuscript, the authors demonstrate a powerful protocol for FACS sorting and sequencing different cell populations from an important evolutionary model organism.

      The transcriptomic analysis presented here demonstrates that both the cephalic EP cells and the non-cephalic TRE cells express components of the photosensory transduction pathway. This observation, together with heterologous cell expression data presented demonstrating sensitivity of Pdu-r-Opsin1 to blue light, suggests that both EP and TRE cells are likely to be light sensitive. The authors also suggest that they observe "mechanosensory signatures" in the transcriptomes, which, together with the analysis of undulatory movements in headless animals, lead them to suggst that r-Opsin in TRE cells functions as an evolutionarily conserved light-dependent modulator of mechanosensation, a conclusion that is not well-supported by the data presented.

      Overall, many of the conclusions drawn from the transcriptome data are inferential and based on weak evidence. Key limitations are listed below:

      1) The apparent overlap between the phototransduction and mechanosensory systems has already been shown (in Drosophila for instance) and the current work adds limited information to this story, and what is added is weakened by the absence of functional and physiological analyses. This is particularly true for supporting the claims of mechanosensory signatures in these cells. For example, genes whose expression is suggested in the text as being indicative of a mechanosensory function (glass and waterwitch) are, in fact, expressed in multiple sensory cell types. Glass (gl) is a transcription factor best known for regulating the expression of phototransduction proteins in photoreceptors. The function of waterwitch (wtrw) is not fully understood, but it is broadly expressed in sensory cells in Drosophila. It would be more compelling if mechanotransduction channels like Piezo and NompC were expressed in the TREs, but there is no mention of this.

      2) The suggestion that the TRE cells share similarity with the mechanosensitive mammalian inner ear is provocative, but lacks strong support. For instance, physiological characterization of the response properties of these sensory cells or identification of anatomical similarities analogous to the stereocilia upon which hair cell mechanosensitivity is based would greatly increase plausibility of this claim. Particularly for a species that diverged from mice and flies many hundreds of millions of years ago, speculation based largely on transcriptome analysis is risky. Careful validation is required as identified genes might not share a conserved function with their assigned orthologs in mice and Drosophila.

      3) The current analysis lacks sufficient power to make compelling claims with regard to potential ancestral protosensory cells. The investigators are examining a single species of marine worm and doing so without detailed anatomical and functional studies of the r-Opsin-expressing cells in the worm.

      4) The behavioral experiments require more functional data to interpret unambiguously. The data indicate that r-opsin1 is required for light to surpress the undulation of decapitated worms. Does this mean that the TREs are photosensors whose activity inhibits locomotion or that the TREs are light-sensitive mechanosensors ?

      5) It is assumed that the TREs constitute a homogenous cell population, but this is not demonstrated. This means that the TREs could be a mixed population (for example, distinct sets of photosensors and mechanosensors) and some of the TRE-expressed genes identified could be expressed in different specific subset of TREs.

    4. Reviewer #1 (Public Review):

      Strengths and Weaknesses. The authors did quite a lot to establish gene expression and function of the annelid's trunk cells and compare them to photoreceptors of the annelid's eye. They isolated the cells with FACS and characterized gene expression in detail, they knocked down r-opsin with TALEN in the trunk and found a significant difference in a crawling response, and they express the opsin in cell culture to confirm wavelength and G-protein sensitivity. As a potential link between light sensitivity and mechano-sensitivity, they report r-opsin function and light intensity influence expression of atp2b2, a gene that modulates neuronal sensitivity in other organisms. Wavelength and G-protein activation data are valuable because I can think of few or no other organisms in the entire group of lophotrochozoan animals, where this level of experimental manipulation could be done. In short, a strength of this manuscript is the detailed characterization of the trunk receptor cells, which express r-opsins. The authors have brought much evidence to the claim that these TRE cells have both light and mechano-sensitive gene expression and function. Based on these findings in an annelid worm, I believe the paper is a significant advance, and of interest to a broad audience by adding to a growing set of discoveries of similar hybrid sensory cells.

      If a hybrid mechano/photo-receptor is indeed an ancient cell type in bilaterians, this would bring many evolutionary implications for sensory biology. However, in these evolutionary interpretations is where I find a weakness of the manuscript. Namely, with only a handful of species shown thus far to have the hybrid cell type - and many differences in detail about these cell types in different organisms - we can not yet make firm conclusions about whether the multi-functional cells were ancestral. I believe other interpretations are equally valid (and still interesting) and should be given more consideration. Namely, it seems possible that photo- and mechan- sensory processes "joined forces" (e.g. through separate co-option events) in new cell-types, multiple times during evolution. The current manuscript loosely indicates ancestral multi-functionality is more parsimonious. However, no detail is given about that. I suppose the authors mean a single origin of hybrid cell types requires fewer evolutionary transitions than multiple origins. However, such a parsimony count does not count the transitions requiring loss of phototransduction in mouse hearing and do not count transitions to loss of mechanosensitivity in eye photoreceptor cells.

    5. Evaluation Summary:

      This manuscript presents an investigation of receptors in the trunk of Platynereis annelids that express genes involved in both photoreception (e.g. r-Opsin) and mechanosensation. This is particularly interesting in light of other work in model organisms like flies that uncovered broadly similar results. The authors compare gene expression of fly Johnston Organ cells and mouse hearing cells to the worm receptors. Because Platynereis is distantly related to flies and mice, the authors suggest this "hybrid" sensory receptor could be very old and homologous across many animals. The question of what role r-Opsins play outside of photoreceptors is an interesting one that remains poorly understood.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #4 agreed to share their names with the authors.)

    1. Reviewer #2 (Public Review):

      NICEdrug.ch integrates well-established previous methods/pipelines from the same group and provides an easy-to-use platform for users to identify reactive sites, create repurposing and druggability reports, and reactive site-specific similarity searches between compounds. Case studies provided in the manuscript are quite strong and provide ideas to the reader regarding how this service can be useful (i.e., for which kinds of scientific aims/purposes NICEdrug.ch can be utilized). On the other hand, there are a few critical issues related to the current state of the manuscript, which, in my opinion, should be addressed with a revision.

      Major issues:

      1) Two of the most critical drawbacks are, first, the lack of quantitative assessment of the abilities of the service and its analysis pipeline. Use cases provide valuable information; however, it is not possible to assess the overall value of any computational tool/service without large-scale quantitative analyses. One analysis of this kind has been done and explained under "NICEdrug.ch validation against biochemical assays" and "Comparison of NICEdrug.ch predictions and biochemical assays"; however, this is not sufficient as both the experimental setup and the evaluation of results are quite generic (e.g., how to evaluate an overall accuracy of 0.73 without comparing it to other computational methods that produce such predictions, as there are many of them in the literature). Also, similar quantitative and data-driven evaluations should be made for other sections of the study as well.

      2) The second critical issue is that, in the manuscript, the emphasis should be on NICEdrug.ch, since most of the underlying computational methods have already been published. However, the authors did not sufficiently focus on how the service can actually be used to conduct the analysis they mention in the use cases (in terms of usability). Via use cases, authors provide results and its biological discussion (which actually is done very well), but there is no information on how a potential user of NICEdrug.ch (who is not familiar with this system before and hoping to get an idea by reading this paper) can do similar types of analyses. I recommend authors to support the textual expressions with figures in terms of screenshots taken from the interface of NICEdrug.ch at different stages of doing the use case analyses being told in the manuscript. This will provide the reader with the ability to effectively use NICEdrug.ch.

    2. Reviewer #1 (Public Review):

      The authors developed a very interesting tool, named NICEdrug.ch, used it to identify drug metabolism and toxicity, and finally predicted druggability of disease-related enzymes and reposition drugs. Comprehensive integration effort based on publicly available datasets and several previous methods developed by the authors (e. g. BridgeIT, BNICE.ch, ATLAS of Biochemistry) results with a resource named NICEdrug.ch. The idea is interesting and addresses a very important problem in the field. The manuscript is clearly written, provides enough analysis of overall challenges and an overview of the most important results. Also, it presents figures that are remarkable.

    3. Evaluation Summary:

      In this study, the authors proposed a new web service/tool and its database, NICEdrug.ch, to be used in the fields of drug discovery and repurposing with the exploration of the metabolic fate of small molecules. The study is timely and will potentially have a high impact as metabolic evaluation of drugs/compounds is a critical topic that is still understudied.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      In this manuscript, Böhm et. al. aim to understand how precise kinetochore assembly is tied to cell cycle progression in budding yeast. In this work, the authors identify CDK phosphorylation sites concentrated in the N-terminus of Ame1, a protein of the COMA complex, and set out to characterize the role these phosphorylation sites may play protein function at the kinetochore. Although phospho-null Ame1 does not affect cell viability, expressing an Ame1 mutant that lacks the phosphorylated domain results in cell death. Interestingly, overexpression of the phospho-null Ame1 mutant accumulates to a higher level than the wild type protein leading the authors to hypothesize that these phosphorylation sites function as phosphodegrons in the Ame1 protein. Through molecular modeling and genetic analysis, the authors determine that Ame1 is a substrate of the SCF E3 ubiquitin ligase and is likely recognized by the Cdc4 F box protein. The authors go on to convincingly show that phosphorylation of what is referred to as the "CDC4 phosphodegron domain" is phosphorylated in a step-wise manner that is cell cycle dependent and that the phosphorylated Ame1 protein specifically is degraded in mitosis. In addition to Ame1 phosphorylation, the authors show that Ame1 degradation depends on whether Ame1 is bound to the Mtw1c (binding prevents degradation), which only happens at a fully assembled kinetochore. Based on these observations, the authors propose a model in which the phosphodegron motif functions to degrade any molecules of the COMA complex that are not incorporated into the kinetochore and in this way prevents kinetochore assembly at ectopic regions of the chromosome.

    2. Reviewer #2 (Public Review):

      Böhm et al. investigated the phosphorylation of the Ctf19CCAN component Ame1CENP-U by Cdk1 which forms a phosphodegron motif recognized by the E3 ubiquitin ligase complex SCF-Cdc4. They identify phosphorylation sites on Ame1 and demonstrate that phosphorylation of Ame1 leads to its degradation by the SCF with Cdc4 in a cell-cycle dependent manner. They also demonstrate that the outer kinetochore component Mtw1c shields Ame1 from Cdk1 phosphorylation in vitro. Finally, they propose a model in which at least one component, Ame1, is present in excess at S-phase in yeast to incorporate into high levels of sub-complexes for efficient inner kinetochore formation on newly duplicated centromere DNA. Then, in mitosis, phosphodegrons serve to mediate the degradation of excess Ame1 (and presumably other CCAN components) and in so doing protect against the formation of ectopic outer kinetochores.

      This manuscript puts forth well-designed and thorough experiments characterizing the phosphorylation of Ame1 and its regulation by the SCF-Cdc4 complex. The writing is clear and the figures are generally easy to understand. The authors succeed in asking pertinent questions, designing experiments to answer them, and considering potential alternative explanations or confounding factors. As a whole this creates a generally convincing study regarding the phospho-regulation of Ame1. However, I also have some important concerns:

      1) The authors begin the manuscript by mapping phosphorylation sites across Ctf19CCAN components but then largely narrow their experimental focus to Ame1 and to a lesser extent its binding partner Okp1. Without mutation of other components, the Ame1 mutant phenotypes are either absent or very mild. This would seem to implicate that, if this is an important process, that other targets for this quality control mechanism must exist. As it stands now, the focused investigation does not make the most compelling case for the broad conclusions that are claimed. More extensive investigation of phosphoregulation of CCAN subunits beyond Ame1 would certainly help justify the claim that phosphoregulation is used to clear excess CCAN subunits and protect against ectopic kinetochore assembly. Is there another lead from their initial mass spec work that could provide some molecular evidence that this is a general process? Failing that, the discussion could at least provide some hint at how the model could be tested in future studies.

      2) The conclusion that the binding of the Mtw1 complex shields Ame1 phosphodegrons is arguably one of the most significant and interesting claims made in this paper. However, the evidence presented to support this claim seems to rely exclusively on in vitro data. Thus, this part is out of balance with other parts of the paper where some in vivo correlations are attempted/made.

      3) The central model mentioned at the outset strongly predicts that the mitotic degradation of Ame1 doesn't impact its abundance at centromeres. That is not the only possibility, though, and some measurement (fluorescence of a tagged Ame1 or a ChIP on centromere DNA) of Ame1 at centromeres before and through mitosis would help instill confidence in the proposal.

    3. Reviewer #1 (Public Review):

      Kinetochores are huge protein assemblies on chromosomes which are used as attachment point for microtubules and allow microtubules to pull chromosomes into daughter cells during cell division. The proteins that form the kinetochore are well known, but the temporal regulation of the assembly of all these proteins into functional kinetochores is less understood.

      In this paper the authors have identified phosphorylation sites in the 'CCAN' of budding yeast, the 'inner', i.e. chromatin-proximal, part of the kinetochore. They characterize in detail the function of phosphorylation of Ame1 (CENP-U in humans), which is part of CCAN. The data support the idea that a cluster of phosphorylation sites in Ame1 is phosphorylated by mitotic CDK1 and serves as phospho-degron for the E3 ligase SCF/Cdc4.

      The authors show phosphorylation of these CDK1 consensus sites in vivo and their phosphorylation by CDK1/Clb2 in vitro. Genetic experiments and molecular dynamics simulations support the idea that phosphorylation sites on Ame1 can serve as phospho-degron for SCF/Cdc4. Even the non-phosphorylatable mutant of Ame1 is stabilized in an SCF mutant background, though, suggesting that this phospho-degron is not the only way in which SCF influences kinetochore protein levels.

      Mutants in the characterized phosphorylation sites do not impair budding yeast growth. This suggests that the degron characterized in this paper may be important for fine-tuning, but is not essential for the proper execution of mitosis. The observations overall add to prior evidence that kinetochore assembly can be regulated by phosphorylation and/or ubiquitination.

      Interestingly, the authors find that phosphorylation of Ame1 by CDK1 in vitro is impaired when Ame1 binds Mtw1, another kinetochore protein. The fact that Mtw1 seems to shield these sites from phosphorylation leads the authors to put forward an interesting model: they propose that cell cycle-dependent phosphorylation and SCF-dependent degradation of kinetochore subunits allows for excess subunits during kinetochore assembly in S-phase (which will speed up assembly) while depleting any excess subunits after assembly, when the kinetochore needs to be functional.

      This is an interesting model. The in vivo evidence is still limited, though. For now, it remains unknown whether the phosphorylation status of kinetochore-bound and free Ame1 is indeed different, whether more soluble Ame1 exists in S-phase, whether too early degradation of Ame1 (or possibly other kinetochore proteins) indeed impairs kinetochore assembly, or whether a failure to remove the soluble pool after assembly leads to mitotic defects. It is an attractive proposal, though, that can now be further explored experimentally.

      In addition to the specific characterization of Ame1 sites, the paper also includes comprehensive data on CCAN phosphorylation sites obtained by mass spectrometry which can serve as basis for future studies.

    4. Evaluation Summary:

      This paper will be of interest to those in the fields of chromosome biology, mitotic regulation, and proteostasis. The authors put forward an interesting model of phosphodegron regulation of kinetochore assembly based on convincing genetic and biochemical data. The novel model will require some additional evidence before it can be considered well-supported, but the paper represents an advance in our knowledge of kinetochore regulation with experiments that are rigorous, well-designed and carefully conducted.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #3 (Public Review):

      Computational models, provide a way to understand emergent network function, and at their best provide a canvas for experimentalists to probe hypotheses regarding function. In this manuscript, Bui and colleagues provide a set of iterative models to describe the locomotor development of larval zebrafish at key developmental stages. These include coil, double coil, and swimming behavior that leads to 'beat-and-glide' behavior. During development, the model steadily moves from gap junction mediated connectivity to more complex synaptic-based network models. In my opinion, this is a very interesting foundation that can be used as a catalyst for future research for experimentalists or to develop more involved models. Like any model it is possible to be critical of the assumptions made. But I expect that it will not be static and be revised over the years. It is important to realize that these sets of models are unique in that they strive to provide models for motor control of a single species across development. The zebrafish is an excellent example since genetic models are widely used, development is swift, and there is active research to understand the physiology of locomotion.

      Strengths and weaknesses:

      The key strength of this manuscript is the detailing of a set of related models detailing the motor output of the larval zebrafish across key stages of development. The models should form a basis for future research. It also a first of its kind - I don't know of similar models focusing on development of locomotor function. The main weakness is the reliance on assumptions of model connectivity. But I suggest that if the model is treated as a basis for the community to refine and validate it will be incredibly useful.

    2. Reviewer #2 (Public Review):

      This study presents iteratively constructed network models of spinal locomotor circuits in developing zebrafish. These models are shown to generate different locomotor behavior of the developing zebrafish, in a manner that is supported by electrophysiological and anatomical data, and by appropriate sensitivity analyses. The broad conclusions of the study result in the hypothesis that the circuitry driving locomotor movements in zebrafish could switch from a pacemaker kernel located rostrally during coiling movements to network-based spinal circuits during swimming. The study provides a rigorous quantitative framework for assessing behaviorally relevant rhythm generation at different developmental regimes of the zebrafish. The study offers an overarching hypothesis, and specific testable predictions that could drive further experimentation and further refinement of the model presented here. The models and conclusions presented here point to important avenues for further investigation, and provide a quantitative framework to address constituent questions in a manner that is directly relatable to electrophysiological recordings and anatomical data. The study would benefit from additional sensitivity analyses, and from the recognition that biological systems manifest degeneracy and significant variability along every scale of analysis.

    3. Reviewer #1 (Public Review):

      The manuscript is somewhat readable but the many acronyms for the cell types in model and biology make it difficult to follow. Is there a reason why the biological neuron names cannot be used in the model? The presentation of data in figures can be more powerful. In many cases, the data in figures and the supplemental videos show apparently different results. This can be an artifact of how the videos were made and if yes, these can be improved. Tail tip coordinates can be plotted to show the behaviors in much better detail.

      Especially for beat and glide swimming, the points regarding burst firing, inhibition, etc. have not been robustly made.

    4. Evaluation Summary:

      In this manuscript, Roussel et al., build models of spinal networks capable of generating coiling and swimming behaviors of embryonic and larval zebrafish. The models use details obtained from earlier experimental studies and insert novel network elements, thus providing testable ideas for rhythm generation. The study will be of high value to those interested in motor pattern generation in general and zebrafish spinal cord function in specific.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      In this study, Tang and colleague report that the multikinase inhibitor YKL-05-099 increases bone formation and decreases bone resorption in hypogonadal female mice with mechanisms that are likely to involve inhibition of SIKs and CSFR1, respectively. The authors also report that postnatal mice with inducible, global deletion of SIK2 and SIK3 show an increase of bone mass that is associated to both an augmentation of bone formation and bone resorption.

      The paper provides novel and interesting information with potentially highly relevant translational implications. The quality of the data is outstanding and most of the authors' conclusions are supported by the data as shown.

    2. Reviewer #2 (Public Review):

      This work tests the ability of a kinase inhibitor to increase bone mass in a mouse model of osteoporosis. The inhibitor, which targets SIK and other kinases, was shown previously by these investigators to increase trabecular bone mass in young intact mice. Here they show that it increases trabecular, but not cortical, bone in oophorectomized mice and that this is associated with increased bone formation and little or no effect on bone resorption. In contrast, postnatal deletion of SIK2 and SIK3 increased both bone formation and resorption, suggesting that the inhibitor targets other kinases to control resorption. Indeed, the authors confirm that the inhibitor effectively suppressed the activity of CSF1R, a receptor tyrosine kinase essential for osteoclast formation. The authors also provide some evidence of unwanted effects of the inhibitor on glucose homeostasis and kidney function.

      Overall, the studies are performed well with all the necessary controls. The effects of the inhibitor on CSF1R inhibition are convincing and provide a compelling explanation for the net effects of the compound on the skeleton.

      1) The ability of the inhibitor to increase trabecular but not cortical bone mass will likely limit its appeal as an anabolic therapy. Indeed, the authors show that PTH, but not the inhibitor, increases bone strength. However, this limitation is not addressed in the manuscript. In addition, the mechanisms leading to these site-specific effects were not explored.

      2) The mechanisms by which YKL-05-099 increases bone formation remain unclear. The authors point out that their previous studies indicate that the compound stimulates bone formation by suppressing expression of sclerostin. However, YKL-05-099 increased trabecular bone in the femur but not spine of intact mice and did not increase cortical bone in intact or OVX mice. In contrast, neutralization of sclerostin increases trabecular bone at both sites in intact mice as well as increases cortical bone thickness. These differences do not support the idea that YKL-05-099 increases bone formation by suppressing sclerostin.

      3) The authors repeatedly state that the kinase inhibitor uncouples bone formation and bone resorption. However, the authors do not provide any direct evidence that this is the case. Although the term coupling is used to refer to a variety of phenomena in skeletal biology, the most common definition, and the one used in the review cited by the authors, is the recruitment of osteoblasts to sites of previous resorption. The authors certainly provide evidence that the kinase inhibitor independently targets bone formation and bone resorption, but they do not provide evidence that the mechanisms leading to recruitment of osteoblasts to sites of previous resorption has been altered. The resorption that takes place in the inhibitor-treated mice likely still leads to recruitment of osteoblasts to sites of resorption. Thus coupling remains intact.

      4) The results of the current study nicely confirm previous findings by the same authors, demonstrating the reproducibility of the effects of the inhibitor. They also provide a compelling explanation for the net effect of the inhibitor on bone resorption (it stimulates RANKL expression but inhibits CSF1 action). While this latter finding will likely be of interest to those exploring SIK inhibitors for therapeutic uses, overall this study may be of limited appeal to a broader audience.

    3. Reviewer #1 (Public Review):

      The primary objective of this manuscript was to examine if multi-kinase inhibitor YKL-05-099 can inhibit salt inducible kinases (SIKs) with the goal to examine a new class of bone anabolic agents for the treatment of osteoporosis. They found that YKL-05-099 was successful in increasing anabolism and, surprisingly, decreasing bone resorption, leading them to investigate why this inhibitor differed from the effects of deletion of SIK2 and SIK3. They found that YKL-05-099 also inhibited the CSF1 (M-CSF) receptor, thus, inhibiting osteoclast activity. This is an interesting manuscript but there are some flaws in the conduct of the experiments and in the analyses which lessen its impact. Nevertheless, it opens the way for another possible oral therapeutic for osteoporosis.

    4. Evaluation Summary:

      This is a very interesting, novel and informative study. The effects of the inhibitor on CSF1R inhibition are convincing and provide a compelling explanation for the net effects of the compound on the skeleton. The study opens the way for another possible oral therapeutic for osteoporosis.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #2 (Public Review):

      In the manuscript Li and colleagues explored the mechanisms that potentially regulated the transcoelomic metastasis of ovarian cancer. By using the in vivo genome-wide CRISPR/Cas9 screen in human SK-OV-3 cell line after transplanted in NOD-SCID mice, the authors identified that IL-20Ra was a potential protective factor preventing the transcoelomic metastasis of ovarian cancer. SK-OV-3 cells with higher expression of IL-20R have lower metastatic potential in vivo. On the contrary, a mouse cell line ID8 with lower IL20Ra expression metastasized aggressively, which could be reversed by over expressing IL-20Ra in the cells. In human, the metastasized ovarian cancers had lower expression of IL-20Ra than the primary tumors. Mechanistically, the authors hypothesized that IL-20 and IL-24 produced by peritoneum mesothelial could act on tumor cells through the IL-20Ra/IL-20Rb receptor to promote the production of IL-18. IL-18 could drive the macrophages into M1 like phenotypes, which in turn controlled the transcoelomic metastasis of the cancer. The in vivo phenotypes in this study were consistent with these hypotheses. The role of IL-20Ra in this setting is potentially interesting and novel.

    2. Reviewer #1 (Public Review):

      The authors used a CRISPR screen to investigate the basis of metastasis of ovarian cancer (OC) cells. Overall, they identified two key genes, IL20RA, one of which was studied in detail. They identify an IL20/IL20RA communication between ovarian cancer cells and peritoneal mesothelial cells to promote M1 macrophages and prevent dissemination of the cancer cells. IL-20 mediated crosstalk is blocked in metastasized OC cells by decreased expression of IL-20RA. Interestingly, IL20RA is also decreased in cells from OC patients with peritoneal metastasis, and reconstitution of IL20RA in metastatic OC cells suppresses metastasis. Moreover, OC cells induce mesothelial cells to produce IL20 and IL24.

      Overall, this is a nice study. It is well-written, and the data are clear. A range of methodologies are used that support the conclusions, with both over-expression and under-expression related studies supporting some key conclusions.

      The overall model is that there is crosstalk between disseminated OC cells and mesothelial cells and macrophages. OC cells when disseminated into the peritoneal cavity stimulate mesothelial cells to produced IL20 and IL24, which via IL20RA trigger STAT3 to produced OAS/RNase L and production of IL-18, to promote an M1 phenotype. The M1 phenotype lowers metastasis. Highly metastatic cells block this pathway by decreasing IL20RA expression.

      These findings are interesting, with potential therapeutic ramifications.

    3. Evaluation Summary:

      The authors studied ovarian carcinoma and identified a potential role of interleukin 20 receptor subunit alpha (IL20RA) and of IL-20 in regulating the transcoelomic metastasis of ovarian carcinoma, where IL-20 signaling in tumors is protective. This leads to the production of IL-18 and an M1 macrophage phenotype, with reduction of metastasis. The study is of interest to investigators in the area of cancer and cytokines and has potential therapeutic ramifications.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

    1. Reviewer #3 (Public Review):

      The authors aimed to develop a 2D image analysis workflow that performs bacterial cell segmentation in densely crowded colonies, for brightfield, fluorescence, and phase contrast images. The resulting workflow achieves this aim and is termed "MiSiC" by the authors.

      I think this tool achieves high-quality single-cell segmentations in dense bacterial colonies for rod-shaped bacteria, based on inspection of the examples that are shown. However, without a quantification of the segmentation accuracy (e.g. Jaccard coefficient vs. intersection over union, false positive detection, false negative detection, etc), it is difficult to pass a final judgement on the quality of the segmentation that is achieved by MiSiC.

      A particular strength of the MiSiC workflow arises from the image preprocessing into the "Shape Index Map" images (before the neural network analysis). These shape index maps are similar for images that are obtained by phase contrast, brightfield, and fluorescence microscopy. Therefore, the neural network trained with shape index maps can apparently be used to analyze images acquired with at least the above three imaging modalities. It would be important for the authors to unambiguously state whether really only a single network is used for all three types of image input, and whether MiSiC would perform better if three separate networks would be trained.

    2. Reviewer #2 (Public Review):

      Panigrahi and co-authors introduce a program that can segment a variety of images of rod-shaped bacteria (with somewhat different sizes and imaging modalities) without fine-tuning. Such a program will have a large impact on any project requiring segmentation of a large number of rod-shaped cells, including the large images demonstrated in this manuscript. To my knowledge, training a U-Net to classify an image from the image's shape index maps (SIM) is a new scheme, and the authors show that it performs fairly well despite a small training set including synthetic data that, based on Figure 1, does not closely resemble experimental data other than in shape. The authors discuss extending the method to objects with other shapes and provide an example of labelling two different species - these extensions are particularly promising.

      The authors show that their network can reproduce results of manual segmentation with bright field, phase and fluorescence input. Performance on fluorescence data in Fig. 1 where intensities vary so much is particularly good and shows benefits of the SIM transformation. Automated mapping of FtsZ show that this method can be immediately useful, though the authors note this required post-processing to remove objects with abnormal shapes. The application in mixed samples in Fig. 4 shows good performance. However, no Python workflow or application is provided to reproduce it or train a network to classify mixtures in different experiments.

      Performance was compared between SuperSegger with default parameters and MiSiC with tuned parameters for a single data set. Perhaps other SuperSegger parameters would perform better with the addition of noise, and it's unclear that adding Gaussian noise to a phase contrast image is the best way to benchmark performance. An interesting comparison would be between MiSiC and other methods applying neural networks to unprocessed data such as DeepCell and DeLTA, with identical training/test sets and an attempt to optimize free parameters.

      INSTALLATION: I installed both the command line and GUI versions of MiSiC on a Windows PC in a conda environment following provided instructions. Installation was straightforward for both. MiSiCgui gave one error and required reinstallation of NumPy as described on GitHub. Both give an error regarding AVX2 instructions. MiSiCgui gives a runtime error and does not close properly. These are all fairly small issues. Performance on a stack of images was sufficiently fast for many applications and could be sped up with a GPU implementation.

      TESTING: I tested the programs using brightfield data focused at a different plane than data presumably used to train the MiSiC network, so cells are dark on a light background and I used the phase option which inverts the image. With default settings and a reasonable cell width parameter (10 pixels for E. coli cells with 100-nm pixel width; no added noise since this image requires no rescaling) MiSiCgui returned an 8-bit mask that can be thresholded to give segmentation acceptable for some applications. There are some straight-line artifacts that presumably arise from image tiling, and the quality of segmentation is lower than I can achieve with methods tuned to or trained on my data. Tweaking magnification and added noise settings improved the results slightly. The MiSiC command line program output an unusable image with many small, non-cell objects. Looking briefly at the code, it appears that preprocessing differs and it uses a fixed threshold.

    3. Reviewer #1 (Public Review):

      In this work, Panigrahi et. al. develop a powerful deep-learning-based cell segmentation platform (MiSiC) capable of accurately segmenting bacteria cells densely packed within both homogenous and heterogeneous cell populations. Notably, MiSiC can be easily implemented by a researcher without the need for high-computational power. The authors first demonstrate MiSiC's ability to accurately segment cells with a variety of shapes including rods, crescents and long filaments. They then demonstrate that MiSiC is able to segment and classify dividing and non-dividing Myxococcus cells present in a heterogenous population of E. coli and Myxococcus. Lastly, the authors outline a training workflow with which MiSiC can be trained to identify two different cell types present in a mixed population using Myxococcus and E. coli as examples.

      While we believe that MiSiC is a very powerful and exciting tool that will have a large impact on the bacterial cell biological community, we feel explanations of how to use the algorithm should be more greatly emphasized. To help other scientists use MiSiC to its fullest potential, the range of applications should be clarified. Furthermore, any inherent biases in MiSiC should be discussed so that users can avoid them.

      Major Concerns:

      1) It is unclear to us how a MiSiC user should choose/tune the value for the noise variance parameter. What exactly should be considered when choosing the noise variance parameter? Some possibilities include input image size, cell size (in pixels), cell density, and variance in cell size. Is there a recommended range for the parameter? These questions along with our second minor correction can be addressed with a paragraph in the Discussion section.

      2) Could the authors expand on using algorithms like watershed, conditional random fields, or snake segmentation to segment bacteria when there is not enough edge information to properly separate them? How accurate are these methods at segmenting the cells? Should other MiSiC parameters be tuned to increase the accuracy when implementing these methods?

      3) Can the MiSiC's ability to accurately segment phase and brightfield images be quantitatively compared against each other and against fluorescent images for overall accuracy? A figure similar to Fig. 2C, with the three image modalities instead of species would nicely complement Fig. 2A. If the segmentation accuracy varies significantly between image modalities, a researcher might want to consider the segmentation accuracy when planning their experiments. If the accuracy does not vary significantly, that would be equally useful to know.

      4) The ability of MiSiC to segment dense clusters of cells is an exciting advancement for cell segmentation algorithms. However, is there a minimum cell density required for robust segmentation with MiSiC? The algorithm should be applied to a set of sparsely populated images in a supplemental figure. Is the algorithm less accurate for sparse images (perhaps reflected by an increase in false-positive cell identifications)? Any possible biases related to cell density should be noted.

      5) It is exciting to see the ability of MiSiC to segment single cells of M. xanthus and E. coli species in densely packed colonies (Fig. 4b). Although three morphological parameters after segmentation were compared with ground truth, the comparison was conducted at the ensemble level (Fig. 4c). Could the authors use the Mx-GFP and Ec-mCherry fluorescence as a ground truth at the single cell level to verify the results of segmentation? For example, for any Ec cells identified by MiSiC in Fig. 4b, provide an index of whether its fluorescence is red or green. This single-cell level comparison is most important for the community.

    4. Evaluation Summary:

      In this work, Panigrahi et. al. develop a powerful deep-learning-based cell segmentation platform (MiSiC) capable of accurately segmenting brightfield, fluorescence, and phase contrast images of bacteria cells densely packed within both homogenous and heterogeneous cell populations. This algorithm, if further optimized and disseminated to the community, will have a large impact in microbial studies in that it will allow for automated analyses of essential every aspect of bacterial cell biology, including cell morphology, cell cycle, cell-cell communications, protein localization dynamics and a variety of cellular processes using time-lapse imaging.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      This paper from He, Y. et al examines how PKC-theta in activated T cells controls RanBP2 nuclear pore subcomplex formation and nuclear translocation of NFkB, NFAT and AP1 family transcription factors. He, Y et al systematically pull apart a molecular mechanism showing that: 1) T cell receptor-activated PKC-theta localises to the nuclear envelope and associates with RanGAP1, 2) PKC-theta deficiency reduces nuclear localisation of import proteins and AP1-family transcription factors in mature mouse T cells and Jurkat cell line, but not primary mouse thymocytes 3) RanGAP1 is phosphorylated by PKC-theta and that phosphorylation of RanGAP1 on Ser504/Ser506 facilitates RanGAP1 sumoylation and is needed for association with other RanBP2 complex components and 4) that wildtype but not Ser504/506 mutant RanGAP1 can rescue nuclear translocation of transcription factors in RanGAP1 knockdown cells.

      A key strength of this work is that, for many key results, multiple methods for validating findings are used e.g. immunoblots of subcellular fractionation + confocal microscopy to show failure of c-Jun into the nucleus in Prkcq-/- mature T cells (Fig 3 G-H). Furthermore, although the majority of the molecular work takes advantage of the more tractable Jurkat cell line for dissection of molecular mechanism, a number of key points are validated in primary mouse or human T cells such as PKC-theta dependent TCR induced association of RanGAP1 with the nuclear pore (Fig 3D-E) and multiple methods of gene deletion were used e.g. siRNA, knockout mouse model and stable CRISPR deletion. The validation of a functionally meaningful phospho-site on the RanGAP1 protein is valuable for further understanding the biology of this protein.

      Immune receptor control of nuclear transport machinery has not been extensively studied but, as is highlighted by this study, is increasingly being understood as an important step in immune receptor control of transcription factor function. The molecular mechanism that is uncovered here is novel and interesting to the immunological community as it links TCR signalling to an indirect mechanism for regulating localisation of multiple key transcription factors for the T cell immune response.

      There are some concerns listed below. Addressing these concerns would add clarity to the manuscript and support some stated or implied conclusions.

      1) The data on the role of PKC-theta driven RanBP2 subcomplex translocation of AP1 transcription factors is largely limited to within 15 min of T cell activation. The broad statements of the paper e.g. line 427 - "PKC-theta plays an indispensable role in NPC assembly" imply that PKC-theta is essential for this process during long-term T cell receptor activation; however, whether PKC-theta deletion has long term impact on nuclear translocation after these first 15 minutes is not established. The demonstration that the RanGAP1 mutant is not able to induce IL-2 production over 24 hrs (Fig 6D) does support the model that a longer-term requirement for RanGAP1 phosphorylation on Ser504/506 is important for translocation and functional AP1 transcriptional outcomes in this system, but from the data presented it does not necessarily follow that PKC-theta is the only regulator of this beyond the 15 min of activation shown here. It is well established that AP1 transcription factors increase in expression for multiple hours after T cell activation and if PKC-theta deletion impact is not long lasting this could mean PKC-theta is important for the kinetics of AP1 translocation but not necessarily for final functional outcome after a longer period of stimulation as is implied here.

      2) It has been shown in the published literature the impact of PKC theta deletion on in vivo immune responses has been varied, with studies showing clearance of murine Listeria, LCMV, HSV. The manuscript currently lacks discussion around how the formation of a largely functional immune response in these contexts fits in with the strong defect in nuclear translocation of multiple important T cell transcription factors that they show here.

    2. Reviewer #2 (Public Review):

      PKC-theta is a critical signaling molecule downstream of T cell receptor (TCR), and required for T cell activation via regulating the activation of transcription factors including AP-1, NF-kB and NFAT. This manuscript revealed a novel function of PKC-theta in the regulation of the nuclear translocation of these transcription factors via nuclear pore complexes. This novel perspective for PKC-theta function advances our understanding T cell activation. The manuscript provided solid cellular and biochemical evidence to support the conclusions. However, nuclear pore complexes regulate the export and import essential components of cells, it is not clear whether PKC-theta selectively regulates the translocation of above transcription factors, or also other components, and whether regulates both import and export. It is essential to provide more substantial evidence to support the conclusion.

    3. Reviewer #1 (Public Review):

      The manuscript by He et al. reveals a novel role for PKC-theta, following T cell receptor (TCR) stimulation, in regulating the nuclear translocation of several key activation-dependent transcription factors by regulating the assembly of key components of the nuclear pore complex (NPC). The authors make use of T cell lines and primary T cells to show that following TCR stimulation, PKC-theta phosphorylates RanGAP1 to promote its interaction with Ubc9 and increase the sumoylation of RanGAP1, which, in turn, enhances assembly of the RanBP2 subcomplex of the NPC that then promotes the nuclear import of AP-1, NFAT and NFB. These conclusions are well supported by a rigorous experimental approach, which included the use of PKC-theta deficient, sumoyltion-defective, kinase-dead, and constitutively active mutants, and RanGAP1-deficient cells.

    4. Evaluation Summary:

      PKC-theta is known to regulate T cell activation, and this manuscript reveals a novel function of PKC-theta in the regulation of the nuclear pore complexes. The work by He and colleagues reveals that PKC-theta is recruited to the nuclear pore complex wherein it serves to regulate the assembly of key components of the RanBP2 subcomplex of the NPC, which in turn enables the translocation of AP1, NFkB and NFAT into the nucleus. However, these results need to be substantiated by additional experiments or by limiting the breath of the conclusions.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #3 (Public Review):

      This manuscript characterizes the additive genetic variance-covariance of behavioural traits and cortisol level in a captive Trinidadian guppy population, in particular to test for the genetic integration of behavioural and physiological stress responses.

      The experimental design, trait definitions and statistical analyses appear appropriate. The main weakness of the study is a lack of clarity on the definition of genetic integration and the statistical ways to characterize, confirm or reject genetic integration (in particular, what defines and how to test for a "single major axis of genetic variation"?).

      The additive genetic variation-covariation is correctly estimated. The presence of additive genetic correlations and the eigen decomposition of G seem to support genetic integration, but the lack of clear predictions makes the the conclusion not completely clear. Another minor conclusion, that "correlation selection in the past has likely shaped the multivariate stress response" is not directly supported by the results as the argument ignores the possible role of other evolutionary forces (in particular mutational input which is likely to be pleitropic for behaviour and hormone levels).

      The nature of genetic (co)variation in behaviour and physiology is poorly known because most quantitative genetic studies of behavioural and physiological traits are still univariate, while it is clear that selection and evolution are better understood as multivariate processes. In addition to presenting some fresh results on the topic, this manuscript provides a mutivariate framework that could be applied in other populations. In particular, eigen decomposition of genetic variance-covariance matrices is not new but its application to the study of stress response integration is original and promising. As the authors mention, such methods could help improve health and welfare in captive animal populations via indirect artificial selection against stress, which is quite an original and stimulating idea.

    2. Reviewer #2 (Public Review):

      This paper addresses a fundamental question regarding the evolution of the stress response, specifically that the action of natural selection on the stress response should promote the functional integration of its behavioral and physiological components. Therefore, the authors predict that genetic variation in the stress response should include covariation between its component behavioral and physiological traits. The results are intrinsically interesting and seem to provide a critical proof of principle that, if confirmed, will prompt future follow up research. However, there are some fundamental conceptual and experimental design issues that need to be addressed, in order to assess the conclusions that can be drawn from the results presented here.

      Conceptual issues:

      1) The authors selected multiple behavioral measures of the stress response but only considered the glucocorticoid response as a physiological trait. In my view this has several problems:

      A) Although, for historical reasons and because they are easier to measure, glucocorticoids have been perceived as a stress hormone, the fact is that they respond not only to threats to the organism (i.e. stressors) but also to opportunities (e.g. mating). In other words, glucocorticoids are produced and released whenever there is the need to metabolically prepare the organism for action. Therefore, glucocorticoids are probably not the best physiological candidate to look for phenotypic integration with stress behaviors, since they must have also been selected to be produced and released in other ecological contexts. In this regard it would have been interesting to measure the phenotypic integration of cortisol also with behaviors used in non-threatning but metabolically challenging ecological opportunities (e.g. mating), and to investigate the occurrence of an eventual trade-off (or of a "phenotypic linkage") between these two sets of traits (stress traits vs. mating traits).

      B) Sympathetic activation is a key component of the physiological stress response in vertebrates. It is thus odd not to consider the sympathetic response in a study that has the main aim of studying the evolution of a phenotypically integrated stress response. I understand that the sympathetic response in guppies is more difficult to study than measuring cortisol, but this technical challenge can certainly be overcome (e.g. techniques for measuring cardiac response to threat stimuli have been recently developed for other challenging model organisms, such as fruit flies; e.g. https://www.biorxiv.org/content/10.1101/2020.12.02.408161v1); or if not, then an alternative model organism should have been used to address this question.

      2) Typically, in vertebrates the behavioral response to a stressor has a passive (e.g. freezing) and an active (i.e. fight-flight) component. It would be very interesting to assess if these two components are phenotypically integrated with each other and each of them with the physiological response. Unfortunately, the authors did not use behavioral measures of each of these two components. Instead they have extracted 3 spatial behaviors from an open field test (time in the central part of the tank in an open field test (OFT); relative area covered; track length) and emergence latency in an emergence from a shelter test. It is not clear how each of the measured behaviors captures these two key components of the behavioral stress response. For example, a fish that freezes in the central part of the tank when it is introduced in the OFT will have a high time in the middle score and eventually a high relative area covered, but relatively low track length. However, if it darts towards the tank wall and freezes there, the result would probably be low time in the middle and low relative area covered. Thus, a fish that has spent approximately the same time in freezing may show very different behavioral profiles according to the variables used here. This could be avoided if explicit measures of fleeing and freezing behavior have been used. Given that the authors have video-tracked the fish, I suggest they can still extract such measures (e.g. angular speed is usually a good indicator of escape/fleeing behavior; and a swimming speed threshold can be validated and subsequently used to detect freezing behavior from tracking data) from the videos. The fact that variables of these two types of behavioral responses to stress have not been used in this study may explain to a large extent why the authors came to the conclusion that, "the structure of G is more consistent with a continuous axis of variation in acute stress responsiveness than with the widely invoked 'reactive - proactive' model of variation in stress coping style".

      3) The authors used a half-sib breeding design, which is the golden standard in evolutionary quantitative genetics. However, and this is not a specific critique of the present study but a general problem of this field, the extent to which estimates of G obtained with breeding designs reflect the G that would be obtained by actually sampling a natural population is questionable, because these designs create artificially structured populations with higher levels of outbreeding and concomitantly also with higher genetic variation than what is usually found in nature. This problem can be illustrated by analogy using the example of heritability estimates, which are typically lower when obtained from selection studies by comparing the generation after selection to the one before selection (aka realized heritability), than when computed from artificial breeding designs.

      Methodological issues:

      4) The authors considered the OFT, ET and ST testing paradigms to be behavioral assays that allow the characterization of the behavioral components of the stress response in guppies, because all these paradigms involve capturing and transferring the focal fish to a novel environment (tank) and in social isolation. Undoubtedly these procedures must have induced stress, however the stressor was not standardized because it consisted in the capture and transfer, and these may have varied from fish to fish (btw are there measures of handling time for each fish? And how to measure "handling intensity"?). In my view a standardized stressor, such as a looming stimulus (e.g. Temizer et al. 2015 Current Biology 25: 1823-34; Bhattacharyya et al. 2017 Current Biology 27, 2751-2762; Hein et al. 2018 PNAS 115: 12224-8), should have been used such that the behavioral measures could have been linked to the stressor in a more controlled way.

      5) Moreover, the authors have measured the "stress behaviors" and cortisol in response to two different stressors: the handling described above and the confinement and social isolation for the GC response. This is not the best experimental design, because the behavioral and physiological expression is expected to be linked and to be flexible, as shown by the data on cortisol habituation to repeated stressor exposure. Thus, when the goal of the study is to characterize the co-variation between traits it is critical to standardize the stimulus that triggers their expression in the two domains (behavioral and physiological) and behavior and physiological measures should have been obtained in response to the same stressor stimulus for each individual. In principle, the failure to do so will artificially decrease the observed co-variation between traits, due to environmental differences (i.e. test contexts and their specific stressors).

    3. Reviewer #1 (Public Review):

      This paper uses a large breeding colony of guppies to measure genetic correlations between hormonal stress responses and behavior in an open-field test. Although we know a lot about the mechanisms of hormone-mediated behavior, we know less about variation in hormonal systems, particularly genetic variation. Understanding how hormones relate genetically to the behaviors they mediate is particularly important because it helps us understand how the entire hormone-behavior system evolves. A priori, we would expect genetic correlations between hormones and the behavior they underlie, such that selection on the hormone would lead to a response in the behavior and vice versa. However, evidence for this pattern is rare.

      Here, the authors show that stress-induced levels of cortisol are repeatable and heritable. Interestingly, they also show that individuals show a lower stress response to later stress and slightly less variation, indicating a G X E interaction. There was a significant genetic correlation between the hormonal response and one of the behaviors measured in the open field test, and the hormone loaded positively in the first genetic principal component along with all the behaviors. This is evidence of an correlated suite of traits that would evolve together in response to selection.

      This is an important study, because evidence of genetic variation in hormonal systems, not to mention genetic covariation with hormone-mediated traits, is rare. The results presented here provide insight into how a hormone-behavior complex might adapt to a changing environment. They are also relevant to ideas about the maintenance of variation in coping styles in natural populations.

    4. Evaluation Summary:

      This is a timely paper on the genetic integration of behavioral and physiological components of the stress response in guppies. Using evolutionary quantitative genetic approaches, the authors show that genetic variation in the cortisol stress response is associated with genetic variation in stress-related behaviors. This result suggests that physiological and behavioral responses to stress should show correlated evolution in response to natural selection, which is of interest to evolutionary biologists and for animal welfare. The reviewers pointed out several conceptual and methodological issues with the definition of the phenotypes under study and and with the definition strong genetic integration.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #4 (Public Review):

      Coombs et al. aimed to establish a pharmacological tool to distinguish calcium-permeable (CP) AMPA receptors (AMPAR) from calcium impermeable AMPA receptors unambiguously. Towards this end, the authors examined the effects of intracellularly applied NASPM, PhTx-433, PhTx-74, and spermine. The authors showed that NASPM completely blocked outward glutamate-evoked currents with a desensitization blocker, cyclothiazide, from outside-out patch membranes from HEK cells expressing GluA1. In contrast, spermine and PhTx-433/74 partially blocked the outward currents in a voltage-dependent manner (Figure 1). TARPg-2 co-expression reduced potencies of spermine and NASPM, and altered shapes of their conductance-voltage relationship (Figure 2) as well as various kinetics of GluA1, including decay kinetics and recovery kinetics (Figure 3). Further, the authors showed that NASPM blocked GluA1 co-expressed with one of the AMPAR auxiliary subunits, TARPg-2, g-7, CNIH2 GSG1L (Figure 4). Finally, the authors showed that NASPM blocked AMPAR-mediated mEPSC events at +60 mV, but not -70mV, in cultured cerebellar stellate neurons from GluA2 knockout mice. Overall, this manuscript provides high-quality data and critical information about TARPg-2, GluA1, and GluA2 knockout mice.

      This provides a solid analysis of GluA1, TARPg-2, 7, CNIH2, GSG1L, and GluA2 knockout neurons. However, it remains unclear whether intracellular NASPM allows an unambiguous functional measure of CP-AMPAR, especially considering many combinations of AMPARs and auxiliary subunits, e.g., GluA1-4 with splicing isoforms, six TARPs, four CNIHs, GSG1L and CKAMP44, etc.

      Strengths:

      The experimental design to evaluate drugs and receptors with outside-out patch membranes and a piezoelectric device provides the highest-resolution analysis and meaningful information.

      Both experiments and analyses are rigorous and of high quality. However, it remains unclear if intracellular NASPM allows an unambiguous functional measure of CP-AMPAR.

      Weaknesses:

      Because the authors tested a limited combination of receptors and auxiliary subunits, it is difficult to conclude whether NASPM blocks all CP-AMPAR unambiguously.

      Slopes of the conductance-voltage relationships are changed upon TARPg-2 co-expression or different concentrations of NASPM.

    2. Reviewer #3 (Public Review):

      Calcium-permeable AMPA receptors (CP-AMPARs) have been shown to have important roles in modulating many aspects of neuronal function. They are distinguished from calcium-impermeable AMPARs (CI-AMPARs) by a property known as inward rectification and block by relatively selective polyamine compounds; this relative lack of selectivity has led to caveats in the interpretations of the roles of CP-AMPARs. The authors here demonstrate that complete block of CP-AMPARs, with no apparent effect on CI-AMPARs, can be achieved by intracellular application of the polyamine NASPM. Importantly, the authors provide evidence that this block is apparently not affected by the presence of auxiliary subunits, one of the key caveats regarding prior interpretations of the effects of polyamines and the roles of CP-AMPARs. The authors hypothesize that this new approach, use of intracellular NASPM, can provide greater clarity regarding the role of CP-AMPARs in future.

      The approach is sound, the experiments are performed appropriately, the data provided is robust, the presentation is clear, the analyses including statistics are appropriate, the immediate interpretations are therefore fully supported, and the overall manuscript outstanding. The authors appropriately used both a heterologous expression system as well as in vitro neuronal preparation to address their hypotheses. The use of intracellular NASPM to unambiguously distinguish CP-AMPARs from CI-AMPARs has the potential to be transformative in future interpretations about the role of CP-AMPARs, so these findings are very relevant and highly impactful to the field.

    3. Reviewer #2 (Public Review):

      This study compares the pharmacology of intracellular polyamine blockers for Ca-permeable (CP-AMPAR) and Ca-impermeable (CI-AMPAR) AMPA receptors in the absence/presence of auxiliary subunits. Spermine is a widely used polyamine blocker to identify CP-AMPARs in native tissue, but the blocking action of spermine varies depending on which auxiliary subunits are associated with the CP-AMPARs. Hence, spermine has limitations. The goal of the present work was to identify if other polyamine blockers would be more efficient than spermine in identifying CP-AMPARs.

      The authors studied CP- and CI-AMPARs in heterologous cells (HEK293T) and in primary cerebellar stellate interneurons from mice lacking the GluA2 subunit. They primarily used electrophysiology to assay channel block by various polyamines. While the technology is standard, the experiments are carried out in a rigorous manner and encompass numerous controls and variations on appropriate constructs (GluA2-containing and GluA2-lacking AMPARs and various prominent auxiliary subunits - TARPs, cornichons, and GSG1L).

      The main conclusion of the work is that 100 uM NASPM fully blocks CP-AMPAR regardless of the associated auxiliary subunit. This conclusion is strongly supported by experiments including testing various auxiliary subunits in the defined conditions of HEK293T cells as well as recording and demonstrating that NASPM fully blocks AMPAR-mediated currents in stellate cells lacking GluA2 subunits.

      I have no major criticisms of the work.

    4. Reviewer #1 (Public Review):

      Using various voltage and concentration protocols in a heterologous expression system, the authors provide compelling evidence for strong block of GluR1 AMPA receptors by intracellular NASPM, and unlike spermine, the block is independent of auxiliary subunit expression. The authors also show that intracellular NASPM provides a more complete block than spermine of synaptic currents in GluR2-KO neurons.

      Overall the manuscript contains high quality data that is clearly presented. It seems likely that this approach will be useful for assessing the contribution of CP-AMPARs in various scenarios. However, currently the authors have fallen short of providing a comprehensive analysis of the use of NASPM to differentiate between CP and CI AMPARs in intact systems containing multiple AMPAR subunits and auxiliary proteins.

    5. Evaluation Summary:

      Here the authors identify that inclusion of intracellular NASPM can fully block Ca-permeable AMPA receptors regardless of association with auxiliary subunits. The distinction between Ca-permeable and Ca-impermeable AMPA receptors is critical to synaptic physiology, and thus these results will be of interest within the field of excitatory synaptic transmission. The data is of high quality and the experimental analysis is rigorous, but the key claim that the approach provides an unambiguous functional measure of CP-AMPAR prevalence has not been fully supported.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

    1. Author Response:

      Reviewer 1:

      In the study by Buus et al., the authors set out to address an important need to understand how oligo-conjugated antibodies should be optimally utilized in droplet-based scRNA-seq studies. These techniques, often referred to as CITE-seq, complement techniques such as flow cytometry and mass cytometry yet also further extend them by the ability to jointly measure intra-cellular RNA-based cell states together with antibody-based measurements. As is the case with flow cytometry, manufacturers provide staining recommendations, yet encourage users to titrate antibodies on their specific samples in order to derive a final staining panel. Based on the ability to stain with hundreds of antibodies jointly, few studies to date have assessed how the antibodies present in these pre-made staining panels respond to a standard titration curve. In order to address this point, this study tests two dilution factors, staining volume, cell count, and tissue of origin to understand the relationships between signal and background for a commercially available antibody panel. They arrive at the general recommendation that these panels could be improved, grouping various antibodies into distinct categories.

      This study is of general interest to the scRNA-seq and CITE-seq communities as it draws attention to this important aspect of CITE-seq panel design. However, it would stand to be substantially improved by not only providing suggestions but also testing at least one, if not more, of their suggestions from Supplementary Table 2, and preferably performing experiments using more technical replicates or biological replicates. As it stands now, the study is largely based on one PBMC and one lung sample, that were stained once with each manipulation as far as can be gathered from the Methods.

      We appreciate the reviewer’s insight into the methodology and enthusiasm for the study.

      We do want to clarify that the study does not use a “pre-made staining panel” from commercial vendor, but rather a cocktail of individual antibodies available from a commercial vendor (with emphasis on epitopes relevant to immunology and cancer research). We have also clarified this in the text of the manuscript.

      We hope that the added analysis, our point by point response to the issues raised by the reviewer, and inclusion of new CITE-seq data from the panel with adjusted concentrations to alleviates the main concerns of the reviewer.

      1) Given the title is improving oligo-conjugated antibody… it would be important to functionally test one of the suggestions. We would suggest a full titration curve of selected antibodies, perhaps one from each of the categories, but if cost is a concern at least two or three antibodies, to identify how titration impacts antibodies, and especially those in categories labeled as in need of improvement. Relatedly, if the idea is that if antibodies (such as gD-TCR) do not have a cognate receptor leading to general background spread, does spiking in a cell that is a known positive in increasing ratios remedy this issue by acting as a target for the antibodies? Does adding extra washes help to remedy these issues of background?

      These are excellent points. Full titration curves have previously been published showing that oligo-conjugated antibodies respond to titration, and in that regard behave similar to fluorophore-conjugated antibodies assayed by flow cytometry (see Stoeckius et al. 2018. Genome Biology; Fig. 3A-D). Our study does not aim to identify the optimal concentration of individual antibodies in isolation but strives to provide the optimal signal-to-noise ratio for each antibody in a cocktail while taking sequencing requirements into account - this is why we don’t focus on full titration curves and saturation kinetics for each antibody/epitope. If we use all antibodies at their highest signal-to-noise ratios, this would drastically increase sequencing requirements of the library as highly expressed markers would use the vast majority of the total sequencing reads. As such, we aimed to get “sufficient” signal-to-noise while keeping the sequencing allocated to each marker balanced.

      Furthermore, as our results show, background signal can be largely attributed to free-floating antibodies in the solution, using high concentrations for all markers in one or more condition would increase the background in all conditions if these were multiplexed into the same droplet segregation. This phenomenon would likely obscure the positive signals and possibly titration response at lower concentrations (similar to what we see for category A antibodies). To avoid this, if full titration curves should be meaningful, each condition should be run in its own droplet segregation making such titration efforts prohibitively costly. We have elaborated on this in the discussion of the revised manuscript.

      We agree that it would greatly improve the study to include results from our panel with adjusted concentrations. In the revised manuscript, we have made efforts to address this by making a comparison between the sample stained with the pre-titration (DF1) concentrations and a sample stained with concentrations that have adjusted based on their assigned categories (from Table 1). We believe that this new data convincingly demonstrates improvements both of the individual antibody signals and at the level of the increased sequencing balance (see new Fig. 5). While the adjusted concentrations could still benefit from further improvements, we show that at similar sequencing depths, the adjusted concentrations provide a more balanced sequencing output and exhibit a 57 % increase in the median positive signal and a 43 % reduction in the median background signal for the 52 antibodies in our panel. The benefit of the adjusted concentration was particularly remarkable for CD86 which went from having 76.5 % to 12.6 % of UMIs assigned to background signal and thus yielded comparable positive signal while using 4.8 fold less UMIs (new Fig. 5G).

      Spiking in cells that express the cognate antigen is an interesting idea. However, as the spiked in cells would be included in all the downstream processes including sequencing of mRNA and other modalities, it would be quite costly to spike-in cells that are not of biological interest – only to decrease background of one or a few antibodies.

      While the results presented in the manuscript do not address this directly, our data strongly suggest that adding extra washing would help reduce free-floating antibodies in the solution captured in the gel-bead emulsions responsible for some of the observed background signal (as can be assayed by the non-cell-containing droplets). For such a test to make sense, the staining conditions should be identical for two samples that are differentially washed (including the exact same cell composition) and would require fully separate droplet segregations (i.e. utilization of separate 10x lanes) which would make it a very costly experiment solely to test the washing effect. However, we have done preliminary tests using short (150bp) cDNA amplicon spiked into different tubes or plates containing ~750x103 PBMCs to determine washing efficiency by qPCR. Here we assayed how increasing the washing volume from 200µl (96-well) to 1.5mL or 50mL for two washes reduced the detection of the spiked-in amplicon in the supernatant as compared to an unwashed sample. While short cDNA amplicons may not behave identical to oligo-conjugated antibodies, they simulate background signal stemming from free-floating antibodies and thus can be used to evaluate different washing conditions for a given set-up. As expected, using higher washing volumes does indeed greatly reduce the amount of amplicon (simulating free-floating “background” antibodies) detected in the resulting suspension. (https://raw.githubusercontent.com/Terkild/CITE-seq_optimization/master/figures/review_washing_test.png)

      2) Another way of improving these panels is through reducing the costs spent on both staining but perhaps more importantly the sequencing-based readouts. Several times in the manuscript (at line 77 for example or line 277) it is alluded to that the background signal of antibodies can make up a substantial cost of sequencing these libraries. However, no formal data on cost is presented, which would be important to formalize the author's points. It would be important to provide cost calculations and recommendations on sequencing depth of ADT libraries based on variation of staining concentration. Relatedly, in the methods, sequencing platform and read depth for ADT libraries was not discussed, nor is the RNA-seq quality control metrics provided other than a mention of ~5,000 reads/cell targeted. This is important to report in all transcriptomic studies, and especially a methods development study.

      Thank you for pointing out the very sparse description of choice of sequencing method and RNA-seq quality controls. We have included additional metrics in the materials and methods and included a new Suppl. Fig. S1 showing number of detected genes as well as UMI counts within the mRNA and ADT modalities in the revised manuscript. We agree that reducing sequencing cost (without reducing biological information) is a major reason for optimizing staining with oligo-conjugated antibodies. We have now added a section in which we elaborate on the potential cost saving, and other benefits of titration of antibody panels and provide some examples from our datasets. Actual savings of optimization of these panels will be very dependent on a given setup, starting concentrations and the depth of sequencing that the particular research questions (and budget) warrant.

      Due to the 10-1000 fold higher numbers of proteins as compared to coding mRNA [16], ADT libraries have high library complexity (unique UMI content) and are rarely sequenced near saturation. Thus, either sequencing deeper or squandering fewer reads on a handful of antibodies, will result in an increased signal from other antibodies in the panel. We found that by simply reducing the concentration of the five antibodies used at 10 µg/mL, we gained 17 % more reads for the remaining antibodies. Consequently, assuming we are satisfied with the magnitude of signal we got from all other antibodies using the starting concentration, this directly translates to a 17 % reduction in sequencing costs.

      In terms of sequencing depth, we are not comfortable giving very broad recommendations. This is due to the fact that sequencing requirements will be very different depending on the composition of the antibody panel as well as the cell type distribution (epitope abundance) (as has been previously noted in Mair et al. 2020 Cell Rep.). If the antibody panel contains only antibodies targeting epitopes that are largely present on a small subset of cells (such as CD56 or CD8 for PBMCs) it would require fewer reads per marker per total cell count than markers that are broadly expressed (such as HLA-ABC or CD45 for PBMCs). However, in a different sample composition (for instance a tissue with few leukocytes) these same antibodies would require fewer reads per cell whereas other epitopes may be more abundant.

      We want to also stress, that aside from cost savings, an optimized balanced panel with low background will yield improved resolution compared to a non-optimized panel. Fortunately, CITE-seq and related methods are very flexible in this regard as you can start by shallow sequencing and then “top-up” the sequencing depth to an optimal level based on the actual data in subsequent sequencing runs (for instance together with the next batch of samples).

      3) One of the powerful elements of joint multi-modal profiling, as mentioned in the title, is to be able to measure protein and RNA from a single cell. This study does not formally look at correlation of protein and RNA levels, and whether a decrease in concentration of antibody either improves or diminishes this correlation. This would be important to test within this study to ensure that decreasing antibody levels does not then adversely affect the power of correlating protein with RNA, and whether it may even improve it.

      We appreciate the reviewer’s suggestion – this is a great idea. Unfortunately, such correlations are notoriously hard to do for scRNA-seq data due to the sparsity of the RNA measurements (which contains high frequency of 0 UMI counts). This is, in part, due to low reverse transcriptase efficiency, and also due to the fact that most proteins have 10-1000 fold more copies than the mRNA transcripts that encode them (Marguerat et al. 2012 Cell). This is exacerbated in our study by the fact that we only shallowly sequenced RNA modality (~4000 reads/cell). Consequently, we see a very high number of cells that despite clustering together within distinct lineage clusters (based on their full transcriptome) and expressing the expected lineage marker surface proteins, do not have readily detectable transcript for the same marker(s). For instance, for all cells that are positive for CD8 at the RNA level, there are at least as many that are negative for CD8 RNA while being positive for CD8 ADT. Importantly, these additional CD8+ cells are still located within clusters consistent with a CD8+ phenotype (see below): (https://raw.githubusercontent.com/Terkild/CITE-seq_optimization/master/figures/review_CD8_protein_rna_correlation.png)

      As such, due to the sparsity of RNA counts, if ADT signal is diluted too much leading to truly positive cells being called as negative, it may actually increase individual cell correlation between RNA and ADT but mean higher levels of “false negative” cells. Direct correlation between RNA and antibody measurements within each individual cells is further complicated by the presence of non-specific/background signal in protein data that is rarely found in RNA data. This can also be seen in the plot above by the fact that positive cells are defined at a cut-off “7” at the ADT level, and not “0” as is the case for RNA. Thus, while having only a few UMI counts for a given transcript is sufficient to call expression, having a few UMIs from an ADT can easily be attributed to background (particularly in an unoptimized panel).

      Due to these technical limitations, we find it more suitable to correlate “positivity” called by either ADT (gated positive as shown in Suppl. Fig. S2) or mRNA expression (i.e. > 0 UMI counts). While this comparison is less quantitative (does not distinguish “high” from “low” expression) it enables us to show whether reducing antibody concentrations affects ADT signal ability to distinguish positive from negative cells (as compared to GEX), which is at the core of the reviewer’s suggestion. The figure below, demonstrates that four-fold titration reduces the fraction of positive cells by some markers (reduction in the blue+red bars by dilution) whereas other markers are largely unaffected both of which is consistent with the analysis in the manuscript: (https://raw.githubusercontent.com/Terkild/CITE-seq_optimization/master/figures/review_protein_rna_correlations.png)

      In terms of assuring specificity, we have also modified the “titration plots” to show more detailed cell type distribution at each rank (by the “barcode plot” to the right of the “rank plot”) as well as the distribution of UMIs among cell types (by the bar plot above the “barcode plot”) at each condition. Finally, to make these “titration plots” more accessible, we have now included a guide to the different components of the “titration plots” in Fig. 2 of the revised manuscript.

      4) How was the lack of antibody binding determined for Category E? CD56 is frequently detected on NK cells in peripheral blood, CD117 should be detected on mast cells in the lung, and CD127 should be found on T cells, particularly CD8+ T cells. From inspecting Figure 1E, it appears as if all three of these markers are detected on small but consistent cell subsets. As the clusters are only numbered and no supplementary table is provided to help the reader in their interpretation, it is difficult to determine if these represent rare but specific binding, or have not bound with any specificity.

      Thank you pointing this out. In light of this comment, it is obvious that we need to annotate the cell types of the clusters. We have annotated all the fine-grained clusters by cell types and re-worked all relevant panels in Figures 1, 2 and 3 (and all their related supplementary figures) to show more detailed and consistent cell type annotation. We have also added Suppl. Fig. 1C, D to show marker genes for each of the annotated cell types, which together with the re-worked Fig 1E, give the reader a clear description of the cluster identity. We do indeed see some signal for Category E antibodies such as CD56, CD117 and CD127 within the expected clusters. This indicates that the antibodies do work to some extent. However, we also find that the signal for these markers is modest, at best, and not present in some populations where we would have expected them (CD127 should be more pronounced in T cells and we are finding an unexpectedly high frequency of CD56-negative NK cells).

      5) References: At 14 references, the paper overall could benefit from a more comprehensive citation of related literature including flow cytometry and/or CyTOF best practices for antibody staining and dealing with background, and joint RNA and protein measurement from single cells.

      We agree that the reference list of the original manuscript was sparse and may have missed important relevant studies. We have done our best to include additional studies relevant for the optimization and titration of mass cytometry panels and flow cytometry staining and added references to a few newly published joint RNA and protein measurement studies. We have strived to reference all studies directly relevant to the present work and do not want to overlook any appropriate publications that should be referenced and so welcome any suggestions of the reviewers.

      Reviewer 2:

      Recombinant antibodies are the most common and powerful reagents in life science research to identify and study proteins. Yet, every single antibody should always be validated and carefully tested for its relevant application, to ensure constructive and reproductive scientific endeavor. I was thus extremely pleased to review the manuscript of Terkild Buus et al, as it provides a careful assessment of oligo-conjugated antibody signal in CITE-seq. The authors tested four variables (antibody concentration, staining volume, cell numbers and tissue origin) and clearly showed that antibody titration is a crucial step to optimize CITE-seq panel. The authors found that, as a general rule, concentration in the 0.625 and 2.5 µg/mL range provides the best results while recommended concentrations by vendors, 5 to 10 µg/mL range, increase background signal.

      In my opinion, the study is well-performed and may serve as a guideline to accurately validate antibodies for CITE-seq, as a consequence I have only minor comments.

      We are very happy that you appreciate the necessity of our work and that you found it to be a useful resource for improving CITE-seq experiments.

      As stated by the authors, the starting concentration used for each antibody was based on historical experience and assumptions about the abundance of the epitopes. This approach may not be ideal, and the optimal concentration may have been missed. Do the authors think that a proper titration would be an advantage? Maybe this could be discussed in the text.

      We agree that using starting concentrations based on historical experience etc. may not be ideal for a completely objective assessment of how oligo-conjugated antibodies respond to the four-variables test. However, we firmly believe that using informed starting concentrations greatly increases the potential improvement of a panel while keeping costs to a minimum (which has to be a consideration for these expensive methods). With that said, we agree that this approach may not reach the optimal concentration (a definition that is a bit complex in this setting). As mentioned in our reply to reviewer 1, point 1, a previous study has shown a more formal titration response for three antibodies using a broader range of concentrations (Stoeckius et al. 2018. Genome Biology; Fig. 3A-D) and we believe that titration for CITE-seq is as much about balancing the sequencing needs of the full panel as it is about reaching the optimal signal-to-noise for the individual antibodies. We have elaborated on this in the discussion of the revised manuscript.

      The authors showed by testing four variables (see above) that they could define the optimal conditions to reduce background signal and increase sensitivity of antibodies and thus this way improves CITE-seq outcome. Nevertheless, the authors rely on the fact that all antibodies used in their panel are specific for their targeted antigens. I am not asking here to test the specificity of every single antibody used in the study as this would be a colossal amount of work. But I feel that this aspect should be discussed in the manuscript, especially when an "uncommon" antibody is intended to be used in the CITE-seq panel; the specificity of this antibody should be indeed tested prior to its use.

      Thank you for this suggestion. This is indeed an aspect of antibody optimization that we have not touched upon. By using commercially available oligo-conjugated antibody clones that are broadly used, the extensive testing of many of these clones by multiple labs within immunology community (for flow/mass cytometry applications) and based on our personal experience with majority of the clones for flow cytometry applications, we expected that the antibodies in our panel should be specific for their antigen. This is supported by the labelling matching what we would expect to find in PBMCs and lung leukocytes, as well as the correlation between expression of the gene encoding the targeted epitope and antibody binding (see our response to reviewer 1, point 3). We have added a paragraph to the revised manuscript discussing that, particularly when using antibodies for the first time or using clones that are unfamiliar, it is important to assure specificity.

    1. Reviewer #2 (Public Review):

      In this manuscript, Dahlen et al. aimed to agnostically investigate the association between ABO and RhD blood group and disease occurrence for a large number of disease phenotypes using large-scale population-based Swedish healthcare registries. Using 2 large subject cohorts, they convincingly demonstrate that beyond the known associations between ABO, infectious diseases and thrombosis, there are other associations with very different diseases. This paper is purely epidemiological with no biological data to explain the observed associations. The clinical phenotypes are derived from hospital coding and probably lack precision, especially in terms of diagnostic certainty.

    2. Reviewer #1 (Public Review):

      The authors aimed to survey a large transfusion database in Sweden to catalog associations between ABO/RhD blood group antigens and a wide variety of clinical phenotypes in a systematic, unbiased and comprehensive manor. They succeed at surveying over 1200 phenotypes in over 5 million people and identify 49 statistically significant associations for ABO blood group and point out a couple novel associations. Their statistical methods are appropriate and help eliminate potential false positive associations. The strengths of this study are the unbiased survey of a large database and the appropriate corrections for multiple observations which allow the authors to explore a large number of associations without loosing site of what is really a significant association.

      This study sheds light on a topic of interest to many scientists. The ABO gene encodes a glycosyltransferase enzyme that has 4 major haplotypes in human populations and results in a specific pattern of posttranslational modification of plasma proteins and blood cells including erythrocytes. Proteins decorated with an H antigen can receive additional carbohydrate antigens from ABO transferase intracellularly. The common A allele transfers UDP-GalNAc while the B allele transfers UDP-Gal. The A2 allele is hypomorphic compared to the A allele and transfers lower amounts of UDP-GalNAc and the common O allele is a null resulting in no transferase activity.

      The allele frequencies of these common alleles varies by ancestry and has geographic differences. Variation at ABO is unconstrained with many rare variants contributing to the four common haplotypes at ABO. Interestingly, geographically specific selective pressures may have led to allele frequency differences. For example. ~40-50% of individuals are homozygous for the null (type O) allele. These null haplotypes are more common in individuals of Latino or African ancestry while 'A' haplotypes are slightly more common in individuals of European origin and 'B' alleles are more common in individuals of Asian and African ancestry. Overall, O is more common than A or B alleles. An unbiased survey of phenotype frequencies by blood type allows for confirmation of previous associations and discovery of novel associations. In this largely European ancestry cohort, blood type A is the most common (45-47%) while blood type O is second most common at 38-39%.

      Limitations of Phenome-wide Association Studies (PheWAS) like the one presented in this manuscript should be noted. Associations with complex phenotypes or those with small effect size will not be detected even in a large cohort such as the SCANDAT. This study is also biased toward associations with phenotypes more common in the Scandinavian population. This may present associations related to the population substructure and not a direct association with ABO. In genome-wide association studies this can be addressed through multiple methods but it is not clear how the authors correct for population structure in this study. Likewise, the insight into the mechanistic reasons for ABO associations is not a strength of this study and will await subsequent studies for many phenotypes. Mechanistic insight might be particularly interesting for the novel associations uncovered by this study.

    3. Evaluation Summary:

      In this manuscript, Dahlen et al. agnostically survey a large transfusion database in Sweden to investigate the association between ABO and RhD blood group and disease occurrence for a large number of clinical phenotypes. The data reported are purely epidemiological associations, with no direct insight into biological mechanism. Nonetheless, these data are a valuable resource for the research community, and offer the potential for a number of important biologic hypotheses and insights for investigation in the future.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

    1. Reviewer #2 (Public Review):

      Landemard et al. compare the response properties of primary vs. non-primary auditory cortex in ferrets with respect to natural and model-matched sounds, using functional ultrasound imaging. They find that responses do not differentiate between natural and model-matched sounds across ferret auditory cortex; in contrast, by drawing on previously published data in humans where Norman-Haignere & McDermott (2018) showed that non-primary (but not primary) auditory cortex differentiates between natural and model-matched sounds, the authors suggest that this is a defining distinction between human and non-human auditory cortex. The analyses are conducted well and I appreciate the authors including a wealth of results, also split up for individual subjects and hemispheres in supplementary figures, which helps the reader get a better idea of the underlying data.

      Overall, I think the authors have completed a very nice study and present interesting results that are applicable to the general neuroscience community. I think the manuscript could be improved by using different terminology ('sensitivity' as opposed to 'selectivity'), a larger subject pool (only 2 animals), and some more explanation with respect to data analysis choices.

    2. Reviewer #1 (Public Review):

      The submitted manuscript 'Distinct higher-order representations of natural sounds in human and ferret auditory cortex' by Landemard and colleagues seeks to investigate the neural representations of sound in the ferret auditory cortex. Specifically, they examine the stages of processing via manipulating the complexity and sound structure of stimuli. The authors create synthetic auditory stimuli that are statistically equivalent to natural sounds in their cochlear representation, temporal modulation structure, spectral modulation structure, and spectro-temporal modulation structure. The authors use functional ultrasound imaging (fUS) which allowed for the measurement of the hemodynamic signal at much finer spatial scales than fMRI, making it particularly suitable for the ferret. The authors then compare their results to work done in humans that has previously been published (e.g. Norman-Haignere and McDermott, 2018) and find that: 1. While human non-primary auditory cortex demonstrates a significant difference between natural speech/music sounds and their synthetic counterparts, the ferret non-primary auditory cortex does not. 2. For each sound manipulation in humans, the dissimilarity increases as the distance from the primary auditory cortex increases, whereas for ferrets it does not. 3. While ferrets behaviorally respond to con-specific vocalizations, the ferret auditory cortex does not demonstrate the same hierarchical processing stream as humans do.

      Overall, I find the approach (especially the sound manipulations) excellent and the overall finding quite intriguing. My only concern, is that it is essentially a null-result. While this result will be useful to the literature, there is always the concern that a lack of finding could also be due to other factors.

      Major points:

      1) What if the stages in the ferret are wrong? The authors use 4 different manipulations thought to reflect key elements of sound structure and/or the relevant hierarchy of the processing stages of the auditory cortex, but it's possible that the dimensions in the ferret auditory cortex are along a different axis than spectro/temporal modulations. While I do not expect the authors to attempt every possible axis, it would be beneficial to discuss.

      2) For the ferret vocalizations, it is possible that a greater N would allow for a clearer picture of whether or not the activation is greater than speech/music? While it is clear that any difference would be subtle and probably require a group analysis, this would help settle this result/issue (at least at the group level).

      3) Relatedly, did the magnitude of this effect increase outside the auditory cortex?

      4) It would be useful to have a measure of the noise floor for each plot and/or species for NSE analyses. This would make it easier to distinguish whether, for instance, in 2A-D, an NSE of 0.1 (human primary) vs. an NSE of 0.042 (ferret primary) should be interpreted as a bit more than double, or both close to the noise floor (which is what I presume).

    3. Evaluation Summary:

      Landemard et al. compare the response properties of primary vs. non-primary auditory cortex in ferrets with respect to natural and model-matched sounds, using functional ultrasound imaging. They find that responses do not differentiate between natural and model-matched sounds across ferret auditory cortex; in contrast, by drawing on previously published data in humans, the authors suggest that this is a defining distinction between human and non-human auditory cortex.

      This was found to be a very nice study and with interesting results that are applicable to the general neuroscience community. The analyses are conducted well and a wealth of results are included, including findings for individual subjects and hemispheres (in supplementary figures). Concerns involved the size of the data set (only 2 animals), and some more explanation was needed with respect to data analysis choices.

      (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

    1. Reviewer #3:

      The authors present the algorithm clearly by comparing it to the most popular SMLM clustering algorithms and showing its robustness in varying density SMLM data, which is a big problem in the field. The presented experimental test on 3D LAMP-1 SMLM data also contributes to the robustness of the paper.

      While reading the manuscript, I missed a comparison with another graph-based SMLM clustering algorithm published previously by Khater et al. in relation to accuracy and computation speed, which is particularly important to demonstrate the advantages of StormGraph. The approach should also be included in Table 1. I also think that a direct comparison in terms of accuracy and computation speed is crucial.

      During the review process, a similar paper has been posted to bioRxiv dated 22. December, https://www.biorxiv.org/content/10.1101/2020.12.22.423931v1.full so the authors could not be aware of this work; however, it would be nice if the authors could comment on this work.

    2. Reviewer #2:

      In their paper "A graph-based algorithm called StormGraph for cluster analysis of diverse single-molecule localization microscopy data", Scurll et al. present a new algorithm to identify clusters in single-molecule localization microscopy (SMLM) data. They use graph-based clustering and show that StormGraph outperforms a selection of existing algorithms, both on simulated and experimental data. The improvement seems not huge, but is convincing, thus this work presents an important contribution to the field. Naturally, not all competing algorithms could be benchmarked in comparison to StormGraph, thus it is not clear if this algorithm is indeed among the best performing algorithms. This is especially true for the cross-correlation analysis. If the applicability of the software included with the manuscript was extended to more potential users, this could be a useful contribution to the field. The manuscript is well written, but quite long. The information content would not be jeopardized if part of the main text and some figures were to be moved to the supplementary information or methods section.

    3. Reviewer #1:

      Single molecule localization microscopy (SMLM) has become an important method for understanding the subcellular distribution of fluorescently labelled biomolecules at length scales of a few tens of nanometers. A critical challenge has been to find out, whether and to what extent biomolecular clustering occurs. While methods have been published which address the problem of identifying biomolecular clusters in SMLM images, they still suffer from many user-defined parameters, which - if selected inappropriately - influence the obtained results substantially. The StormGraph-3D method proposed here addresses these issues, based on a comprehensive mathematical framework which reduces the number of user-defined input parameters. The method was evaluated using comprehensive simulations of data, which show its robustness compared to alternative approaches.

      The methods part of the paper would benefit, however, from more realistic data of single molecule blinking behavior, and the evaluation of the consequences on the performance of the method. As the authors acknowledge, overcounting due to blinking has challenged data analysis previously, and gave rise to artifactual localization clusters that do not represent the underlying protein distribution. It would be of particular interest, which results in the method yielded for a random biomolecular distribution.

    1. Reviewer #2:

      This is a very interesting study, examining the properties of different types of neurons in the primate Frontal Eye Fields. It is commonly assumed that a serial processing of information takes place in the frontal lobe, from visual representation, to working memory maintenance, to motor output. However, some evidence to the contrary has also been reported, creating a debate in the field. The authors have characterized meticulously FEF neurons receiving V4 projections, by means of orthodromic stimulation. They report two main findings: that visual-input recipient neurons in FEF exhibit substantial motor activity and that working memory alters the efficacy of V4 input to FEF. The paper provides an important addition to our understanding of FEF processing. Although the first result is unambiguous, and goes against the traditional view of the FEF, the interpretation of the second is less straightforward and would need to be qualified further.

      1) Orthodromic activation of FEF neurons via V4 stimulation increases the percentage of FEF events that lead to spikes and decreases their latency during working memory. Such an effect appears expectable if FEF neurons are at a higher level when a stimulus in their receptive field is held in memory compared to a stimulus out of their receptive field. Are the authors suggesting something special about working memory? Would the same outcome not be expected during fixation or smooth pursuit for FEF neurons that are activated by these states? It was not clear that the efficacy of transmission itself improves by working memory, just the likelihood that the spiking threshold would be reached.

      2) It would strengthen the author's thesis to discuss the existing functional evidence (in addition to anatomical evidence) that motor FEF neurons receive visual input and can plan movements accordingly. See for example Costello et al. J. Neurosci 2013, 33(41):16394-408.

      3) The authors match the receptive location of FEF and V4 neurons to maximize the chances of identifying monosynaptically connected neurons between the two areas. However, a negative finding of ia orthodromic activation does not entirely rule out that the FEF neuron under study receives V4 input, from another site. Some discussion is warranted on this point.

    2. Reviewer #1:

      The authors of Working Memory Gates Visual Input to Primate Prefrontal Neurons studied how working memory influences information transmitting from V4 to frontal eye field via extracellular recording and electrical stimulation on behaving primate. They found that V4 neurons target FEF neurons with both visual and motor properties, and its synaptic efficacy of V4 to FEF was enhanced by working memory. These findings are interesting and important to our understanding about how our brain acts during daily WM-related activity.

      1) In classical working memory tasks, the task periods usually consist of fixation, cue, delay and then a response period. The neural activity during the delay period is typically considered to be a working memory-related signal. However, in the current study, the authors didn't point out whether only delay period activity was included in analysis when they compared synaptic efficacy between stimulation and non-stimulation trials, in Figure 4a. Because the differences of neuronal response during fixation period cannot be viewed as relevant to information held in working memory, it may be better if only neuronal activity in the delay period was included in their analysis.

      2) Did the 96 visual-recipient FEF neurons exhibit working memory-related activity in their memory guided saccade task? The example neuron in Figure 3a didn't show significant difference between In and Out trials during the delay period. If the visual-recipient neuron didn't present working memory related activity, how could the authors say enhanced synaptic efficacy from V4 to FEF was caused by working memory?

      3) Did the two example neurons in Figure 4c show adjusted values (subtracting the same measure during non-stimulated trials)? The authors mentioned in Method that Figure 4 showed adjusted values, but it may not be applicable for raster plot in Figure 4c. It may be helpful that using adjusted values show stimulation effects on evoked spike counts during memory In and Out trials.

      4) Did the authors find some FEF cells showing elevated firing during delay period in outside-RF trials compared with baseline firing? These elevated firing was not caused by RF cue, may underlying working memory signal.

      5) The sample size should be indicated in Figure 3b Venn diagram.

      6) It's better to indicate electrical stimulus protocol in Figure 1.

    3. Summary:

      This is a brief paper documenting the properties of neurons in the frontal eye field (FEF), a cortical brain area traditionally thought to receive visual input and transform it internally to motor commands. The authors used extracellular recordings and electrical microstimulation in behaving non-human primates to add to this view by showing that visual input to FEF from visual area V4 appears to be gated by working -memory activity in FEF. The study was considered well done and interesting, albeit with several important concerns about the interpretation of the findings.

      Reviewer #1 opted to reveal their name to the authors in the decision letter after review.